You are on page 1of 249

A, D-A247 719

AG.ARD

AGAR.-783

ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT. 7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE

AGARD REPORT 783

Special Course on

Engineering Methods in Aerodynamic Analysis and Design of Aircraft


(Les Methodes d'Ingenierie Employee lors de lAnalyse et de la Conception des Aronefs)
kw rnaterial assembledin thiv book was prepared under the combined sponorshipof the,FluidDynamics Panel,the Consulwa andExchange PrormmeofAGARD. and reLI Kdrmn Institute, and w, preswntedas an AGARD Speci Courseat the vote Middle East Technical Unirveriy, Ankara, Turke', 6th-Orh Ma 1991, at the

1;j])and at the UniversitadPolitecnicade Madrid.Spain 20h-24th May 1991.

." Kdrndn Institutefor Fluid Dynamis, Rhode-St-Gense, Belgium 13th--17th May

-~NORTH

ATLANTIC TREATY ORGANIZATION

92-06961

FPubismhed January 1t90


L3f ttwi;

n ar d Ava labdit, nLAck

Best

Available
Copy

AGARD-R-783

ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT 7 RUE ANCELLE vi2200 NEUILLY SUR SEINE FRANCE

AGARD REPORT 783

Special Course on

Engineering Methods in Aerodynamic Analysis and Design of Aircraft


(Les Mtthodes dlng~nidrie Employde lors dIe l'Analysc ot do la Conception des A~ronefs)-

The material assembled in this book was prepared under the comrbined sponsorship of AGARI), and the F'luid Dynamics Panel, the Consultant and Exchange programme o)f the von Kiinnin Institute, and was presented as an AGARI) Special Course at the Middle East Technical University, Ankara,'llirkcy,6th-l10th May 199 1, at the von Kirmnin Institute for Fluid Dynamics, Rhode-St-Gen~se, Belgium 13th-l17th May 1991 and at the Univesitad Politcoica de Madrid, Spaiin 2th-24th May 1991.

Treaty Organization North Atlantic du TraI16 de lAtlantique Nord Organisation

!-

The Mission of AGARD


According to is Charter, the mission of AGARD is to bnng together the ladmg personaltes of iheNATO nations i 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

conmon benefit of the NATO community,


- Providing scientific and techmcal advice and assistance to the Military Committee in the field of aerospace research and development (with particular regard to its military application), - Continuously stimulating advances in the aerospace sciences relevant to strengthening the common defence posture, - Improving the co-operation among member nations in aerospace research and development: - Exchange of scientific and technical information, - Providing assistance to member nations for the ptrpose of increasing their scientific and technical potential, Rendering scientifiL and technical assistance, as requested. to other NATO bodies and to member nations in .onne.tnon with research and dcelopment problems in the aerospace field Tht highest authrity ssithin AGARI) is the National Delegates Board .onsisting of officially appointed senior representatioes from cavh member nation The mission of AGARD is earned out through the Panels which are composed of experts appointe by the National Delegates, the Consultant and Lxt hange Programme and the Aeroslace Apphations Studies Programme The results oa AGARD work are reported it) the member rations and the NATO Authorities through the AGARD series of Pblications of hich this is one Partilipation in AGARD artisiies is b) invitation only and is normally limited to citizens of the NAIO nations

The content of this publication has been reproduced directly from material supplied by AGARD or the authors

Published January 1992 Copright O AGARD 1992 All Rights Reserved ISBN 92-835-0652-9

PriniedIn silzstd I'nnnng Sre(es Limited 4) ( higull Lane Loughton, t s.et OUs) IZ

Recent Publications of the Fluid Dynamics Panel


AGARDOGRAPHS (AG) Testing of High-Performance Parachutes Design and AGARD AG-319, November 1991 Experimental Techniques Inthe Field of Low Density Aerodynamics AGARD AG-3 18(E), April 1991 Densilt Techniques Expitrimentales Likes i l'Arodynamique i Basse AGARD AG-318 (FR). April 1990 ASuns of Measurements and cy Measuring Techniques in Rapidly Distorted Compressible Turbulent Boundary Layers AGARD AG-3i5, May 1989 Re)ynolds NumberEffects In Transonic Flows AGARD AG-303, December 1988 REPORTS (R) Alr.raft D)namics ait High Angles of Attack: Experiments and Modelling AGARD R-776, Special Course Notes, March 1991 Turbomachinery Applications Inverse Methods InAirfoil Design for Aeronautical and AGARD R-780. Special Course Notes. November 1990 Aerody namics of Rotorcraft AGARD R-78 1,Special Course Notes, Nosembr 1990 Three-Dimensonal Supt rionie/1l)personic Flows Including Separation AGARD R-764, Special (rturse Notes, January 1990 Advattces InCryogenic Wind Tunnel Technology AGARD R-774. Speciat Course Notes, November 1989 ADVISORY REPORTS (AR) Air Intakes for High Speed Vehicles AGARD AR-270, September 1991 Appraisal of the Suitability of Turbulence Mfodels In Flow Calcutatlous AGARI) AR-291 . Technical Status Restew. July 1991 Rotary *ltatnce Testing for Aircraft Dynamics AGARD AR-265, Report of WG 11, Dcember 1990 Calculation of 3D.Separhted Turbulent Flow4 InBoundary Lay er Limtit AGAiU) AR-25S. Report ofi 10, WO May 1990 Adaptive Wind Tunnel Waits: Technology and Applications AGARD AR-269, Report of WGI 2,April 1990 CONFERENCE PROCEEDINGS (CP) Effects of Adverse Weather on Aerodynamics AGARD CP-496. Decmber 1991 Nlanocuiring Aerodynamics AGARD CP-497, November 1991 Vortex Flow Atrod nzamic 3 AGARD CP-494. July 1991 Missile Aerodynamics A(,ARD CP-493 October 1990

Aerodynamics ofCombat Aircraft Controlsan frudEXet AGARD CP-465, April 1990 adofGud Computational Methods for Aerodynamic Design (Inverse) and Optimization AGARD-CP-463, March 1990 Applications of Mesh Generation to Complex 3-D)Configurations AGARD CP.464, March 1990 Fluid Dynamics of Three-Dimensional Turbulent Shear Flows and Transition AGARD CP-438, April 1989 Validation of Computational FLA Dynamics AGARD CP-437, December 1988 Aerodynamic Data Accuracy and Qualitr.Reqtuirements and Capsebllities in Wind Tunnel Testing AGARD CP.429, July 1988 Aerodynamics of Hypersonic, Lifting Vehicles AGARD CP-428, Novemnber 1987 Aerodynamic and Related Hydr dynamic Studies Using Water Facilities AGARD, CP.4 13, June 1987 Applications of Computational Fluid Dynamics in Aeronautics AGARD CP.4 12, November 1986 Store Airframe Aerodynamics AGARD CP-389, August 1986 Unsteady Aerodynamics - Fundamentals and Applications to Aircraft Dynamics AGARD CP-386. Novemiser I 85 Acrod;Tuamlcs and Acoustics of propellers AGARD CP-366. February 1985 Improvement of Aerodynamic performance through Boundary La~cr Control sod High Lift Systems AGARD CP-365, August 1984 Wind Tunnels and Testing Techniques AGARD CP-348, February 1984 Aerodynamics of Vorsical Type Flowis Three Dimensions In AGARD CP-342.lJuly 1983 Misesile Aeeod~namlcs AGARD CP-336, February 1983 Prediction of Aerodynamic Loads on Rotoreraft AGARD CP-334, Scptcmbe; 1982 Wall Interference InWind Tunnels AGARI) CP-335. September 1982 Fluid Dynamica of Jets swith Applications to V/STOL. AGARD CP-308, January 1982 Aerodynamics of Power Plat Installation AGARD CP-30 1,September 1981 Computation of Viscous*Inslscld Interactions AGARD CP-29 1,February 1981 Eubsonlcfrrasonic Configuration Aerodynamics AGAPJ) CP-285, September I198,1) Turbiulent Boundary La)tra Eaperimens, Theory and NIodelling AGARD CP-27 1,January 1980 Aerodynamic Chsaracteristics of Controls AGARD CP-262. Spmber 1979 Hi1gh Angle of Attarl, Aerod~nanslcs AGARI) CP-247, Janu'y ! 979

Foreword
Enigineertng Asork preliminary designof new projects is basedi,alarge degree. on basic fundamental in to experimental tests, empirical procedures, and low level (fast, inexpensive, and easy-to-handle) computer codes restricted to potential flow with simple correction terms for viscous effects. There isa need for training young engineers joining industry to weork these with simple engineering tools Without skillful use of these tools. the art .t cost-effective preliminary design of new aircraft will be jeopardized 71ceobjective ofthusspecial course istopresent proven enigseening methods used duringconsceptial and prelimnary design and development of new aircraft concepts 'Ile course ,III focus oii simple computational procedures for conceptual and preliminary design, low-level anatysi, compaier codes. and experimental techniques for aircraft perfoitnance predictions. PWSacher Special Couric Dircc,or

Avant-Propos
I s Ira%cux d'iugnrc entrepris aunivrau des etudes primnaiies d'un usiuvean soul bd~,en~i pro~et grande paie. surdesessais esperimentaus foudjinentasis. de% procedures empiruques. et dc-, cules machine du preieir cheilon fiipidne. peu co~itcux ci euinsislaus) imite a-is tcoclcmcnts poieuiiels asec simpiles des factcurs decorrection pour leseffets visquetis.
Usajeunes ingemesirs qui debuient dlans lindustre doniveu formes 4niplot decea5 tre outils siesples d'aide ala conception. Sinon. letude preliminaire des nouveaasavionsuian desouions derentzdilite accepiables sera forteinent compromise. Lobjet de eecours special estdepresenter de:s mihodes d'ingntine quiout fsu lcntm prcuvss lot, d'etudes lireliminaires ci conceptuelles cntrepnises en sueAk ddseloppe des nouveaux concept, d'aetonefs, Lc cours incurs 1*4ccent desprocedures de ealcul simples [xwur iur l'etude preliminaist ei conccpinelle. des codes machine d'analyse iniale et dos techniques eaperimentales pxsui :aprevision dus performances8 desaironefs PW Sacher

Special Course Staff


Special Course Director~Dipl. Ing. P.WSacher Deutsche Aerospace Meserschmitt-lkow.Htohm GmbH Military Aircraft Division Advanced Design Dept, PO. Box 80 1160 D-8000 Munich 80 Germany LECTURERS M."P.Perrier Aviation Marcel Dassault llreguer Aviation 78 Quai Carnot 92214 St. Cloud France Mr ILWN Iloeijmakers National Aerosace, Laboratory NLR Anthoi-y Fokkcr%%seg 2 P0 Box90502NASA 10li t B'-X'nsterdam The Neihcrl~inds M r 11 amnes 3 liromhan Road Ilidcilnhr, IBedfo~d NIK40 4AF United Kingdom Mr CW.Boppe Manager - Technology Development B/35/35 Grumman Corporaion Becthpage, NY 11787 United Satcs Dr I E Lamar Mlail Stop 361 I-anglcy Research Center flamswon, VA 23ri65 United States Mr DP Raymer President. Conceptual Recsearr'. P0 Box 923156 Sylmair, 91342-31% CA United States LOCAL COORD3NA1ORS

ProfessorMN Caiboiialo YonKarinan Institute for fluid Dynamics (hauw.s/e deWateilso, 72 !640 Risode-Si-Gecs PCuin

Professor R.Maitincz-Val Uiiisridad lioliteesim de Madrid r TS I Arronautico lilaza Cardenal Cisneros 3 28040 MIadrid Spain

lirofessor C.Ciray Aeronautical Eng Dept Middle Exit Technical Unisecsity Inoira Ilus ar Pk. 06531 Ankara Turkey

PANEL EXECUTIVE lDr Goodrich W Mall from Europe. Mall from US and Cansda: A*.-ARD)-0TAN Attn FDP Fxctv AGARD-NATO Attn F'DI'ixecunve

Contents
Page Recent Publications of the Fluid D)namics Panel Foreword/Avant-Propos Special Course Staff iii
v

vi Reference

Introduction by P.W Sacher Computational Procedures for Preliminary Design by P.,Pcrner Configuration Dcvelopment by DP Raymer Suney of Experimental Techniques for Performance Prcdit~on byU. M~ines Panel Methods for Acrod) namic Anal~sis and Design byIt WN1I lljmakcrs I ligh Angle of Atlack - Acrods nainics byi r lamar Aircraft Drag Anal~sis Methods byC WV ttppe

I 2 3 4 5 6 7

[7
INTRODUCTION by P.W.Sacher Deutsche Aerospace Meserschmo tt-Bblkow-Bloh-GhibH Military Aircraft Division Advanced Design Dept. P.O.Box 801160 D-L300 Munich 80 Germany 1. Rationale for the special course In 1986 the AGARD Fluid Dynamics Panel organized a special course on the subject of "Fundamentals of Fighter Aircraft Design" at the V.K.I. Brussels, the AF Academy Athens and at the METUAnkara. More than 200 young engineers attended this course. It seems to be timely to rcpeat a similar approach within the AGARD technical prograrvne and with respect to the scope of the previous course three major modifications were approved : (a) Aerodynamic analysis tools used in conceptual and preliminary aircraft design should be included (b) Extension to civil aircraft should be allowed and oJrae %c) Addressing mostly conceptual and preliminary design, the scope of the should be restricted to fast, inexpensive and easy-to-handle design and analysis tools. First the terminus "Engineering Methods" should be defined more in detail. It is understood that this methods shall represent proven engineering procedures most commonly used in industry during the conceptuai and preliminary design nd development of any new aircraft concept.

PH.,ASE I DESIGN CONCEPTUAL

P14ASE rl DESIGN PRELIMINARY

N|iASE M] DETAILDESIGN

STRENGTH . LOCAL S REOMT REOMTS * AEROELASTIC -BASIC MISSION REQUIREMENTS REQUIREMENTS .SPTEO -FATIGUE ALfII~UOC . RANGE -FLUTTER REQUIREMENTS * PRODUCIBILITY 2 BASCMATERIALPIRj(TI REOMTS I STRENGThREQMIS FUNCTIONAL -OVERALL a/p EIp S/La INTERNAL ARROMT BASIC GEOMETRY DESIGN '* OBJECTIVES COMI1NSTEEXTII',L COFIG In AIRFOILTSIE ER.T.ST -DRAG LEVEL *CAM 'e~ CSTTR'UTIONS 1.44 ,I/C S Ot PrIORLII * *ltIG.IT OALI 'LOCAL .X A DETAILDESIGN MECHANISMS C N,& N * JOINTS.FITTIh'.. AOAHUNO

AS REFINEMENTS *COSTGOAL$ STRESSES. * DESIGN MAJORLOADS, RESULTSOF TESTB OPER DEFLECTIONS Design (L.H.NicoIal) eiental Aircraft of Ref.,Funda Fig 1 Major design phases in aircraft development

1-2 In this sense engineering methods are characterized by o basically fundamental orientation * they have to be fat, inexpensive, easy to operate, flexible and validated in the limits of its applicability and mostly for the last rason * their efficiency is strongly dependent on skill and experience of trained personnel. Ic consequen, young engineers, starting their professional carreer in industry have to get acquainted to work with this "Workhorses" of the aircraft design business. This course shall contribute to this "Training Process". It is left to describe the state of "Conceptual Design" and "Preliminary Design" of a new aircraft project. As Fig. I outlines schematically, conceptual design is phase I in the overall des~gn process. Based on desired mission requirements, the first impression of the new vehicle is achieved by using iterative design-sizing programmes starting from sxls'i g similar "Baseline" designs with known performance. In the following phase II, an optimization process follows resulting in the complete definition of the external configuration and the database for geometry, major laods, stresses and performance. This geometric shape will be "frozen" for the detailed design (phase III) and more sophisicated design tools, e.g. complex viscous CFD, will be applied. According to the restriction to conceptual and preliminary design, contributions to the special course have been selected. In addition to preliminary design (Chap. 2) and configuration finding (Chap. 3), surveys on basic potential flow codes and experimental verification techniques follow in Chap. 4 and 5. Due to progress in using more and more the extended nonlinear range of angle of attack, Chap. 6 was included, taking also account for high speed aircraft designs, having large leading edge sweep and vortlzal type leading edge flow separation. The aircraft drag analysis methods conclude this selection of fundamental surveys on engineering methods for the daily work of the aeronautlcal engineer during conceptual and preliminary design Ir Chap.7 2 Levels of aerodynamic flow simulation

The 'lasical way t, get conflden, on a new a icraft design is the experiment using windtunnels. This neyperimental flow sinulatlon" has led to the aeelopment of the airrtaft of today. Bit in recent years the extension )f the flight e'ivelope of new projects has reached flow regimes whert the flow stoulatlon in ground test facilities has become questionable. Toe small Reyncldsnurlbers, achieved in wlndtunnels have always been a problem, but now, In addition, the flow simulation for .gh speed concerning temperature, "real-gas" chomistty and hot nodel test techniques play an important role. So more and t.ore numerical flow simulation contributes to the extrapolation from windtunnel to real flight data It has to be inderstood clearly, that CFD will never replace windtunnel experimental work, but CFD will give a strong support to analyze windtunnel data in a complementary way. The result is more confidence in a new design before first flight. There is a long list of attractie features provided by CFD when applied parallel to experimental work . (G) Increase of design broadness. An increased number of configurations will be investigated by using CFD in addition to baseline experiments. (2) The guarantee of Compatibi~ity of derived similar vehicles. (3) Quality assurance of data obtained will be independent of personal skiil. (4) Reproducability, transparency and standardization of the overall design process will be achieved. (5) Last not leas the complementary us. f CFD will result in a considerable r~ductl-i - tie design risk

1-3 Fig. 2 shows some major characteristics of experimental and numerical flow simulation.

Computational flow simulation


+ realgeometry + no limits for variation of parameters + known boundary conditions + realRe-number + short response + cost decreasing with time
-

Experimental flow simulation


scaled geometry - model flexibility limited - not always defined
-

Re-number too low

- long term (time consuming)


-

cost increasing

errors not known systematical errors (equation' good reoroducablilty /objectivity

+ accuracy of measuring technique known ? sometires hidden ? questionable (experimental "skill")

flow representation by model approximation


computer speed and memory limited

+ real flow (flow quality?)

rig. 2

Compilation of major characteristic fo&Lures in computational (CrD) and expe-

rimental (W7D) flow simulation Expetimental investigations during an early design phare require modular models with a high degree of flexibility to get all effects of major geometric parametera. As Fig. 3 demonstrates, such a modular model requires an extensive test campaign

MODULAR MODEL

Fig

3 Complexity of modular models during experimental configuration opt Lization

Therefor d nurber of good argumerits speak for the increasing importance vf computational flow simuiaton but the big "unknown* today is the demand for "code-validar.on" or the question of confidence in predicted data. Aer~dr,%nic computational ccaes used it aircraft industry for ana)ys!s and design can
vP groured into three major categories Fig 4 shows a somewhat arbitrary, but ne-

vert'eles representatlve collection of cudes used in the .MBB advanced design departrent

1-4

I
Sim'
OATCO -hanbook method,, SCHEMENSKY *handbook method for HYPSEUPLEX numertica Si l imuattin, modified newtonian theory iSES mas tpeIa - htgher ordxr sub-snd_ -~ I

P5105

to' sueb-,svpor-.hyierao panlt method derivattves loade and LAMAR vortex lattce method tod.drig mtntmtzaton vortex lift HARRIS -Wove analysis drag ofi tion rig. 4

HtOFS

iflo destign otion of HtSSS unotsody flow -EULER eqoationa potential theory IVtOC frE ' eotenston boundary to layer NAVIER.STOt'>S viotus WAINtee flew type laminsr, :OW tu, Searated botlt .viaetd/tnvrld totereettona -shock .tscous Interactions zional version

tIut.EX

Aerodynamic computational tools for analysis aind design

Some getneral comments on Fig. 4: (1'Empirical methods have to be Simple, cheap, fast and easy-to-handle. These are handbook methods but also simple linear 1. order singrularity Methods like linear 1. order panel and vortex lattice methods. In many cases this simple codes are based on engineering experience and "rules", (e.g. "area rule", "leading edgesuction analogy"). Viscous flow corrections, slender body approximations and propulsion system induced effects may often be pre-estimated within this catcgr-y. (2) "Advanced' higher order (still linearized) potential flow codes take account for vortical separated flow, nonlinear wake Interactionsa and corrections for boundary layer dovelopoc it. Iavier Sto(3)On the "Highest level", full potential flow codes, Euler solvers annl kes codes belong to the third category, often understood as the real domain Of CFD. This last Category of codes nay not oe used during conceptual and preliminar-y design because Of the need of timeconsuming input requirements and comput ational cost. Sopercritical wing design, high angle of attack aerodynamics and viscous/invisoid lr.teractions can only be simulated using Such flows with strrrmq

Following the terms of reference of this special course the content will be restricted to procedures of the first group, the simple, cheap and fast Methods (1).3. Level of confidence in &MD and CID ttsirg computer codes a general tools. We distinguish remark has to made on the status 'if computational

R. esearch codes They produce teat results which have to be validated by teat or flight data. In garterel this codes could be used only by the originators. (2) 1-ilot codes Are ready for in-house applications by several engineers having the possibility to discuss questionable results with the originator of the code. (3) Production codes Ready for transfer to other places. They have already been validated and detailed documentation is available for external applications. Our empirical methods In most cases belong to the third category. But the validation of the codes has been often replaced by "calibration'. Sofar some remarks on the problem of code "Validation, have to be made. It is understood that code validation is to insure that the mathematical and numerical schemes employed in the code

accurately model the critical physics of the flow field. This may not be necessarily
the case for empirical methods where the mathematical model representing the flow physics is poor. Effects of mesh resolutions mathematical algorithms turbulene models and gas models are ofren negligible. rig. 5 identifies some of-the major sources of errors in computational procedures.

ltSCRIpOIC( of Flow
IonsJ by Ewa~t -NS e - Eu FPE

Error sus 'Simulation to ofFlo-'

Nurerical Grl TSP-Meh Alioritrnt-4 Error ofDisretization - Derivates -A D0I. Elemrets Fbvla .', veVokS e.tste -Control Point tat of - Reores, Ion Unmetry eratboc) Solution of of ,ibsear) Systems - StiOn
-

Ofs"baI.Enr~r

SLOR - -WAY~hd ADI -e.t..


rig. s

IReore$entatton of Data J inCoomftrrsr

off error

Sources of errors in computetional flow simulation codes

In consequence many attempts have been undertaken to validate computer codes using carefully selected "test-cases'. The prediction of drag has been proven to be still the most critical problem. rig. 6 shows a compilation of data obtained in an early attempt (GAY241981) to validate computer codes for a simple NACA 0012 airfoil at transonic speed. Eveit for the prediction of pressure-drag results obtained from various classes of solutions (non-conservative, full conservative full potential flow same category soluand Euler equation so.vers) differ significantly, but even in t.l,e within l00t deviation. tions of different codes predict values for drag

CO

600 ,I
_ _

__0.

oI5..

1.01

od .I,

- ,."ta,
rig. 6

..

et

50

59'

Test-cases for code validation (NAMA 0012) Prediction of drag using CrD codes

Since that time the situation has improved but even using Navier-Stokes flow solvers, the prdiction of drag remans the toughest challenge for CFO. On the other hand experimental work has also been done to =validate" experimental data obtained in diffgrant windtunnels. The situation is not so different from theoretical work. As Frig. 7 shows, pressures and coefficients differ significantly for the same simple 2-D profile secticn.

1-6

I
xO.76,.z, *

0.51. COO? 110011.SctIofo 7

voo

: I.0...,'

7R.0 '

H Mimi,

rig. 7 Experimental approach - uncertainty and sensitivity for identical models in AG02) different windtunnels (CGARTEUR The list of activities concerning code validation during the past ten years is very impressive and AGARD h4s played an active role 9 * a
a

1979 1981 1982 1984

AGARD AR-138/FDP WG04 Experimental data base for computer program %ssessment GAHM workshop on 2D test-cases
Full potential and Euler flow codes

9 e
a

1985 1986 1986 1988


1991

a a

AGARD AR-702(1984 addendum No. 1) Compendium of unsteady aerodynamic measurements NASA NTF delta wing model Database for computer code development (AIPA 84-2150) AGARD AR-211/FDP WG0 Testcases for inviscid flow field methods ACARD AR-226/FDP WG 08 Aerodynamics of aircraft afterbody International vortex flow experiment on Euler code validation FFA Stockholm AGARD CP-437 CFD validation
AGARD AR-2?0/FDP WG 13

Air intakes for high speed vehicles

In this "environment" cf CrD and experiment the engineering methods during conceptual and preliminary project work will be outlined in the following chapters. Major emphasis will concentrate on the applications. Regarding detailed theoretical basic asaumptions underlying engineering methods, the references will be given. It is the intention of this special coure to initiate interest in the overall design procedure of a new aircraft and to give young engineers and students the opportunity to get acqueinted with the "workhorses" of daily routine in aeronautical engineering.

1-7

a. Conclusion This Special Course on "Engineering methods in aerodynamic analysis and design of aircraft has been orgonized at - The Middle East Technical University-(METU) of Ankara from,6.-l0 May,1991 (36 Attendants) - The Von Karman Institute (VKI) in Brussels from 13.-17.May 1991 (38 Attendants) - Politechnical University of Madrid from 20.-24.May 1991 (60 Attendants). In addition to the technical presentations a Round Table Discussion with the lecturers was scheduled at the end of the course. Some preliminary technical evaluation of the course Yas given by the course director as follows. Six major presentation have been given during the Special Course - Preliminary design - Configuration development - Experimental techniques - Potential flow codes - High anglo of attack aerodynamics - Drag analysis methods The first question at the end of the course is concerning the completeness of the content. Accord.ng to the comments from the audience during the final discussions no recommendations for additional topics came up. The second question addresses the aim of the course. Did we attract a sufficient number of attendees and did we reach the "young engineer" who is about to start his profenslonal carreer in aeronautical engineering? Due to the number of attendees and the contributions to the Round Table Discussion also the second question may be answered in a positive sense for all three places. The recommendation case from the floor that a similar course should be repeated each second or third year. rome major findings from the presentations may be highlighted for better recollocio,.

Computational procedures for preliminary design Pierre Perrier introduced the audience to the different definitions of design levels in the environment of the "magic triangle" of Real Flight - Experiment - and CFD. In this sense EFD stands for the simulation of the -Real World" in contrary to CFD simulation of the "Soft World'. He described the 'Rendez-Vous' procedure in terme of levels of quality versus time for development. According to this phi losophy the state of conceptual de3ign using simplified engineering empirical tools develops to the stage of feasibility using much more sophisticated experimental and computational tools before approaching the state of manufacturing the .aw aircraft. Configuration development Daniel Rayer stressed first of all the necessity of design trades. Basic design trades e.g. canard- versus aft-teil configuration or wing planform trades have to be repeated for any new project design. A second group of basic trades deals with "Requirement" trade-offs, e.g. max. speed versus maneuverability or maneuverability versus detectability. It is obvious that these timeconsuming trades could only be performed using automated design programmes. A major role dLring the application of design programmes is the definition of a socalled "Baseline-Design" configuration with known performance. To save computing time these "aircraft sizing" programmes rely to a large degree on simple empirical engineering procedures. Reference to these methods applied in design programmes have been given in detail. The result of the ap

plication of

this design

programmes will

be the

evaluation

and transparency of

"design-sensibilities" depending on systematic parameter variation concerning GrossTake-Off-Weight (GTOW),. Experimental,techniques for performance prediction Barry Raines-reviewed the present state-of-the-art In experimental testing as a means of prediction of aircraft performance. Standards of accuracy arf defined. The first part of the presentation deals with all aspects of data acquisition systems and discusses all effects contributing to uncertainty of measured data, (flow quality in windtunnels, windtunnel-wall interference, modelsupport interference and scale effects). In his second part the lecture discusses the types of models and test rigs used in determining the propulsion interference effects on both transport (turbofan and turboprop) and combat aircraft. Especially for recent designs the engineering problem of optimum engine- airframe-integration plays a dominant role both for civil (due to econonic reasons) and for military (due to drag-performance) projects. Even arbreathing space transportation systems emerging in the near future rely to a large degree on the interdisciplinary "integrated design" or the engine components like intake and afterbody-nozzle. Panel methods for aerodynamic analysis and design Harry Hoeijuakera presented an extensive and complete overview on aspects of panel methods used in aerodynamic analysis and design of aircraft. This solutions of the linearized potential flow equations are today the most important "Workhorses" in the "Tool-Box" of an aeronautical engineer. They ha':; now reached the level of personal computers for practical application. Being more complex as the methods generally

understood as "4empirical", the use of panel methods requires a great deal of engineering experience and personal skill. Limits of applicability have to be understood (and explored) by each individual user. The lecture starts with the detailed outline of the theoretical approach for the approximations made for the flow field and the discretisatlon used to deal with complex vehicle geometry. It reflects all major issues of existing panel methods and shows examples for applications both for simple and for complex geometry. Propulsion integration, viscous correction procedures nonliner vortical separation is referenced. High-angle-of-attack aerodynamics John Lamar discusses been identified : the different regimes of the CL-0l plane. Four *C-segments have and

low moderatehighpost-stall -

attached flow dominates


combination of attached and separated or vertical flow separated or vortical flow dominates vortex break-down or massive stall

Depending on the wing planform and Machnumber this segments extend to different size. The paper deils first with engineering methods for the prediction of vortical separated ilow (e.g. Sychev similarity, Vortex Lattice Method-Suction Analogy, Digital Datcom and Free Vortex Filaments). Second the high angle of attack range is stressed for stability and control. The effectj of different wing planforms and the effectiveness of control devices (including "vortex flaps") is discussed. Finally the subject of Post-Stall-Flight is addressed, including aerodynamic control devices, thrust vectoring and dynami, stall.

|V

Aircraft drag analysis methods

1-9

I
! "

CharlesE oppe structured'his lecture into two major parts (1) 'Evaluation of drag and (2) Reduction of drag First-the sources of different drag contributions are discussed. Handbook metbds are indispensable for prediction of drag contributions. The engineering methods are considered to be the bridge- between empirism and windtunnel testing. C.)nsiderable timesavings are achievable by careful analysis and understanding the drag mechanism. For the second subject engineering methods are applied to achieve higher performance. Insight into the complex drag mechanisms is required for the desired goal. Reduction or windtunnel test time and flight tests is achieved through the use of empirical engineering drag prediction tools.

Summary Three major statements characterize the major fndings of this special course (1) Engineering work in aeronautical analysis and design is traditionally performed in two ways * Experimental approch, characterized by limitted simulation of flow physics (e.g. Re-Number, Hachnumber, Real-Gas .. ,). " Numerical analysis, characterized by trend to higher level codes, high cost for viscous 3D codes, production of high quantity of flow field data, lack of code validation. (2) Interdisciplinary approach is mandatory in conceptual and preliminary design work : * Experiment will not be replaced by CFD, in addition to traditional configuration testing the experiment has to provide data for CFl-code validation.
* The role of CFO is a complimentary one with respect to the experiment,

extensive use of CFD leads to quicker and more reliable selection of the most promising configuration. (3) Engineering methods are indispensable because : * "High-levol*CFD analysis is excluded in conceptual/preliminary design a Experimental work is not (or limitted) available for configurational conception * Empirical (e.g.-Handbook-), low-level (potential-) flow code analysis on PC's and Workstations and extrap6lation from engineering experience obtained during
previous design work is the logical consequence

(4) There is an obvious need for training young engineers to get acquainted with simple engineering methods.

10

ACoDNOLEDGMU

As director of the AGARD FDP Special Course on "Engineering Methods in Aerodynamic Analysis and Design of Aircraft" I like to express my appreciation to the lecturers which have performed their part during the course in a most professional way. It was my intentionto bring together experiencedexperts from industry and, research institutes, experts which are-recognized worldwide to work successful in the related field of aerodynamicanalysis and design. They all have reactedspontaneously in apositive way, shortly after having been a-3ked contribute to such an extensive effort of the to AGARD community Mr. Mr. Dr. Dr. M. Mr. Charles W.Boppe from Grumman Aircraft Systems, New York Barry Raines retired from ARA, Bedford Harry W.M.Roijankers from NLR, Amsterdam John W.Laaar from NASA Langley Research Center, Hampton VA Pierre Perrier from Aviation Marcel Dassault, Paris Daniel P.Rayaer from Conceptual Research, Sylmar CA

In addition I wish to express my gratefulness to the local coordinators. They have provided excellent local arrangements and perfect organisation of the Special Course during the meetings : Prof. Dr. Ca hit Ciray from the mtiddle East Technical University in Ankara Prof. Dr. Nrio Carbonaro from t.he Von Karman Institute in Brussels Prof. Rodrige Hartinez-Val from the Polytechnical Univertrty in Madrid. This Course could have not been organized without the outstanding support of the Fluid Dynamics Panel Executive Mr. Wimston Goodrich and his Secretary Nd.. Amnemarie Riveault, from AGARD Headquarters, Paris.

2-1

forpreliminary design procedures Computational P.,PERRIER, dhhrodynamique Thorique Chef du Opartement DASSAULT AVIATION - FRANCE 300<- 92552 SAINT-CLOUD CEDEX

0. - INTRODUCTION evolved has, design of aircraft Preliminary on largely the past ten years.The mainoriginof evolution came from rationalizationand broadening of the preliminary phasis of Has to be evaluated of development a new project. sooner the interestof the project in its of capabilityto fulfill the requirements a for program- has to be proposedthe strategy with : makingItsdevelopment with what funding, collaboration, what national or international the level Moreover withwhat long term calendar. : is involved to be evaluated not of technology but also only as the stateof the art available in effortto be included an as requiring ad-hoc the totalcostor timeschedule. or Rationalization of national international of a large programs has led to definitions of levelsof freezing of succession well-defined the design; of comon use is for examplethe of succession fourdefinitions following tentative 0 Definition is the firstcomplete all include the for definition whichthe drawings necessary for freezing the main ingredients

Such passengers .>.). accommodation of be can definition therefore the firm basis for a for the trade-otf study : it includes complete aircraf.as a system and an evaluationof to problems be studied. aerodynamic critical of 1 Definition is the product an improved of analysis of design; it results the complete testi,9 0 the definition afterfirstwind tunnel of and firstanalysis the criticalpoints.It study can be the basisfor the firstfeasibility ; aircraft it can give valuable on the nominal support to the first evaluation of main It uncertainties. is clear that it is the first of evaluation the wherea realistic definition neededfor probable performances and of efforts can be having them, plus risks associated. not may given.Performances be determinated as but a estimation stateof the art preliminary comprehensive set of data comingboth from CFO and Development. Research and from experiments. of to are related criticality the needed Efforts problems compared to the objectivesof the for technology solving and on available program them. Risks assessmentare basea on the of solving the problemswith the difficulty and on the funding and the time scaleallowed,

solutions. of architecture in details - volumefor equipments. existence alternate of correctgeometry main parts fixedor moving 2 Definition can be frozen when iteration ; (undercarriage removable parts, external including with detailed requirements ( ,). It is highly hoped that the center stores.. trade-offs. maintainability, economic of gravity be at a reasonable position relative fabrication requirements...) cle - the way to centerof pressurein the flight to aerodynamic firstdetaileddesign of major or long-cycle envelope ; in, the same effort of having a of freezing Progressive partsof the aircraft. are main requirements to firstdesign. reasonoble performances definition2 can allow some extra time for on be fulfilled flyingqualities, work, mainly on the more complex (volume of tanks) weapon system (antenna aerodynamic shapesneed to be phtnoeia ; but geometrical of (volume, center gravity, payload locations), costly charges.Final stablenow for avoiding 3 data can be a definition near the geometrical 2. definition

2.2 Here we will covermainly the preliminary design work needed for the production of -definition Howevera part of the trade-off 0. studiesand of the continuing work of refinement of definition and 2 can be'donewith'thesame I tools. Analysis of critical problems on definition can lead to a reassessment sone I of alternate design ; again such variants 1.1, 1.2... have to be bui!tquickly, with the same so tools, One main philosophy that has emergedfrom repeted experience such preliminary of cyclesof design the concept "samelevelof quality is of of design comparison".There is a clear difficulty to extractvaluablecomparisons of data obtained on different designs with different tools and different levels of convergence it leadsto mixingof evaluation ; of different designs with the evaluation quality of of the output of differentdesign tools. A conservative design may seem poorer than a promising not yet compromised design.It but new is pirticularly trn , but in the opposite direction direct for experimental comparisons of designs that are not at the same level of improvement krO : If designfor example wing by a has been optimized, a large variation of main parameter (sweepangle, location nacelle...) of can be selected withoutnew optimization but ; the derivative will be probabiy poorerwhen the derivative are issued from pralinlnary design toolsor when designis sensitive qualityof to design (e.g.transonic. laminar...). Anyway the computational proceduresare central in the evaluation a designnot only beforebut also of for analysis afterW.T. testing. 1.1 - Center pressure of evaluation It is the first main aerodynamic characteristic that Is needed from the beginning. Without it the work of the design office cannot be realistic. Some trends are needed for'balancing general the architecture of the aircraft.The size of the wing can be deducedfrom rough estimates the realistic of wing loadings and the size of the fuselage is generallycoming from volume constraints, the balance the mass require but of quotation from the start of the project definition. a orogressive So approachby three procedures (each being more complex and more accurate that the preceding one 4eem necessary and haveto be used successively. The first Is an old but efficient rule of determination approximate of subsonic centerof pressure the drawing on table.It relieson the assumption that the repartition of tn lift on the different elements of the aircraft is ellipticor slightlydistortedfrom elliptic distribution given by figure I versus the as aspect ratio. sweep angle and taperratio.And we can assure that wing + fuselage characteristics are obtained from wing alone plus Interaction. Conventionally KW is the factor wing liftIncrement of when fuselage-body is present, Kb is the percentageof lift transferred the body. Figure 2 gives an to estimation Km and Kb (coming of from low aspect ratioor slender body estimation) versusratio of equivalent cylinder wing span. to The aerodynamic center can be built oy assuming that its localposition for a slicein spanis on a 25 % position current on chordand I -GENERA S CO UTATIO PROCEDURES is distorted 25% 1 "112 by the symmetry to conditions. the curveupon whichcan e put So the load givenby figureI and 2 can be easily approximated. r, o
,,

1.10 - We will cover successiaelythe computational procedures the firstiteration for of designbeforethe selection definition of 0, as referred before,and in the preliminary the

freezing definition main importance after 0. Of


are evaluation centerof gravity of position and of thrust-ninus-drag, and L/O for general performances. first. So these Items will be covered

N.*

2-3
-

such,.jo&,.s to retain-on a.P.C., with, extended


core,memry . and arithmetic, coprocessor two, ke : one,,in~subsonic, useof simpllfied,codes mthod, the second one is yortex-lattice Tinearised supersonic planfou,,evaluation. Both requireanlimted timeof computation but also precise rules fcr input of-geometry for the qualityof the results. On the vortex-lattice method it Is absolutely required to well known the following rule : a one quartersingularity plus three quarter control point is mandatory

I' there is a tall. the same approach is useful but a reduction efficiency of comingfrom def;ection needed and Ft . 3 gives typical is fig res vs separation distance betweentail and wi'g.7he absolute valueof tail or wing alone li'tis usually not far formDiederi(h formula 2 t( ACL ........... I +(u.ajA Such first level of aerodynamic center estimation also used for quick evaluation is of trade-offsduring first interaction on the design. The hand procedure here described can appear in a codefor a PC such code is useful for avoiding errors but with it engineers may loose understanding the complexity interaction of of and of theirapproximation moreover inputsmay ; be complex to theirnumbers may be source due and of errors.

for each cell retain in the discretisation process. the supersonic On linearised method, ;n efficient way is to use the quadrilateral meshing alongthe characteristics lines; if the mesh is regular the matrix of influence coefficients be in-ersed priori and such can a furnish a very cost-effective computation procedure however such regular predetermined ; qrid (with an affinity factor for span adjustement) implies difficulty with irrnular planforms ; that can be improved by local refinement. Another advantage of such linearised procedure to give an approximation is of the Cp distribution chordand span and of in lift and pitching moment; such data are useful for first determination of some critical problems . it supposes a first definition of a camber control or deflection needed for balance at a givenangleof attack. it Is the first So way of making evaluation camber or twist. of Inputs become more complex with an important numberof celles ; that can be the source of (often complex analyse) to errors data. in

The second level of estimationof the position of the aerodynamic center Is based on the use of small computers. A goodway of doing

2-4 Thethiadievel of estimsiiidn of pousition of 6center Wi6ibg to 3D coputation the aeri~dynic withb~t ilneiriisation on 'the plifd4. 'Such aircraft requlied for -aiy pioceddie is 'soon difficult to balatici. In th past, the costof review the tools .gainst helow. bat we can stannrize from now the three levels in the following table.

itools v:iw~~ ,09 t des~ .:.lwwl..ita is bf now it is io more _____________________ such app~oich waseeisi~i PP ltt-P.C 1.4.1 7 a .t are able to notice that the tyia true and wie ta. iitl. Curirelit capacity of' in-houise workstation. codes - use otta'. 2 et6 devote such workstation LCYOL Linesrsaoi prcie ilno

~t

irt

to the design gr~ap for specific task as the suich

tt'rA
-t..1.
xatw

C
Mil air,,.C
witwtfwtitltltidr0 x ittotttt o r .Iwttairo
*.b

aerodynamic and stress analysis at the levels of the preliminary design. Probably the ccoplete transsonic design will be excluded of suRh preliminary work 1,however the main limitation comes frcom the interface with geoetrical 25 definition and the time for having a goadmesh for computations. It is clear that singularity on is panel method well adapted to such wourk,this is because such panel miethod requires only the discretisation of the surface of the aircraft,. Subsonic and supersonic panel methods are the basic tools. bat morecomples 3Dcomputations may be moreuseful. In the third level of codes to be used in

" I.e haoia ....


aa.

Complt
witeio ,va.0 wl.

4 dlm

Iid

ihilrtti ttwtral

,wtai

sit. 11..~
I tt
it~r'

wh

... 1 t

soalP

~
....
iil;!i

MMIii~tl~t
li.

10.0.t
tth

I ........ I tI
thia

1.2

Lift evaluation

preliminary design inthe level of 3Dcomplete


Inthe suve manner,the codes used in the viscous codes ;their use is questionable. In we associated 1.1can.deliver elements for lift. However, cost fact, their effective isdirectly between3 different : Of the of with cost preparation computation have to distinguish the data aerodynaltic : linearlift angle of mesh definition on geometrical data. subdomains with lift attackrediction, mainlyobtained definition, checking of the quality and correct of lift. high and high-lift-devices angle attack meshrefieement as locally required. So finite for Thethree are of main inmportance different element method,taking into account the total the cover lWe flight envelope.will parts the of to seems adapted ofthe complexity geomeotry, well successively. three is fast answer except If mesh too costly ur long to obtain, Good linear lift vn.Incidence prediction Is achieved by the three level of codes described Intermediate codes, for that point of view, are before. It Is clear that such prediction Is no the 30 codes Involving direct meshing, that mesh, useful for high dynamic pressure rectangular Is the Typically meuh a regular the solver generally takes advantage of uuch corresponds to low values of angle of attack. Imes appear In such flight reg Some concern mtay reguarity-it can be linear finite difference or aeroelastic effects ouch aeroelasticity for of *. spectral local application boundary with beam ane procedure thelong effectssimplified a conditions can be dune directly or with local approximation :flexion and torsion of wing and regular refinement. At the other extremity of of by fuselage be estimated projection can boundary conditions are the codas devoted to the on efforts and moments neutral line of the around any irregular complex eon structured mesh Iterative procedure A equivalent beamus. Newton on solver element for willask finite body .that deflection. final towards helpto converge geometry, as needed on final will any ccomplex Hain contributors are coming from twist-indaced We shapes. will aircrafts ccW utations, on final with of of lines boa wings of by flexion neutral fig. S. sweep

2-5 On figure 6, we have put the non-linear inviscld valuzof,2O lift and-the experimental We wingsection. camber-of for increasing values to related haveput alsothe angleof incidence maximumliftand the equivalent angle of attack thesame maximm lift as the section but having on- the, inviscjd curve. A lot of empirical of criteriafor determination these two data We have been proposed. will retain the two procedures following
"7

are computations available, If no viscous results on similar use of experimental wing section will help to define the "state the art* lossof liftand angle of for achieved stall. of attack of the computation If can be fulfilled of the boundarylayer on upper ,urface wing sectionwe can make the following assumption - on one elment section the when separation maximumlift is obtained occursat 85% of the chordwith inviscid pressure distribution. Such figure is a mean value but can be very useful at the designlevel. preliminary can samefigure airfoil, On multi-elements be retained for the main section Revaluable at separation the lastelement. with a valueof 50% on data are obtained its own chord. But when the camber does not the Cp distribution increases, any morewith tigleof attacknear change on edge : total separation the trailing the slot is the best criteriafor stall prediction.

withoutor with High lift characteristics, of mainlyrely on dissipation devices, nigh-lift m~in wakes and mixing of viscouswakes and of So separation. it withor without layers bourda,'y flow computations. be cannot predictby inviscid can be done in the Howevera firstassessment line. if we returnto So of approximation lifting of procedure liftingline wing + the preceding we + interaction fuselage can use the following computation iterative - compute the 2 0 l'fting correction due to + viscosity stallestimatirn
-

lifting line lift non by It distribution. is obtain iterative linear spanwise induced downwash compute the 30 computationsuntil convergence toxards equilibrium.

can be used : 1st So a threestep procedure to contribution high lift of estimation inviscid by the previously defined codes : e.g. with non linear boundary singularities, conditions.

,1.

2-6 Betterway of designis to'use a complete iterated code taking, into' account' shape of the the separatedwike and interaction between bondary layerand wakesdownstre of each shot. If the work is donet n2;60 (that"is to say with account sweepangleand localtaperirgof the fbr wing)a complete set of lift (and noment) versus local angle, attack is available. of Adding a lifting linecomputation induced of velocity will allowto be simplified code to give a very 3D precisepredictorof high lift devices when aspect ratiois not too low and sweepangle not too high (say AR)3, If 45). On fig. 8 is given such a rebuilding of a typical subsonic aircraft by a genuine Dassault code. For very high sweepangleanotherapproach is to use the Polhamus approximation where the succion is assumedlost and transformin a vortex lift. Best procedure'for succion predictions can be done by a vortex lattice method makingthe succioneffective however ; maximum lift related to vortexbursting only is empirically chosen. In transsonic, fast znswercan be obtained with wing aloneor wing + simplified body finite volume computation. time of cosputation The is not so large and a complete set of data can be obtained by survey of maximumlocal Mach number normal to the shock-wave. simplifiedrule A assuming shock-wave/boundary layer separation

when Machnber Is larger than 1,4 gives a good


approximation of the buffet angle of attack (fig. 9). If the complexity of the aircraft does not allow reasonable answer by wing alone Zconputation (andin orderto fix the CL aircraft versus CL wing) sone cowplete aircraft parel computations are needed, for examp e b

-'.~

mthrd.

,!i r1

-Drag evaluation
Drag evalation is the moreccmaplex and

|i

IL: ,
I I4 L1 ",t

,I
--

difficult task of any engineer In charge of preliminary design. Some probable evaluation can conponents as those drags related to integration and at the drag of engine airframe miscellaneous. be highly empirical the can at

be done for friction and Induced drag. but more

I4complex

preliminary stageof definition an aircraft. of

i1

2-7

with good is and Friction formndrag obtained are curves, ;,-such booksfrom accuracy, the ,data (see well established, for examplethe friction drag, of Van Oriest data book). But soe 'are needed ; they,are. directly corrections taking ratio forof re)atedto square thickness larger than in accounttrue local velocities are to be given to value.Some concern infinite for particularly low altitude the roughnessodrag, taking in A missionprediction. drag breakdown accountlocal chord and their Reynoldsnumber effect is importantin the selection of of with large variation chords. configuration of estimations frictiondrag only Preliminary except at the basedon wettedarea is dangerous stageof study. verypreliminary Ww'jedrag can be obtained At low cost in the flowfor wingsend of of approximation linearized equivalent area flow for axisymetric ca distribution the body. Dut an interaction for area rulingeffectrelated is process needed Fig. 10 interaction. and to transonic supersonic process obtained by a gives such rebuildiug design. code itedIn preliminary genuine Dassault for highly It is to be noticedtha. correction pylons.,.. for canopy. Is effect needed non linear is If such correction to be added,howeversuch particularly givesa muchbetteranswer procedure in transonicrange than the transonic or supersonic area rule formula based on ; trans-supersoni: area rule distribution it was is shownthat such formula only applicable with (variable successto very slenderconfiguration geometry aircraft with high sweep angle without troubles coming from configuration) edgecontributions. trailing

of the evaluation drag related We will discuss in installation the nextchapters. to engine

PROCEDURES ENGINEERING DESIGN 2 - DETAILED design of 2.0 - Evaluation first preliminary withouta qualitycan no more be done actually area of design.We quick survey of separated of evaluation local will coversuccessively'the non-separated between area,frontiers separated includes Such evaluation regions. and separated the necessary first survey of air-intake integration and of afterbody integration. Many timesit is at the levelof induced separations with non the interaction that one has to predict requireents as those coming from terodynamic RCS signatre reduction. areasevaluation 2.1 - Separated to It is of main importance surveyfor scoe criticalpoints of design the boundariesof areasof the wetted totalarea ; all separated of all angles versus be the skincannot examined attack and mach number of interest for layers. of separation boundary designhas to lead directlyto Preliminary : configuration deviously of selection aircraft is of one main element choiceof configuration of design or the generallythe cleanliness from a, aerodynamic of boundary suchcleanliness designtool pointof view.The best preliminary afterthe other of is the survey one stretiline layercode ; fig. 11 gives with a 30 boundary of an example a flow survey at the wall for a separation rear fuselage towards Falconoriented Suchcode can analysis. by estimation streamline be operated on a workstation using an inviscid distribution coming from panel methods in subscnic or finite element in transonic. Of of main importance are the ability the code to give indication of the local determination of and 30 boundary layer by shape parameter A and shearangle versusthe localconvergence Easy curvature parameters of the streamlines. surveys of the orgine of streamlinethat is separates needed.

&I -pressure

'a

2-S
Particular, insistence has'to be put on accuracy of such-finite difference codefor evaluation of the drag because-the too-rough" evaluation of 'succion" recovery directly extracted of ID imnmentuu equation, so-called as "additive drag", is dangerous. It is better to rely on integration pressure such code which take of of correctly in account the, internal-external "recovery" the lips.Conventional drag, on ram as put in the definition the thrust of delivered by enginemanufacturer generally given in its brochures, to be compared true pressure is to integrals. Equivalent axisymetric air intake can furnish betterdata if careful duplication of local slope and duct area distribution are done. On supersonicor transonicdesign, the probl)m is generally much more related to existence high intensity of shockwave and os correspondingupstream shock-boundary layer ir.teractions. Checking of validity of criteria of design by direct Navier-Stoxes solution with odelling is out of the budget turbulence simpie of preliminary designand has to be replaced by

I
FA=CN 5 -1-1

empirical evaluations.
2.2 - Air-intake Integration Large difficulty in design cones from engine-inlet integration soon in the designis ; the necessity define preliminary to stage at leastroughlyfor the boundarylayer diverter. The necessity evaluate volume to the and position to be reserved to the air-intake is much mandatory for internal architecture of any project. simpleone dimensional A analysis code is neededfor evaluation the areadistribution of of the dict and of the throatarea required In different flight regimes.Anothercode has to help prediction of supersonic recoveryfactor taking account of losses In the external or Internalshock waves and boundary layers. A simple axisywetric code Is needed that uses 20 the correctarea distribution the duct for of preliminary design of possible internal divergence, out of design external spillage and drag. Fig. 12 shows a typicalresult of such axisymmetric code that helps a lot in the preliminary design phasis when intake area, external and internal devices are to be selected,

i
r

Complementary work has to be perfomed from the beginning the designrelated Incoming of to flowfleld. Some external reconpressions of the flowmany timesare comingfromthe shapeof the aircraft. Effort are to be devoted.from the beginning the design, clarify of to what is the interaction the forward has' fuselage wing on or the flowfield the entrance the air intake. at of Selection frontfuselage of shapecannot done be without such preliminary study. It can be fulfilled simplefinite difference by code as described 13. in

2-9 with the,streis effects or for interaction of Howeverthe selection analysisdepartment. position, shape, volume to be devoted to are a part of the-same effort towards antennas preliminary at~the Maxwell integration complete or weapons and,external phasis.Internal design
A. WA, n-fkMintegration

effort for tanksare also partof suchgeneral but specific tools are not needed evaluation separation of exceptfor preliminary problems.

3-CONCLUSION
tools all the engineering We can summarize designin the following used in the preliminary

J- I.

table Engine
Integration 2.3 - Afterbody AC/CP Lift Drag Inletexhaust fore and

Symmetric

tork on afterbody

has

to

be
-

[after-body
SG.C S.G.C I L.M FAX 2.SPand FON Level3 30P and 4 3 DP and F SGC L NLC + E.R I INVISED CODE E.R I Level1 S.G.C I Level 2

it However is clearthataxisymetric fullfilled. or monodimensional codes are not convenient for such study for twin engine integrated afterbody. For such study the delimitation of separated areas are to be done systematically with the procedure of 2.1. For more complex shapes the analysis is out of the scope of simplified flows. viscous-inviscid inviscidor incoupled cdseitFE Some codes eint that can take advantage of criteria basedon reattachment simple correlation or on mixig-layer development. but there are generally of limited values.Progress are to be done,but they will come from sloplification f much more comp~eveNavier-Stokes solutions. Such and in are results to be validated wind-tunnel in Flight work are in progress. ; 2.4 - Interaction with requirements non aerodynmic

3 Of and t FN

30P Ff 30 S and S

: Simplifiedgraphic data-sheet and conputing; R : Expences'srules ; L.M Linearized method , P.M. Panel Method , FU : FEN : Finite e'sment Finite difference method method ; L u NC: Linearized + non linear corrections AN : Approximate methods ; Ff : Flow field separation. S and s : Streamlineand

of integration ACS or IR reduction aerodynamic design. It

More are to be done In future design for in the is to be assumed that

of reduction cost of It is clearthatrecent computationby the minisuper computer and advanced workstation has shift the CFO work to the analysis from detailed computation Numerous toolsare now phase. design preliminary used in suchphase,it quality of the 0.0 project. will improve greatly the definitionof any new

Maxwellsolversare at the equivalent simplified and that Interaction can of disposal designers and take place betweenaerodynamicists Maxwell in Integration the same team is specialists. more As an exople this is probably mandatory. had been than in the past when efforts important of for push forward integration the aeroelastic

"JT

2-10

Rdfdrences

"

1;.d.R. Sears':'ighspeed Aerodynamics -and propulsion, Vol. VI General theory of high s e cwing

11- H. Carlson and iii'e 0. Iet ., tand

The influence of Cambered

design for Supersonic cruiseoAlIA Paper 81 -16 S6 - 1981Aerodynamics Ahsedo Vi 12. F.W. Bursham, D.R. Bellman A flight

2. A.F. Donovan, ,'onn of n o

H.R. Lawrence-: . Lawcr e t en

tinvestigation

of steady state and dynamic pressurephenonena in the air inlets of sopersonic aircraft AGADCP 91 - 11. 71 P9 GR

3. Royal Aeronautical Society Aerodynamic book.suesncarrf

data

4. E.G. Covertand all : Thrustand Drag : its predictionand verification. Progress in Astronautics Aeronautics and Vol. 98.1986

13. W. Schmidt: Aerodynamic Subsonic/Transonic aircraft design studies numerical by methods AGARDCP 285 - 1980.

S. M.L. LOPEZ : Aerodynamics and performance characteristics wing l;ft augmentation of system. VKI lecture series 60,1973.

14. L.R. Harper The subsonic Performance of P-actical Military Variable area convergent Nozzles AGARDCP 301 - 1981.

6. B. Halises :Aerodynsic interference . A roneral overview AC.VD lecture Series Report 712.1983

15 - l.i. Rettle : Aerodynamicdesign for overall vehicle performance AGARDreport n712 - 1983

16 - Rossow. troll. Radespiel, Schen : ce Rdsil rl. 1 osw - AP -re : PROGRESS 2Investigation 7 - P. Perrier M. Lavenant -a MECHANICAL P. n PGESS REPORTON REPO3 0T of the accuracy of finite ApriCAl A 3 4volane methods for 2 and 3 dimensional April1974 flowsAGARD n' 437 - 1988. CP
8. P.

Perrier J.J. DEVIERS : Calculs tridimentionnels d'hypersustentation Colloque d'afrodynamique appliqu~e de I'AFITA# (AA9F) 1972.

17

P.W. Sacher : Fundamentals of fighter Aircraft Design AgardReport740 1 1986

9 - R. Eppler : Airfoil Springer Verlag- 1990

Design

and data

IO - B. Oillner and C. Koper The role of computational aerodynamics In airplane configuration development - Agard C.P. N280 1979.

ECONFIGURATION DEVELOPMENT Daniel P. Raymer


.P.'O. Box 923156 Sylmar, CA,.,USA 91392-3156 NOTE: The followig materal, presented~aekpart of the AGARD FDP SPECIAL COURSE ONENGINEF.RING .METHODS-IN AERODYNAMIC ANALYSIS AND rzedzfrom the author's textbook, DESIGN, 7is xcerptedands's rAIRCRAFT- DESIGN: *A Conceptual Approach" (Copyright C 1989, Aeronautics and by the Aiserican. Institute- of published Astronautics, 370,L'EnfantPromenade, Washington, D.C., U.S.A., 20024)*. 'The lecture charts ire part'of-the-five dayShort Course on Aircraft Conceptual.Design which is regularly presented by the author., The authorretains full copyright protection of this material, and further publication or reproduction beyond this AGARD sh6rt'courseis'strictly forbidden without prior written approval.

3-1

INTRODUJCTION & SUMMARY


Aircraft conceptual lesign ie a
eo..., .
"""'N

complex, involving

process multidisciplinary science, history, art, and

magic, in sometimes equal proportions. In this AGARD special course, we dre focused upon the aerodynamic aspects of aircraft design, but the overall configuration of good the aircraft must both provide aerodynamics and reflect a wide variety of lecture other considerations. In this

I
'
,..

, .

I . .

number

three,

we

will

discuss

configuration development and its key role in aerodynamic design. CONFIGURATION DEVE.OPMENT PROCESS Aircraft design can be broken into three major phases depicted in figure one. Conceptual design Is the phase where the basic questions of configuration size and weight, and arrangement, performance are answered. first question is "can an The affordable aircraft be built which teets the requirements?" If the answer seems to be "no", the customer may wish to change the requirements. This is not too unusual, for the customer sets the requirements as a compromise between what experience says is feasible and what the end-users of the new airplane would like to get. fluid Conceptual design is a very process. New ideas and problems emerge as in evera design is investigated increasing detail. Each time the latest design is analyzed and sized, it must be redrawn to reflect the new gross weight, fuel weight, wing size, engine size, and other changes. Early wind tunnel tests reveal problems requiring some often changes to the configuration. Preliminary design can be said to begin when the major changes are over. The big questions such as whether to use a canard or an aft tail have been resolved. The configuration arrangement can be expected current to remain about as shown on drawings, although minor revisions may occur. At sose point late in preliminary design, even minor changes are stopped when a decision is made to freeze the configuration. ,.-no".

[
"''

"

".

'

'"

'

..

.4a..

preliminary design the During specialists in areas such as structures, systems will landing gear, and control design and analyze their portion of the aircraft. Testing is initiated in areas as aerodynamics, propulsion, such structures, and stability and control. A mockup may be constructed at this point. for Assuming a favorable decision development, the entering full-scale the detail design phase begins. Here, fabricated are actual pieces to be designed. For example, during conceptual and preliminary design, the wing box is designed and analyzed as a wholo. During detail design, that whole is broken down into individual ribs, spars, and skins, each of which must be separately designed and analyzed. Detail design ends with fabrication of the aircraft. Frequently the fabrication begins on part of the aircraft before the entire detail design effort is completed. Hopefully, changes to already-fabricated pieces can be avoided. The actual design effort usually begins with a conceptual sketch (figure 2). This

Copyright C 1991 by

DANIEL P. RAYNER

EA

ALL RIGHTS RESRRVED

3-2

'WING PLANFORM SELECTION


can be isfo're the design layout started, the wing geometry must be selected; 4including parameters such as aspect ratio, sweep, taper ratio, dihedral, and thickness. While all these parameters- will be numerically optimized -at somelater dit&, that optimization will proe d 'fro' '*'-baie21ie' aircraft

arraiege
s

inital gue-sa to these parameters. hus; designers hive evolved a number of "first-order" methods whic are provided The "reference", or "trapezoidal" wing is the basic wing ge)metry used to begin the layout'.Figures 4 and the, reference parameters' of 5 show the key

'n'd aseline 'ust include Kat

J .geometric . IIdll is the "back of a napk)n- drawing of aerospace legend, and gives a rough indication of what the design may look like. The sketch is used to make a ,first estimate of the required total weight and fuel weight to perform the design mission, by a process called "sizling" The "first-order" sizing provides the Information needed to develop an initial design layout (figure 3). This is a scaled three-view drawing complete with the more important internal arrangement details, including typically the landing gear, payload or passenger compartment, engines and Inlet ducts, fuel tanks, cockpit, major avionics, and any other internal

wing. ot hat the reference wing is ficticious, and extends through the fuselage to the aircraft,centerline. There are two key 'sweep 'angles, as shown in figure 5. The leading edge sweep Is the angle of concern in supersonic flight. To reduce drag it is common' to sweep the leading edge behind the mach cone. The sweep of the quartor chord line is the sweep most related to subsonic flight. Airfoil pitching moment data in subsonic flow is generally provided about the quarter-chord point. That is the point about which the airfoil pitching moment is essentially constant with changing angle of attack (io, the "aerodynamic center").

components which are

large

enough

to

In a similar fashion, such

a point

is

affect the overall shaping of the aircraft. Enough cross-sections are shown to verify that everything fits. T1,is initial layout Is analyzed to determine if It really will perform the mission as indicated by the first-order sizing. Actual aerodynamics, weights, and installed propulsion characteristics are analyzed and subsequently used to do a detailed sizing calculation. Furthermore, the performance capabilities of the design are calculated and compared to the requirements mentioned above. Optimization techniques are used to find the lightest or lowest-cost aircraft that will both perform the design mission and meet all performance requirements.

defined for the complete trapezoidal wing. This is based on the concept of the "mean aerodynamic chord". The mean aerodynamic chord, shown in figure 6, Is the chord "c" of an airfoil, located at some distance "y" from the centerline. Figure 6 illustrates a graphical method for finding the mean aerodynamic chord of a trapezoidal wing planform. The entire wing has its mean aerodynamic center at approximately the same percent location of the mean aerodynamic chord as that of the airfoil alone. In subsonic flow, this Is at the quarter chord point on the mean aerodynamic chord. In supersonic flow, the

f-

1-

$It 3

Cs-tng.,

1..o

3-3 The first.to investigate aspect ratio idetail ware-the Wright~brothers, using a wind tunnelthey_:constructed. They found "Z -'

-71 !, '

_that 'a ,,lon, 'kiiinyk wing -(high aspect ratio) 'has less drag'for a,given lift'than
-ls'die' tthe ing;,(lowaspect ratio). This a short, fat threiie;d lensii6nail "bffecti" a When a wingjis geneating"lift, it has reduced pressure on-tbiiupper. surface,

ACTUAL %ING

R!a. z I

and an

increased.pressurelon ',thelower,
The air would like' to "escape"
'

z': surface.

Ifronthe'bottom of~the'wing,m oving to~the, I oaround


oC = 1Mfl1C L TOAaT

A
t/e

:ICTA.TIO
AMOLI

S-.A,

two in is not p6ssible This top. dimensional. flor" However, for a real, three-dimensional wing, the air can escape the wing tip. When air escapesaround thewing. ,,tip, the pressure difference between the upper surface is the ,lower surface and decreased. This reduces lift. Also, the air flowing around'thetip' flows in a

, .. wPw 6*4 S

C--.23

.1e

c M-Y.

C-

circular tip. This

path when seen from the circular, or' "vortex""

front,

no. 4

and in ffect;,Ipushes downi'on the %wing near'thetip, which ,reduessthe effective angle 'of attack oftheairfoils'near the continues downstream behind the

S -------

Atpattern

flow

wing.
A wing with a high aspect ratio has the wing tips further apart than an equal area

wing
A Aamount

with

a low aspect ratio,

so

the

U.

of the wing affected by the tip vortex is less than for a low aspect ratio wing, and the strength of the tip vortex is reduced. Thus, the high aspect ratio and wing suffers less loss of lift incriase of drag due to tip effects than aspect ratio wing of equal area. Vne i-'w

's most early wings were rectangular in


shape, the aspect ratio was initially defined as siwply the span divided by the chord. For a tapered wing, the aspect ratio is defined as the span squared divided by the area (which defaults to the earlier definition for a wing with no ataper). A.t coo
-

- s.,, *

. e-w... . $ .,s

T..

p-

7V

~layout,

The maximum tubsonic lift to drag ratio of an aircraft Increases approximately by the square root of an increase in aspect ratio. On the other hand, the wing weight aspect also increases with increasing -. ratio, by about the same factor. Later in the design process, the aspect ratio will be determined by a trade study in which the aerodynamic advantages of a aspect ratio are balanced against Lhigher % AUOOP-AmIcthe increased weight. For initial wing

the values and equationca

provided

in table one can be used. Thee were determined through statistical analysis of a number of aircraft, using data from Jane's 11 The World's Aircraft. Wing sweep is used primarily to reduce the adverse effects of transonic and

C,

Th60 A tird I

a.
Fit. 6

is.,ci~m--sc

H ,,,lW6

60

U,.- .4),..k

tie3

Akbd~aCt

7.5

aerodynamic center moves back to about 40 percent of the mean aerodynamic chord. The mean aerodynamic chord and the resulting to aerodynamic center poirt is used properly locate the wing. The shape of the reference wing is taper determined by its aspect ratio, ratio, and sweep. These will be determined now, along with the desired dihedral.

PIA..S j"W, jdlt" inf "(19) M. In.clot1h.)1, WAVY5.uo.

to EW ,k&$1$RA-AW__
A C -0919 -0621

4737
416

10 5$0

-0622 0

3.4

."l#mZC~

~,e US

-~~

~W

1011MT~ 03 TM
?StAM'PIIO#. 425 5.E

5151AT
ATT",

i't 8 sW .. ,
Supersonic fl

............
h
,..

ATT ...
a~ ow FU. 9 T-84ft m

Theoretically

0 .0 20 f~oralo 0 oa a et~wing. it determined no yhe passing- overactua~volocity of the, air thewing, but by the air
velocity, in a direction Pe tersonc f theo e ancua This allows to an Increass,in critical mach number by the use of sweep n t supersonicbspeyds the lose of lift associated with supersonic flow can be reduced the sweeping the wing (crcyln( leading aftof by bcchycone angle no)ot Figure -8 how. . historical trend for wing leading edge sweep versus *,jge line
P

Ie

An elliptical wing planfo is difficultWing to build is to build. The and expensive easiest the untapered rectanglar wing. However, the untapted wing has constant cho length along the span, tip so hascompared to the toward and when excessive chord the idea elliptical wing. This "loads up" the tip, auing the wing to generate more of its If towards ithanis p Ideal. The end result is that rectangular wing has aboutan sevenuntwisted percent wore dra due to lift than an elliptical wng of the se aspect ratio. a When a hectangular wing is tapered, the tip choas become shorter, alleviating the undesired effects of the constant-cho rectangular wing. In fact, a taper ratio of 0.5 almost completely elisinatea those for an unswept wings and procens s ift distribution very close to the elliptical ideal (figure 10). This results in a drag due to lift which is lers than teipcrc e higher, alleiea uoiael ical e s h t o rectng rswept at tends to divert the air outboard, creating the tips. this loads towards more lift outboard up the aps. than foran equvalent unswept Wing. To return the lift distribution the desired elliptical l ft distribution, It in necessary to increase the amount of taper (is, reduce the'tapor ratio).

number. The historical trend differs from this theoretical result for two reasons In the high speed rangec it becomes euctur. y impractical to sweep the wing Pat the each cone. In the transonic speed regime (roughly ach .9 to 1.2), the desire for subsonic airflow velocity over the thn leading edge) Is more rsportant tor airfoil (when measured perpendicular th teeccts each con affect, which would Indcate zero sweep for ach onec The Wing swop aci aspect ratio toether have a strong ffect on the winge one p(tchup characteristics. p)tchupe dsthe highly undesirable tendency of aircraft, upon reach sean aperoenicuar near stall,eangle of attack. increasthe to suddenly and uncontrollably The aircraft continues Pitching up until it stalls and deais totally out of control. Figure 9 provides bounaries pe pitchup avoidance for combinations of wing quatter-chord aspect ratio. Pitchupsweep angle and avoidance should beal, considered for military fighters, arobatilc aircraft, qeneral aviation aircraft, and trainers, af Wing taper ratio is the ratio between the tip hord and the canterlne root chord. Most wings of con sweep have taper ratio of about 0.4 to 0.5, whileas art andct t iners, tono moat swept wings have a taper ratio if o)

tb

dt

ih h ut

of

about 0.2 to 0.3.


along the span of the wing. As proven by the Frandtl wing theory early in this century, ainimum drag due to l0.t, or bnduced" drag, occurs when the lift is distributed in an ellipticl fashion. For an untwisted and unswept wing, this occurs when the wing planform is shaped as an ellipse. This result was the basis of the graceful wing of the Supermarine Spitfire.
"

* U
A

.3

I
4

see?*f00
1, 1 U.1. 16W Wf~Iq,.

A 4

'.

LOCAT)09 ileo,,,,

nr

3-5 Figure 11 illustrates the results of NACA wind tunnel tests to determine the taper 'ratio required to approximate the ellipticallift~distributionfor a" swept, untwistedx.,wing.,However, -it- 'sh6uldbe noted 'that taper~rati6s-much A*-wlerthan 0.2 should be avoided~forlall "but' 'delta wings, as'a very lowitaperratio'tends"'- to6 . promote tp~stll. . Wing dihedral i"the"angle of the wing with.,respect-to thehorizontal'when seen from the-front -Dihedraf tends to'i6r1*the aircraft level~wheneverit is banked. 'Tfis explained is-frequently, and incorrectly, as the result of(atgreater'projected' area The reverse isitrue- ;for low aspect suchas'-'a'delta '-iing', ratio 'swept'wings,4 Here,, a. sharper '-leading edge, 'provides' maximumilift due to the formation greater 1, of vortices, which dolay stalling. Thickness 'also affects the structural weight of-the,,wing. Statistical -'equations " p the" ,wiii4 t -VtraI -eg r6 ately for, wing weight show thatWi inversely with' the square root' of the thickness ratio. For,initialselection of the thickness in ratio, the historical trend shown figure 12 can be used.

" ca which is',lowered. for ,the'wing 'Actually, the-rollinguoment Is caused eI 'lACA 5M

C1731

tLcj~f~PftAF02

"

'"'

( m -"-"II1I1i
S'TA55#It

51,1.l
____

.o

TU.

0kWU MMC.MAC14

d 11. 1 m-t,.,

"

Me"

sknl.h .w

by a sideslip introduced by the bank angle. The aircraft "slides" towards the lowered wing, which increases the angle of attack of the lowered wing. The resulting approximately is moment rolling proportional to the dihedral angle. Wing sweep also produces a rolling caused by the moment due to sideslip, change in relative sweep of the left and right wings. This creates an effective dihedral which is added to any actual geometric dihedral. Roughly speaking, ten degrees of sweep provides about one degree of effective dihedral, In addition, the position of the wing on the fuselage has an positive influence the the effective dihedral, with on greatest effect provided by a high wing. initial estimates of Table 2 provides dihedral. wing airfoil thickness ratio has a direct effect on drag, maximum lift, stall characteristics, and structural weighth with Increases drag subsonic The increased increasing thickness due to separation, and the critical Mach number The reduces with increased thickness. thickness ratIO affects the maximum lift primarily by its and stall characteristics effect on the nose shape. For a wing of fairly high aspect ratio and moderate sweep, a larger nose radius provides a higher stall angle and a greater maximum lift coefficient. T.db 2 OM.d. .ridh. Po' VAi4 U..d(dl) SAMk S" ts Swloo ,We. t________ t t7 32to OtwS 3w4 t -l2 -'Sto0

W WING LADING AND THRUST-TO-WEIGHT RATIO and The thrust-to-weight ratio (T/W) the wing loading (W/S) are the two most important parameters affecting aircraft performance, directly The thrist-to-Weight ratio affects the performance of the aircraft. An aircraft with a higher thrust-to-weight ratio will accelerate more quickly, climb maximum sre rapidly, reach a higher speed, and sustain higher turn rates. On the other hand, the larger engines will consume more fuel throughout the mission, which will drive up the aircraft's takeoff design the weight to perform gross mission. closely is ratio Thrust-to-weight Table three related to maximum speed. provides curve fit eoustions based upon for mach number or velocity maximum differnt classes of aircraft which Can be estimate for thrust-toa a first used weight ratio. designed For an aircraft which is primarily for efficiency during cruise, a initial estimate Of the required better thrust-to-weight ratio can ba obtained by -thrust matching'. This refers to the comparison of the selected engine's thru t available during cruise'to the 'ustizatad aircraft drag. In level -unaccelerating flight, the Likewise, the thrust mlst equal.the drag. (assuming that weight mulstequal'the lift is ali'gned with the flight the thrust

thrust-to-weight ratio Mpath). Thus, inverse of the lift-to-drag L. ..... must equal thethe
0to2 .3t -2 -UO,

An estimate of L/D 'obtained ratio. through one of seviral methods ia thus used to determine minimum T/W for cruise.

317
4 ,
o~ll+ 064S 0514 3 + 05% -0,141 ,+ l~t Jdfi~htf(dOlnlhtf)! + Jeidigflu(dt))
Jrfp,+0+

loading, tice tly such that the, paras tragis ea the induced drag. ,This~yields equation 3 for wenqa,oadongnselcton for-constantthrust range'optimization.
-q1.AICWSt Jr Rstge: W/IS MdaximumT An aircraft,-designed for, air-to-air dogfighting must be capable of high turn

range' by

There are many other criteria which can

set the thrust-to-weight ratio,

such

as

rates., When

air-to-air,missiles

are -in

climb rate, takeoff distance, and turning performance. These other criteria also involve the wing loading and are described later. The wing loading is the weight of aircraft divided 'by the area of reference (not exposed) wing. As with the the the

thrust-to-weight

ratio,

the

tern

"wing

use, the firstaircraftto turn towards the other aircraft enough to launch a missile will probably,win. In guns-only dogfight, the aircraft-with a the higher turn rate will be able to/maneuver behind the other. A turn rate superiority of two is considered second per degrees

significant.

loading" normally refers to the takeoff wing loading, but can also refer to combat and other conditions. Wing loading affects stall speed, climb rate, takeoff and landing distance, and loading wing The performance. turn determines the design lift coefficient, and impacts drag through its effect upon wetted area and wing span. Wing loading has a strong effect upon sized aircraft takeoff gross weight. If the wing loading is reduced, the wing is larger. This may improve performance, but the additional drag and empty weight due to the larger wing will increase takeoff gross weight to perform the mission. To maximize range during cruise, the be wing loading should, if possible, to provide % is a function of selectedconditions. L/D high L/D at the cruise dynamic pressure. The wing loading for best 140 increases directly with increasing dynamic pressure. loses propeller aircraft, which A thrust efficiency as speed goes up, gets the maximum range when flying at the speed aircraft best L/D, while a jet for maximizes range at a somewhat higher speed the L/D is slightly reduced. The where for speed best L/D is that speed at which the the parasite drag exactly equals induced drag. Therefore, to maximize range a propeller aircraft should fly such that equation one is satisfied. qx. $

There are two important turn rates. The "sustained" turn rate for some flight condition is the turn rate at which the thrust of the aircraft is just sufficient to maintain velocity in the turn. If the aircraft turns at a greater rate, the drag becomes greater than the available thrust so the aircraft begins to slow down. The "instantaneous" turn rate possible, rate the highest turn is ignoring the fact that the aircraft will slow down. "g-loading", The "load factor", or during a turn is the acceleration due to the lift expressed as a multiple of standard acceleration due to gravity (32.2 ft/sec squared). Load factor ("n") is the by the lift divided to equal aircraft's load required turn wing attain loading to weight. a The required factor can be solved as follows: W .I4 _ also is sustained turn rate The important for success in combat. Sustained turn rate is usually expressed in ters of the maximum load factor at some flight condition that the aircraft can sustain without slowing. For example, the capability for sustaining five "g's" at 0.9 Mach number at thirty thousand feet may be specified. The wing load-ng to exactly attain a required sustained load factor "n" using be of the available thrust can all deterined by equating the thrust and drag, and using the fact that the lift coefficient during maneuver equals the wing loading times "n", divided by the dynamic pressure. This yields equation S. W (T/ a (r1

During cruise, the lift equals t19 weight, so the lift coefficient equals the dynamic loading divided by the wing pressure. Substitution into equation one allows solution for the required wing loading to maximize L/D for a given flight condition. This result (equation 2) is the wing loading for maximum range for a propeller aircraft. MMImsm PropasSC WIS -q s"4 As cne aircraft cruises, its weight reduces due to the fuel burned, so the wing loading reduces during cruise. To optimize the cruise when the wing loading is steadily reducing requires reducing the dynamic pressure by the same percent. This

The still speed of an aircraft may also define the required wing loading, and is directly determined by the wing loading and the maximum lift coefficient. Stall speed is a major contributor to flying safety, with a substantial number of fatal accidents each year due to "failure to maintain flying speed". desigi, military and Civil

is can be done by reducing velocity, whzich


undesirable, or by climbing to reduce the optimizing range This density. air technique is known as a "cruise-climb". A jet aircraft flying a cruise-climb at a constant thrust setting will maximize

Specifications establish maximum allowable


of speeds for various classes stall aircraft. In some cases, the stall speed is axplicitely stated.

3-7

TAKEO0ff

El

W L q,SChDV%.,St.,_

~
BALK5ICE5

Equation
weight

6l states

that lift

equals

LLOT

in level, flight, and that~at stall speed, the aircraft is at maximum lift coefficient. Equation 7 solves 'for the stall speed with,a.c2rtain maximum lift coefficient. The air density, %gr, is standard typically the sea level orsoele value (.00238 slugs/cubc~foot)' th 5000 foot altitude, hot day value (.00189) to ensure that the airplane can be flown into Denver during summer. The remaining unknown is the maximum very be coefficient. This can lift difficult to estimate. Values range from 1.2 to 1.5 for a plain-wing with no a-,t flaps to as-much as 5.0 for a wing with large flaps immersed in the propwash or jetwash. Maximum lift coefficient depends upon the wing geometry, airfoil shape, flap geometry and span, leading edge slot or slat geometry, Reynolds number, surface texture, and interference from other parts of the aircraft such as the fuselage, force or pylons. The trim nacelles, will provided by the horizontal tail lift, or reduce the maximum increase depending upon the direction of the trim force. If the propwssh or jetwash impinges upon the wing or the flaps, it will also have a major influence upon maximum lift during power-on conditions. For an initial estimate of maximum to rert to lift, it is usually necessary data. Figure test results anw historical 13 provides maximum lift trends versus of classes angle for several sweep aircraft. Note that the maximum lift using typically be the takeoff flap setting will about 80 percent of these landing maximum

required

wing loading to attain

a given
4

A o

J_

TAIFehVPlAR WTU -w U. 14 rktofrdl~wmlat.

Landing distance can also sometimes determine the required wing loading. Sling which loading affects the approach speed, determines the touchdown speed, which in which must turn defines the kinetic energy be dissipated to bring the aircraft to a halt. The kinetic energy, and hence the stopping distance, varies as the square of the touchdown speed. In fact, a reasonable first-guess of feet, total landing distance in the is clearance, obstacle including of the approximately 0.3 times the square knots. in speed approach better a provides 8 Equation distance approximation of the landing be which can wingused to estimate the maximum term loading. The first landing represents the ground roll to absorb the kinetic energy at touchdown speed. The constant term, Sa, represents the obstacle 6learance distance. _ S,
-

values.

Frequently the takeoff distance will loading. the required wing determine the 14 permits estimation of Figure takeoff ground roll, takeoff distance to clear a 50 foot obstacle, and PAR balanced foot field length over a thirty-five obstacle.

). ] $fop) 100(ah t(ype *fTie( I0ff ioa3oh) &, 600 SAItlope) 7.&S S (STOL.

OTHER AERODYNAMIC CONSIDERATIONS


! 6.1,Lq.

1IeS1p,50)t s tf 4W 41D L 4at*

The overall arrangement and smoothness the fuselage can have a &ajor effect efficiency. A upon aerodynamic can have excessive poorly flow designed aircraft and transonic drag rise, separation,

of

supersonic

wave drag. Also, a poor wing-

9.o)Jm.*u 9 ------------N

Ai "A'

'a"
e e w

fuselage arrangement can result in lift desired or disruption of the losses distribution. elliptical lift good for requirement major The aerodynamic design during fuselage layout the maintenance of smooth longitudinal is contours. These can be provided by the use
of smooth longitudit.al control lines.

Generally, longitudinal breaks in contour should follow a radlu at leact equal to the fuselage diameter at that point.

9I

IS

d,'3,.

Mf 113ll lf

3-8

'. -,
Zr'~~i
-1, A

!?

'

" I

~~Fie, Hb.wk'Ohme de6lbelO0. It S


c..li, g9."l Fig.IS L..gRidl.W To prevent separation of the airflow, the aft-fuselage deviation from the freestream direction should not exceed 10 to 12 degrees (figure 15). However, the air inflow induced by a pusher-propeller will prevent separation despite contour angles of up to 30 degrees or more. A lower-surface upsweep of about 25 degrees can be tolerated for a rearloading transport aircraft provided that aircraft's the longitudinal change in total cross-sectional area. In fact, wave second drag is proportional to the derivative (is, curvature) of the volume distribution plot. volume distribution from Thus, a "good"A a wave drag viewpoint isons in which the is required total internal volume distributed longitudinally in a fashion which minimizes curvature in the volume distribution plot. Several mathematical uolutions to this problem have been found for simple bodies-of-revolution, with the

the fuselage lower corners are fairly sharp. This causes a vortex flow pattern
which reduces the drag penalty. In general, aft-fuselage upsweep should be minimized as much as possible, especially for high-speed aircraft, For improved aerodynamic efficiency, the wing-fuselage connection of most aircraft is smoothly blended using a "wing

"Sears-Haack" body (figure 18) having the lowest wave drag.


If an aircraft could be designed with a volume plot shaped like the Sears-Haack volume distribution it would have the minimum wave drag at mach one for a given volume. )ength and total internal However, it is usually impossible to exactly or even approximately match the

fillet"

(figure 16).A wing

fillet

is

Sears-Haack

shape for a real

aircraft.

generally defined by a circular arc of varying radius, tangent to both the wing and fuselag2. Typically a wing fillet has

Fortunately, major drag reductions can be obtained 'imply by smoothing the volume distribution shape.

a radius of about 10 percent of the


chord length.

root

As

shown

In figure

19,

the

main

contributors to the cross-sectional area are the wing and the fuselage. A typical luselaqe with a trapezoidal wing will have ,*n irregularly-shaped volume distribution area with the maximum cross-sectioial located near the center of the wing. By

"squeezing"

the fuselage at

that point,

Sthe

the volume d1stribution shape can be smoothed and the maximum cross-section area reduced. This design technique is referred to am rduce the wave "coke-bottling" and can "area-ruling" or drag by as such as fifty percent. Note that the volume removed at be center of the fuselage must provided elsewhere, either by lengthening thetfuselage or by i,creasing its crosssection area in other places.

will Low wing, high-speed aircraft root frequently have a modified wing airfoil to further minimize fuselage drag interference and shock-inducd Increases. This modification takes the form of an uncambered or even negatively-

o-'..

-,,

cambered airfoil set at a angle of Incidence.-

high positive-

nf

For supersonic aircraft, the greatest aerodynamic impact upon the configuration layout results from the desire to minimize supersonic wave drag. Wave drag is a pressure drag due to the formation of shocks, and Is analytically related to the

3-9

0C
'. ep9A. 5EF55NcZAISJ

Tx, (X

.qS.,

+ C..+

C... ,

(X-,

X X

X~sT

S.',"I. ArTINLE.,-1,qS,
TT z+

xr
F1 20 Lteoeriii min.. STABILITX AND CONTROL CONSIDERATIONS :Imean My The basic concept of stability is when that- a stable aircraft, simply disturbed, tends to return by itself to velocity, etc.). "Static stability" is present if the forces created by the disturbed state (such an a pitching moment due to an increased'angle of attack) push in the correct direction to return the iriraft to its original state. Most aircraft are symmetrcel about the ctnterline, so that moderate chonges in angle of attack will have little or no influence upon the yaw or roll Iof the aircraft. This permits the stability and analysis to be divided anto control longitudinal (pitch only) and lateraldirectional (roll and yaw) analysis. Figure 20 shows the major contributors to aircraft pitching moment the the center of gravity, iraluding aboutwing, tail, fuselage, and englne contributions. contribution The wing pitching moment includes the lift through the wing mean

F 1(X) qS. '

To simplify the equations, all lengths can be expressed'as a fraction of the wing "chord'(c)._These'fractional lengths are denoted 'by a bar. This leads to equation II. For a static "trim" condition, the

its original

state

(pitch, yaw,

roll,

pitching-moment must equal zero. The

main

the flight conditions of concern are takeoff and landing with flaps and landing gear down and the maximum speed. Usually the most-forward CG position is critical for trim while the aft-CG position is most critical for stability as discussed below. For static stability, any change in angle of attack must generate moments which oppose the change. In other words, the derivative of pitching moment with respect to angle of attack (eq 12) must be negative. Note that the wing pitching moment and thrust terms have dropped out a they are essentially constant with rs pect to angle of attack. . -W.) , -, -.(-X)=-1
-_

aerodynamic

chord ("MAC"), and

the

wing

moment about the MAC. Another wing moment term is the change in pitching moment due to flap deflection, The long moment arm of the tail times its lift produces a very large moment which is used to trim and control the aircraft. While this figure shows tail lift upwards, under many conditions the tail lift will be downwards to counteract the wing pitching moment, produce fuselage and nacelles The pitching moments which are difficult to estimate without wind tunnel data. These moments are influenced by the upwash and downwash produced by the wing. The engine produces three contributions to pitching moment. The obvious' term is the thrust tiaes its vertical distance from the center of gravity. Less obvious is the vertical force ("Fp") produced at the propeller disk or inlet front face due to the turning of the freestream airflow, jet..induced the propwash or Finally, flowfield will influence the effective angle of attack of the tail and possibly the wing. Equation 9 expresses the sun of these of moments about the CC. The effect elevator deflection Is included in the tail lift term. Equation 10 expresses the moments in coefficient form by dividing all ters by (q Sw c) and expressing the tail lift in coefficient form.

Due to downwash effects Che tail angle of attack does not vary directly dith aircraft angle of attack, so a derivative term is included which accounts for the effects of wing and propeller downwash, as described later. A similar derivative is provided for the propeller or Inlet normal force term (Fp). The magnitude of the pitching moment derivative changes with CG location. For any aircraft there is a CC location which provides n-j change in pitching moment as angle of attack is varied. This "airplane aerodynamic center", or "neutral point (Xnp)" represents neutral stability and is the most-aft CG location before the aircraft becomes unstable. Equation 13 solves for the neutral Equation 14 then expresses the point. pitching moment derivative in terms of the distance in percent MAC from the neutral point to the center of gravity. This percent distance is called the "static margin", and is the term in parenthesis in equation 14. S C j..
-

. ,
C -.

C,

-. . t.

C. +

C..,=- C.

-T.)

3-10 C.. pno sumb+ . ~tn4 -i .5. . TAIb 4 TM 0e.w .Tooci< ',/ Horizontal r c 0.50 rL-W AID I'WL+TA~rJ'Icultr -I FRALAAII Tw5Lli~rn ti~ou0 ofkk.l *aiu,$ VnI",Ic., 002

-I+Hocbdtf
"t / , ,-

0 50
0 70 020 " 90f 0.70 0 40 100

004,
004- 00 0 00174
I

~mn
M TAKE . 0 1 3

bolt "0.70

2
SS I'

006
0 06 0 07 009

a RfIen TM i IMht

Je tri44<o

",.d.Won. 11g. 21 TypM p0blo. e-iLC I

Lht

Z tfheCG is ahead of theneutral point (positive static margin), the pitching mosent derivative Is negative so the aircraft is stable. At the most-aft CG position,,atypical transport aircraft has a positive static margin of 5 to 10 percent. While current fighters typically have positive static margins of about 5 percent, new fighters such as the F-l6 are being designed with "relaxed static stability (RSS)" In which a negative static margin (zero to -15 percent) is coupled with a computerized flight control system which deflects the elevator to provide artificial stability. This reduces trim drag substantially. Figure 21 illustrates pitching moment derivative values for several classes of aircraft. These may be used as targets for conceptual design. Dynamic analysis during later stages of design say revise these targets.
In a fashion siilar to the above, and is discussed in my textbook, For the initial layout, a historical approach is used for the estimation of tail size. The effectiveness of a tail in generating a moment about the center of gravity is proportional to the force (is, lift) produced by the tail, and to the tail soment arm. The force due to tail lift is proportional to the tail area. Thus, the tail effectimes il omeional to the tail aeftivees isepo tail ar Then product has units of volume, which leads to the "tall volume coefficient" method for initial estimation of tail size. The "vertical tail volume coefficient" is defined by equation 15. The "hnrizontal tail volume coefficient" Is shown by equation 16.
Sn,., 1

1. - sire N5 S b.- WID a. o WIN. MEAN o.M

ii,. 2 Ir n

An .

Lateral-directional

analysis

proceeds

One of the most important aspects of handling qualities is the behavior of the aircraft at high angles of attack. As the angle of attack increases, a "good"

bS,/t,

C.n,

The definition of tail moent arm is shown in figure 22, along with the definitions of tail area. Observe that the horizontal tail area is commonly measured to the aircraft centerline, while a canard's area is commonly considered to include only the exposed area. If twin vertical tails are used, the vertical tail area is the sum of the two. Table 4 provides typical values for volume coefficients for different classes of aircraft. Those values are conservative averages, and are used in equation 81 or 82 to calculate tail area.

t warn the pilot, retains control about all axis, and stalls straight ahead with immediate recovery and no tendency to enter a spin. If a spin is forced, the "good" airplane can be immediately recovered. A "bad" airplane lose- contrl in one or *ore axis as angle of attack increases. A typical bad characteristic is the loss of aileron roll control and an increase in aileron adverse yaw. When the aircraft is near the stall angle of attack, any enor nar sll angle inboard wing enough th to stall it. With only one wing gonereting lift, the "bad" airplane will suddenly departs into a spin or other uncontrolled flight mode from which recovery is impossible. There have been many criteria proposed for good departure characteristics, based upon various aerodynamic derivatives. One useful one is the "Lateral Control Departure Parameter (LCDP)", sometimes called the "lateral control spin parameter" or the "aileron alono divergence parameter" (equation 17). The LCDP focuses upon the relationship between adverse yaw end directional stability.
LCDP a,-C C'

airplane

Perecsmlsuftig

CM,. -9C ',

3-11 Equation 18 shows another departure _which 'C-n-beta-dynamic', parameter, includes the effects of the mass moments parameters of inertia. Both of these should be positive foe -goods'/departure' '! resistance. Figure 23 showsfa cros'spl6t of theALCDP and 'C-n-beta-dynamic' as angle'of attaick, is increased, showing the boundaries for as, departure - resistance acceptable determined from high-g simulator tests earli6r using experienced pilots. The Weissman criteria is also shown, UCTUAL CONSIDERATIONS I The primary concern in the development of a good structural arrangement is the provision 'of "efficient "load paths". A loadtpath'is the structural .elements by which opposing forces are connected. The primaryforces to be resolved are the lift of'the wing and the opposing weight of the major. parts of the aircraft, such as the engines and payload. The size and weight of the structural members is minimized by locating these opposing forces near to each other. Carried to the extreme, this leads to the Flying Wing concept. In a flying wing the lift and weight forces can be located at virtually the same place. In the ideal caser th~weight of the aircraft would be distributed, along the span of the wing exactly as the-lift isdistributed (figure 25). This is referred toas "spanloading". rarely ideal spanloading is While

SF4

possible,

the spanloading concept can be

applied to more-conventional aircraft by spreading some of the heavy weight items

Pool Ri MI 14"AILS to . P00* B%off i $

lo

once an aircraft has departed into a Lspin, recovet, becomes a high priorityl The -ertical ta, plays a key role in spin recovery. Figure 24 illustrates the effect of tail arrangement upon rudder control at high angles of attack At high angles of attack the horizontal tail Is stalled$ extending a turbulent wake producing forty-five ,it approximately a upward degree angle which can blanket the rudder. It is considered desirable that at least one-third of the rudder be un-blanketed. An empirical method for estimating if an aircraft will in fact recover from a spin is provided in my textbook.

W.

T " 33-MGTmlo.

I/

* :S 2 4p**.s*

toi

/"

//

'-7 ,

/'--;"

such as engines out along the wing. This will yield noticable weight savings, but must be balanced against the possible drag increase. If the opposing lift and weight forces cannot be located at the 'eme place, then some structural path will be required to by structural membrs can be carry the load. The weight reduced those of provising the shortest, straightest load path possible. structural Figure 26 illustrates a arrangement for a small fighter. The major fusela, loads are carried to the wing by /lonqevons", which are tyI cally "-" or

/H -shap d extrusions run


d.d heavy, minimized by

g form and aft


should aircraft be so

their weight designing the

7_
r-" F%24 Tilse I- qu
*i'* f

that they are as straight as possible. For aircraft such as transports which cutouts and concentrated loads have feer than a fighter, the fuselage wili be of number with a large constructed are or "stringers", which longerons, approximately evenly distributed around the circumference of the fuselage. Weight is minimized when tLe stringers are all straight and uninterrupted.

3-1

vm

1W

CTimai M00

4,6W,

00Cm

WGIn

Hamm05

0f%4F4IC26511U00t11

C 1.

0111% TUN,

II

&II T CO

F%

37

Ain

C"1141tMWWIM

The lift force on the wing produces a tremendous bending moment whore the wing attaches to the fuselage. The means by which this bonding moment is carried across the fuselage is a key paramoter in the structural arrangement, and will greatly influence both the structural weight and the aerodynamic drag of the aircraft. Figure 27 illustrates the four carrythrough of wing types at mor structure. PROPULSION CONSIDERATIONS propulsion. All engines operate by compressing aircraft outside air, mixing it with fuel, burning the mixture, and extracting energy fr)m the resulting high pressure hot gases. In a piston-prop, these steps are done Intermittently in the cylinders via the reciprocating pistons. In a turbine engine, these steps are done continuously, but in three distinct parts of the engine. The selection of the typo of propulsion system, i, piston-prop, turboprop, turbofan, turbojet, ramjet, or other is usually obvious from the design requirements. Aircraft maximum speed usually limits the choices as shown in figure 29.

MO
74.25

-nstess

. nw u

'"-'n

.1rLt1

S05

11

figure 28 illustrates the major options for aircraft

W
5K.l.

CIP*'*ePn01150W*A

WIS

CIMUW4

'A.14
115u0o&

=r

,fos
I

.
I 3 5)5.5.(1501 I S

tlt 29 P,...WtIo.

V,

3-13 Turbojet and turbofan engines are

L.

the air'enterng'them is slow"d to a'speed of about.mach 0.4, to 0.5. Thii-is to- keep the tip, speed of the compressor blades below sonic speed-relative to'the incoming air. Slowing down the incoming air is the

incapable

ofU efficient

operation

unless

'"

primiry purpose of an inlet system. The instilled performaes' of a jet


engine 'greatly depends upon the air

.,
'

systi.

inlet

,Roughly speaking, a one percent


in inlet pressure recovery

reduction

(total ,pressure~delivered to

the

engine

.....

divided by freestream total pressure) will reduce thrust by about 1.3 percent. There are four'basic types 'of inlets, shown in figure 30. The NACA flush inlet was used by several early jet aircraft but is rarely seen today for aircraft propulsion systems due to its poor large losses). pressure recovery (is, The pitot inlet is simply a forward facing hole. It works very well subsonically and fairly well at low supersonic speeds. This inlet is also called a "normal shock inlet" when used for supersonic flight ("normal- meaning perpendicular in this case). The pitot inlet is seen on most subsonic jet aircraf.. The remaining inlet types are for supersonic aircraft, and offer improvements over the performance of the normal shock inlet at higher supersonic speeds. The conical inlet (also called a spike, rouad, or axisymmetric inlet) is based upon the shock patterns created by

05 1 30 IW .

that the second ramp has a variable angle, and can collapse to open a larger duct opening for subsonic flight. Figure 33 summarizes the selection criteria for different Inlets, based upon design mach number. Note that these are approximate criteria, and be overruled by special considerations. 1IS*rAI(IO.NT
"o"""""'

supersonic

the two-dimensional ramp inlet (alto called a "D-inlet") is based upon the flow over a wedge. External compression inlet types are shown in figure 31. The greater the number of oblique shocks employed, the better the premrurs recovery. Figure 32 illustrates a typical threeshock external compression inlet. Note

flow over a

cone,

Similarly,(
It
"o '",'e

J
IIItm

IMM

"

s.w.eAtsls

______

1e5

The inlet location can have almost as great of an effect on engine performance as the inlet geometry. If the inlet is located where it ths fuselage or a can ingest a vortex off separated wake from a wing, the resulting inlet flow distortion can stall the engine. The F-Ill had tremendous problems with Its inlets, which were tucked up under the intersecticn of the wing and fuselage. The A-10 reraired a fixed slot on the inboard wing leading edge to cute a wake ingestion problem. Figures 34 and 35 illustrate the various To design the inlet for a particular application, capture area must be known. Figure 36 provides a quick method of estimating the required inlet capture area. This method is statistical and is based upon the design mach number and the engine mass flow in pounds per second. A sore detailed discussion of Inlet location options and capture area estimation is available in my textboo%, The aircraft's forebody builds up a thick boundary layer. If this low-energy, air is allowed to enter the engine, it can reduce eagine performance oubsonically and prevent proper Inlet operation supersonically. Unless the airnraft's inlets are very near the nose (2-4 diazoters), some form of boundary layer removal should be used just in front

options for inlet location.

31

-A

-,k.

.10 a,"t

1er2Ttuta~tet.,o,

(."M W

(,..

..

, *

.. I...U5 /:...,

iturbulent ""24 0el.

IllIsi Z %rwW

"

of the *nlet.

K3-1

03M

es-cu

IDIIm~

WV5ArV 1%l 17 o

0ALA1YfM l a-U Ism .,.

--.
t Fle 34 I

A3... 3
ta u--W

.A

kft

The channel boundary layerdiverter is the most common diverter for supersonic aircraft. It provides thebest perfomance and the least weight in most cases,

and ducted to an aft-facing hole.

OBSERVABLES CONSIDERATIONS
(Nots: The following material on observables has been approved for release by the U. S. Air Force.) 01IMI AFT Man Ever since of military aviation attempts the dawn made to have been reduce the detectability of aircraft. During wwi, the only "sensor- in use was the human eyeball. Camouflage paint in mottled patterns was used on both sides to reduce the chance of detection. Rodar is the primary sensor used against aircraft today. "Radar" is an Radar consists of Detection And Ranging. 4 for RAdio a transmitter antenna which broadcasts a directed beam of electromagnetic radio waves, receiver antenna which picks up theand a radio waves which faint bounce off objects datect~on, khe airc.-aft must return such a t o e by .he radiosr"illuminated" t beam. To avoid low amount of the transmitted radio beam that the receiver antenna cannot distinguish between it and the background radio static. The extent to which an object returns electromagnetic energy is the object's "Radar Cross Section (RCS)". RCS Is measured in square decibel square meters, with meters or in "zero dBsm" equal to ten to the zero power, or one square 5w'ter."Twenty dBsm" equals ten the second power, or 100 square meters. to There are many phenomena which contribute electromagnetic to the RCS of an aircraft. These require different design approaches for RCS reduction, and can produce conflicting design requirements. Figure 38 illustrates the mI or RCS contributors for a typical, untreated fighter aircraft.
Al 050*tV *A'5

TI.L0O1.005 % 35 : , coo, w. * ,, , ,*,. " & -0. -

.ow

qacrunAy

.M. t,

'Act ft .*D*-

Ausually

KM ote pset

The four major varieties Of boundary layer diverter are shown In figure 17. The step diverter Is suitable for subsonic aircraft, and relies only upon the boundtry layer itself for operation. The boundary layer bypass duct is simply a separate ilet duct which admits the boundary layer air and ducts it to an aftfacing hole. The suction form of boundary layer diverter is similar. The boundary layer air is removed by suction through hole, or olots Just forward of the inlet

(,l O

(01

1M%0

3 O 38

M$ RCS -h

.t

3-15 tin addition to reshaping the aircraft, detectability can be reduced through' the use of skin materials which absorb radar Such-'materials,,-called' yradar energy. absorbingmaterials"( A), are _typically -

go

composites, infrared

-'-uch' -as 'fiberglass detectability

or
or

sw,ea

graphite/epoxy embedded..with ferrite particles. concern to the aircraft

carbon

designer.-'Many

is also

of

Le. PCs Fla.39 Fat RS

One of the largest contributions to airframe RCS occurs any time a relatively is aircraft the of surface flat perpendicular to the incoming radar beam.

rely uponIR seekers. Modein IR are sensitive enough to detect not only the radiation emitted by the;engine exhaust and engineahot parts, ' but :also that emitted by the whole, aircraft skin due to aerodynamic heating at transonic and supersonic speeds. Also, sensors can which the solar IR radiation detect cockpit and the skin off reflects transparencies (windows). missiles reduction of IR detectability. The most potent is the reduction of engine exhaust temperatures through use of a high-bypass engine. This reduces both exhaubt and hotpart temperatures. However, depending upon in result application this may the selection of an engine which is less than which sizing, aircraft for optimal increases aircraft weight and cost. Emissions from the exposed engine hotparts (primarily the inside of the nozzle) can be reduced by cooling them with air bled off the engine compressor. This will also increase fuel consumption slightly. Another approach is to hide the nozzles from the expected location of the threat IR sensor. For example, the H-tails of the A-10 hide the nozzles from some angles. Plume emissions are reduced by quickly the As mentioned, a high-bypass outside mixing the exhaust with theengine is air. best way of accomplishing this. Mixing can also be enhanced by the use of a wide, thin nozzle rather than a circular one. Another technique is to angle the exhaust the upwards or downwards relative to

Imagine

shining a flashlight at

a shiny

There

are several

approaches

tor

aircraft in a dark hanger. Any spots where the bea is reflected directly back at you will have an enormous RCS contribution, Typically this "specular return" occurs aircraft the the flat sides of on fuselage, and-along an upright vertical the abeam (when the radar is tail aircraft). To prevent these ACS "spikes" may slopi the fuselage sides, the designer angle the vertical tails, and so on, so that there are no flat surfaces presented towards the radar (figure 39). Another contributor to airframe RCS occurs due to the electromagnetic cLrrcnts when skin the up on build which illuminated by a radar. These currents the skin until they hit a flow across discontinuity such as at a sharp troillng edge, a wingtip, a control surface, or a Jsoatr", hecro~ aIicnitiy crack around a removable panel vr Join. At or radiate electromagnetic brcrgy, some oi which is transmitted back to t o radar (figure 40).

freestream.
KAr :1

This

will

have an

obvious

thrust penalty, however. CONFIGURATION LAYOUT METHODS conceptual The process of aircraft statistical numerous includes design predictions, and estimations, analytical the optimizations. However, numerical product of aircraft design is a drawing. While the analytical tasks are vitally important, one must remember that their only purpose is to influence the drawing, for it is the drawing alone which is ultimately used to fabricate the aircraft. The design layout process generally of conceptual with a number begins sketches, Figure 41 illustrates en actual, sketch from a Rockwell fighter unretouched conceptual design study. As can be seen, these sketches are crude and quickly done, but depict the Major ideas which the designer intends to incorporate into the actual design layout. A good sketch will show the overall the concept and indicate aerodynamic internal major the of locations

9'h.

' '.

This effect is much lower in intensity than the specular return, but is still sufficient for detection. The effect is is discontinuity when the strongest straight and perpendicular to the radar such as at beam. Thus, the discontinuities the wing and tail trailing edges can be swept to minimize the detectability froe the front,

3-16

K71*

.r '! _ ! -x,:_ 1
-

of the conic is wide variety of curves which can the be represented, and the ease with which the conic can bo constructed on the drafting table. A conic curve Is constructed from the desired start and end points (-Aand "B"), and the desired tangent angles at those points. These tangent angles intersect at point "C". The shape of the conic between the points A and B is defined by some shoulder point "s-. Figure 44 illustrates the rapid graphical layout r' a conic curve. To create a smoothly-lofted fuselage using conics it is necessary only to ensure that the points A, B, C, and S each of the various cross-seutions can in be connected longitudinally by a smooth line. Figure 45 shows the upper half of a simple fuselage, in which the A, B, C, and S points in three cross sections ar, connected by smooth longitudinal lines. These are called "longitudinal control lines" because they control the shapes of

The

landing gear, crew station, payload the or passenger compartment, propulsion systms fuel tanks, and any unique internal components such as a large radar. The actual design layout Is developed using the techniques to be discussed below. Such a design layout is shown a6, figure 42, courtesy of International's North American Rockwell Aircraft Operations. This drawing is typical of the initial design layouts developed major airframe companies during by the design studies. "Lofting" is the process of defining the external geometry of the aircraft. For an initial layout the overall lofting of the fuselage, wing, tails, and nacelles sust be defined sufficiently to shown that these will properly enclose the required internal components while providing a smooth aerodynamic contour. The traditional form of lofting is based upon a mathematipal curve known as the "conic". A conic is a second-degree curve whose family Includes the circle, ellipse, parabola, and hyperbola. The conic is best visualized as a slanted cut through a right circular cone (figure 43).

components.

These

Should

include

C "'"

Fl43

kt

&

great advantagA

'

'

'

'

.... .

l ,.

'.k h).

3-17

SsIl,

111155111
X.

I IIIIl ~ lIMM II-III


5I

5,.511 . 5

11? W.aV

II

.3

154l

545

t110II .111. 511 515 C5il llIS

I M 5 $11114511

L 151 II

55*I l.

the11l~l

coni

Figur.e

nom

1511setins 4651o lin1es. onr. naM longitudi iI control lo ni. Ii the54115$ 46 Ilmig in i d. .reate fll I i f desrb1111 d II c e s l s I 1111 1.$ $1 I.5111 111 Ml~ll1115J~

Is m gl 511

51 5.1
Ide4Illl S 51l 5.1111

5
II

$11
MI

11
511

the conic cross-sections. Figure 46

shows

the

side

and top

views

of

these

TSS5

longitudinal control lines. In figure 46, the longitudinal control lines are used to create a new crosssection, in between the second and thirdis

for Five control stations are required this example, comonar 47ieih ofs applicationaai staions Five ontro re rquireeomatics this~o stndesgn. , I .dep

also savings. Note that the biggest savings beow figusrese inur fuiea D sseseolaer uh fori woftnn and sse. oi frseg mndt capabt a esin o ai t te ue oreaio, sstet smooth c T displayor and plottin, irviso cockaellCoeerioft lopmngutodio ains nigurt ed haiat isd descibn aig.Nted aecomple nwmautera oTis decreibed.I a coewhn Thsdeilu landing ytote rsaigaf the fo th inlsrt coceptalatis ovie Noteo th Conceptualt maingirra cosncptan inmaircraft ins partiiat prtianti gear ln course. lou reshaigs Dein horit cioures eo.Shw

CONFIGUaATION DEuELOte
IO

Te SYtM

capabiiies

ndfSEM DEELPMN SYg

(CDS)

ln

61e

-o

d1h1

4d

e 11II 4

~ll
lit h+

.51,

47 15,IIs5s

3-18

WING & FUSELAGE CREATION,

AUTOMATIlC SMOOTHING

7 - ..i.

CROSS SECTION RESHAPING

COCKPIT REQUIREMENTS

xr|O(I1

7ISON1

(Il Aw. J~l|

fl

4'

3-19

Fgu. 53

LANDING GEAR KINEMATICS

I-TUNION AXIS

3-20

T. a
DeS.a f:ras

TY*tMkWdeAri.e
Reqaeetnet tradesGot
eo.t~o

TIW,A WIS
A.A

Da dweSto
LeteeeC,,ae

1&. XSro,

fitt Hgh de S
M1,60% 1 Coeflgratto

BPR.OPR.TIT.etr.

.sied , P Crmr,-e T SfttwayIet=h ,


DesIsmo-rent

c*,O,
P 'i

C Iroaled lhrl ad SFC

Wrm1e type etnivws rod of


-. lMllhly (at#-I~

palteltlee 1t11" teett

TRADE STUDIES Trade studies produce the answers to design questions beginning with "what if..." Proper selection and execution of the trade studies is as important in aircraft design as a good configuration layout or a correct sizing analysis. It is only through the trade studies that the true optimum aircraft is determined. Table 6 shows a number of the trade studies commonly conducted in aircraft design. These are loosely/organized into design trades, requirements trades, and growth sensitivities. Design trades are those which are conducted to reduce the weight and cost of the aircraft to meet a given set of mission and performance requirements. Requirements trades are conducted to determine the sensitivity of the aircraft to changes in the design requirements. If it is found that one requirement is resulting Itt a large Increase in weight, the customer say relax it. Growth sensitivity trade studies determine how much the aircraft weight will be impacted if various parameters should increase between conceptual design and production. These are typically presented in a single graph with percent change in the paramaters on the horizontal axis, and percent change in takeoff weight on the vertical axis. It has been assumed in the above discussion that the measure of merit for trade studies is always takeoff gross weight, even though cost is the final selection measure in a design competition. Using minimum weight as the measure of merit is usually a good approximation to minimum cost because the acquisition cost is so strongly driven by the weiqht. However, life cycle cost is driven largely by fuel costs, which say not be minimized by the minimum weight airplane. In such cases, trade studies with life cycle cost as the measure of merit can be conducted.

4-1

SURVEY OF EXPERIMENTALTECHNIQUES FOR PERFORMANCE PREDICTION

by
A.B.Haines Consultant Aerodynamicist

Aircraft Research Association Ltd Bedford United Kingdom

This lecture reviews present the state-of-the-art experimental In testing In large wind tunnels as a meansof predicting aircraft performance. Desirable and attainable standards of accuracy are defined and the lecture lists and discusses depth the factorsthat contribute this accuracy. Many In to referencesare quoted to enable the reeder to obtain more detailed Information. The lecture discusses (I) (11) (111) the balances and pressure scanners used for measuring the forces and pressures. the significant issues the qualityof the tunnel flow that can In affect the accuracy the testdota. of the methodsused for correcting the test data for the effects of tunnelwaill Interference subsonicand transonic at speeds up to near 9-1.0. the methods used establishing corrections the for nodel support Interference In both lowand high speed tunnels. a methodology for simulating as far as possible In the model tests, the viscousflowbehaviour over the full-scale aircraft and then. for extrapolsting testdata so fuii-scale the Reynolds numgets. the types of models and test rigs used In determiningthe propulsion Interference effects on both transport (turbofan and turboprop) combat aircraft and Particular attention Is paid to the use of powered simulators and to the difficult!s In obtaining reliable afterbody dragdata.

(iv) (v)

(vI)

0 British CrownCopyright 1991/OD Published with the peralsslon the Controller Her Britannic of of Majesty's Stationery Office

4-2 wing advanced modelfor the basicteststo develop for transports the sakeof designsfor new subsonlc that the most reliable the higher test Reynoldsnumber (Ref II). The accepted It Is generally aheadof particularproblems of half-modeltesting are performance aircraft of method predlcting In of the firstflight a new aircraftIs on the basis discussed detailIn 11. large of of the results modeltestsIn a reputable complexity effectsleadto even greater wind tuonel. This doos not mean that all wind Propulsion aircraft.For many years,It testdat. are reliable.To obtainreliable In the caseof combat tunnel to (Ref-12) test a suite practice greatcare in both has beenstandard exerclse one needsto' results. model the of and In the Interpretation the of at leastthreemodels: norml complete the actualtesting as aircraft linesdistorted the addresses Issues with the full-scale results The presentlecture model tested a as that have to be borne In mind and containsa little possible; specialintake with a partial or testingpracticeIn both,either in Isolation preferably of description current and part of fuselage of high and low speedtunnels. Much of the material representation the forward is to be found In the published literaturethe wing and finally. an afterbody model to study reports; the material has the jet effects and the effects of the Inevitable Including earlier AGARD and been updated where necessary, finally, the text distortion of the rear fuselage on the cotplete from the testson the. three nmber of referencesmodel. The results by Is supported a substantial furtherdetails. The models have then to be combinedto predictthe which can be studiedfor hevehowever aircraft Combat performance. are broadly aircraft as techniques described experimental and now, In many coupled such as the RAE 8 ft x I becomemuch more closely thosein use In UC tunnels thatone can validto assume ft and 5 metre tunnelsand the AM 9 ft x 8 ft cases,It Isno longer but It Is hopedthat the general treat the Intake, wing and Jet effects as tunnel, transonic simulator needsa powered One really much of the detail. Independent. and philosophy Indeed. testing but this is aircraft In of is a fair reflection testingpractice other (Ref 13) as for a transport more difficult because of the geometrical such as the US and France. countries constraints. The developrentor appropriate problem to to techniques copewith thismostdifficult practice It With mast new aircraft, is standard research still being addressed In models in both high Is complete test representative Impliestwo establishments. and low speed tunols: this probably scale. The test models at a different different most of the has Introduced form and, Th, above discussion resultsare reducedto non-dimensional but It Is assuming It has been possible to test at topics to be covered In the lecture alms the to and Incidences anglesof appropriate startby considering likely Mach numbers, appropriate required they can then be used to predict the of the wind tunneltests,the accuracies sideslip, In and the Instrumentation use for aircraft performance This may suggestthat all from the tests, the measurements thct Is required Is to develop and use making that Instrumentation will measurethe force,and TESTAIMS of 2 TYPICAL standards on moment, tne moaolto the necessary accuracy L.t this would be an over-simplified test aim for the more to the story. The Clearly the most Important picture-there Is muck, Is of that is prediction aircraftperformance to measure been tested In an airflow modelwill Shave of standard L/D to the required wails,the flow over the drag or strictly by constra1ned the tunnel the only testaim' or accuracy.This Is not however modelwill have been affectedby the presence set are usually of rear sting(high speedtunnel)or the limits the flightenvelopo the supporting otherthan drag. To definethe del struts(lowspeedtunnel)sod finally, by considerations under-m testalms In more detail numberwill likely In most cases,the modeltest Reynolds be far belowthe valuefor the fullscaleaircraft. has aben be all those three problem can Research or at least greatly alleviated,for a tranuort aircraft be eliminated adaptive walls (Refs 1-4) to reduce wall To measure the absolute drag and the drag mSgneticsuspension(Refs 5,6) to I Interference. increments between different configurations In and prousurised, eliminatesupport Interference and In the cruise conditions the specified tunnels(Ra 7) to achieve or approach cryogenic segmnt climb, second nuibers. Theseconceptsare Renolds full-scale for routinetestingand howevernot yet available and, In boundary To define the buffet-onset have to be applied for wall and so. corrections particular, the margins to buffet In terms of support Interference and methodologies (Ref 8) have or CL at the cruiseMich number Mach numberat to be devised to control the boundary layer CL. as the cruise over the modelIn orderto simulate, development flow TheseIssues the far as possible, full-scale over the distributions the pressure 3 To measure In In are discussed detail this lecture with theoretical wing partly for comparison In predictions and, more particularly, to Propulsion effectsar* anotherproblemare. understandthe flow In Iportant operating with merely modelsare tested the complete general, and peprorate should be how theios nacelles although tests with turbine free-flow iover powered simulators are sometimes undertaken In Improved, lage io speed tunneis suchas the EV tunnel. At high speeds, the normal practice with a subsonic 4 To obtain the slope of the lift versus underuing aircraft with pylon-mounted transport the Incld-ne curve to lp in Forecasting and to teat Is naceiles to use a largehalf-model flight. to response gustoIn cruising (Refs9,10)as welI or with a powered blosnnacelle nacelleas on the complete as with a free flow CIux. the between the resultsfor 5 To establish usable dpo with d without The differences model tihe hIg-IIft devices deployed at speeds for the Jet these two casesprovidecorrections to appropriate tae-off and landing interfertce that Is not representedin the the sa$m approach modeltest In theory, complete with aircraft transport can be used for a subsonic aircraft although For a combat mounted the rear fuselage on the nacelles about tbe use of the this raisesmore questions technique In passing,it may be noted I To measurethe drag in iong rangeand loiter half-modeI conditions, thatthere is a growingtrendto use a largehalf-

4-3

(c) Accuracy In the sense of forecastIng the drag ofith fullscaleaircraft.Thisls;eve -mor ",the Introduces difficult Abecause it uncertainties, of -predictIng the scale effect which 3 Toassess thelikelyusable liftboundary betweenmodel and full scale and allowing for buffet-onset-as, for will-,be determilned not by dlstortions,-ofthe model and stability -:the.aeroelastic but the civilalrciaft, -probably-,by Also, one'has to allow full jscale aircraft. -and control considerat ons such'.as.pltch-up, present the on wng, drop, nose slice- ioss of .directional for the dragof the excrescences aircraft butenot represented onthemodel. -of postponing and to suggest -ways stability, Realistic claims itbout-thasttainable~standsrds of theseeffects. as expressed (a.lfc)-can'be in accuracy reipect"of 4 To determine the post-stall behavlour,folIows: Mach numbers, part Icularlyat low and moderate In can be discriminated the (a) Drag differences best tunnels 'to an accuracy of 10.0001or of stores on the 5 To determine effects external better CD, In overalldrag and to assess the store release behaviour. as drag (b) Thesabsolute of the modelconfiguration tested In the tunnelcan be obtainedto an over the distributions 6 To measurethe pressure of accuracy tOMO0S In CD, wing for the same reasonsas for the civil aircraft(and,if course,to obtainthe loads can be (c)The drag of the full scale aircraft stresing conditions), incritical In of 0.0010 CD. to forecast an accuracy the low speed stalling 7 To determine drag characteristicsIncluding the drag and (a) implies that one must be able to meaosure this high as and control characteristics a ot 0.00005In C D or better. To achieve stability to have to be developed remove techniques mans of forecasting the usable CLx, with and standard, In or of any effects variability unsteadiness the deployed. devices withoutthe high-lift that axialforceor flow. It Is not simply Therewill.of course,be othertestalm but the tunnel above lists give some Idea of what Is required In drag has to be measured to this standard: other have to be measured to similar high performance.It will quantities of the Interests predicting og be aeen that drag Is the moat Ilportant measurement standards, effects and alsounsteady control and but stability Totaland StaticPressures. Tunnel .re all relevant. N and p : 0.1% STANDARS 3 ACCURACY M Mach number. : t0.0001. The moststringent accuracy requIrements as regards t5 This is unlikely be achievedin the takingof transport are sot by civil prediction performance a should Include program aircraft.One drag count,1# 0.0001In CD. can be the data but the computer the for correcting data to this accuracy. routine Impacton the as having a significant regarded and for prospe.t* a new aircraft on the competitive CL Liftcoefficient, : 0.001. of rangeand fuel economy the aircraft. The most on statement the accuracyrequired authoritative testsIs that prepared by the Wind It should be noted that this Is an order better from wind tunnel In of Techniques (TES)Suboumsittee the than the figure theACARD report (Rof 14) quoted Testing Tunnel this for demanding ACARDFluidDynamics Panel and Issued (Oaf 14) In earlier. Thereare two reasons the when considering drag higherstandard. First, 1982. This statedthat the accuracyrequirements the conditions, wave drag Is likelyto by as moment, suggetted In cruising for lift,dragand pitching to be sensitive smallchangesIn CL and second, in are: sources. and Industry research various normal and by general, drag Is obtained resolving the fact axial forceinto lift and drag. Despite ACL - 0.01 coefficient Lift that aircraft now tend to cruise at near-zero : c0 - 0.0001 Dragcoefficient my still be the tare Cy sin Incidence, 4 m - 0.001 Pitching soment coofficient of in significant the cruisebecause a difference In general discussions about attainable accuracy. In angle between the balance and mind axes. viewsare often expressed. conflicting apparently s Incidence. : 10.03" or If possible, 0.01". On the one hand, some wind tunnel tast engineers of claimthat theycan *nsure drag to an accuracy of example a in CD. i halif a drag count, while others This Is very 1portant. For a typical 0 00005 at CL - 0.51 10.03,is cruising any claim to measureto better than 10 civil transport ridicule as C0 (again a resultof to This confusion arises from equivalent 0.00003In drag counts. falaunderstandlngas as to *hat Is meant by the word the CL sin a term). betw n distinguish One 'accuracy'. can and should Cpb Base pressure, : 8O.002. mesanings: three 2 sustained manoeuvri and To measure the drag In. high speed dash cndil'tlons,
--

base area of (a) Accuracy as regardsability to obtain drag This value Is based on a fuselage eg Increments, differences In drag between two 0.015 x wing area and should be scaled for area ratios. Clearly, different configurations. but different similar to this Is, to tha firstorder,equivaient a definition of repeatability although, as noted Formulae for the dependence of CD on these and are below, knowledge of. for example, wall otherparameters derivedIn detail In Raf 14. that the claimsIn (a) can be achieved Is Interference and support Interference effecte Evidence by provided Figs I, 2. Fig I shows the current relevant, may stillbe highly a In standard of repeatability measuring drag polar tunnel. run In the ARA trmnsonlo drag of the In a given tlest the absolute In (b) Accuracy obtaining of inter-test correctedFig 2 shows the current standard having modelas testedIn the tunnel, are taken the for supportand wall Interference.This is repeatability; threepolarscompared on from the three differenttest series spanning than (s): It depends clearlymoredifficult bias almost a year with the model derigged and precisely; knowingall the corrections The between the three series. are reassembled standards errorsas sell as repeatability from Fig I and other examplesthat conclusions relevant

4-4 DESICN SES4TVIY AND CA!IRRATION BALANCE M needsto discrialnatlon rule,balance As a general accuracy. In thanthe required be-an corder greater this, mans 'that the resolution particular, be shU-ld for capability drag,and~lifr coefficients Thsise llievd 0.00001,and0.0001 respectively. In.both the ARAtransonic and-RAE 8jft x 8 ft qS It Is Important to note that accuracy In tunnels for a typical, (ie prodct"of dynamic area) and modelreference even.differences in dragcan depend Cn pressure data acqufition system of 8000Newtons. determining does not often The basic tunnel wall could, have, been, presented are that, In this ARA to in tunnel,it is~possible, a giventest series, repeatthe polr shapeto anaccuracyof -O,00002 to ani-to repeatthe polar level' 10.00003in C. The claim that one can, wlthcare, discriminate to0.0001 betweenconfigurations drag differences reasonable. entirely appears orbetter therefore 4
knwirgwhat corrections to appiy,,for It Is unwise to assume and support Interference. pose A limiltation. ,A.broad, account of the system

remain the same for two that these corrections This may be true In the similar configurations. case of wall interference (although even here It is important associate drag Increments to the with the correctMach number)but supportInterference can between two undoubtedly change significantly configurations the same model. This will be of discussed detailin 9 but, even at this early In stage,It may be helpfulto give an example. The of affect Interference a rear stingcan seriously the drag of the engine nacelles theyare mounted If on the rear fuselage. Stingcorrections for the aircraft modelshown In Fig 3a, with and without the nacelles, presented Fig 3b. It will be are In seen thatthe difference between the curves. the Ia error,ir the stingcorrections not applied, are In amountsto due the drag Increment to the nacelles 0.0004,ie 4 drag counts, the cruise at Mach number 0.76.also,the errorvaries with bothMach number and CL. The prlmatyreasonwhy the nacelle drag Increment reducedby the presence the sting Is of is thatthe taperof the stingreducesthe velocity and localMach numberover the nacelles, for a 4-engined aircraft such as the VCIO, Fig 3c. the effectcan be oven greater: typically WCi due to the nacelles for a 4-engined aircraft could be reduced 0.0010or say. 30% leading a serious by to underestimate the drag of the aircraftIf the or st ig correct ons are not applied.

in up In use in the ARA tunnel to 1989is contained Ref 15. This has since been replaced by a more modern system. The main improvements with the new systemare that in Is generally more robust with for and stability with facilities amplifier greater of calibration the ampliffers. regular, automatic On paper,the figures for diacriminstion given in system but the Ref 15 still apply to, the new. figures important point Is that the theoretical should now be obtained consistently practice. in typical of Improvements thisnature of are probably what Is currently happening othertunnels in when and if theirsystems updated. are

Ceneral practice manytunnels to use Internal In Is strain gauge balancesmanufactured the Task by but in the UK. RAE and ARA have, for Corporation many years,used balances manufactured In-house to a design orfginally developed RAE. Fig 4. taken at from Ref 16. showsone of thesebalances.It Is In regularuse In the RAE 0 ft x 8 ft tun.el for accurate drag mesurewentsIn tests at total pressures to 2 bar at high subsonic up speeds. In the ARA tunnel, where testsare only possible at totalpressures near I bar, a similar but smaller balancedesign Is used; this has a diameter of 57.15am or 2.25'and a normal forcecapacity of 7120 Neatons. These balances are machined. generall" maragingsteel, from a solid block In with no Internal Joints. The positions of the It my be helpfulat this point, even at the strain gaugesare shownin Fig 4; the axialforce expanse of soe repetition, to list the main Is determinedfrom the strains in the centre factorsthat contribute to success In obtaining flexures; other5 components obtained fro the are highaccuracy fromwind tunnel teststhe strain gauge bridgeson the frontand rear cages *head of and behind the axial force unit. accuracy The dennd over the yearsfor evergreater of I Resolution basicinstrumentation, for forcesand moments has led recently to a reassessment of the basic 2 Sensitivity balances of and of pressure transducers pressures, for design with the aid or finite element methods. The 3 CapabilityIn calibrating balances and In weakestfeature the existing of design In that it allowing drifts. for is oftendifficult achieve perfect to a slcp-free 4 Standards pressure of scanning equipment, fit in the taper Joints where tha balance Is attached to the sting and to the model wing modelattitude, 5 Abilityto measure In one recent 6 Ability to cope with any variabilityof mounting block (see Fig 5). refinment of the design the forward taper Joint in unsteadiness tursei flow,. by has been replaced a flangeJoint: in another. 7 Geometric fidelity modelas a representation cf tha balance has been made Integral with the sting. of the fullscaleaircraft, the length are beingmade to increase Also,efforts for tunni wall of 8 Knowledge how to correct between the measuring element and the end Interference, 9 Knowledge of how to correct accuratelyfor fixations. support Interference. 10 Abilityto correctfor nacelleInternal drag, It is most important that these balancesare before and after and It Knowledge how to fix transition of how calibratedregularly:ideally, of the position, each test program*. The full balance matrix as to determine transition 12 SLIII In simulating the full scale boundary generally deterined in the past Includes direct 6 layer behaviour and in extrapolating the factors, firstorderand 126 second 30 orderterm results fullscaleReynolds to numbers, although sae of thesecan be takenas zero. The *vry few should be undertaken distortion the full calibration of 13 Knowledge the aeroelastic of a monthsand in the RAE g ft x 8 ft tunnel, check model, 14 Abilityto cope with other specialissuesin of the direct factorsand the most significant Interactions oade at the end of each test with is half modeltestiS, US Finally - and most Important the skill. the balance still Installed the model. fis: in care and dedication of the tst 6ab give two examplesof results from balance experience, to at recallbrations ARA: Fig 6a is considered be a engineers. evidentIn resultbut the hystereals satisfactory Thu special problema propulsion of testing will be rig 6b was not accepted and the balncO was of regauged. The ala is to achievean accuracy in addressed 12,13. x 10.33 Newtons;this correspnds to tO0.05% full scale, at the very least, one should aim for
tO.151.

4-5

I'-

The advent of cryogenictunnels Implies that obviouswhat gradientcontrolsthe variation i,n bal nce callbratlon becomesan even more onerous zero., This difficulty Issnccessfully bypassed In requirement: one hasto deierolne the dependence of the ARAtransonic and RAE ft x 8 ft tunnels, but 8 the matrixon temperiture. This Is leadingto the not necessarily all tunnels. relating the in by all in. data to results obtained In a special traverse development. of automatic calibration machines, the machine described In, Ref 18, callbration loads through the test Mach nsmber -range at a. given are -applled to the non-,metric (sting) end of the Incidence or CL. This special, traverse ' balances andthese loads are, measured bya machine undertaken;as the last traverse In, the test when which Is similarIn design to, ex, *I balance temperatures tendedto stabilise, an have thesedata pointi'beingcom uted with,respect to the zeros used In low speed tunnels. commonly Such as those has This machine been designed as an itemin the measured at-theend of tbe test (although theremay Technology Programme In support of the ElI but, be occasions where this appearsto be the wrong although, cryogenic tunnels.,providd the -pur,to approach: -no-general recommendation on this point ,his development, the mchnes ,whn-mnufacosred can be as good as the exper.ence of the skilled and available will,no doubt, beused in support of operator who knows his own equipment). In testing in Conventional tunnels. The, different unpressurlsed tunnels, repeat traverses are often principles the new automatIc of and traditional carried -out In a special additional run and if calibration equipment are illustrated Figs 7a.b necessary, In theseare repeateduntil satisfactory taken fromRef 18. In, the conventional rig. the repeatability (as defined earlier) achieved. Is balanceIs enclosed a sleeve whichthe.loads In to are applied; each Isading. realignment the Drifts In the zeros-are particularly at a of troublesome rig Is needed In order to ensure that the loads are with hAlf-model balances,. These balances are appliedIn the correctdirection relative the invariably to situated outside the tunnel working balanceaxes: a laborloun procedure. In the new section and can therefore be affected by scheme, the model end of the balance is mounted to temperature gradients between the model and tunnel the 'external balanca' whichIs positioned have structure. In the ARA tunnel, the half-model to Its reference centre at the same position as the balance Is submerged In a temperature-controlled reference centre of the balance being calibrated, oil bath but perhaps the only really satisfactory The 'external balance' a very stiffdevice;I approach to Insulate Is Is the balance and to ensure measuresloadsappliedthrougha systemof sever. that the balance chamber freeof draughts.This Is load generators whichare sufficient permitthe may be viewed as Idealised to advice:it is not easy srpilcatlon of any single load or load combination.to follow. In the RAE 8 ft x 8 ft tunnel a new Interferences due to any misalignment are also half-model balance recently installed Irroved has measured by the 'external balance'. With the the situation Lut it Is stillstandard practice to conventional rig, the first and second order applycorrections duringthe computing the test of Interaction factors are evaluated but there are data. for the effects of a temperature Ltradient caseswherethis does not appearto be sufficleat between the setric and 'on-metrlc parts f the to represent the non-linearity in the calibration, balance. With the new scheme. an algorithm developed the at Technical University of Darmstadt extracts third The problems of balance drift have not yet been a order calibration mtrix. for a six-component fully solvtd In large low speed tunnels where the calibration, this matrix meds a data set of 1300 balances are,of course, much larger.This Is why to 2000 different loading conditions. A special modeIs In these tunnels are still often supported computer program has to be used in the tunnelsince on under-modelstruts despite the consequent one cannot invert a third order matrix. To aerodynamic Interferences 9). (see sumarlse, the primary alms of this development of are to provide finally,one should note that humiditymy be a significant source errorif suitable of measures to (a) a totalaccuracy about0.021, of combat It are not taken. Precautions that have (b)a repeatability at least twice as good as the been found to minimise these effects Include: required ae'curacy. controlling the humidity in the tunnel. (c) resolution ast least five limes better than waterproofing the gauges, providing power to the accuracy, balance at all times when the model is in the (d) a rig that does not need any realigrxent during working section. and finally, storing the balance. a calibration, when not in use. in dry conditions end with the poweron. Even when theseprecautinsi taken. are Perhaps the most crucial Issue In balancedesign traditional thinking and experience suggests that and operating practice lies In how so avoid or at one should start a test on a now model with a least. how to allow for drifts in the signals due shake-dom or warm-uprun. Recent eAidence has to temperature effects during a test run. In the indicated however that such a run my, n fact, be ARAtunnel, the drift in the axial force balance an excellent method taking of the Initial aerofor zero can be equivalent a drift in CO - tO.0003 the maIn testat a momentwhen the gradients at to are based on a q - 25000 Nfetons/MZ. Balance drift Is theirmost several Even so. the practice could ImportantIn all tunnels but particularly In still be Justifiedon the grounds that the pressurlsed tunnels shere there is often shke-down run Isa meansof exercising the balance appreciable delay between taking the Initial zeros and the Joints over the range of test loads. It and takingthe firstdata point. Is is standard will however, be realised that there will be practice to thermally watchthe balancebridges. occasions when It falls because of the possibly This eliminates change In sensitivity to a adverse effects on the Initial zero for the main any due uniformchange In temperature but It does not run. compensate for changes In Young's modulus or for the really important point that the balance zeros 5 rcOUI =a FOR pgrSSURtE MEASURDAM S are always sensitive,to a greater or lesser extent, to temperaturegradients across the .L.EEM1. e Transducers balance. Measurements the localtemperatures of at points on an Internal balance have shownthat the For many years, pressures have been measured by changes In these local temperatures lag various types of pressure transducer. These considerably behind the changes in tunnel total convertpressure Into the position a needle on a of temrature. There Is therefore no virtue in mechanical pressure gauge or into an electrical relating the balance drifts to the tunnel output such as voltage or current. in wind tunnel temperature. In any case, It Is a teaperature testing, the voitage output type of sensor Is used gradientthat matters but It is not mmedlately almost exclusively The sensing eltment In high

4-6 of diaphragm to two or more !nputports, a calibration the Is transducers a silicon pressure quality Is circuit. pressure transducer perforned with every scan a normal Wdheatstone that fors Unfortunately, the electrical characteristics of Various establishments therefoee invest ed heavily silicon aie highly dependent on temperature.in Scanivalves Pressuresat ore than 600 at on and offset tappings the wing have been measured ARA In 'in both the 'sensitivity resul-lng is tests on complete aircraft models using 16 voltagevaryingwith time If"the temperature Instilled the fuselage.The D- and In changing. Variousmethods have been used'to ScanilvalVes used most frequently The Stype'Scanivilves-- the'types overcome these temperature problem. of in transducers can be calibrated -situduring a Inwind tunnels have diameters 3.18 and 2.30 cm respectively. known, calibration test: a -numberof accurately are appliedtol the transducer and -at pressures nature that the physical to a least, 2-pint calibration performed establish One has to admit,however, leadsto soew suchas a Scanivalve of the sen.qitlvlty andoffset on-line. Alternatively, an MSP sensor the for differential transducers, two pressure problems. The rate of takingthe data is not as type of connected fastas one wouldlike. This Is partlybecause sensing ports can be pneumatically design and partly becauseof the of Its mechanical together makea meeasurementthe actual.offset to to everytime the makingthe assumptlon thatthe sensitivityneed to allowthe pressure settle voltage, volumein the rotor A third method that is not so valveIs stepped. The Internal has not changed. cavity is the main reasonfor the against and transducer ofte-used is to calibrate the transducer settling time. Whenthe rotormovesfrom and at temperature to measurethe temperature the pneumatic moment making the measurement. At 1990prices, one port to the next, a trapped volume of air Is of of resulting an errorat the Irstant In transducer costsabout retained a typicalqualitypressure connection to the second port and the scanner 1350. requires provides's reservoir whosepressure itself Particular care has to be taken about the choice of finite time to settle to the value or the external The usual method for reference pressure to be measured. Instruments to masure the tunnel pressures.In the RAE 8 ft x 8 ft tunneland the chezking whether the scanning speed Is acceptable
ARA trtusonte tunnl, they are not measured by the or not Is to repeat a given pressure measurement on

sam tp of transducers thoseused for pressure two successive ports on the valve and arrange for as measurements on the a.tual models. In the RAE the previous port to be connectedto a very tunnel,the referenc, pressures are measured by different pressure. As an obvious exanple, let us self-balancing capsule manometers. The Instrumert Imagine that two reference pressures - tunnel total static pressure - are (%otal) pressure has A pressureand free-strea measuring stagnation of Sresolution of 0.34mbaror 0.017% the stagnation connectedto successiveports with the static If the pressur* at two ataospherea pressure and those pressure repeated on the following port. this have a resolution of two values of the staticpressuredisagree, recording static pressures In the ARA tunnel. the reference will Indicate that the scanning speed is too rast 0.17 abar. portscopingwith pressuresare measured by Rusks type DtR6000 to be able to rely on sue-essive as total differences gleatas thatbetween preasure gauges (0 - 2.5 bar) which have a pressure static pressure. The experienced accuracy s 0.04 mb, The readingof and free-stress of specified speedIs my stillfeelthat the scanning these gauges Is matched In the data reduction engineer process the outputfrom DruckDP1140precision acceptablefor all other pressure differences to a in This encountered the test and the onlyactionthatIs baromtsr at the start of every run. necessary is to ignore the first measurement of barometer has a specified accuracy of 50.1S abar. Thus, the maxlmume.rors in dynamic pressure and tunnel static pressure,I* the one letedlately Mach numberarisingfrom the use of thesegauges followingthe total pressure. The acceptable speedis likely vary fromonc fscillty to scanning and t0.0001 respectively are 90.02% to smother bcause of thu different tube lengths. and with some models, for bass In some facilitles in It Is alsonecessary everytest to measure the Scanivalves are meunted outside the in example, pressures high accuracy.Theseare measured to In the ARA tunnel, exporience has shown the ARA tunnel with 345 abar Oruck type PDCR22 tunnel These have a specifled that generally,It is acceptable to scan at 10 differential transducers. accuracy of 0 06%full scale which, when converted ports per second, thus giving about 5 seconds for scan at more than S prts per second. to C D with a typicalvalue of base area to wing the complete reference area of 0.015, gives a possible error in Even a time skew of 3 tconds can degrade the consistency and hence, accuracy of the data. This C of only0.00001. D bhilnd the grosing trend in Is the dol,.ant meotlv# the world in general to abandon lSPs in favour of L1_1. UAJ. lictronically Scanned Pressure Sensors (ESPs). In general, there is not enough space to mount many PretuqgjSenrgESPI Scanned model. It SA Electronically individual transducers in a Olnd tunnel to Is not entirely satisfactory mountthemexternal of tubing USpi et fundamentally different from MSPs. It Is to the modelbecausethe length pressure tapplngs where tha masurement not simply that the scanning Is carried out between the pressu-e rather than mechanically. each is required and sh transducer Itself leads to electronically porthas Its on separate transducer and, To avoid they* problems, pressure significantlags. (MSPs)sore with an ESP. it Ix the output from these Mechanically ScannedPressure Scanners that Is scanned eectronically either in San transducers Corporation deeloped by the Scanivalve to or on sequentially randomlyand then amplified thereare others the market, Diegoalthough noiseproblems. nitils9 electrical a A Scanlvalve provides a means of connecting in portsto a single ESP sensors were firstdeveloped the aid-2970s numb*r (typically of pressure 49) a A motordriverotates shaft to which by several transducer manufacturers and research transducer the Scanlvalve These Included Is connected rotor Into which Is cut a channel agencies, a which pnesmatlcally connects the centrally mounted Corporation, fullt, NASALangley Research Center Pesearch Center. The ailm were to In and NASAAmes transducer to the various Input connec-lons many cases, a single motor drive unit can operate produce compact units capable of giving good rates. Two of the above and highscanning rotorunits The factthatthereis only a accuracy several and Corporation NASA single transducer brings several attendant organisations the Scanlvalve viableproductswhichire now It Langley- produced advantages First, reduce, cost,second, It the NASA e model and availableon th commercialmarket, greatlyreducesthe spae neededin thia finally, knowncalibration If pressures applied Langley are designhaving been developed and marketed

4-7 by Pressure System Ircorporated. Earlydetailed Te0 research establishments NLR In the descriptions of the NASA Langley design are Netherlands RAE Bedfordin England have been and contained In Refs 19-22. The approxinateparticularly Involvid developing In techniques for dimensions the spacerequired a modelfor a the measurement unsteadypressures. In the of In of single 48-ay unit are 4.6 cm x 6.8 cm x 2.9 cm. original approach at NLR, a large number of 96 It Is possible convert to sucha unit Intoa -way pressure tubes were connected a salllnumber to of unit althoughthesehave only been used on rare scanning valves (Ref 23) and each valve was occasions. Using these dul units,wind tunnel connected a groupof tubes in sequence. This to testshave been modeon on- relatively smallmodel approach was relatively cheapbut -the Infornatlon equipped with alnost800 pressure tappings and It obtained was sonehat limited: was not possible It is realistic Imglne that testswith 1000 or to maure transientpressuresor to perform to more tappings now possibli. are cross-correlatIons onlyone scanning when valvewas but was mote expensive used. The RAE approach The outputfromthe separate transducers scanned provided Is muchciore Infornatios; this approach, In a by a digitally addressedanmlogue multiplexer.largenumber transducers of (typically NuliteXCQL Since the transducers are being electronically 093/25A transducers) are mounted In the actual scanned, data rates In excess of 20.000 model surface: each transducerhas Its own measurements second per are possible Data skew Is amplifier and simultaneous measurements are sda or component coherent the unsteady effectively eliminated. Since every the man pressure, therefore pressure be masured Is permanently connected to to with the nodalmotion and the random conponent of a transducer, thereIs no pneumatl settling time pressure everypoint. Thisapproach at can provide of data. Details and cross-correlation of by otherthan thatImposed the volume connectingtransient are problems. the technique to be foundInRef 24 shareIt Is pipingand so. thereare no 'carry-over' to had to be devised compensate Temperature drifts have still to be addressed but notedthata method the units contain a built-in calibration facility, for the fact that the output from the transducers valve In depended slightly but significantly on temperature actuated calibration A preunatically Included In each unit. This valve has two oth as regads zero and sensitivity,The RAE one to abandon the sonewhat bulky When in the technique ,.Iows positions; noml and calibration with the transducers supplied resistor calibration node, a known calibration pressure compensation ard so to take full advantage of the very snIl (knownby reference to a Ruskagauge or barometer) The data Is then of the actual transducer. by to can be applied all the transducers; applying size acquired and processed on-line into coefficient a series of say, 5 such pressures, the zerooffset, Is form by the Presto system described In Ref 2$. of each transducer sensitivity and non-linearity are results obtained this approach by typical determined It Is good practice to connect at SAt.e In to pressure at leastone of presented Ref 26. leastone calibration the transducers thro.yhout the test to monitor (Ruf 27) a combinedsysem whether a recalbration Ia required. LimitedUK KL later Introduced to be aide between Is that a recallbration alsays hich enables comparisons experience suggests with the two approaches necessary the end of everypolar In a typical results obtained at coarplete modeltest. The IssueIs crucialfromthe point of view of accuracy: withoutthe repeated recalibrations. errors the orderof 2-3 abar or of morewouldbe coonoplace as compared withA target shouldnot be takento imply The tim takenfor an The abovediscussion of accuracy 0.2 - 0 3 mar have to be measured to dependson the volume of the that unsteady pressures In-sltu calibration boundary of pressure sourceand obtain a ,,sdlctIon a buffet-onset the calibration tubingbetween vary between Othermethodsthat are more likelyto be used in the ESP sensor itA can therefore testinginclude to secondain a closer,coupledsituation several routine enutes in a reall , wite, tunnel envirorment.In wing root bending tle is (a) aeaureents of the unstesdy the case of the A, tumunelthe required about 2 5 - 3 ainutes These recalibretIons, moment by meansof atrain gauges mountedIn of pukets In the wing surface, slightly erode the besicadvantage therefore, [SP sensors for speeding up the rte of data near the point in favour (b) awasurenenta of the steadypressures taking Anoter very significant wing trailing edge. of [P sersore thatthe onlymoving Is part in them Is tile valvefor changing the calibratlot to eode. this holdsout the hpe that they will need far (c) noting the departuresIn the lift versus loss maintellwe than s.Pr Incidence cures from a basically linear trend, and The author conscious Is thathis personal knowledge of experience wilh ESP sensorsis nuch less than (d) noting the breaks in thae axial forceversus that of eany tunnel engineers other countries In Incidence cv.., it eemis fair to conclude beever that the claies for speeding the rate of data takingare fully up Justified with tire qualfication noted above Til skill and only sorI of cautionis that clearly, greatcare All these wethodsrequireconsiderable applied In sillhave to be practised experience Interpretation'kinkology' discipline and technique prone to and mintain the to the lift curves Is particularly If we are going to use thtel givinga becauseflow separation to which we have beco. misinterpretation of accuracy standards buffet) one part of on loss in lift (and possibly accutoned sha sing ay be asked In the overall results by some other change In fMw on another part of the L . a . . winj, giving altIncrease In local lift The best methods advice Is to realise that no one mthod will be concerns experiaental Although this licture therefore, applyall in a to it for performance, is stillrelevort include successful everysituation. few words about the aeasuremt4tof uSteady possible methods,comparethe results, Interpret pressures buffet onset for civil aircraft and any discrepancies term of the flow behaviour In buffet kenetratlonfor Cilitary aircraft are over the wing and. In crucial and difficult cases. Important considerations hrten delermining usable masure rmtsteady pressures In appropriate lift 'oundaries locations

4-8
6 DETERMINATION MODEL OF ATTITUDE 7 WIND TZN EL FLOE t~iVtROsinTr

It was noted, 3 that It is vital to be able to It is self-evident In that the accuracy of th, measure angle of attack to a high degree of performance data obtained fro- wind tunnel tests bccuracy Simple examination the equation of depends the reliability applicability t lo on and ol calibration of the flow in the empty tunnel and on CD - C sCins + CAcoso whether sound techniques have been developed for copingwithany variabilitylor unsteadiness the In shows that ax has to be knownto an accuracy of flow. The discussion below Is not Intended to be s0.03" .in order to achieve 10.0001 In CD at a comprehensive; merely highlightssome Issues It the that hve been foundto be.particularly Important typicalcruiseCL of 0.5. This is therefore minimum requirement for a desirable accuracy In In ti(, context of model testing to obtain modelattitude measurement: ideally. one wantsan performance data. even better resolution than t0.01".
7A1 Empty Tunnel Flow Calibration

ARAhave, for many years, used a Sunstrand QAP900 the meter' (Ref 15). In the emptytunnelflow calibration, flow In as accelerometer aa 'incidence Experience shownthat,with standard has filtering the workingsection is relatedto two reference techniques, these Incidence reters can still be aressures which usually approximate to the free pressures. In a mar butfet-onset,stream static and stagnation in used successfully conditions SoMe refinements the technique In havehowever been tunnel with solid walls, the referencestatic Introduced since Ref IS was published. For pressure usually Is measured the tunnel on wall at exampie, the accelerometer Is now mounted Integral a hole which Is sufficiently far upstream of the ulth the balance to give added rigidity: modelstationfor the pressure not to be affected temperatures sensed on the Instrument are Itself by the presence the nodelwhen It is present. of and a systematic pitchcalibration from 0* to 90' In a ventilated tunnel, a tunnel le with slotted or Is carried out before and after each test. perforated walls. the pressure in the plenum Corrections for the change in zero and sensitivity chambersurrounding working the section usually Is of the Instrument based on the measured taken as the referencestatic pressure The temperatures appliedIn the computing the reference are of stagnation pressure usually Is sensed at test data The change In zero Is the more a hole In the wall of the settling length upstream significant effect,typically, this can amountto of the contraction ahead of the working section. 0.00061' per degree C and a typicalchange In The static pressure and hence, Mach-number Instrument temperature duringa run can be about distributionalong the length of the working IS.C with these refinements, the resulting section Is obtained most accurately by measuring accuracy in and near the cruise condition nowmeets the pressures along the side of a long tube of the target set out In 3. as circular cross-section mounted In the tunnel with the rear end In tire normal model support prd with Is the RAE8 ft x 8 ft tunnel, on the other hand, the for*ard end extending far forward ahead of the model attitude Ismeasured by the more traditional contraction (Ref28), In thisway, the presence of method of measuringthe quadrant attitude and the tube does not modify the flow through the applying corrections for the deflections the working section of Typical targets for this sting under load The quadrant attitude Is distribution are 10.002 In Mach number at subsonic measured an absolute by encoderwith a resolutionapeeds and t0 005 at transonicspeeds In t0 001 and calibrations have shown that the drive addition, the calibration normally Includes is sensibly liner with no measurable hysteresis measurements of the distributions of static lhe total deflection of the model-sting assembly pressure and flow angle over the working section can be of the orderof I at a stagnation pressure cross-section at various stations along the length of 2 bar and at highsubsonic speeds Typically in likely to be occupied by a model a calibration, there are sal11 shifts between anglesfor Increasing decreasing and loads owingto This description a tunnel of calibration appear may hysteresis effects in the joints The men to be simple textbook material but several calibration is used to determine angular Important points shouldbe noted. miealignments between the balance rollaxes,on and (I) Mny tunnel calibrations sere made a long the oxe hand, and the roll axis and the fuselage time ago shen standards were possibly not datua on the other In both cases at zero-gravity as stringent ae they are today For conditions example,Ref 28 (written the present by author?) in discussing the original One should not dismissthe different approaches in calibration the ARA transonic of tunnel. the two tunne as Implying difference opinion ls a of suggests that the Mach number based on between t*o groups of tunnel enginers It Is In plenus chamber static pressure can be used fact a logical consequence of the different as a reliable Indication of the freeengineering characteristics of the two model stream Mach number at the model provided supportrigs The quadrant In the RAE tunnel Is that the convergenceof the working very stiffand specifically much stifferthan the section wallsdoes not exceed12 minutes. model curt in the ARA tunnel On the otherhand. however, graph In Ref 20 showsthat at a modelbounce one approaches an! enters buffet Is as 12 minutes convergence. there Is a muchmore noticeable the RAE tunnel Both these In discrepancy of approaching 0 003 In Mach characteristics favour the use of the traditional number This specific point is methodin the RAE tunnel and of an Incidence moter unimportant because the ARA tunnel, In in the ARA tunnel The general messageIs that the routine testing, is never operated with best method of determining -,del attit.de can vary the sails converged but It Is quoted to from tunnel to tunnel and should be chosen In the Illistrate how standards have becomem re light experience each particular of In facility strict over the years.today,corrections would certainly be Included for differences of 0 001 or even 0 0003 in Ability to measure model attitnde to high accuracy Mach noaber Also. it Is now recognlsed is. of course, only part o. -he story, one also that tunnels should be recallbrated on a needs to know the tunnel flo angle to the sam regular bas's and that customers need to order of .- crcy This oil1 be discussed In 1 checkthe. toe calibration sufficiently is bolos,tire r ,-ai approach ,n a complete model test comprehensive to satisfy theirparticular is to tet ulth tie node) both erect atd Inverted requirements

aThe development of the roundtry laye hnl along the walls of the unnel controls not only thee locity gradientthrough the tunn a (and hence the empty tunnel buoyancy corrections) but also the relationship between the free-stresm Mach
number at the model and the value based on the reference 'pressures. it follows that in a variable density tunnel, this relationship should be determined at all

flng has been detected In other tunnels. The explanation for the presence of these /artless ay'vsryfrom tuhnelto tunnel,but It Is of Interest to not that, Io tA ho nnl, completely eliminated tunnel flow Ilas-esn mhl 'feature In the
h b h'nrdcin fahnyobI .settling Chamber downstreaml of the 4th corner. A flow angle 'distribution such as that show- In Fig 9 modifies the twist of

(Iii)

stagnation pressures likely be used for to a wing of a half-model mounted on a testing. This pointhas not alwaysbeen balancebelow the tunnel floor. Tests appreciated but It is now often quoted as were made in the NASA AmesIt ft x 11 ft a leading example of what has become known tunnel on a symmetrical wing half-model as a pseudo-Reynolds effect (Ref 29). with the results shownin Fig 10. The Thin pointis particularly important when mean derived tunnel flow angle over the the mia of the test Is to determine the wing was appreciably different according forces on merely part of the model to whether one used the lift-incidence or Installed the tunnel.For example, in drag polars to derive the figure and When testing an afterbody model, the hence. one cannot remove the effects of ac uracy of the afterbody drag Is the empty tunnel flow angle by a simple ritically depondent on whether the change In Incidence datum. presurs on the front face of the afterbody has been related to the correct Ceneral experience shows that this fre-stream staticpressure. it can be flow-angle problemIs less serious when shownthat an errorof one drag count In testln complete models: the discrete afterbody drag will resultfrom the very vort'lesare generally not presentnear smail errors, 6M. In free-stream Mach the centre of the tunnel stream and, to numberas givenby the curve plottedIn the first order. one can remove the Fig Oa. ExtremeaccuracyIn the tunnel effects smailvariations flowangle of In calibration therefore is required for this across the span of the model wing by type of testing, Ignoring the possible testing the modelerect and Inverted. In variation the tunnelcalibration In with effect, one uses the modelwing as a pitch Atagnatlon pressure a variable In density meter to determine the man flow angle tunnel can result in completelywrong over the model. Again,it is or interest conclusions being drawn about the to note that the insertion of the variation afterbody of drag with Reynolds honeycombIn the ARA tunnelappears to number This Is shown by the exampleIn have had the efrect of producing mean flow Fig 8b takenfromRef 30. it willbe seen angles that,at a givenMach number, are that Ignoring the change in the virtually Independent of the wing calibration with stagnation pressure Is planfore. thiswas tot the case before the sufficient to change the sign of the honeycomb was Introduced. Titls suggests variation afterbody of drag with Reynolds that even nar the tunnel centre-plane, number The surprisingtrend In the the Introduction the honeycombhas of Incorrect results puzzledresearchers for Improved the floiangledistribution many years before the error was discovered, The best discussion the of possible effects of not calibrating a variable density tunnel at all test Reynolds numbersIs givenin Ref 31 One 7 2y.jjajbi tv and Unsteadlnoss Tunnel of Flow should of courue not go to the other extree of dismissing all changes with As noted earlierin S.3, the ai should be to Reynolds nuaher as pseudo-Reynolds maintain the testMach numberIn a testpolarto an effects Thos discused laterin 10 are accuracy of 10 0001 and. If this cannot be genuinet achieved, correct the data to this standard to In the post-processing routines fig II presents The emphasis in many calibrations In evidence to support this statemnt At high (iL' transonic tunnoissas originally placed then savodrag (or possibly. flow separation) is marely on the standard of the longitudinal beglnnt.. to appear, the sesitivity of the drag to distribution of tch nu ber and arguably. snall changes in Mach nuaboarIncreases rapidly. th there *as not enough eaphasis on the cruise condition Is likely to be near or jus uniformity the flow,particularly beyondthe breakin thisC o - CI curve of as regards flow angle, oer the cross-section of the working section the trend, Regarding tot effects of flow unsteadiness, already ment ioned to assess tho fltuat Ions at high frequency ar* generally perforsance of ew civ;ilaircraft by filtered electrically this leaves out the effects testing relatively large half-models, has of fluctuations at low frequency Takingthe ARA strengthened the need to look at this tunnel as an example, fig 12 shows that the flow uniformity, or lackof It. with a critical oscillates low frequency, at notably 0 25 Hz and at eye it has been reallsed that, In ca.y at 0 3 it Particularly the higher lift at tut not all high spoad tunnels, the coefficient, the model Incidence and the forces distribution fio angle is far from respondto this flow oscillation To meet this of to ARA It A goodexample of this potential situation, Is standard practice record48 p. "c t prob % is to be foundIn Ref II Results data point samplesover a poriod of 4 & - 6 are presented for the NASA AmesII x II seconds. The variation CD withinthesesamples Of Unitary Wind tunl, Fig 9 shows the can be quitssubstantial (seeRef 15) but even so, average has thata simple anglewith height longexperience Indicated variation cross-flow of =hove the tunnel floor, a variation, of up of such samplesgenerally gives A repeatability of the figs I. 2 provide The NASAAmes betterthan 10 00001 In CD to s0 03, Is Indicated Is turnl is certainly not unique in this uit nate evidvne thatthe procedure successful. She existence of two vortices In at least up to and beyond the likely cruise respect the flow above the floor sd below the conditions

73 Stream Turbulence 'nd Acoustic Spectrum The flow In any tunnel always contains a sall amount of unsteadiness in the form of both velocity and pressure fluctuations. Low speed tunnels are generally assessed In terms of their velocity fluctuations, Is their turbulence, but It has generally been assued In many papers that, at transonic speeds, the acoustic no!se spectrum is the controlling variable. The nost obvious effect of the streotm turbulence and nolse spectrum Is on the position of boundary layer transition on a model under test. Comparative tests have therefore been made In all the major transonic tunnels In the Western world to determine the position of transition on a 10' cone (Refs 33. 34). The observed transition Reynolds numbera from these tests are shown plotted against pressure fluctuation level In Fig 13 This picture reproduced from Rer 33 appears to establish an approximate correlation with the pressure fluctuation level but It will be noted that there Is a 120% scatter about a mean line and probably, this should not be dismissed as scatter. Indee', a later re-analysis of some of the data In Ref 35 has cast doubt on the original conclusion. In Fig 14. taken from Ref 35, the results for 4 leading NASA tunnels are plotted against bath velocity and pressure fluctuation. This fl;urte ma appear difficult to understand at first sight, but the authora of Ref 35 argue that it shows that when the results are plotted against the pressure fluctuation. toey show considerable scatter whereas. hen they are plotted against the velocity fluctuation. they correlate much better. The authors suggest"n relationship of the form: a - tr f(pu) rStream Rt, shere (pu) Is the momentum fluctuation and n - l/ for the begnmning and a - 1/6 for the end of the transition region. Further research appears to be 'e. the needed to clarify the subject- fur os correlation In ig 1h4 in jet fr the correlation in Fig 14 Ia proposed fur the rall Mach-humber range from 0 I to I 2 shre,., if the data are analysed In terns of the tunnel noise one often finds that It, tend, to decrease with Mach number up to M - 0 end then to increase rapidly through the transonic speed range, as might have been expected since, In many tunnels, the pressure flucutations are found to reach a maxils 0 9 and then to decrease near M -

tunnel at FFA, Sweden (Ref 36). The latter Is a particularly Interesting recent example of the detail'that has to be addressed. Ref 36 shows that It Is not sufficient to have a second throat at the start of the diffuser: this leaves the possibility that appreciable noise generated in the model support region can still propagate forward Into the working section. Close attention has therefore to be paid to the longitudinal distribution of the tunnel cross-sectional area opposite the model support to avoid as far as possible severe decelerations In the flow that might Induce a flow separation. Differences in turbulence and/or noise in different tunnels are liable to lead to differences In natural transition position on the model under test (Ref 37). However, the recomended standard practice in transonic tunels Is to test with transition fixed artificially and this removes the risk that results from different tunneln will appear to be Inconsistentbecause of differences In transition position. It does not follow that dorterences In turbulence and noise are unimportant. As will be discussed In detail In 10. a technique In comon use for simulating the behaviour of the full-scale boundary layer ie to test with a transition position on the model that is further aft than that expected on the full-scale aircraft. It Is desirable that the stream turbulence does not place any limitation on the use of this technique. The ability to maintain an extensive length of laminar flow will be even more Important when testing models of laminar flow aircraft.

turbulence also has an effect on the development of a turbulent boundary la)er This has been studied by Green (Ref 33) who suggested that it was possible to transform turbulence Into ffective Reynolds number. Thia led to the usuggestionIn scme quarters that Increasing the turbulence of the stream could be one method of Increasing the effective test Reynolds number The difficulty with this suggestion however Is that Increasing turbulence only Increases the effective Reynolds number in respect of the boundary layer shape factor (and hence. boundary layer separation onset) in terms of boundary layer skin friction and hence, drag, it reduces the effective Reynolds number. This IsIllustrated by the results In rig 15 reproduc.d from Ref 39 In nost transonic tunnels, the turbulence level Is far less than 1% and so. the effects shown In Fig 15 can be dismissed as trivial It has however been recognsied (Ref 40) for many years that accepting too high a level of tunnel stream noise can degrade the accuracy of buffet data. Cambey suggeaed that. to obtain data uncontaminated by any interaction with the tunnel noise, the value of (nf(n))i should not be greater than 0 002 where n is the non-dlmensional frequency for say. the model wing fundamental bending mode and ihere F(n) is pressure related to the non-dimensioraI fluctuations by the equation q, o Qthre p' - acoustic pressure signal r(n) dn

For most axlslng transonic tunnels, tihe transition Reynolds number for the I0' cone at N - 0 8 1lit In the range 3 x 104 - 5 . l04 factors that can affect the precise value Include (I) the noise and turbulence being propagated from upstream, eg from the valves In a blowdown tunnel and %hether or not there has been any treatment In the aettling chamber aimed at damping these dlsturbhnces. the nature of the tunnel walls. eg whethar they are solid. slotted or perforated and shether there has been any attempt to nolse-generation their alleviate properties, shether the tunnel design contains any feature such as a second throat to prevent the upstream propagation of noise frum the downstream diffuser

(it)

(111)

With the Increased interest In laminar flow aircraft design, all ths Issues are now receiving close attention both In modifications to existing tunnels, eg the honeycomb In the ARA tunnel (Ref 32) and In the design of -.* tunnels, eg the TIS0O

Fig 16 presents an exampte of how the unsteady wina-root atrain can be Inflenced by reducing the unsteadiness of the tunnel stream In thiu example. the stream unsteadiness was reduced by a change of slotted working section wall from one having a hard surface to one with a laminate. as . consequence, buffet onset becoe more clearly defined and the buffeting measurements ahoed much less scatter

t
" TUNNEL WALL INTERCFENCE
-

4'll

I:
I

|,I The Classical AsoroachS The . Closed tunnels at subsonic sneeds presence of the tunnel allsmodiffes

the

(11)Evans,showed that most wings can be represented by a uniform non-tapered wing,-havlngthe Isae volume ,mean sweep and thickness ratio as the original wing but with a span equal to 2(3)1k x where k. Is the radius of gyration of the original wing about the x axis. (il The value of P in the denominator of the above expression should be based on the corrected Mach number. This my seem to be a trivial point but early experleniceIn the 1940s showed that if 0 was based on the uncorrected Mach number (the more straightforward procedure), the blockage corrections could be seriously underestimated. This is an Important point which was not always remembered'In later years Having determined the Interference velocity, dU MU, corrections to the stream quantities and force and moment coefficients follow as set lut In both Refs 42 and 44.

effective angle of, Incidence and the effective speed of flow over the model. These effects are known, respectively as, tunnel constraint and blockage ,and the measured data from tests In a conventional tunnel must be corrected accord' ely. In the classIcal approach to a prediction mthud, the model, is replaced by singularities and the wallsaby a doubly-infinite set of images. These methods aredeveloped In detail In Agardograph 109 (Ref 42) which Is the major reference on the subject. A full oat of formulae and graphs are given in this reference for closed, open and ventilated tunnels. Different standards of approximation will be needed for different types of testing In various tunnels but In the author's experience, the formulae discussed below sumarlse a reasonable set of corrections staying within the limitations of this classical approach which, It will be realised, Is based on the assumption that the flow Is uniformly of the snall-porturbation type. These formulae can be listed as follows:

(c) Interference at hlsh lift The corrections for tunnel Interference described above can be applied to the results of tests when the flow past the model Is attached. When the flow is part:ally separated, however, a less rigorous approach bze to be adopted. The general practice In the XX and elsewhere has been to adopt the method put forward by Maskell and described In Agardograph 109 (Ref 42). It Is not possible to represent tliewake as a plane sheet of streamnlse trailing vortices. Maskell based his approach on a study of the flow past a bluff body. Experimental measurements described In Ref 47 confirmed that, for wings of soderate to small aspect ratio, the localised regions of separated flow that develop as bluff-body wakes and iaskell concluded that the tendency to axial symmetry In the separated flow region could be assumed to be universal, applying to most wings of practical Interest The formulAe derived from askel's model of bluff body flow are applied to the separated-flow part. CD . of the total drag. ultimately giving a blockage correction In the form 2;q

()Tunnel

constraint equation for

In the simplest approach, the basic the Interferenceupwash angle Is


*O-

6j scjc

(2) such wings beginto stall,resembleaxisyssnetric

values of the factors 6a and 61 are presented In Ref 42 for square and rectangular working sections with aiternatively 4 closed, 4 open and 2 closed/2 open wlls For a square section, go - 0 13 and 01 - 0 25 if the walls are closed. This simple formulation should not be used If the model wing span Is greater than about 0 5 x tunnel width From the author's experience. one should then use ts, relations In Rsf 42 in serms of a paramet.e If the span/tunnel width ratio Is 0 8, the (o) value of 0 for a square section with closed saIls then becomes 0 162. ie an Increase of 2S% relative the value for a small model

2 C C

x2

(CD

CD

- "CN)

(3)

there CD, is given by an extrapolation of the dragdue-to-i fs in the attached flow range (see Fig 17) and q. Is the corrected value of the dynamic pressure, q The exaeple In Fig IS taken from Ref 42 shoms that, for this case at least. the formula Is very successful

The presence of thu tunnel walls modifies the flow around the model even at zero lift In a closed tunnel, the flow around the model Is speeded up and vice versa In an open turinl This Interference is due to the volume of the model and Its wake. In general, It is acceptable to treat solid and wake blockage as independent of each other, this is not necessarily true at high lift (see (c) below) for the calculation of blockage by the classical rathods, the model Is replaced by an appropriate distribution of sources and sinks Simple formulae are listed In Agardograph 109 but. In the Wk, the standard method that has been used for many years is that produced by Evans In 1949 (Ref 44) This method was based on earlier sork by Thoa and Toimpson (Refs 45. 46) A fe points about the method are worth noting (1) Eapirical terms are Included to allow for the effe-t* of wing thickness/chold atin and body fineness ratio

8.1.2

Tunnel with ventilatedsals AL sgbonl sgeeds

Ventilated tunnel walls were Introduced In the early 1950s In general, the walls have either longitudinal slats or perforations, the primary aim is to alias air so pass between thy sorking section and th surrounding plenum chamber and so. to reliev the choking of the flow that would othrwise occur before reaching M - I 0 A supplementary s Is to reduce and. If possible, eliminate the nall Interferenceat subsonic speeds

Early studies showed however that It was very unlikely that complete elimination of this Interference would be possible with either slotted or perforated saIls

~4-12
!n this early work for slottedtunnels.' was it reasonswhy Ref 49 concluded that slotted walls assuaedthat the real wall couldbe replacedby an equivalent homogeneous boundary having a similar Influence the flow ear the modelas that of the on real wall. The Ilnearlsed condition for byte eprse anb boundary wal his euvln following equation: 0 *K an -# + 04 (4) were preferable perforated to walls for testsat subsonic speeds As notedabove,thin Is however not a clear-cut Issuebecause the viscous effects witha real slotted wall mightproduce similar (but probably smaller) effects. The results In Fig 19 are for a rectangular working sectionwith two perforatedwalls; subsequently, AR calculated

In usingthe formulae tunnel circular f ... f 8; numerically, the values are slightly different but In principle and Indeed, general In where 9 Is the perturbation potential,x Is magnitude, results the are very similar. Finolly, masured In the streamdirection and n alongthe Fig 20 0 shows the variationof the constraint outward norms to the surface. The boundary factors and 61 with P/P for perforated walls. condition relatesto Inviscid flow past a slotted Compared with the Idealslotted wall,this picture wall;on thisassumption, thereIs no pressure drop at firstsight looks encouraging that 60 pases In acrossthe wall and this Is In direct contrastto through zeroat a valueof jl/P similar the value to the porousor perforated walls where there is a thatgiveszero blockage. notedbelow, As however. pressure drop -hroughthe wall giving a boundary this does not moan that one can Ignore constraint condition the form: of effects existing In perforated-wall tunnels. Despite the fact that all this material had bosn ax Ppublished by 1966,many operators of tunnel with ventilated walls continued for many yearsnot to whereP Is a porosity parameter defined by applyany corrections theirresults. Theyhoped to that if the models were kept small (le blockage ap 6)are ......tio lss than 0.5), the corrections would P---, (6) be trivial exceptcloseto i - 1.0. They felt that they could not apply the corrections outlined as Several different typesof wall can be Identified. above because they did not know the porosity factors, and P for the wallsof theirparticular K Closed wall : K o 5, P - 0 tunnel. Determining these factorsdirectly by Open Jet K - 0, P .smeasuring the pressure differential and flow idealslotted wall P Im through wallswould indeed a difficult the be task. Real slottsd wail . bothK and P termspresent However. Ignore to the existence the corrections of Perforated wall K - 0, P dependent wall simply becausetheremay be some doubt over the on geometry wall boundary precisevalues and alwaysseemedto the present author layerthickness to be the wrong attitude ARA almost from the outset applied lift constraint corrections, or a perforated wall with normalholes.P varies blockage and blockage buoyancy corrections These with the pressure-differential throughthe wall; corrections werederived follows as earlytestsat AEDC showed that,withnormalholes. the valueof P was very dependent whetherthere on (I) porosity factors for the wallsof the ARA wes Inflow outflow or through the wall to obtain a tunnelwere obtained by Interpolation of sensibly linear characteristic for the wall the AEDC data contained Ref S0 for the In porosity, needsa wallwith the holes Inclined one characteristics of various perforated at 60" In the direction the flow,this reduces of plates with differentplate thickness, the resistance outflow to holediameter and open-area ratio, Agardograph 109 (Sef 42) containsmany figures (11) showinghow tunnel blockageand lift constraint varywithK and P In different typesof slotted and tunnel It will be realisedthat perforated-call theseestimates were made by the methodsavailable aheadof 1966and, numerically, couldbe Improved the today Nevertheless, figuresstill serve to illustrateso"e Importantconclusions. For example,for an Idealslottedwall,the open-area Is from that ratiofor zeroblockage very different eg needed for zero lift constraint. In a working and 2 solidwalls, tha sectionwith 2 ventilated open-arearatio giving zero blockage Is still factorof 70% to calculated give a liftconstraint of that for an open wall However, allowing for (iii) the viscousflow !n the s*otsIn the real slotted wall.Increases chances finding open-area the of an ratio that will give completely Interference-free flow (Ref 49) However. the viscousflow In the slots Is also predicted to give a longitudinal gradient throughthe workingsectionand bnce, a buoyancy correction. TurnnS to a perforated tunnel, Fig 19 shows the longitudinal distribution of the blockage effect This Is a most Important graph. It will be seen that the longitudinal distribution for f/P - 1.28. which giveszero blockage the modelmid-point, at Is strongly asymmetric This can lead to a sizeable buoyancy effect Calculations for a typical subsonic transport model might show that this buoyancy effect would Increase the drag coefficient high subsonic at speedsby as much as KD - 0 0010 - 0 0020 This Is one of the main thesevalues P were thenused to obtain of lift constraintfactors and also. the blockage at the aid-point the model of The derived valuesshowedthat the tunnel was too open to give zero interference As regardslift constraint, the factors nere about70% of those that wouldapply In an open tunnel The blockage corrections were predicted to be about for -0 2S x thosethatwouldbe calculated tunnel For the corresponding closed-wall transprt model with subsonic a typical about 0.7% blockage area ratio, this thatam - -0.005 N - 0 65, at Implies finally, blockage buoyancy corrections were derived the basisof Fig 21 To on understand this figure. one has to be awareof the open-area distribution along the wallsof the ARA tunnel opposite the forward part of the model, the open-area is stillclotbing to its finalvalueof up ?2% shich Is then held constant opposite the rear of the model The suggestion In Fig 21 that the buoyancy effect Is not felt by the nose of the model was conflrmd in a pressure-plotting on a test civil transport model. comparing tie pressures measuredIn two testswith the holes In the walls respectively open and sealed In effect, this meansthat the buoyancy correction Is only half that It wouldhave been If the open-area ratiohad ben 22% along the full length of the

4-13 In model; eveq so, the correctiont Is still Inthe UK In the late 1950s, %whichseveral models " scaleswere .highly significant;if. It were nut to the same design but'at dilfferent In at'RA,and the ARA amounting testedIn two slotted-tunnels a applied, spulous drag-creep, of tunnel. The'results thesetestsare , t6 ,mie than 0.0005 In CD, would be transonic reportedIn Ref 53. The model was,a,wing-body result. misleading a present: seriously wing having symmetrical with a 61'thIck combination testwith the holesIn the an aspectratioof 2.83,a taperratioof 0.33 and In the samecomparatlve of'blockage open and sealed,the 45'sweepon the O.Sc line. The values tunnel walls alternatively tests are, of These near M- 1.0 revealedby these, were-measured. pressures wing trailing-edge ratiosof the of a that,fots modelof a reasonablecourse, function the open-area a suggested results of the tunnels being compared. It Is one size,eg 0.5 - 0.7% blockage, couldassumethat walls, to of blockage therefore more generalinterest comparethe correction(note:not the, the blockage of derived the analysis the experlmental by This values was- zero up to M - 0.85. buoyancy). that may be predictions beliefthat at H - 0.85. data with any theoretical the contradicted earlier, methodavailable appeared the tine available.The only theoretical at AM - -0.005.The new evidence by and It became standard In 1959was thatproduced Pageof NACA Ames (Ref to be unchallengeable corrections54). The formulaeproposed by Page for the at practice ARA net to applyblockage AM. at M - 1.0 are given for this size of model up to U - 0.85 and to blockagecorrection, be below:
subtract 0,005 from the values that would

methodfor Mach numbers by calculated the previous (r*/x*)'/l aboveN - 0.85. It was feltthat It was betterto AMo - -0.9g(rt/h)S/7 accept the direct evidencefrom the comparative tunnel, and slotted whichwas for rectangular test than to relyon the earlier method (7) asuption thatdata fromthe basedon the unproven (r*/x*)?/? plates couldhe used AMo - -0.82(r*/R)S/7 on AEDC experlments perforated of the ARA characteristics the to forecast porosity tunnels perforated wmthod for circular In tunnel walls. The weakness the original asasumption that on was that it rested the unproven ratio, on on layerthickness the walls of the whereg depends theopen-area the boundary of r*,x are the coordinates the sonicpoint on wth the thickneaas the was comparable ARA tunnel body of on the noseof the equivalent If geometry; this was true, of AEIC plates similar to revolution the modelundertest, it would be somewhat of a coincidence. The semi-height h Is the tunnel to the blockage be zero up to M of practice taking tunnel. of since1968 for the sake of and R the radius the circular - 0 85 has been retained althoughthere consistency date-bank maintalning In has alwaysbean someuneaseas to whetherthis was Valuesare compared the tablebelow the comparison the correctapproach. For example, RAEslotted ARA perforated on measured a distributions wing pressure between wall tunnel wall tunnel model of the SuperVCIO had showngood agreement, as reportedIn Ref 52, with those masured In AM derivedby Predicled 0 for corrections even thoughblockage flight -0.020 model (1) -0.007 the original method had been applied; to have 0.05%blockage assned that 454- 0 up to M - 0.85 would have . it Predicted for of reducedthe standard agrelemt. Recently, *0.049 model (2) .0 016 of that the results the 0 S% blockage has howeverbeen reallsed testcan. versussolidwall comparative perforated thatby M - Difference between on in fact.be challenged the grounds 0.029 0.009 tunnel are not (2) snd (I) 0.85.the resultsIn the solid-wall by to correctable, use modern terminology, slmple There should also be a Difference as derived Ace and AM corrections. 0,020 0.010 wal-induced camber effect which, for a given from exporlmntal data a., would increase the auctionsnear corrected and mid-chord hence, the adverse pressure gradient The estimates In the above table wor* obtained by valuessuggested Page.g This camber effect usingthe numerical back to the trailingedge over 0 35 for the slotted tunnel and a numerical layerdevelopment couldmodify the boundary of the rear of the upper surfaceof the wing and constant 0 02 for the perforaed tunnel The that the valuefor the slotted coparisons suggest hence,It nay be wrongto expectthe trailing-*dge In the solid and tunnel Is reasonable but that a smaller value than pressure to be the sam suggest that that proposed should be used for the perforated tunoeIs Calculations perforated-wuli On tis and other evidence obtained later, tIhis could account for the discrepancy discussed tunnel be above It should stressed that for the majority ARA have used 0 6 as the numrici constant In an rewritten termsof tunnelseml-height in transport models. %here the expression of tests on civil tItan radius cruise Mach number Is near U - 0 $0, this rather but It has 11 be seen from the above formulae that Is uncertainty of trlvlalimportance le sone length here been described atthat can arise In to IllustrateIt applying the blockage area ratio Is no longer a relevant the difficulties classical methods Thin Increases the inportance parater at A - 1.0 The variation AM with of .1selcalwthds Tis
icreses he lporsme ar. size is mtuchmore In sympathy with the linear

and the distanceol the and takingfull dl nsionsomth mordal a of adopting moremodernapproach that the ia follows t den f the moel of the developmentsIn CPD mthods. advantage Isflosttth tunne -si model from ithe laterIn 8 Thesemethodsare discussed for Is stillsignificant very tunnelinterference eog smallmodels, even for a pilottube while, on the other hand. Increases In model size can be 8 2 IAllInterferere as cSgeeds tolerated without as much pnalty as one might Near end AboveU4- I0 point Important have expected.Another intuitively close to M - 1.0 aboutthe interference at apeeda 8 2 I alockae correctlons can is that the Interference be greater for a to approach the calculationslender model than for a model of low fineness Clearly, the classical by corrections *hlch,for a ventilated-ratioand of the sam size This was pointed out of blockage Is In salt tunel, AM- a factorx (4)closedhas to be by iarndt Ref 55, the reasons that the lteral M beforereaching - I 0 To obtainsome decay of the flow field Is less and hence, the abandoned the at interference the wailgreater lit, gudanve as to the Interference closeto and above potential model *as programie launched slender 4 - i 0. a majorcooperative

4-14
Strictly, As not possible obtainmeaningful It to ,results of the 1.0. The
results, at literally M -

flow fieldaroundthe actual

high subsonic speeds near Ml'- model. At 1.0, recent

comparatlve, tests discussed above also showed-that the aft movement the terminal shock which should of re~o Cho base of ,he Lmdel just
labove M 1.0 was

US work (Ref 57) has suggested that one has to undertake Navier-Stoke's calculations; the results of'
Inviscid

considerably delayed on the 0.05% blockage models.


This means,that, In a test at an uncorrected Plach number of say, M - 1.05, the flow over the front

Euler (I)

calculatlons

can

misleading,

be

complete)y

part of the modelgenuinely resembles what would be expected in free-air at 4 - 1.05 but the flow over the rear

part of the model is more representative

the second type, known as two-component methods, do not require a calculation of


the local flwfedaroundth oebt

of what mightbe expected at N - 1.0. Such data are completely unrepresentative the fra-air of results and theycrnnt be corrected.It canot be elaphas load
too st-'onglythet ame should not test at

Involve the measurement, a relatively of large number of streamelseand normal velocities near the walls. Methods the of
second type art therefore easier to apply

Mach numbersver near to and just above M - 1.0. All test programmes should omit the rangebetween say. - 0.98 and N - 1.05dependent the sizeof M on the mdel. This situation wouldbe Improved It if worepossible reduce to the wall open-area ratioto a very low valuewhen testing near M - 1.0.

in the case of solid-wall tunnels where one can make the assumption that the flow near-the wall Is parallelto the wall (strictly, parallel the boundary to layer on O'e wall) and hence, one still only neat to masuse one flow comp'onent. .

At higher Mach numbers, tunnel interference takes Methods of the first type were developed by Smith ttne form wave reflections of fromthe tunnel walls, of NLR (Ref58) and Capolier, Chevalier Bouniol and Clearly, In a solid-sall tunnel, one Is not In the at ONERA(Ref 39). In the US, the Initiative came clear until the reflection the bow shock has from Kemp at NASA Langley(Ref 60) followed of by passed behindthe baseof the model. The situation urman(Ref 61) and recent work Is described US In is such the same in a slotted-wall tunnel but Refs 62, 57 and 63. The second type of method was perforated wallsprovideso e alleviation. 22% developed Ashill The by and weeksat RAE Bedford (Refs open-area ratio, normal holes of the ARAtunnel ar* 64,65) and currently, serloos use of this technique successful largely In cancelling reflections the of is being explored In the RAE 5 metre tunnel (Ref incident shockwavesat M - I 15 and above(seafig 66) Historically, can ho arguedthat the idea It 22) but, with normal holes, expansion flow fields of using wall pressure measurements as a guide to reflect discrete as shockwaves. It followsthat model blockage corrections was Introducedby the data In the ARA tunnel do not becom Coethert long ago as 1940(Per67). as effectively Intsrference-free the reflections relatively until A simple mthod of the first type Is of the forebody expansion flow field have passed beingused to correct data fromthe RAE 8 ft x B ft behind the base; inclined holes would improve this tunnel (Ref 17) Measurements of static pressure situation, for the reasonsexplained earlier. In are made at fourpointson the tunnel walls - two any new perforated-wall tunnel, one would choose In the roof close to the model centre of volume. walls with Inclined holes of variable open-area and a corresponding pair In the floor. The model ratio. is sulated by a distribution pointsources of and sinks and calculatiolns are madt of beth the L 3 The Modern Aenroach Increment In streammlet speed at the position of the wall holesdue to both thesesingularities and itwas notedin 8.1 thatthe classical approach to their Images, and of the blockage Increment In the calculation of tunnelInterference subsonic velocity the model. Thisprovides at at the ratioof speeds suffered from several Important weaknesses the blockage Incremeont to the arithmetic man of To listthesebriefly the calculated increments in speed at these four holes it is thenassumed that thisratioapplies (a) the methods rely on a small-perturbation te real tuseel flow and henc* one obtains the in representation the model, of blockage from the wall pressure measurements the homgenous Results from applying this approach ha-e been found (b) for tunni wihslotted wails, with results (Ref 64) to be In good agreement (e wai borundars onditon Is is u i kown (Ref 56)obe to obtained by a potentially more accurate twounrepresentative, component method and It Is believed that the (c) for tunels with perforated sails, the porosity approaching the choking value technique Is accurate up to Mach numbers are characteristics uncertain, (d) the methods do not take proper account of the fact that the wall Interference be ver) can dependent on the boundary layer development alongthe salls, and (e) as one approaches M - I 0, It Is no longer valid to assume that the interference Is correctable torsoof simple In corrections to endS Since 1978, there has been a major effort at many research establishments to develop sme, Improved methodsof estimating wall interfersnce Most of these inolv, the wmasuresent of pressures on or near the tunnel walls, most involve the use of the powerful CfD tools that have now become avaltable Rroadly, methodscan be dividedInto the two types (I) the firsttypecan be described 'mdel as representation methods' These require only a relatively limitednumberof wall pressure measurements but need a reasonably accurate calculation of the An obviousapplication a method of the second for type Is to the correction dataobtained high of at model lift in a low speedtunnel. The flu.around the mdel beingpartly separated difficult Is to simulate mathematically any accuracy with blt this Is not needed for a method of the second type The method developed by Ashill and Weeks (Ref 64) has therefore been applied the results testson a to of very largehalf-model a landing In configuratisn In the S metre tunnel (Ref 66) Measureento of pressures ers made at aboutI5 tappings on each of 10 streamlse rows and upwash, sldewash 'and streasweash corrections have been derived Typical results are shown In Fig 23 Results obtalned' by using the standard correction technique are also shown for comparison Reasonable egreemer -s 2 shownfor the Incidence correction Rig 3a but a in significant dlscrepecy stremeash correction fig 231evident in that In Is It appears the the standardtechniqueleads to an appreciable overpredlction and these results constitutea warning that iLaskel approach may not be i's satisfactory sow realistic for casesof partially

4-15 Individual pressure has been corrected by termswhichvary alongthe chord. In Interference other words, the AN, Ac approach has been abandoned and thereisnow somehope that correction methods, that have been developed beth pre- and post-test will remainvalid up to very-close to M4- 1.0. wll have to However, researchand development continue for some time to come before one could for method Is available claim that a correction use. Ref 63 notesthatthe WIACprocedure routine corrections of an Ref 64 contains example blockage with was apparently not completelysuccessful in by calculated methodsof both typescompared data fromthe NASA 0.3 the value obtained by the classical method correctingsome experimental thatthe but It ispossible Tunnel, Is earlier.This comparison shownIn Fig metreCryogenic discussed of full agreement between a corrected 24: the application relates to a two-dimensional lack aerofol| test in the 8 ft x 8 ft tunnel at RAE experimental and theoretical result, cay be due to Bedford. Cood agreementIs shown between the Inadequacies In the turbulence modelling In the Navier-Stokescalculations, rather than any and representation the twofor the odel results method. flaw in the correction method&but all these resultspredict fundamental component method. thanthe classical blockage greater notably Walls over the aerofoll 8.4 AdaotlVe particularly Ctl for when the flow issupercritical. tunnels used for None of the major transonic The recent experience at AEUC In applying these performance testing are fitted with adaptive walls of about the development at Interference and so a discussion perforated-wall modern methodsto the scopeof this walls Is reallyoutside high subsonic Mach numbers near M - 1.0 Is adaptive Nevertheless, for the sake of are lecture. rwo types of technique In described Ref 57. It completeness, should be noted that. In many discussed: been ents, there has research establish progresswith adaptive wails since In technique which the considerable (a) A pretestpredictive based on a 1975, Achievements are described In detail In the wall boundary condition Is not Croup12 (Ref4) global approach as In classical methods but finalreportof ACARDFlDPWorking and on many references the subject, In allows for local variations the porosity whichcontains is characteristics. This meansthat the slope or a ssmary of some of the main achievements In but available Ref I constant a Is the characteristic no longer is a function of the boundary layer thickness wind tunnel of on the wallsas shownIn Fig 23 takenfrom Ref The basicconcept an adaptive-wall flow-disturbance 7his graph refers to the 60' inclined Is to match two Independent 62 In at an Interface the tunnel measured wallsbut similar quantities holesof the AEDI perforated for an computed to graphscould be createdfor othertunnelsIf experiment the same quantities made interference-fr-a outer flow beyond the Interface were experlments necessary the has been greatly helped Increasing the lift on the mde will Increase Application of the concept wallby advances In wind tunnelInstrumentation. the pressure variation Induced on the wall at echanlsms, control technology, computer high subsonicspeeds and. as a result, the control and CFD algorithms and.more particularly. on layerthickness the top wall. This hardware boundary floe. many In two-dimensional Is the major reason for the Increase of AM with codes to have shownthat It Is possible In the pretest establishments lift referred to shove the aftershaping Interference predictive method, the tunnel flow-fieldreducethe residual Th. residual is made with the AEDC boundary walls to a very low level calculation by th, methods can be calculated interference wall. on spoclfied the tunnel condition already earlier using the Information discussed has the walls The concept for shaping correctingthe available (b) The US VIAC approach for applied to Group 2 Flows resultsIn which the flow aroundthe been successfully measured -o flow *vtondlng and beyond the (supercrltical measured with the pressures modelis calculated waills) In the context of the present on a boundary close to the tunnel walls test-section has development been the most significant lecture, condition the boundary defining ing that two-dimensio.,al the evidvnce sulgest can be used succesfully to liniwlst walls were made by an Euler adaptive the calculations Initially. code whlch was expocted to be more than adequate the interfer.nce in tests on three-dimensional This evidence Is presented In detail In In which the model was a wdels for an application wing-body com1bnation with 30' swept wings with Chapter 4 In Ref 4. the aspct ratio or the tunnel Important parameter is an section However, these working 0OO 4 synmetrical sections NACA In failed that the derived corrections xperieste at supersonic speeds Is limited .t calculatIons present overcorrected the results for the large seriously blockage model In the small tunnel when compaled against those obtained with the same model in a INTFPFrIgENC code, 9 ndit,sUPPORT Use of a Navler-Stokes larger tunnel however,goes a long way towards bringingthe tInr Interference corrected results from the two tests Into 9-1 R In are agreement Thesecomparisons presented Figs the tunnes.. modelsare 26a.b. r In this figure Is the waillopen-area In high sFeedand transonic on sapported a sting from the rear of the ratio 1he authors of Ref 57 draw the conclusion usually either from the centre of the to that It is necessary allowfor viscouseffects model rrotrudlng rear fuselage or as a blade from underneath or In the model flow-field calculation This may well As already be true but the present author believes that the occasionally from the top of the fin effects Interference note, In 3, the consequent couldhavebeen significantly Eulercode comparison and model calculationscan be significant They arise for two main Improvedif the free-air had been made for different Mach numbers, the reason. First the presence of the sting itself to a and particularly, of any taper on the sting can difference In Mach number corresponding on M first-order correction In case the procedure have a forwardInfluence the flowover the rear Is the flowvelocity reduced Is of these WIAC calculations not clear, It Is fuselage In general, Panel mthods can be used to the by anid drag reduced north noting that the values of CL are obtained Secondly. the rear In distributions whlcheach estimate these effects of Integration pressure that,despitethe separated flow. It seemslikely need for a large number of pressure measurements extending far upstream and downstream of the model, methods will find increasing two-componenit for particularly models in application the future, with complexflows,eg modelswith bluffshapes, with rotor simlation ASTOVLmodels,helicopters at and models high lift.

4-16 fuselage

has to be truncated distorted admit employed In the design of the new rig. the stings and to the sting; in this respect, calculations~are less are very slender and have reduced torsional because viscous effects are paramount, stiffness; the model 'is mounted further forward effective relative the yoke. As with the previous to rig. as will be determined the In The technique regularuse, the IK to-obtain the sting corrections in th(sestingcorrections experimentslly to mount difference between the results for two Is the modelon twinstings from the wingsand thento configurations with alternatively the true measure the forces on the rear fuselage with and afterbody and the distorted afterbody and dummy 27 without simulation of the rearsting. Fig a is sting. a a dagrammatic picture the rear of model layout for Such a test. The balancemeasures the forces on Some typical sting corrections were discussed the rear fuselage with the dumy central sting In earlierin 3, on the basisof the results figs In position as shown and with the sting removed, the 3a,b. It should notedthat theseresults be relate bore filledand the rear fuselage restored to the to stings designed for tests at a stagnation correct aircraft shape. The difference betweenthe pressure of I bar For tests in a pre-surlsed two sets of balance readings gives the sting tunnel,It Is likely that the degree distortion of corrections. This may sound simple but much of the rear-end would be greater. A recentpaper development testing had to be undertaken beforethe (Ref 69) from NASA Langley has quoted experience technique gave satisfactory, repeatable results, which Indicates that stingcorrections can amount Allowance has to be made for the pressureforce to 9-10%of total aircraft drag and can vary In a acting the Internal on fuselage surfaces or the non-linearfashion with Mach number. it is aft splitand,when the dummystingIspresent, for the therefore likelyto remain a majorproblem a as in pressure forceactingon the seal plate. Accurate pretsurised tunnel. determination these terms is vital; In an of example quotedIn Ref 15, the forces acting in the 9.2 strutInterference Low SpeedTunnels In drag direction were: For mostof the testing largelow speedtunnels In (I) force external on wetted surface: suchas the RAE S metretunnel, Fl tunnel Le the at CD - 0.0022, faugaand the DM tunnel, modelsare mounted the on (11) force Internal on fuselage surfacestiuts frombelowand the forces are measured an on CD - 0 0015, underfloor balance (Ref 70). Two different types (III) force sealplate: on of mountingare commonlyused: either a 3-strut Co - _0.0016. mounting with two underwing one t $l strutor a and single central strut. The Interference is Thus, the unasntedpressure forces are each of determined experimentally a similarfashion In to similar magnitude the actualrearfuselage to drag, that described above for sting Interference.In Thosecorrections obtainedby measuring are about otherwords,the model Is mountedIn a different S0 pressuresInside the fuselageand about 10 way, le eitheron a rear stingor on a strutfrom pressures the seal plate. Variousprecautions on above, and comparative tests are made with and have to be taken.the gap betweenthe forwardand without dumny replicas of the standard strut rear partsof the modelhas to be kept small,the supports Possible layoutsfor such tests are modelhas to be designed Inhibit to flow In and out shownIn Fig 29. for the 3-strut arrangement In and of thisgap and to givenar-uniformity In pressure Fig 29b for thv central strutscheme Such tests over the cross-section at the gap. the response to determine the interference can be laborious and characteristics on both sides of the pressure time-consumlng so, there Is a great Incentive and diaphragm the transducers In have to be carefully to findwhether these Interference corrections can matched and finally, has to be able to ove the be predicted a theoretical one by method This has led part of the dummystingb) meansof a small to such activityIn recentyears and It Is worth forward Integial actuatorto locate it correctly the Including briefsuemmary what has been learnt in a of bore. from these studies. Further detailsare to be foundinRefs 70-73 The success thistwin-.ting of technique depends on A typical testprogramme the 3-strut for case would 4 number basicassumptionsof be (a) one can Ignore the possible effects the rear (a) testswith the modelsupported a rear sting of on stingon the flowover the forward part of the but with ne representation the guards or of m.odel, struts. (b) one can Ignare th possible inorerence (b) testswith the model supported a rear sting on effects the twinstings the fIowover the of on In the presence of all three dum y guards Ong - at least to the extent, that these moMotedon the floor,but without any struts effects mightaffect the differunce betweenthe (Fig29s),
two tests.

with duany front together og one can calculate, by a panelmethod,the (c) tets onthe tmodel, on supported the stingIn the presence possible interfrstut effects f tra yoke osiingle ntw n e of all threedu.myguards mounted the floor on joining twin stings the rear, the at for these tests, the struts would be represented replicas the upper part of by of On a closelycoupledconfiguration such as that the realstruts,thesewouldbe hung from the shownIn Fig 27, assumptlons and (b) are open (a) wing and would terminatejust Inside the to question. Te technique practised the as In guards. past Is only viable if thereIs a fair lengthof uniform flow upstream and downstream the split (4d) of tests on the exposedstruts moutted on the and one cannot met this requiremnt with a underfloor balancewith the guardsmounted on configuration such as Fig 27b ARA are therefore the floorbet with no modelpresent, this test developing modified form of the technique as a serving establish basicstruttares to the Illustrated fig 28 The model Is stillmounted in on twin stingsbut now, forcesare to be measured Such a test programmse clearlyextensive is and of'the cusplete modelwith balances fittedIn the addedcomplexity arisesfrom the fact thatwhenever forward end of the pair of stings The balances the Incidence Is changed, It is necessary to will be calibrated Individually with the modeI readjust fitting eachduaeystrutand possibly and e in Installed the rig finite In elementanalysis was alter the fore-and-aft positions on the doamy
(W)

4-17 guards to avoid any contact betweenthe dummy will be more serious than with the 3-strut struts and guards;.hence, a multiplicity short arrangement. the strutIs circular, of If partof the runs are required, undersideof, the model will be exposed' an to interference flow fieldwhich.In principle, could A test programeas set out above recognises the be sensitive to changes in Reynolds number needto separate the- effectsof the struts and of according to whether the flow around the strut the guards. The near-field Interference of. the contains laminar turbulent separation. Such a or struts largely depends on viscous effects and Is an effect, greatly Increasing the strut not readilyamenable theoretical to calculations. Interference low Reynolds at number, has been found However, the far-field effects the guards. of which in the test range of the RAE 5 metre 'tunnel. are generally the more Important effects Experience suggests that the Interference depends numerically can be calculated panelmethods, by strongly the localgeometry on and is greatest for is where the underfuselage notably for a configurations This Is not easy: a typicalcalculation non-circular (Ref 73). Therecan be a significant 3-guard/model configuration couldneed approaching 4000 panels. However, as shown in Fig 30, interference with the aerodynamic lateral relatively good agreement with experiment can be characteristics; can be minimised reducing this by obtained to near the valueof C L at which the the strut diameter.Ideallyto 0.2 x fuselage up wing stalls,this appliesto the interference on diameter or less. both CL and CD A full panel calculation can therefore be succeasful but there Is still a need 10 BOUNDARY LAYER SIMtOATION SCALE AND EFFECT to find whether any simplermethod will give comparable results.Ref 71 presents sucha method. 10.1 The Need to FixTransition The standard practice in moot transonic and low speedtunnels operating Reynolds at numbers the In range up to R - 15 x 109 is to test with boundary layertransition fixedartificially near the wing leading edge and body nose. The case for adopting (i) an upwash due to the strut guarC thisapproach has been established many years. for displacement effectgiving a term of the Thereare two maIn reasonsform,ACL - constant, (i) allowing transition to occur naturally (i) a streameash, again due to the guard would mean that the transition position displacement, giving a term of the form, could vary with both C and Mach number, L ACL proportional CL to , Extrapolation the data to full scale of would be difficult unless the transition (1ii) an upwash Induced the effectsof the by positions at all test conditionswere trailing vortex wake from the strut determinedaccurately To date, this guards. This wake is associated wi,,, the would have been very laborious although side force Induced on the guards b. the there is now some hope that this nay be lifton themodel The ACL fromthisterm possible In the futurewith the use of is proportional the lift coefficient, to liquid crystals, CLon the guards. (il) It Is Important ensurethat, to as on the (lv) and finally, sidewashAnd streamnis. a full-scale aircraft, i is a turbulent effect again due to the guardside ferce. boundary layer that interacts with the In this case, ACL is proportional to the whack The need to avoid a laminar Cr product CL x CIC of transitional boundary layer interaction as was established long ago as 197 (Ref It follows that the total lift Interference Is of 74). A separated laminar boundary layer the form can reattachas a turbulent layer, thus giving spuriously optimistic results CL - KI 4 K2 CL . K3 CL' relative to those with a turbulent 1 boundary layer ahead of the shock where K and the upwash dependent contribution to I K are proportional to the wing Wlft-curve slope Exanples of misleadingresults obtained with 2 and the sign of K3 depends the wing s*eepbeing naturaltransition shownIn fig 31 The bucket on are negative for a swepthack wing and positive for a In the CD - M curve is not a genuine bucket, it Is septforward wing Fig 30 shows that the new due to transition moving aft on the wing upper Features thisanalysis, in viz the Introduction surface the localsupersonic of as region extends aft term (iv) and the empirical use of tie measured between - 0 72 and 0 7S. The retent ion of a high M lift-curve slope including its non-iirearity at lift-curve slopeup to beyond a - 2' with natural high C L produces reasonable agreement with transition is related to tie ability of a laminar experiment even at and beyond the stell This separation reesinas a closed to bubble and for the success, to quote from Ref 71, "isolds out tie boundery layer to reattach as a turbulent layer prospoct predicting leastsoe aspects the There is a danger that these results of at of Could have model support system* lift interference on sings been seriousiy misinterpreted
For some aerofoilis.

Ref 71 Is Illuminating In that It contains a detailed description the physical of natureof the interference, Four significanteffects are identified:

through the us& of fairly simple panel method calculations' One has to adult, however.that this simplified method cannot provide a sufficiently accurate predictionof the drag interference to the guards This is thought due to be due to the relativelylarge changes in interference over the area of the wing these effects cannot be averaged accurately in a simple fashion Also, there is significant viscous drag Interference due to the struts that has to be determined experimentally Turning to the central single strut mountIng arrangemnt, In general, there will be no side force on the support but, on the otherhand, the blocikage Interference effects of tie strut/guard

buckets in the CD - Mi curve have been found In transition fixed resaits.these sould have been genuine features of the aerofoil design but a bucket due to transition movementsin transition free results has to be dismissed as haing no relevance to th e full scale porfornance. Similarly, without a clear understanding of what can happen wit a lazinar boundary layer/shock Interaction one might have been tempted to treat the differences th liftcurves an exatpie in as of genuine scale effect. in fact, it is likely that the lift-curve slopeIn the transition-free results near a - 2 is higher than the value that would be obtained with transition near the leading edge at any Reynolds number

4-18

The general advice, therefore, Is to test with There are however some cases fixed transition. where this advice does not necessarily apply: (I) as noted later In 10.4, transition-free tests can be Included In test pregrames for diagnostic purposes, my be the transition-free tests appropriate choice If It Is known that, for reasons of either relatively high test i5 x 10), Reynolds number (say. R relatively high tunnel turbulence or simply adverse pressure gradients In the pressure distribution, transition will occur naturally rnar the leading edge, Is to cases where the test obective measure the hinge moents on a trailingedge control; for these, It may be Important to obtain the thinnest possible boundary layer over the control, tests on medels of aircraft designed to achieve extensive laminar flow In flight. For these, new model test techniques will have to be developed an discussed In 10.7.

(11)

transition at buffet-onset than in the cruise in tests at subsonic speeds. -Roughness height Is not the only significant ylraeter; the width of the roughness band and the density of particles In the band are also'important, The width of the banys Is The re4 ulred usually either 2.5 m- or 1.25 wi. roughness height to fix transition depends on the Interpretation the wind tunnel engineer places on Even a the phrase 6a sparse roughness band'. change In density from 4% to 16% can be -avea greater significant: the .4% band has to height to fix transition on a given wing at a given The desire to use a very sparse Reynolds number. band (to avoid a substantial drag penalty) appear to lead. In general, to a need to use a roughness height greater than suggested by the Braslow and Knox criterion. Traditionally, the roughness drag penalty has been predicted by a relation such as ACD - 2 t A/c where a is a magnification factor that can be eatimated by Ref 79, c Is the local wing chord and 40 Is the Increment In momentum thickness at the trip and Induced by the trip. However, recent trends In aerofoll and wing design are such that the effects of the trip should net be thought of The increase in simply as an Increase of drag. give a layer thickness can also boundary reduction In rear loading and hence, significant often, an Increase of wave drag for a given total lift. An approximate relation for the Increase In monentum thickness at the trip is l N Ar t CDR

(iii)

(Iv)

12.2 Methods for fIxine Transition The basic requirements are to fix transition with the mintmum disturbance to the flow and In a the In the l, consistent, repeatable mannr favoured method Is to apply a band of glass balls Thiese are preferred to known as ballotini. carbotundam because the) offer better control of Tha ballotinil balls are sieved roughness height and stuck to the model surface by blowing them lightly on to a tacky cement such as Araldite for consistency, 103/951 In the search alternatives to ballotlnl .ro favoured In certain quarters, eg (I) IOe Brough have used transfer characters devised for graphic w-rk (Ltraset) to produce regular transition strips, RAE have developod a technique in which a row of holes Is drilled In a tape at regular Intervals and the minute munds so formed provide a consistent distribution of roughess. Boeing% have devised a somewhat similar method %hereby a tape with a row of beies drilled at regular Intervals Is stuck to the wing surface and then an epoxy-ased filler such as Isopon Is spread over the tape; the surplus filler Is removed and the tape is lifted froa the sing leaving a ro* of xcrescenc s

(11)

here N Is the number of excrescences per unit the frortal area of individual area, Ar is excrescences and t Is the streamwise width of the CER Is the drag coefficient of transition trip each excrescence based on Its frontal area. Ther for an accurate available evidence Is little estimate or Cm but clearly, CDR - 1.0 Is an upper bound. On the assumption that CDR Is unlikely to vary rapidly with lach number, the values in Ref 80 The effects of the trip can then be can be used estaleted by mans of a CFD calculation Including Ad as an Input parameter The choice of a suitable chordwlse position for the transition trip lll be discussed later In 10 4 but, for the present, one can note that, to obtain Interaction layer/shok a turbulent boundary without any undesirable Interactions between the flow over the trip and the shock strength and position, the trip should always be at least 0 10c and preferably 0.15c ahead of the shock Another technique that has been used successfully (ef 81) in research exporiments is to Inject air Into the boundary layer In order to fix transition in a This is a much more elegant technique tso-dimensional test with on-line ienltoring of the data, It *1i1 always be possible to see whether one Is being successful In fixing transition. One does not have to be very precise as to how much air one uses. In contrast with tests with distributed rughness, the penalties of using more than the mlnimum required amount of air are trivial 103 Mthds for Dtetrmnns TraMLEi. lLLLLW

(III)

Various criteria are available to determine the Of these, the best required roughness height Inoan are those due to Braslow and Knox (Ref 75), Van Driest and Riuaer (Ref 76), Evans (Ref 77) and The Braslow and Potter and Whitfield (Ref 70) Knox criterion states that - 600 -k' is the Reynoldo nuxber based on the wh"e Rk' roughness height. k. and the floe conditions at the All the criteria forecast top of thi. roughness tht the required roughness height Increases with I 0, 33% at M 15.-201at M Mach number - by about - I 5 a,,d 80% at H - 2 for the Braslow and Knox criterion (with the values someshat depondent on This Is an Important point not Reynolds number) nre'y for tenting at supersonic spedst, It Is also the explanation why general exporlence has shown that one needs a greater roughness height to fix

The standard method in most tunnels of determining the transition position and of checking whether a been effective In fixing roughness band has transition Is by mns of a sublimation test with say, a 1011solution Of acenaphthone In Inhibisol. Closed circuit television is used to judge %hen the in areas shere the sublimate has evaporated boundary layer Is turbulent and photographs are

4-19 taken at regular Intervals. There Is some 1: Collection relevant of Infoation difference of view asuto whether one should spray the entire wing surface or merely the areas For example, Some elleve downstream of, the -roughness band. Whatare the alas of the test? that ,to spray ahead-of the band and In'the'bend reduces the effectiveness, of the band. This Whatare the important design and operat.ng conditions? Implies that lf the wing has been sprayed ahead of the band, one should accept some turbulent wedges What transit-ion Reynolds numbercan be downstream of the 'bend; otherwise, the roughness achieved the tunnel? in size that Is acceptedwill be largerthan that theoretical calculations on needed fix transition a cleanwing. If the 2: Preliwinary to wing is merely sprayed downstream of the band, a wheredrag Is ImportantCalculate by the most advanced theoretical method 'goodtrip'In a condition Is probably one thatgives somevery smellwedges conveniently available the wing pressure but In a buffet-onset condition, It my be distributions, boundary layer development and, If preferable to choose a band that leaves no wedges. possible, the wave drag and viscous drag at the One should always choose the smallest possible Important operating conditions. The general of aim an Is as heightthat meetsthe criterion agreed thesecalculations to give the testengineer roughness between the wind tunnel engineer and the customer earlyIdea of whether and wherethe flow over the for the testsand one should he consistent what model Is likelyto be subject scale-sensitive In to one accepts. viscouseffects, eg It may he difficult to apply a sublimation is therelikely be a rearseparation the to in technique In a pressurlsed tunnel because of the modeltests? time needed pressurise tunnel.A plotof CD to the Is therelikely be a laminar to separation near versus Reynolds number and a comparison of this the leading edgeaheadof any possible plotwith theoretical predictions willhe usefulIn transition tripposition? Identifying ranges In which the transition fixing has not bean fullyeffective cannot regarded These calculations but be will also provide guide to a as a complete answerto the problem. A decrease whereto locate forward a trip,eg it should not be with Reyn-Ids numberIn the excessprofile drag on placed at or inmediately behind the peak suction one part of a three-dimensional wing could, for position, and to the range of positions where a example, mask a failure to fix transition on trip would still meet the requirement mentioned another part of the wing. A visual aid for usa In earlier of heing 0 10 - 0.15c ahead of the shock pressurised and cryogenic tunnels Is therefore needed:hence, the Interestin liquid crystals.3: Initial datumtestswith forward fixed Other novel techniques are being explored. For and with freetransition example, Cartanherg Old Dominion at University (Ref 82) Is using an Infra-red Imaging system. Use of The alas of the test with a forward trip are such a system becomes difficult at low temperatures, partly because the differencein (a) to establish absolute the drag levels freefro. recovery temperatureIn laminar and turbulent any uncertainty due to an unknownlengthof regionsdecreases at low temperature and partly laminar flow becausethe sensitivity also decreases. However. thereseemsto be somehope thatthesedifficulties to compare (b) with the results the theoretical of can he overcomeby monitoring the reactionto a calculationsundertaken In step 2. any transient heat flux thisshouldbe more rapid In disagreemeont should he explored as it may the turbulent areas indicate the existence of a strong viscousinviscid Interaction, 10,4 Siemlation Methodology. PrRgAm Test (,.) to find how the shock position varieswith CL Complete simulationof the full-scaleboundary and uch n.cber and so to define what will he layer behaviour In a tunnel test at reduced possible In later tests with different Reynolds number Is obviously neverpossible All transition positions. thaton can hope to do Is to devisea methodology that will placethe testing and the extrapolation (d) to study the nature or the pressure to full scale on a sound wclentific hasis. ACARD distributions aft of the slcik to establish recognised this med and In 1984set up a Working whether any separation *hen presentextends Croup to review the subject and propose an rearward from the shock(class flow)or A appropriate mthodlogy This Croup reported it spreads forward from the trailing edge (class 8 1988 (Ref 8) It Is hoped that. its the future, flow), the distinction originally Introduced by increasing use will be made of the AGARD Pearcey (Ref 83) Particularly If It Is a methodology It Is based on the best of currenit class D flow, strong scale effects are to he practice and It represents an at-empt to use the expected and It Is o.tly recently that wind tunnel and CFR as partners In an Integrated theoretical methods hase been developed that approach The de3cription blow and In 10 5 and are capable dealing of with these(Ref84) 10.6 sumsaarlses the waSi.features, for further details and a background study of the subject, the Ibe test with free transition Is Included largely reader should consult t" compteh.tive treatment for diagnostic purposes for example. will show It In Rf 8 (I) whther a laminar boundary layer can be maintained backto the shock, There Is m.Dc more to a viscous simlatIon (11) ,tethodology than making declslons about whether, how and %here to fix transition the methodology as proposed requires action before, during and afterthe tests It contains six steps whether any rear separation observed In the transitton lxod test Is still present, so. it will not be possible if to avoid this at the model test Reynluds nuber.

4-20
(iII) z the furthest aft shock position that can be achieved at'the test Incidence and Mach number, the boundary layer thickness being less In a transItIon-free test than In any test with a trip. 4: In-depth study of viscous effects atrip S el Steps 1-3 have In a sense all boon preliminaries to step 4. The data taken In Step 4 will be the form the basis for the definitive data that wll prediction of the full-scale aircraft performance. In practice, steps 3 and 4 may frequently be combined In a single test programe In the tunnel. Step 4 will be described below, 5Itweroretatfon of the data after the tests a spurious hump In the drag polars as Illustrated In Fig 32b. The excess drag In these humps arises because of a local Interaction between the trip and the development of the-supercrltlcal flow as the shock passes over the trip as It moves downstream with Increasing Mach number. As shown In the upper picture In Fig 32b, theflow accelerates over the and a second shock Is furmed downstream of the trip; with a small furvher Increase In Mach number, the two supersonic regions combine to give a final shock wave that Is stronger and lies further downstream than If there was no Interference from the trip: hence, the extra spurious wave drag. Similarly, on the approach to the boundary OB, the forward movement of the shock Is arrested and the shock hesitates downstream of the trip and this hesitation can be recognIlsedby a slight Increase In lift-curve slope and, generally, a nose-down blIp In pltchl 3 moment. Ideally, one needt 3 or, If possible. 4 points on a transition sweep to establish a trend. This Is only possible In a relatively small part of the CL - M plane s shown In Fig 32c but fortunately, for a subsonic transport, this Includes the cruise conditions and most of the buffet-onset boundary. The shock pattern over a three-dimensIonal swept win$ Is likely to create further limitations near the wing tip and root but on a transport wing, this Is unlikely to be too troublesome because separation-onset generally occurs Just outboard of the Intersection of the 3-shock patters near aidsemi-span. The three-dimensional mature of the shock pattern will pose mote serious problems on a combat aircruft wing of moderate aspect ratio but the technique has still been practised successfully. The above description has bosn somewhat simplified. ag as noted earlier, trips of different height my have to be used In the cruise and near buffet-onset. UsIng a trip that Is effective (but only Just effective) In the cruise, will be excellent for drag but It liable to give a spuriously optimistic estimate of buffet-onset, while using a trip that Is adequate for buffetonset will give a pessimistic idea of the drag. Other qualifications are to be found In Ref S. .15 Simulation M4ethodology.Interpretation 2L IIgiDAU Fig 33 presents results from some two-dimensional serofoll tests soef 8s) shere It was possible to undertake both Reynolds nusber and transition sweeps The results eoreobtained before the ACARD Group wa set up bt they Illustrate the way in which results should be Interpreted It step 5 of the methodology A' Is a Reynolds-number sweep wlth transition near the leading edge: fixed artificially at R - 2,3 i 108 but occurring neturally ear the leading edge at higher Reynolds nsmbers Is i transitIon sweep at R - 2 3 x 0l: it appears that, by testing at R - 2 3 x 10 with transition at 0.30c, It Is possible to obtain results comparable with those that would be obtained with forward transit ion at about R - 8 x lOg This example suggests that all test results from both Reynolds number and transition sweeps shauld be plotted as In Fig 34 against either Reynolds number or an effective Reynolds number havIng found hw to convert transition position Into an effective Reynolds number One would not necessarlly choose CD as the 'simulation criterion' because the drag will Include a strong skin friction contribution and there Is no Intrinsic reason why Cf should vary with R In the sane manner as the wave drag and other scale-senltive parameters LookIng at tha past literature, one might be tempted to choose say. shack position, hat recent research suggests that shock strength or some function of the boundary layer over generally the wing upper surface my be a better choice Wefort discussing the major Issue as to how to

This step will be discuassedIn 10.5. 6: Extrapolation of the dat. to eredLct the full-scale verformance This step will be discussed In 10.6, Returning to step 4, there are two approaches to an In-depth study of the viscous elfects one can eIther, If possible, conduct Reynolds number sweeps with transition position held constant at the position forecast for flight at the full-scale Reynolds number or second, one can conduct a sweep through a range of transition positions at a given test Reynolds number. It will be reallsed that, In both cases, the tests are rosily a sweep through a range of boundary layer thicknesses. Mhnever possible, both types of sweep should be Included In the programme . There are limItations on the use of both approaches. Reynolds number sweps can only be accomplIshed satisfactorily In a variable ontslty (or variable temperature) tunnel. Admittedly, two models at different scale. og a complete and a half-model can be tested but, since It Is unlikely that one could obtain precise agreement between complete and half-model data at the same Reynolds number, one still needs to be able to vary stagnation pressure (or teaperature). le one uses the half-model to extend a trend as Indicated In fig 35 The range of transition positions that can be covered In a transition sweep Is limited by the ned (i) to ensure a turbulent boundary layer/shack Interaction, to avoid any local Interaction between the trip and the floe near the shack, to mInilise. as far as possible, aty serious disturbance to the superurllcel flow development over the forward part of the wing surface, to ensure that one can claim that there Is laminar flow upstream o the trip In all test conditions, otherwise. interpretation of the results will be laborious

(II)

(iII)

(iv)

These lIgItation Imply that with any one transition trip, one can obtain valid data In a 32 corridor (Fig s) between two boundaries, AA and RB. corresponding to the Mach number (AA) at which the shock wave moves 0 10 - 015C downstream of the trip as Mach number is Increased and second, the AcAch number at ohich the shack wave haa moved forward to 0,15c behind the trip under the Influence of a shock-Induced seperalon If the wing Is being pressure plotted, these haundarle can be determined easily but, even If only overall forces and moments are being measured, they can be detected witn fair certainty AA I*s Just beyond

4-21 convert transition position to REFF, the aim In plotting graphs such as that Illustrated dlagramneatlcallyin Fig 34 must be discussed Graphs such as Fig 34 are plotted as a prelude to the extrapolation to full scale Reynolds nmbers In step 6. The primary aim Is to compare the measured trends with the computed trends from the These preliminary calculations In step 2. calculations were rade by 'the most convenient method readily available'. At the tm the ACARD methodology was published, It was assumed that this phrase Implied that the calculations would not be able to allow for any form of strong viscousMethods (Ref 84) have Invlscld Interaction however now become available that are capable of a lImited separation near the trailing allowing for This does not Invalidate the maIn deduction edge from Fig 34 that below RcrIt, where there Is a major divergence between the measured and computed trends, It Is probable that a strong viscousInviscid Interaction Is present In the experiment, Extrapolation of the results to full scale has therefor, to be based on the measured trends up to Reit but can be based on computed trends above Ref 8 Identifies 5 simulation scenarios Rcfit. according to the relative values of Rfllght. Rerlt. In and the maximum R or REFF In the tests. practice, one Is most likely to encounter scenarios 3 and 4 which are defined by the relations: Scenario 3. Rzrit < axtimumR or REFp In tests < Rfllght Scenario 4: Maximi R or REFr In tests < Rcrlt < Rfllght Clearly. extraolation Is easier in scenario 3 because Rcrlt Is within the test range. Indeed. one could describe the vim of the aft-fixing technique as being an attempt to bring test data nhlch would otheraise be In scenario 4 Into the orbit of scenario 3 The greater certainty In the scenario 3 situation relati., to scenario 4 can be appreciated frca a study of Figs 55a.b Returning to the issue of how to convert a transtiln position Inti an effective Reynolds number, one must consider the nature of the scale effects that may exist Rlserar Introduced the concept of direct and Indirect scale elfects In wave drag. The fundamental Importasce of these Indirect effects suggests that an appropriate parameter on which to base the equivalence of a transition position and an effective Reynolds number would be the boundary layer displacement thickness at (or near) the trailing edge on the upper surface. This has been confirmed In research undertaken sInce the publication of the ACARD methodology (Ref 86) Results from this research are presented In Figs 36a,b,c. Tests had been made In the RAE (Bedford) 8 ft x 8 ft tunnel on a 14% thick aerofoll (RAE Tests were 5229) with appreciable rear camber. made at Reynolds numbers of R - 6 x 100, 10 x 10 and 20 x lOs with transition at 0.05c and the range of data was than extended by calculations for other The Reynolds ntbers and transition positions. first picture. Fig 36a shows (1) and good agreement between measured calculated results at R - 6.05 x 106 with XTR - 0.05 at M - 0.735, CL - 0.65. the design condition for the aerofoll, appreciable Indirect scale effect between and R - 30 x 106 with R - 6.5 x 1O transition at 0 05c, a reasonable but net perfect correlation between the c- ustedresults for R - 6 05 , x 104 xTR - 0 40c and R - 30 x 10. XR The significance of this result Is 0 0Sc that this correlation of XTR - 0.40c is what would have been predicted using the zero-level simulation criterion proposed In Ref 8 This criterion, the boundary layer momentum thickness on the equivalent flat plate. is often remarkably successful and, in this case, as in many others, it gives a good match as regards shock position However, bearing In mind that the criterion Is not related to the boundary layer development over the real wing. this most be somewhat coincidental It will b sees that it does not product close agreement in the rear loading

..

(ii)

(111)

Turning to the more soundly based criterion suggested above. Ie the boundary layer displacement thickness on the upper surface of the real serofoll, comparisons based on this criterion are (I) direct Reynolds number (or viscous) presented in figs 36b.c This criterion yields t effects arise as a result of changes In REf - 20 x 106 and 30 x 10 for x"R - 0 28c and the bundary layer (and wake) development 0 3)c respectively Fig 36b shows that in for a fixed or "frozen" pressure subcritical, attached flow, this criterion gives distribution xamples Include the perfect agreement In the pressure distributions and variation of skin friction *Ith Reynolds this Is maintained at the design condition, CL number and changes in the length of a 0 65, except in the supercritical region on the shodk-indutced separation bubble for a upper surface Analysis has confirmed that the given pressure-rise through the shock, and diflerent supercritical flow development can be it terms of the different boundary layer eoplained (ii) Irdleect Reynolds number (or vixcous) development In this region The shock save is effects associated with changes It, further forward and weker I the aft transition. prnssure distribution reaultintg from too Reynolds number rosal, then in tie forward changes with Reynolds nsmber In the transition, high Reynolds nmber distribution Fig boundary layer and wake development 36d shoss that no othr choice of xTR would have helped in giving agreeaneot in aave drag This The indirect effects are surprisingly Important In complicates the Interpretation of aft transition the context of scale effects on aircraft wings In results by just converting to an effective subcritical. attached flow, the only significant k*ynolds nuaber, one cannot equate with higher scale effects ( tvlng aside th, changes In skin keynolds number antsers, oe ls to Include a It is friction) are due to the changes in pressure correction to the measured save drag distribotion that fol:ow from the changes in Ioseer possible to calculate this correction In theoretically and there Is no doubt that this Is boundary layer displacement thickness In other words, the the correct ph)sical approach suporcritical fNow. these changes become icre significant as increase In Reynolds number procedure has to account for to distinct effects decreases the boundary layer displacemnt thickness a change In the viscous mvelopoent at the rear. and this leads to an increase in rear Ioadcg. a hich Is alloted for by the choice " xR and a reduction In the lift contribution that is needtd change is, suporcritical flow development which Is from the forward upper surface to achieve a given alloued for by a correction to CD.WAVE MAG total lift and uence. In many cases to a reduction

1I~
4-22 At higherCL, when the shock Is strongenoughto determining Rorit It shouldnot be takenas the measured is number which the extrapolated at a the approach to Reynolds Induce separation, Instinctive the computedtrend This position to an effective trendwould intersect convert a transition valuesfrom Reynolds number on the basis of obtaining a wouldimplyblindfaithIn the absolute separat bubbleof the saw. length. The bubble the computed ion results whichobviously wouldnot be length is a function of the boi.ndary layer momentum Justified.Rather, one should extrapolate curves (or displacement) thickness at the foot of the of, for example, skin friction near the trailing shock(Ref 87) There is no Intrinsic reason why edge againstREFF to find the valueof REF - at use of the bubble length a correlating as parameterwhichCf - 0 In practice, the relation set out should give the same relationshipbetween aboveshouldprobablybe reshapedif the results transition and effective Reynolds number as would are In scenario 4 so that terms (1) and (3) are be obtained with the boundarylayerdisplacement determined not for Rcritbut for the furthest aft thickness the trailing at edge He*ever. the transition In position the modeltest programe and in example discussed in Ref 86, perhaps a further correction has to be ncludedIn term coIncidentally, proved he true. this to (4). This extracorrection termhighlights the why extrapolation Is uncertainIn scenario4. The In the example quoted above, It was assumed that correction Is. In fact, an estimate of the anount when the transition pooition was noved aft at low by whichthe results the furthest at aft trAiItlon Reynolds number, It was moved aft on bothsurfaces position In the model tests are affected because Bearing in mind that the main effects are this value of REFF lies below Rcrlt In a very associatedwith changes In rear loading, the approximate fashion,this can be estimatedby ;eneralconclusion that If transition not observing difference Is is the between the slopes the of movedaft on the lowersurface(a practice adopted measured and computed trendsand allowingfor the in some test programmes), change in transitioneffectof this difference It would affect the the as position on the upper surface has to be extrapolation to Rcrit. up correspondingly greater Details of a nodified The computed trendsin CD with Reynolds number have criterion allowfor this pointare givenIn Ref to allowfor changes bothviscous to In and wave drag. 86 Refs 88-91shouldbe helpful. 10.6 Sloulatlon Methodoloty: Extranlastlon Procedure 10.7 Simulation Mehg.tbocy. Laminar FlowAircraft

There Is only one paper (Ref 92) In the open The general principle in the extrapolation literature addressing the particular problemsof procedure is that one should follow the measured obtaining wind tunnel data for aircraft designed to trends up to Rcrit and then the computed trends maintain extensive laminar flow. There are two from RcrItto Rflight As notedIn 10 5 above, typesof problem first, extensive laminar flowhas this Is much easierin scenario than In scenario to be achieved the tunnel 3 In sostaand second, one 4 To take C as an example, In scenario 3, the has to be able to forecast and simulate the fullD full-scale valueIs ohalned fron a simple relation scale transition movements with CL and Mach number of the fore In off-design conditions CD.filgit (1) s (2) - (3) . (4) The abilityto maintainextensive laminar flow In the model tests on a suitably designed shape where depends on (i) - measured valuefor transition position that (I) tie tunnel flow havinga low turbulence converts o Rcrit. level, say, les than 0 13, R valuefor flight and transition (2) - computed position. as (3) - compued value for test conditions In (I), (il) the tunnel being a quiet tunnel with the (4) - the sues ion of a series corrections t or for valueof Cpirs ideally it or less a 0 certaini) no marethan 8, This depends 0 (a) excrescences theaircraft on but not on more than Just the design of the represented the eodel. on tuntl, there Is evidence(Refs36, 92) affects not represented a In that It can e adversely affected tie by (b) propulsion normal complete model teas, presence of the model support. (c) trimmsing effects I flight. the modelclearof effects the (ff) on the successIn keeping (d) differences aeroelasstic in aircraft and the model contalnation frogsite Impact particles of in the flow The allowable roughness Reynolds is addition allowance has to be made for the height is hated on a roughness number defined by the height of the increase is drag and lois In rear loading due to the presence of the roughness band Th:s can e roughness and the flow conditions at the done by either allowingfor it In (3) or as a top of the roughness Critical values of or correction (I) to about600 for three-ditlensionl10 for two-dimnslonal disturbances are usually It shouldbe notedthat the conversion fromXTR to quoted but experience suggests that R plays no part in the final prediction or ite semeshat larger values can be tolerated full-scale value, it has merelybeen used In the near site leading edge,presumably because procedureto identify that the results are in of favourablepressure gradients To scenario 3 The. is Just another way 'f Poin1ing achieve these standards, It may be out the advantages of using art fixing to bring the necessary filter to the flos. results Into scenario 3 Ir nevertheless the results are still In scenrlio 4, the extrapolation (Iv) the success Inserting in pressure holes in ia necessarily more uncertain The measured trends the model that do not trigger transition have to be corrected for tite fact that, with (Ref93) forward transition at R - RfF. the supercrltlal fio dcvelopmnt and hence,the wave drag would be Ref 92 by Elsenaar contains a detailed discussion different ard then.the trends are extrapolated to of how natural transition likely varywith CL Ia to Rcritand thismeansthat the conversion IF Is and U at both flight and model test Reynolds to involved in the calculationto ploduce the nuabers. assuming tlitthe points In the preceding full-scale value It controls the slope of the paragrapit have been negotiated successfully measured trend There are also problems in forecasting the natural transit ion position is

4-23 generally undertaken ltoing e method the but there is stillSr at uncertainty overwhat valueof N to choose, if the aims of the preceding paragraph have been met, naturaltransition likelyto be Is furrFier aft in the tunnel than In flight both at the design condition and at strongly off-design conditions, thus allowing one possibly to use noratl tripping techniques. the Intermediate In range Of CL, however,the forwardmovementof to with CL is likely be delayedand to transition occur core abrup:ly on the model than In flight, If these comparative movementscan be predicted, may aft tripping havea part to play In givinggood simulation
N

Half-odels have frequently been used for tests with powerednacelles where one is merely seeking the drag Increment due to the nacelle Installation. Experience shownthatsuchtests, has even on wingmounted nacelles, should be made with a metric someearlyreports half-fuselage Thiscontradicts on the subject. above, It seems Despitethe problems discussed will continueto be used likelythat half-models for

(I) tests for determining differences in drag two configurations, between nacelles, (11) highspeedtestswithpowered In not discussed Ref 92 arisesfrom (ill) tests aimed at achievingthe highest One majorissue possible test Reynolds number and, In flow the fact that In flight with a laalinar extending the range of a particular, by Is to transition likely be triggered a aircraft, Reynolds somber sweep. interaction. Little boundary/shock laminar research has been undertaken to determinethe TEST ES: 12 PEOPTSIO INSTALLATION TECHlOIL scaling lawsfor such Interactions
SUBSONICTRANISPORTS

HALF-5M)DEL TETING. SPECIAL PROBLEM 12.1 Jet Turbine Enelnes 12,11 Tyes of simulator The ulscussion thispar& Is takenfrom Rer 10 In Different types of simulator are shown diagrasmatically fig 37 in

There have been several referencesIn earlier of to practice lestinglarge paragraphs the growing half-models subsonictransport of aircraft a. a meansof obtaining higher test Reynoldsnumber a accentuates problems that (RefII). This practice have to be considered all testing. In (I)

asywentrically Most completemodelsare fitted in with through-flow the lsrgemodelmounted Is to (TFNl) see top picturein Fig 37. This the workingsection likely have more nacelles They can effect the steadiness the airstveam, Is the simplest form of simulator. on of providethe correct inletgeometry and the correct (11) the sall Interference corrections will be mass flow ratio If the exhaust geometry Is more difficultto predict, particularly enlarged. Alterustlvly, If It Is felt important for tunnels with ventilated wallsbecause, to retainthe corlectexit geometry, one has to Inletmass flowand It may thenbe one wallwill have accepta reduced for half-model testing, to be solidto act as a reflection piame, necessary to modify the Inlet geometry to avoid spillage drag the- would not be present, full (111) buoyancy effects - both empty tunnel scale With an usolerwing nacelle.It Is probably buoyancyand In perforated-wall tunnels. Important not to modify the exit geometry for the blockage buoyancy effects will be larger sakeof obtai cKthe correct Interference with the and more difficultto predict without flowover the wing extensive pressure plotting the actual in of The weakness a TEN Is that the exhaustflow is tests. unrepresentativ In both total presu-e and (iv) In many tunnels. the standardof flow temperature consequently exhaust and In plumeshape angle uniformity Is mot as good near the and stream shears Attempts to Include a hard wallsas In the centreof the stream(see shapedextension the nacelleto represent to the 7 I) correct plune shape have not generally been the TFt should successful Whenever possible. there are also problems directly associated with retain the geometry of the separate fan and core the miod, of testing (Ref 94) There was stream cxlts considerable activityin the 1960s to solve the The generalpracticeIn high speedtestingis to problems the root by sealing schemesbut correctfor the lackof Jet effects with a 1FN by at lealage tests on a corresponding comparative approach undertaking thesewere oftenunsuccessfulThe normal is to mount the modelwith the aircraft centre-li e half-modl ith both a IlN and one of the powered below described away from the simulators wall but displaced ot at tie tunnel wall by a distance equalto aboutthe wallboundary A blownnacelle has both practical advantages and layerdisplacement thickness Opinions differ as practical disadvantages. the one hand, since On to ubether one should measure the forces on the all the air that exhausts from the exits has been extra piece of fuselageInsertedto create this fed Into the aodel, the Inst.tsmntation displacement Whatever one does in this respect, requirements simplified are but,ot te otherhand. it It unlikely that tile meaturedlift-curve slope the air requirements be verygreat,ag perhaps can will agree with %hat would be masured for the three times those for a turbine powered simulator corresponding complete Idel There are too many (lP$) Also,problems arisefromthe fact thatthe reasons for satll differences eg the tunnel wall exhausttotalpressure bothprimary of and fan are may met ant as a fullyeffective reflection plane, very low in comparison withthe supply pressure (by there wIll he a reduction in the dynamic pressure perhaps I 15) Complex pressure-dropping systems close to the wall and. in somecases,theremy be can lead to non-unifor pressure and teeerature some leakage between the model and the wall Most distributions The apparent advantage of a blown tunnel operators adjust the lift-curve slope to nacelle to permit oer-blowing for calibration agree with that mleasuredwith the complete model purposes, to provide is the correct flight pressure and have developedsel-empirical metbods for ratios at the mzzles In a static calibration, my modifyingthese corrections for the next model also proveto e an illusion It is dangerous to havinga differnt geometry Clearly. this Is not assume that the flow distribution the nozzle it a fullysatisfactory approach It mroally leaves remains the same in ,hesooverblown renditions and a residual errorin pitching moment amounting 0 01 also.thisappro..h to requires that the dependence of - 0 02c In aerodynaic centreposition mozzle thrust and discharge coefficient internal on can be quantified ReynoIds number

4-24 There are also practical aerodynamic difficultiesthe tank, thus enabling the upstream nacelle In the use of blownnacelles.A shapehas to be stagnation pressure and the nozzlestaticpressure designedfor the fairingover the front of Che of the tunnel tests to be reproducedin the nacelleand the flow over this shapehas to be calibration whilstmaintaining quiescent conditions representative over a reasonable rangeof CL and in the flowsaroundthe Inlet and downstream the of Mach number. The displaced intake streaatubenozzles. The aim Is to calibrate grossthrust the entersthe gullybetween wing (or fuselage) the and and inlet=aso flow In termsof the sane reference nacelleand the effectsof this are difficult to pressures and temperatures will be used In the as quantify. tunnel tests and to use the Internal Instrumentation calculate net thrust to the and ram The greatadvantage an ejectornacelleover a drag. of directblownnacelleis that It requires much less high pressure aIr. Assuming sMat one can achieve The original NSTI was designed for high flow rates an ejector mass ratioof about 1.5,the inletflow and relatively large models; It Incorporates with an ejectornacalleshould about60 to 65% critical be venturi monitoring mass flow ratesand of of the design operating value. The real challengetwo slx-component balances. ST2 was developed to with an ejector nacelleIs to obtaina consistent. neat a need for greater precision for the smaller repeatableflow at the nozzle Instrutmentation models; it uses single axial component force reference plane, This reference plane Is always balances The aim of this MST2 designwere to likelyto be nearer the ejectorplane than the obtain rules of the complete mixing would allow. Proponents the ejectornacelle of wouldclaimthat (I) force measurements of 00.1 lbf thisdistance nevertheless is acceptable others but repeatability, and believe that It Is too close to guarantee (i1) enhancedmixingof the TPS exhaustflow repeatability The accuracy of an ejector ahead of the tank ames flow simulator dependent the repeatability the Is on of Instrumentation flow fros a multitudeof minute condl ejector nozzlesand their mixing with a distorted flow To produce enhanced mixing,the flow for UST2 Is field, first extractedfrom the tank Into an annular mixer/plenum priorto flowing alonga high velocity Finally, turbinepoweredsimulators (TPS). these feedduct and Intoa further mixerat entryto the have been used extensively.Som establlshmants. '-mas flow,plenum, shownIn fig 38. The model as notablyARA. tiNW and OI4ERA have acquired large axis is vertical a and threesingle component Bofors amountof expertise theiruse. Initially. In In 'ahearforce'high precision load cell are used some quarters, therewere some doubts abouttheir for the measurements. speciallayout metric A of use in view of the largenumber rotating of parts and non-metric components featuring annularcells but in practice the units.designed and built by was devisedto compensaste the pressurearea for Tech Developen: Inc. have proved to be very tera arising the modelmounting on zonedue to the robust Most TPS units are associated with a basic tank external to internal differential particular full-scaleengine and appropriate pressure The tank top is equipped with a novel cladding is manufactured In the testing arrangement rolling of diaphragm seals specially establilhment or by the customer to suit a manufactured the patentholders, by Bellofram Inc. particular Installation Technically, TIS units Flat diaphragm sealstried Initially not give did have several distino advantages. t eg the required accuracy The overall uncompensated load on the basic metric area at 6 psid Is (I) both Inlet and exhausteffectscan be approxlmately 1200 lbf but the arrangement of adequately represented the sam test compensating In cellsreduces the net metric load to A typicalfigurefor the intakeflow Is less than IS lbf at 6 psid The RDS give an 50% of fullscale, essnlally linearresponse with ... hysteresis and good repeatability Calibrations with exlernal (11) a linked accounting system can be used to loadsup so I0 lbf and with 6 paid differential estimate the ram drag and the gross have produced less than tO 05 IbF forces data thrust This alnlasea he potential for spread, largeerrors be present a reault so as of having to subtracttwo relatively large Repoated use of MST2 has shownthat it Is possible termsIn obtaining external the drag. to achieve 1s accuracy boththrustand mass tO In flow calibrations This is a notable achievement Crest care has nevertheless be taken in the but. bearing mind that typically, drag,fan to in ram calibration of the units with much diagnosticgrossthrust and coregrossthrust are respectively Instrumentation seek out faults. These 300. 500 and 230 counts compared to with a nacelle procedures discussed the next para are In externaldrag Increont of say. 20 counts. this and is standard accuracy repeatability necessary of One detailed feature worthnotingIs thatuse of an If external drag differences are to be epoxybasedfibreor glassclothlaminate material discriminated the accuracy to discussed 3 In (Tufnol) provided acceptable baa an solution the to problemsof Ice formations due to the very low Detailed monitoringtechniques have had to be turbine exhaust temperatures daveloped ensure the safetyof the TPS units to during the testsand to be ble to diagnosethe ILLL2CallbrII Lacn chnlcusg sources any apparent of Inconsistency the data In One particular feature the reduction the data of of All siamulators with their claddinghave to be i that they are 'poser corrected' Ref 10 calibrated tanksthat resemble altitude In the test containsan examplesheft the measurements, then chaubersthatare used for the full-scale engin e sampledat a finite set of duct locations, to led Boeings were the first to develop such a nozzle coefficients *hich shoved apparent calibration tank but theynow existat uanysites variations with rpm at fixed values of the fan eg C0c, OIRA, M88 (Brwen). NASA Aes and ARA nozzle pressure ratio The 'powercorrections' Th discussion belowIs basedon a description approach Is based on the assumption of that these the facilities ARA. (Ref10) at apparent rpm effects the nozzlecoefficients on are due sampling variations opposed to real =o an The Mach Simulation Tanks (MST)at ARA are shen effects Ref 10 describes methodfor correcting a diagiamrstically in Fig 38 It will be seen that for those apparent poer effects and, In the %he unitsare mounted partly In and partlyout of exaspie discussed the spread of the nozzle

4-25 coefficients at different rpm at a given fan nozzle pressure ratio was reduced from about 0.7% to less than 0.3%. This was a cast where a large nuber of doct pressure and temperature samples were taken. The published literature (Refa 95,96) contains examples with fewer samples where the apparent power effects before correction were as great as 3%. Mach Simulation Tanks can, of course, be used and are used for the calibration for a forms of simulator includingTFs. The calibrations of TFPs are undertaken to determine the Internal drag correctilons. 12.1.3 Accountinn technioues truncated afterbody but primrily. It Is a rig for refining the shape of the forecowl and for checking that no avoidable spillage or wave drag Is present in the Important operating conditions. It can also be used for designing the modified shapes of cladding to use with powered simulators with limitations on the maximum available mass flow. It Is not suitable for studies on afterbody shape because of the effect of the downstream support mountIng. The second picture shows a rig that can be used for checking the performance of the afterbody/nozzle design. It Is a two-stream strut-mounted rig. The rig has been used extensively for tests on axisymmetric nozzles, the effects of a non-metric wing panel on nozzle performance and full nacelle/pylon configurations. Som typical test results are prsented In Ref 10.

Fig 39 shows schematically the essentials of the MST and In-tunnel bookkeeping process. The calibration phase yields a set of nozzle coefficients which represent the characteristics of The main test sequence Is that Illustrated In the the nozzle and Instrumentation combination. it is three pictures on the right The general practice important to recognise that the nozzle coefficients Is to create a test programe including tests on. will change If the Instrumentatlon Is changed. It Is essential, therefore, that the Instrumentation (a) TFN/pylons mounted on a long strut. These remains the same In the tunnel as In the teats, taken In conjunction with evidente from calibration The ram drag and the fan gross thrust the Isolated forecowl rig, will reveal whether are computed using an Identical mas flow term. there Is any nacelle/pylon Interference. Care Multiple methods are used to estimate the fan has to be taken to avoid or at least, allow for nozzle mass flow; comparison of the results bulds any buoyancy effects with respect to the up confidence In the results and helps in non-metric part of the strut, fault-finding (b) TPS powered nacelles/pylon combinations again It Is Important that the thrust/drag bookkeeping on the long strut Results from these tests sche m Is defined clearly and agreed This applies will form a datum for the later test data from even In the simplest case of tests with a throughthe Installed tests but also, comparison of flow nacelle Several different definitions for results from (a) and (b) will be of Interest as the Internal drag of such nacelles are In use an Indication of jet effects !n a free-stream according to whether one Interprets it in terms of environment. The comparison also serves to the change of momentut from upstream to downstream confirm whether the design of the forecowl for Infinity or morely to the duct exit. Any the TPS unit is satisfactory Finally. deflni'ion can be used provided that other terms in carrying out these tests first will man that the full thrust/drag scheme are adjusted to suit one arrives at the start of the installed programe with added confidence, Statistical analysis techniques are applied to the results These techniques are discussed In detail (c) a half-model fitted with alternatively TFNs and In Ref 10. IPS units and on a corresponding complete model with 7T"s The data from these tests can be 12.1 4 Prooolsion Inteeration high sind used in several ways (I) comparison of results for different test DroLrgmf builds of the TPS nacelles on the A number of different models and rigs have to be half-model will indicate how to used in a typical test prograerse to study and optimise the complete Installation optimise the propulsion Installation on a subsonic bearing in mind that all effects are transport The description below Is again based on represented In these tests. the facilities at ARA but similar approaches would be followed elsewhere (11) subtraction of the results alIth the TIN and IPS nacelles on the halfThe range of possible rigs and models Is model will yield corrections that can Illustrated In Fig 40. The pictures on the left be applied to the data from the show two special rigs for Isolated component complete model tests to allow for the testing on respectively the external cowl shape and jet effects not present with the second, the nozzle and afterbody lhe three TINs. pictures on the right shoe a nacelle/pylon model being tested In Isolation at the top of a long (ili) suboraction of the results with the swept strut, on a half-model and on a coaplete TPS units on the half-model and those model In the last three pictures the nacelle can. obtined In (hi will provide an Idea In princiOle, be any of the simulator types of the aerodynamic Interference described above present is the total Installation and The coal models tested on the Isolated rig shown in hence of the Improvements that may. tha left top picture are appreciably larger than lit theory, be possible. In this those used for the installed teeta. Thia enhibla connection, it should We noted that the tests to be made at a higher Reynolds number zero Interference is not necessarily and also, the models can be a more feithfei the beat that can be achieved: representation of the full-scale nacelle Imluding favourable interference Is a real asymmetries sch as Intake droop and any exters;al psslbillty Fig 41 Is an exaaple bulges to house accessories on tie full-scale taken fr Ref 9 Tbe aim should be engine The external drag is obtained from to design a propulsion installation pressures measured on a rotating raks aft of the and wing that together give optlmu nacelle and the mass flow Is deteralnd from performance at Ia almost axiomatic pressures measured on raes rotating in the dwcts that this Irllies that this Tests can be made on a complete cowl with a performance Is better than the sum of

4-26 the clean wing and nacelle cell as with the jet engine. This Introduces Installation performances when engineering problems in the design of rotating determinedseparately. This is balances and, with modern propeller/spinner obviousIn the case of aft-mounteddesigns, thereare difficulties separating In drag nacelles where the presence of the and thrust and, as regards propeller efficiency, and experiment In comparing of nacelles major Influenceon can thereare difficulties have a downstream the wing the theory because it Is often not practicable to position of the shock waves on the measure what one can calculate. Theseadditional wing hut it Is also trueof underwing issues are considered some detail In below. nacelle sntallations. Ref 97 identifies differentsources of In the UK, Industry S and Government cooperated in interference such Installations for the development new facilities of (Refs102-105) at high speed and Ref 10 adds a for modeltesting withpropellers' special a Test further4 sourcesImportant low Housewas built at DA for proving at trialson the speeds. modelpropellers aheadof the tunnel tests,three compactand powerful *lectri motors for testing Even with half-models, high pressure the air feed relatively largemodelpropellers (up to 3 ft in to a TPS simulator has to be taken throughthe diAmster for single rotating propellers) were balanceoutside the tunnel wall but the associatedacquired and have been used in tests in both the problems not as serious theywouldbe If the ARA and MW tunnels, are as air motorswere obtainedfor TIPSunits had been Installed in a complete model, tests on aircraft models with both single and Such tests are however feasible as was shown In a contra-rotating propellers finally, linerwas and a resoarch experlmnt on a 2-engined Lockheed lOll designed to create an 8 ft x 7 ft acoustically model withoutits rear engine. An air transfer treatedworking section for the ARA transonic bellows system was designed successfully and the tunnel. For convenience, the author will use the comparative testdata for two configurations showed ARA experience comment the problem mentioned to on reasonable agreement with flight data (RefI0). At abovebut It is, of course, recognised thatvarious low speeds. it Is howver more important to use other establishments have tested model propellers coplete models and, by now, DKl have acquired either in Isolation on aircraft or models, the eg considerableexperience In this field (Refs de Havilland (Canada) 30 ft x 30 ft low speed 98-101) tunnel (Ra 106). the ONERASI tunnel (Ref 107), The twa complete model testshave been meade with the NASALewisfacilities (Ref 108),the facilities two types of large TIPS unit, one designed and at United Technologies ( Rer 107). NASAAmesand one should Douglas (Ref109) and finally, manufactured by TDi and one by ND, The aim of McDonnell an wo were the firstto develop mentionNoelngs, the testshavebeen' acousticworking sectionfor a transonic tunnel (I) to establish the jet interference effects (Rer110). on the drag In the secondsegmnt climb. the drag Is needed to an accuracy of tl Propeller testsat ARA can be mode at two different aircraft dragor about1i0 drag counts, mdel scales. The engineering and aerodynamic problems can be Illustrated a briefdiscussion by (11) to determinethe jet effects on the of the hardwareand typicaltest progranes for stability and controlcharacteristics testson contra-rotating in propellers the smaller at ground effect during tak-off, scale (111) to investigate the trrust reverser characteristics Including braking capacity, handling qualities and relngest boundaries, ion and to studythe nature the Interference of by mans of surfare pressures and flow field measurements (I) fig 42a shows the rig that has bees developed obtain the basicpropeller to thrustIn the presence the realspinner of shapebut followed an unrepresentative by nacelle, being merely the miniaras body requiredto house the 101700 air motor This combination Is mounted on a metric struthoused in a non-metric shield The aIm of the design oere to provide sufficient rigidity coupled with aerodynamic cleanliness and as sual a tare drag as possible on the underfloor balance readings the achievement of a satisfactory comprostise between these conflicting requirements was a far froa easy task The propellers cleared are for tunnel testing by monitoring the output of blademounted straingauges%hichprovide a mans of assessing rotor dynamic disturbances, critical points and flutter Also, the rig Itself is fitted with accelerometers.Ihe compressed air for the air mtor enters the metricmodelIn a dire .tion perpendicular the thrustaxis to hut It is still necessary insert the to naceIle/nzz ee system in the Iach SimulationTank discussed earlier to determine the notee thrust and discharge coefficients a quiescent In environent, fig 42b showsthe hub design scheme, the attachments to both the shafts and the componentpropellerswere designed to produce sysuntric a low-ttress torque and thrustpath to the balanceflexures The balance electrical signals were transmitted by slipring for the front rotorand by telemetry the rearrotor. for

(Iv)

The crucialengineering problemin the complete model testing with IPS unlhiis iw to bring the high pressure air scross the balance In th fuselage The air pipe aust be flexible to mininlse interactions with the blance measurements but stiffenough to malntainits positionIn the modeland to withstand the high pressure of about 40 bar The MW design Is shon In Ref 98. there are two air bridges,one on each side of the balance, in the models to allow for independent control to engines of

IL2.tr

Milu

tharehas been considerable Investment recent in years In new facilities and techniquesfor the testing of modelpropellers and of asrorwit models fitted with propellers both low and hgh speed in tunnoeIs ilany of the isn,,es discussed above in connection with aircraftwith jet engi-s still Apply but thre are none additional problems Ill) First, model tests are needed on the propeller It.lf to determine its performnce ar1 asroacoustic characteristics and second, in the testson aircraft models,the propeller thrust has to be measured directly te wind tunnelrather In than Irdirectly mans of reference by prstrates and temperatures alliedwith a calibration a test In

4-27 (1ll) the shaft strain gauged balancedesign, In a Fig 42c, comprises set of flexures a layoutwith strain basically symmetrical gauges placedso as to providesymmetry (ii) of and duplication the key elements. A of repeatability 1% was achievedIn a calibration, static the development of the telemetry to provided be a challenging(ill) Installation task: the high CF loads dictated a configured housing with specially rings. After various coposite retention structural and electrical refinemients, a i%1 standard of signal processingwas obtained, may and methodology the test programme but. In fact, the appear complicated as procedure set out at the bookkeeping for bottom has been somewhatsimplified thispresentation. betweenthe contextare the differences curves, various differencebetween the the substantial In blade and net thrusts the theoretical of tl'e importance the resultsconfiro.s on the effectson the pressures blade spinner, and betweenthe apparent the differences net thrusts Is significant In the minor In the but relatively experiment calculation. The most obvious explanation of this Inconsistencyis that the calculations do net include the skin elementIn the effectson the friction nacelle but It ray alsobe due to detalied differences, blades-on versus blades-off gap. More the spinner In the flowthrough it prosaically, may merely reflectthat the theory has its limitations. These In compalsons have been discussed som that, at the detail merely to emphasise time, one is in some difficulty present trying to use theory to settle the experimental In uncertainties whether data have bean handledcorrectly. The may wish to know the designer propeller of efficiency his designto I% or better eitherby this Ia difficult but at present experimentor theory' It Is another example where comparativeaccuracy Is better than absoluteaccuracy but even accuracy is likely to be comparative issues sensitive to the Interference between propeller and spinnerdiscussed Is above. More research needed.

(Iv)

(v)

For the aircraft nodal tests, the precise definitionof propellerthrust Is unimportant provided one retains consistencythrough the accountingin Fig 42d but. In tests on the A (phase In Fig 42d).aimed In propeller Isolation thrustand efficiency the at determining propeller the prediction, and comparingwith theoretical becomecrucial As we have seen, In definitions are measurements made with botha the experiments, balance The difference shaft and an undsrfloor shaftthust and blades-off between the blades-on thrust The net yield%an apparent uesurenents thrust can be obtained In two ways either by and blades-off the blades-sn differencing underfloor balance thrusts or by correcting the above related to an apparent thrust for the difference. blades-on and The model tests discussed Similar blades-off,In the rear nacelle axial force aircraft with wing*mounted engines has been undertaken by McDonnell Douglas obtained by integrating the pressures measured on activity can In connectionwith a design with aft-mounted that this Integration nacelle Assitsing this Tests on powered complete to the requiredaccuracyand that engines(Ref 109) be performed drag of the modelsat both lowand highspeedsweremade For Is allowance made for the skin friction the nat the high speed model, the high pressureair was nacelle,these two methodsof obtaining routed forwardalong a sting locatedbelow the givethe sameresult should thrust Into the forward modeland then up a bladesurport to the fuselage the fuselage and then backthrough of an efficient elementin the design An essential for operation at high powerednacelleat the rear A contra-rotstlng propeller installation was simulated Poer-Induced of the flow pusher arrangement speeds Is the slowingdown subsonic were powerand overal stability by of over the thickrootsections the propeller a effectson taill total and also,a n clle-based rotating surfaceopposite ad ahead matsured wa stlngof the spinner (seeFig 42b) At the very least, pressure wake rake *as used to explore the propeller of tite thrust slipstream of estimate the propeller any theoretical should allow for this effect but and efficiency propeller witit rig tls Ix net enough to produce a theoretical The finalpicture, 44, concerned with either testing, shout the large scale tV rig with a directly that can be compared estimate iln Installed Installation propeller Any contra-rotating the apparent or net thrusts from experiment tite acoustic working section of the ARAtunnel calculations by strip theorystill nd refinement by allowln for tie effects of the pressures on the Points to note Include Inner surfaces of the blades adjacent to the propellers are driven by two AC Induction spinner and for ith pressures Induced on thit (a) tite These electric motors mounted In tande 7batse eta spitter by the presence of the blades
obviously part of the apparent thrust from the motors are designed and manufactured by the

of Able Corporation the US and are eachcapable suggestthatthey and results calculations measured of generating 660 SHP at 7000 rpa. they are All can accountfor 5-10% of the final answer In verycompact eachcan be contained a volune suggestthat a field method these complications 25- longby 14' diameter. with experiment should be used for a comparison usingthe Denton The resultsof such A comparison the of system threeshaftsconnects fieldmetbodare shown in Fig 43 taken from Ref (b) a co-axial rotors and, in this case, stationarynose Variouscomuentscan be made about this ill to bullet the Ablemotors. comparison (t) of primaryinstrumentation the rig was a of ovarpredlction the theust (c) Sits the apparent torque shaft-mcounted thrust and pair of at a given blade to balanes and a set of surfacepressures issue anglemay not be the most Important to enablethe rotorperformance be evaluated, It shouldperhapsbe thoughtof as being rig and iotor"health' coupledwith extensive in due to a discrepancy bladeangleeither Instro.ntation including strain gauges on the because of an undetected blade twist in blades, effects viscous or te soperimnts because Of In been ignored the calculatoss tave more direct Interest In the v-esent
and poser coefficients

4-28 (d) the larger size of this rig carries many The standard approach Is to test a stite of at advantages- wore acceptable a Reynolds number leastthreemodels: on the blades and the abilityto attempta wider rangeof tests, eg dynamicand steady (a) a normalcomplete model of the aircraft with pressure measurements with transducers flush the best possible representation the intake of mountedon the blades and a lasersystem to through-flow the leastpossible and distortion monitor the twist and camber of the blades of the rear fuselage, underload. (b) a partial modelof the Intake a larger to scale (e) despitethe relatively largeblockageof the and Including faithful a representation the of modeland the microphone traversing In the rig Intake ductsback-o the engineface,and tunnel with the acoustic liner, tests were possible to M - 0 9 up In passing,it is (c) a partial modelof the trueafterbody shapeof worth mentioningthat Ref 112 is a useful the real aircraft Including the facilityto reference the effect propeller on of thrust on blow the exit nozzles at representative jet tunnel wall Interference presureratios. The distorted afterbody shape of model (a) Is also tested this afterbody on In the performance tests with this large rig, model and the differences between thatcan the results corrections for the two shapes provide supplementary tests were made with beth laser anemometry and laserholography. For the laser be applied testdata from(a). to anemonetry, flow the was seeded plumes 0.5 pm by of particles mineraloil Injected of from a special Clearly, configurations become as have more closely struterectedIn the tunnel settling chamber Two coupled, doubtsarise about the validity of this pairsof focussed beamswere used,the orientationthree-pronged approach As a smallmove to meet and meeting pointsof thesebeamscould be rotated thia point, both intakeand afterbody models now and translatedto enable velocities to be ten Include relevant parts of the rest of the determined different In regions the flow field, aircraft whatever of to extentIs permitted model by eg the tip vortexregion, the In-rotor between blockage (ie considerations. However, with a closely blades) region, the between-rotors region and coupled layout It may be wrong to assume that can be studied separately and jet effects of upstreamand downstream the rotors. In the intake holographic tests, features that were observed Included the tip vortexpath,the chopping the The techniqueswill be discussed under two of frontvortex the rear rotor,the shearlayerin headings. Intake testing, afterbody testing by a of the wake region and supercritlcal flow features Including verybriefdescription the prospects such us the expelled bladeleading edge bow shock for fullsimulation the propulsion of effects The measurements testsare decribedin detail and in Refs 105 and 113 This wag a majorexercise run 13.1 Testson lIntakModels on a cooperative basis with Rolls-Royce the as major partner The brief description has been 13 1 1 Scour flitake of pets Includedhere since this programe is a good of In exampleof %hat Is going to be possible on a more To quote from Leynaert Ref 117. the purpose regular basisIn the future. Intake testing to qualify the flow they deliver Is to the engine, and to determine the effect they 13 PROPULSION INSTALIATION. TECHIQOUESTEST have (In terms of drag, lift and moment) or the C21.ULASR external aerodynamics of the aircraft Most testing Is carried out without any engine being for many year., the ACARDFluid Dynalcs Panel has present In otherwords,the conclusion Ref 110 of devoted much attention to engIne-airframe that the engine has littleeffect on the Intake interference a.l. in particular, to the test pressure recovery, levelof turbulence, the the and techniques thatare required this field An ad mxlessa instantaneous in distortion Is accepted. hoc committee reported (Ref 114) In 1971 on the However, at a relativelylate stage In the results a detailed of studyof the testingmethods development a new aircraft-ensgine combination, of In use at that time and this was followedby a testsare sometimes made on the real Intake-engine lectureseries on the same subjectIn 1973 (Ref in the very large facilities ONERAModane (the at 115) In 1974.the FDP field a 4-daysymposlum on S1 tunnel)or AEC Tullahoma (the 16 ft PVT) (Ref airfraie/propulsion Interference (Ref 116) as 119) regards Oind, tunnel teating, this symposlum almost esclusivelyconsidered problems In afterbody Specific alm of Intake testing Include testing, eg strut Interference,temperature effects More recently. two WorkingGroups have (I) to obtain the pressure recovery/mass flow considered different facets the subjectWorking of characteristics, Croup 08. whlch reported(Ref 12,13) In 1986, dealtwiththe aerodynamlcs aircraft of afterbodles (Ill to obtainthe spillage drag as a function and Workling Croup 13, *hlch Is reporting In 1991, of mass-flow. Mach numbarand Incidence. Is concerned with intakesfor high speedvehicles and speclflcally Inludes a chapterIn Its report (1i1) to refine the details of the intake devoted to testingmetlds and techniques (Ref design, to optimise eg the boundary layer 117) This part of the presentlecturedraws bleeds cnd/ordiverter. extensively thisrecent on material lv) to study the unsteady Copared with the subsonic transport discussedhis respect, one characteristicsin should diutinguish above, combat aircraft introduce 4 major between unsteady distortion and the coeplicatloss surging of the Intake DistortionIs related Instabilities the intake to In (I) a greater speedrangeup to supersonic Itself and so. thereIs little deponderce speeds, on how the internal flosvalving systemIs (iil greaterrange angleof attackat high a of arrangeddownstream but when surgingIs subsonic transonic and speeds, being studied,or any other relatively (ill) a morecomplicated closely and coupled low-frequency characteristics Involving geometry, wave propagation tlmesbetween the engine (lv) a lack proven of enginesimulators that con and the intake,a valve system must be be used In routine testing largely to due provided at the positionof the first severegeomtrical constraints compressor stages in order to reproduce

4-29 p as the phenomenaexactly, particularly be should (which frequency reduced regards model Inversely proportionalto the scale), Recovery. Pressure Performance: 13,12 Internal Distortion Dynamic Swirlwith the local deviation measurements deviation measuremnts found by measuring the total pressure and cornertingto axial velocity fluctuations has sot approach but fluctuations this elaborate
yet been adopted.

the can When a floometer be usedas In Figs45sa,b, and precisely. directly mass flowcan be measured assumedthat the static pressureIn the For many combat aircraft, the primary design It Is and Is Is of chamber the flowmeter uniform for the intakewill be at supersonicsettling condition at the as measured a hole In the side of the chamber speeds. In two respects, this simplifies In are speeds,the flow into the and calculations made as described Ref 117. at supersonic testing: the sonicthroatarea When the test by can only be Influenced the shapeof the to determine intake plane and hence, Mach number Is not high enough for the flowmeter of the intake upstream aircraft on need not be represented throatto be choked,or when the pressureloss the restof the aircraft flow leads in a the mdel and second, the ratio between the accompanying sharp reduction mass set-up can the same of flow and the to an unchoking the throat, of the Internal pressure stagnation as the flowmeter a venturi. is such that It Is stillbe used,treating external static pressure at pressure the throathas easy to capturethe naturalflow with In this case,the static relatively the Intake Figs 45a,bshow two testset-upsfor, to be measuredalthough ideally,a calibration floweter. a be an Isolatedand installedintake should madeagainst reference respectively, In both cases, the aim Is to obtain the test (2) Stations botis to If it is unacceptable Include highest possible test Reynolds number. Ti., limit the sections, mas flow has to test Is set and (3) as separate Intake on modelsize in the Installed by the need to keep the Intakeplanebehindany be measured at the station at which it Is Is ot at shocksfrom the tunnelwills. In the controlled If the throat this statior reflected duct the Internal flow ducts choked,the static pressurein the Isternal example Illustrated, with the and combined has to be measured upstream contain area to area of the throat cross-sectional facestation determinethe mass flow. Ref 123 should be at Instrumentationthe engine (I) of for diagrams sonicplugsthat can be consulted (I), to movement vary the mass by adjusted translational a station (2) at whichthe flow through flow. A unit incorporating one of these plugs Is (ii) There Is no settlingregion either by an shown In Fig 47 the duct can be controlled sonicthroatas shown or by a upstream of the throat in this case and the adjustable definitionof the throat Is less drop suchas a aerodynamic pressure variable non-sonic not may precise The flow easurement therefore be valve,and finally butterfly section as as accurate when there Is a separate throat can be choked and (III) a station(3) at which the mass flow is (3) When no downstream one can use the Is with the flow profile not uniform, by, for exaple, a flowneter measured at measurements the engine face station(I) to a sonicthroat, on thendepends the compute mass foe, The accuracy In other cases. It maynot be possible to use two the detail in which the flow at this station has where a detailed tho space been explored In one example eitherbecause throatsinseriesat (1,2) rake supported was pressure exploration made by a rotating or is not available becausethe Internal for hypothesis the waIl boundary dropshave to be limitedto achievethe desired by a flowmatching of an Ref at layers. 124 claimed accuracy 1%. flow,In such cases,the mass flow is measured (2) station In contrastso supersonic speeds, the Intake flow 46 shows three examples of the engine face at subsonic and transonic speeds is Influenced by Fig one instru entation that was adopted by lAe as d 118 In the entire aircraft Strictl, therefore, modelsbut by comparing fullIt is generally shouldtest complete one of their test prograres believedthat about 40 stagnation pressures are model and partial-modei tests. MAch number and mean pressure Incidenceconditionscan be defined for the to sufficient obtain a reasonable tests that will bring their itse can partlal-model transducers 40 and recovery similarly, pressure resultsclose so those of the cmplate model be used so obtain the maxlmum Instantaneous distortions Variouswetbodshave been proposed Testsat thse speeds have to be extended up to This Is noraily a very high Incidences for eastilating the unsteady distortion fro Ref 120 accomplished by transferrlng modeIs designed for of saller number wasurements and, Indeed, quotes a 'rule of thumb' to use In cases here no testing in moderate size tunnels to larger tunnels For example, a measurements have beenmade this for the high incidence testing pressure unsteady with the partial nodel of the Rafale without wings and with distortion the unsteady rule correlates In thathas been tested the ONRA canards drop throughlb. truncated pressure stagnation man Internal area diffuser. This approach $2 tunnl (cross-sectional 3 an) was converted duct startingas tire for tests at modelof the aircraft withthe generaltrendshichis Into a complete however, conflicts, In high incidence in the Si tunnel (cross-sectional to take more measurements than in the past it is often considered mandatory that area40 a') particular. the test instrumentation includes measurements of swirl, eg in the case shown In Fig 46. sixteen 13..1 E.et&aL.ilriS A paramter for probes were used fve/hioe quantifying the rotational deviation has been in addition to Che normal tests on a complete model testsare made on the air intake This Is defined In an of the aircraft, proposed in Ref 121 the definltion, swirl alom or ot the forwardpart of the aircraft mannerto the 0(260 analogous of the Intake She purpose these tests value In a 60' including averaged S60 Indexis the maximum changesin of tho effects detailed of the Is to establish component sector of the circumferential velocity, divided by the axial velocity As long Intake design and of Intake ass flow on the model allo*s usteful partial ago as 197'. Carriere(122) rooeoended that the externaldreg the differences In relative Incidence of the engine Increase in Reynolds number and a mere faithful of face fluid stream with respectto the rotating representation the detailof the Intakeof the problem aircraft The majortechnique blades- In the differemes that give rise to the full-scale the determine how to 4 from a itt such a test IlesIn knowing distorton effect - shouldbe considered
global point of view by combining the fio doonstream momentum accurately Fig 8a shows one

4-30 miss flow was periodically injected by possible experimental set-up for which another negative at the compressed air supplied through the vane. lies In obtaining adequate stiffness problem showedthata good with flightevidence Jointbetweenthe live and earthedparts of the Comparison and Its at model. In thiscase,the Internal momentum this simulation of the engine surge phenomenon of from the measurements the effect on the Intakewas obtained In this way. Joint is calculated flow at the engine face On the assumption that Another similar deviceis presented Ref 128 In the flow Is being mesured more accorately elsewhere, the probe measurements at the engine agreement with to facecan be corrected givebetter this more accurate value and a corresponding 13.1-6 Conciuding Remarks correct can then be appliedto the momentum ion Fig 48b Illustrates another arrangement (Ref 123) A major comparative test programueof Intake and the of measurements has been organised by AGARD at plug Is placed the outlet in whichan earthed beinganalysed.Models to a are currently a the Intakeduct and the momentumIs evaluated results the of upstream the plug by measuring static comon design have been manufactured to slightly Ilttle at scalesand testedIn wind tunnels RAE part of the different In and totalpressures a cylindrical and unsteady ONERAand DLR. Both steady that Is neededIs for (Bedford), ductand the onlycorrection the drag of the cylindrical part of the tube measurements have been made over a wide range of over a at between the measuringsection and the outlet incidence M - 0.8 and at low incidence, The section a smallfriction term. The mass flow Is wide range of Mach number up to M - 1.8. pitot models representa simple subsonic-type calibration in obtained a separate and cross-section bluntlip. with a circular Intake used Detailsof the instrumentation In the various are given In of tunnelsand examples the results of and 13.14 Acouisition analysis distortion from this exercise Tb, first Impression Ref 117 MgILrNU In of standard showan Impressive is thatthe results In evenwhen differences the absolute Indices comefrom the fortyunsteady consistency, The distortion the are observed, Indices of transducers the engineface rakes The highest valuesof the distortion similar The thathas to be considered of the order trendswith mass flow are broadly Is frequency from thiswork will be presented by finalconclusions of 1000Hz fullscaleand thishas to be divided symposium September 1991. It is In One way of obtaining a at an ACARD the scale of the model exercise will be cooperative IndexIs to buildan analogue likelythata similar distortion particular to on Intake design cumputerwhlchdeliversa signal proportional proposed a morecomplex the desireddiatortIon Index. These are wide!y used,they provide result realtine but their 13-2 Testson Afterbedy a In .odels usefulness is limited the one distortion Index to for which they were designed Hence. as a Wind tunnel testing to determine the true complement, the transducer signals are recorded on aerodynamic characteristics over the afterbody of a magnetic tape for off-line coaputar analysis full-scale combat aircraft Is perhaps the most areasdiscussed task In all the problem Better performance can be obtainedwith a PO4 difficult (Pulse-Code Modulation) record, The distortion In this lecture She normal complete model test on Indicas can also be computed digitallyusing with the model supported a solidstingfrom the In the Jet analogue-digital conversion with a suitable array rear Is deficient two respects first, processor coeputer effects are beingignored (since isonly In rare It an provides acceptable casesthat the stingItself The aircraft designer will want to know the maximum representation of the jet plume)and second, the values of the various distortion Indices as shape of the afterbody has probably been distorted recorded duringa given length timeof orderI appreclably admitthe sting for example.If of to minute, full scale, for the flight conditions at the aircraft afterbody has two nozzles with their shlchseriousdistortion present The recorded exitsat the bodyrear and, the gullybetween is these distortion can be analysed statistically to define two nozzles my wellhave been partly filled In to these maxlajatvalues according to a given admitthe sting,thiscouldhave a dramatic effect prohahlilty (Ref 125) A detailed map of tie on the viscous flow development affecting not instantaneous may also be of interest flow but the merelythe external drag but also the effectiveness Irportunt Issue Is knowinghow to selectfrom the of any tallsurface the vicinity Earlyflightin vant amouni of data that Is typicallytaken tunneI comparisons for aircraft such as the Systems for doing this have been developedat Lightning reealed aerlous differences In the various establishments and In Industry, eg at AEDC directional stability characteristics which sere (Ref.126.127) subsequently traced to the unrepresentative shape of the model&fterbedy Supplementary model tests If no analogue computers or high speed data most therefore be made but many questions then acquisition systems available, firstestimate arise,eg should are a one test a partial modelor ahouId of the variousdistortion Indicescan be made, one test the complete modelwith the trueafterbdy based on ihe Wl5t values of each Individual total but iaunted In a different mnner and howshould pressure probe i the rake The etatisical either Of thesemodels be supported? Also,what analysis techniques employed are described In Ref should one leasure. eg should one test a partial I1? model or should on test the complete model with the true afterbody but sounted In a different eg mannerand what should one measure, total thrust ntakefiatdynlac study L and drag for thrust balances minusdragor separate To design a control systemfor a variahle-geotry or drag by means of a pressure Integralion? intake. dynamic characteristics the Internal Further, how should the jets be simulated, by the of eg flowbase to be studied It Is alao Important to high pressure air ducted to the nozzles and direct identify the levelof the wall pressurerise,In blow or by use of a turbine ejector or simulator? transient flo. for the design of the intake Thereis no simple unique or answerto any of these structure To perform such a study, high speed quest ions It all depends ot the aircraft shape rotating vanescan be Installed at the compressor and on the test requirements A study of the facestation producea periodic to variation the literature of suggests that every conceivable answer reducedmass flow The Intake Is equipped with has already been triedbut It Is stilldifficult to unsteady trassducers which measure the amplitude arrive at any general conclusions The chapters and the phaselag of the pressure waves One such (Refs 12,13) by Bowers and Carter In the -CARD device Is described Ref 115,in thisexample. Sorking Cioup WCOS report are the mat recent in a

4-31 text the sceneand the following to attempts review Is broadlyin line with theirconclusions.The are discussed in ieading questions outlined above that they are turn ht it iI soon be realised Is to le very Interrelated, the answer one question to likely dependon what answerhas been givento one of the otherquestions models? or 13.2.1 Complete partlal One cannotgive a generalanswerto this question have except to commentthat, as combat aircraft a the coupled, case for testing more closely becone complete model or at least a closer approxlmation to a complete model has strengthened. Jaarsma In Ref 129 gave a good review of the relative partialand of and advantages disadvantages testing one model$, of models. In favour partial complete accuracy, betterpotential scale, can quote larger a more more abilityto Incorporate Instrumentation, faithful representation of the primary and, when jet stream secondary even tertiary and necessary, and probably cheaper testing for parametric On the other hand, the common investigations practice of using a cylindrical forebody may mean that the afterbodyis being tested In a very environment In favourof testinga unrealistic complete model, the principal advantages are better of nozzle external simulationand duplication mutual environcent, better accounting of influence, moreaccurate Interferences forebody and and plume simulat ion of aircraft aerodynamics On Interference. the other hand,the models are (it) more and smaller, toe Instrumentation limited It Is and tertiary air difficult to Include secondary of to It Is tempting say that It Is not a question either/or t.t tather of decidingthat oee needs both the complete model for overall effects and a of modelfor the development the nozzle/aft partial end It Is arguable that the past literature on this subject does not place enough ephasis on the need layerahead for a representative approach boundary Is of the afterbody The flow over the afterbody as strongly viscous and it Is Justas Important on an advanced wing to have a good slmulation of the full scale boundary layer The advances In CFD methodsnay nake it possible do this much more to than In the past This concern scientifically about the boundary layer does not necessarily favour use of a partialmodeldespiteIts larger that one can more easily (1l1) size and the probability modifythe approach boundarylayer The pressure gradients Imposed the flow fieldof the restof by the aircraft way be core Important factors In the boundary layer development determining Increasingly. the trend Is toward a compromise 1IMle is between a partial and a complete model shown by Fig 49 which Is a picture of a nw rig being developed for RAE (Pyestock)by ARA Is a strut-supported partial Strictly, this afterody model but it can Include a correctly scaled representation of the aircraft forebody, forward lift ing surfaces and th Inner wing thich, o the underside, forms the upper part of the support strut (I) or the modelon an annular a pair mounting the obvious stings is perhaps of annular method of testing a complete powered is model. it certainly the best approach and, for minelsingsupportInterference my be the only way of for this reason. datacloseto M- 1.0. However, obtaining there is one Important Interference the free jet as consideration to whether correctly In the plune Is represented has of presence the sting This question been studied at AEDC (Ref 130) where stings have been Used to support annular such as the F-16. modelsof aircraft large In a research exercise at AE., It was foundthat the stingflare shouldbe at least 3 body dianeters downstream of the nozzle base with a 10" boattall and 5 diameters downstream of a cylindrical couldbe The stingInterference boattail. determined experimentally by mounting the model, with alternativelythe true rear-end and with the modified rear-end strut on and dumeysting, a sweptsupport one The decisive final pointas to whether can use an annularstingsupportis the has to be extent to which the afterbody It is distortedto admit the sting distortion. likelythat to avoid serious can only be used this methodof mounting for testsat low Incidence. mounting Is thatwing-tip Ref 12 concludes a viable alternative especially for afterbody/nozzle testingbut Incremental M only if the Mach numberrangebetween 0,8 and ii- I I can be avoided. In one unpublished case. however,the subsonic up successfully to M rangewas extended designed 0 92 by the use of specially The wing planform at bodies the wing tip geometry has to be modified near the tip for structural reasons The rig has on Interference to be judged the basis of comparative tests with and without hardware with the dumuywing tip support model mounted on a slender sting Hopefully,this interference would not significantly between change a configurations *hile testing seriesof different but similar afterbodles/ozzlos Clearly,a strut mounting is the best sciteme From the point of view of support strength,rigidity and duct space for Instrumentation and high pressure air but the overriding Issue Is whether the aerodynalmc Interference Is acceptable Also, obviously, the technique cannot he conditions used for testing undersideslip at and also, there nay be difficulties high Incidence Close to M - I 0. very large Interference drag values have been reported. eg 20 or even 80 drag counts, but the magnitude of this Interference and the Mach ntuabar range over hich It Is unacceptable will depend on the geometry or the strut support and on how much of the Installation Is tmetric A side range strut geomet es ri were of different comparedIn Ref 131 R.f 12 concluded that despite all the problems, the strut wouldcontinue be used but all to support new arrangements should be based on erpirical guidelines, past exprience and Increasingly, theoretical OFDtools should be used In the design It will be noted that, In the most recentdesignshown In rather Fig 49. the strutIs sweptforward past than swept aft as In mny Also, forces are being arrangements measured on the entire odel/rtg by means

ILI'

e thods of Colel

pqn

Many different types of model support have been such Ideas as proposedand explored,Including mountit.g the complete model from the nose at the tube tall-end of a long Interference-free stretching from the tunnel settling ctamber It all general, however, one can characterise rigs In types, commonuse as beingof one of threepossible viz sting-mounted an annular sting(s) wing-tip on k (rigs 50a-cl mounted and strut mount, Considering cact of these in turn

4-32 OIAPS technology In test programes was still In Its Infancy and, apartfrom an early programme of operating tests at AEIDC, the only recorded application OIAPSIs thatdescribed Ref 132. of In Fig 53 showsthe Instillation OAPS In the model of for thisapplication: will be seen thata single it 13.2.3 Tyseof measurewents supportstingbringsthe driveair intothe model and takes the turbine air out of the model. The To illustrate the potential variety,fig 5Ia-f sting entr) Is from below the fuselageat the shows six different strut-supportedmodeI centreof gravity intoa manifold up system which configurations afterbody forces being measured distributes high pressure with the air ductsout to the withone or more forcebalances The totalmetric two earthedsimulators the wings. The live in -o dei thrust-minus-drag arrangement (A) Is modelshellIs mounted the liverear end of the on considerably different from the three afterbody-balance Considerable effort has been put Intothe only thrust-minus-drag configurations (B,CD) and developmnt of OIAPS and appropriate calibration the afterbody drag balance models (E.F). Ref 12 facilities NASA Amesbut the vitalfeature at which cocucots that there is no convention standard Inhibits or its application models VSTOLcombat to of for afterbdy/nozzle force balance arrangements aircraft Is its geometry: an Inlet to nozzle whichIs separation length 9.5 engine of diameters Multiple forcebalances often used to determine much too long to allow Installation a closeare In sefrtrately forces on different the pirts of the coupled configuration Design studies for the model A typicalsyst,,c uld consistof a main design of a powered simulator more relevant of balance to detern ne lift and thrust-minus-drag of geometry are being made in the I and one can the afterody together with a thrust balanceto expressguarded optimismthat,In the long term, measure nozzle normal and axial forces; the theseefforts will be successful weakness of this approach Is that drag Is determined as the differencebetweentwo large The other possibility to developan ejector Is quantiti.s Another possible way to determine the poweredsimulator CarterIn 1986concluded (Ref drag Is to pressure plot the afterbody. thiswy be 14) that these were capable of providingan a sensible Ideawhen testing ralatively simple, eg adequate simulation the flow requirements of of cylindrical aeterbadies but, in general, It way be current engines hut thatthe length/diameter ratios of an acrate utderfloor balance It Is hoped that the relatively largesize of thisrig will enable studies be made of to not -erely the afterbody drag but also of the unsteady flowover the afterbody
difficult to include enough holes on a complex of these slmulatcrs were likely to be even higher

shape to obtain an accurate integration,

than for 04APS Hene, at the present time, for routinetestson model%of combataircraft, there The force balances used Ili arrarck nt such as is no practical alternative to the direct blow soe of those illustrated Fig 31 ate often approach Ir complicated theiroutput osl-is orrections and
to provide data with acceptable a*u stibllity In finally, It should be noted that none of these

addition the balance Interac to

os

det,-1i wed is

a calibration of the bare balanre. cor'cction.-. 4re needed for pressure area tare turcee istlualtg cavity, metric seal and base pressurs and f w mosentu tares Any bellows 3wu.m als, hex , b-

simulators providesslmulation the full-scale of jet tempt , ure Modern turb.-Jet and turbo-fan
effluxev operate in the region 1500' to 1500' R while tise ajority of model afterbdy testing is conducted with a cold Jet near 500' R rhereare tso says In *hih the jet affectsthe afterbody

calibrated These corre-.ions ny svtsl, A greater than the final rorrete4 dr. If t1l model Includes both metric and tn-Imetrlc orts
the position of the breal bets.e- Ciese h.- to c chosen carefully, thinking of n-th arol)ytsv sd structural considerations It swis d be Ir a egion of relatively uniform proesarc and at i pe-.t.-o there any flexing of the model ur er lod cots ot Introduce a step at the break

drag first the drag is reduced by the Jet plume forward pressure Interference and second, the arma
is increased by the jet entralinuent Experimental work reported in Ref 134 studied how the jet plume shape and *ntralnoent depended on the physical properties or the exhaust Carter in Ref I3 concludes that te effects of jet temperature are likely to be most significat in tlietrassosir flow range where latse erect o' base or separated afterbody flows occur Th effects are greater for

Ref 12 concludes that new msodel cnfigurations and balances should be desigted and built to suit the teat objectives. tIhe test facility and the configuration Itself and In the light of past eperierrce 13,2., lot Ulatic

The above discussion has conce-trated on wdeis with direct blow exhaust s but with Imuiatlo the
close coupted nature of many modern aircraft and

large afterboy angles and can be as large as 35% of the jet.off afterboly drag for high jet pressure ratios 20% Is perhaps mor, typical a figure: the cold jet results will give pessimistic afterbody drag predictionsTh eff..t$ temperature of are therefore significant a- configuration-dependent but It Is unlikely thatany serious attempt will be adr to develop methcdafor heating the jets In re, tue model testIng Thle effects shouidhowever
-' he 1. ored 0

evenware fien looksforward ASTOVL one to designs, It is cx lunger trictlyadmissibleto ignore LL.[ possible Interact Ions between intake and exhaust floss The need to test Inlet and efxaust floe- In Tht lecture
combiat ion s recongnlsed more than 10 years ago techniques

has revie-ed the expe. maenta ittuse Its eind turnZi t.odol tests for

In tim US and this led to te developmets of the modeI engine unit known as OdAPS (Compact Mullimission Aircraft Ptopulsion Siwalator) ihis Is shoon n Pig $2 and descriptions to be fotnd are In Ref 132 and 133 and various other refernuces quoted In Ref 13 This unit has a 4-stageaxlal compressor drive by the power generated by a single stageturbine file turbineis driven by the high pressure inletair which Is then mixed with cospressor air througt a alxer/eJector module to provide the exhaust nozzle total pressure air supply Petforware details ace summuarised Ref In
13 Carter In 1986 notes that the application of

performac. predictfion Ike r-o0. Is inevitably tonged with personal blia bat I' Is hoped that use subj(I. .overed In detail%nd the largenu ber of reftrontcas will be of contitued benafit Looking to the future, ale clearly ti-s thrsold of a oe at largeexpansion the use of non Intrusive In optical techniqueslaserholography, laseranamowetry and particle imagevoelocietry to study and .. asure tite flow field Thesedevelopments vital for the are salidat Ion of new CIO methods and so, one can claim that experlotetal techniquesand their further refinewantare th key to future progresa In
aerodynsalcs isl geeral

4-33 Acknowledgements 22 The author wishes to acknowledge gratefully the helpof Mrs A Ell and Mrs M Connolly preparing um In this lecture, various useful discussions with 23 members of ths staff of ARA and RAE (Farnborough and Bedford) and the permission of MOD(PE)to publishthe lecture. 24 Referencs I 2 3 WolfS W, Status adaptive of wall technology for minimisatlon wind tunnel of boundary Interferences, 1990,ICAS-90-6.2.1. Coodyer J, The selfstreamlining tunnel. M wind 1975.NASATM-X-72699. Canzer Igate ZIemann Design U, Y, J, and operation TU-Berlin of windtunnel with adaptable walls,1984,ICAS-84-2.1.1. HornungH (d), Adaptive wind tunnel walls. technology application. and 1990. Report of ACARDW112,AGARD-AR-269. Barnwell W, Edwards L W. Kilgore A, R C R DressD A. Optimum transonic wind tunnels, 1986. AIAA 86-0755. Britcher P. Progress C towards largewind tunnel magnetic suspension and balance systems. 1984,AIAA84-0413. Eilgore A. Lawing L. Cryogenic R P wind tunnels for high Reynolds nunber testing, 1988, ICAS88-38 4. Laster L (ed). Boundary M layersimulation and controlin wind tunnels. 1988. Reportof ACARD W.O9.ACARD-AR-224 iarris E. Carter C. Wind tunneltestand A E analysis techniques usingpoweredsimulators for civil nacelle installation drag assessment, 1981.AGARD 301-24. CP Harris E, Paliwal C, Civil turofan A K propulsion systemIntegration studies using powered tosting techniques ARA Bedford. at 1984, AIAA 84-0393 ColdhassnerI,Steinle W, Design M F and validation advanced of transonic wingsusing CID and very highReynolds nuaboer wind tunnel testing. 1990.ICAS-90-2 2 6 Boners L, Stateof the art assessment D of testing techniques aircraft for afterlodles, Aerodynasics Aircraft oF Afterbodles Report of KOS. 1986, ACARDAR 226 CarterE C, Jet simulation, Aerodynamics of Aircraft Afterodles keport of 08. 196. ACARD-AR-226 Wind tunnel flowquality and dataaccuracy reqolreents., 1902.ACAOD-AR-184 CarterE C, PaIliate; C, oeveiopmnt of K testing techniques a largetransonic in wind tunnel achieve required to a drag accuracy and flowstandards wdern civiltransports, for 1985. ACARDCP 419. Paperno !1 WoodM N. CappsD S. The accurate measurement of drag in the 8 ft x 8 ft tunne 1i9835. AGARD 429,Paperno 9 CP AshillP k. Private comunication. 1990 EsaldB, Clescke Polanski Schenck P. L. C, Automatic calibration machlno for cryogenic and conventional internal strain gaugebaiatce, 1990. AIAA 90-1396 CrossC luanarena B. A miniature 0 48-channel pressure u.,sor module capable In-situ of calibration, 1977,IS 23rd International Instrumentatlon Symposilu Juanarena B, A distributed D processing high data rate metiport winJ tunnel pressure measurement system. 1977.ISA 23rd Internatlionai Instrumentation Symposlum JuanarenaD B, A multiport sensor and measurement systemfor aerospace pressure measurements. 1979,ISA 25th International Instrumentat IonSymposlum 25 26 CrossC, Juanarena B, Electronically 0 sc.nned pressure scanners, 1981,Transducer/Tempcon '81 Conference, Wembley, England BerghH, Tijdeamn Theoretical H, and experimental results for the dynamic response of pressure measuring systems, 1965, NLR-TR-F238. WelshB L. Pyne C R. CrIpps E. Recent B developments the measurement of timeIn dependent pressure, 1983. ACARDCP-348, Paper no 36. WelshB L. McOwatD M. PRESTO- a system for the measurement analysis timedependent and of signals, 1979,RAE-TR-79135. MabeyD C, WelshB L. Measurements and calculations steady of and oscillatory pressures a lowaspectratiomodelat on subsonic and transonic speeds, 1987.Journal of Fluids and Structures. 445-468. Horsten J. Recent J developments the In unsteady pressure measuring technique, 1981. NLR-TR-81-0SSU. Haines B. JonesJ C M J, The centre-line A Mach-number distribution auxiliary and suction requirements the ARA 9 ftx 8 ft transonic for windtunnel. 1960. R&N 3140. Binion WJr. Potentials pseudo-Reynolds T for number effects. 1988,ACARD-AV-303. Chapter 2 KennedyT L. An evaluation wind tunnel of test techniques aircraft for nozzle afterbody testing transonic at Mach numbers. 1980, AEDC-TR.80-8(AD-AO91775) Aulehla F, Pseudo-Rynolds number trends. 1987. AIAA Sth Applied Aerodynamics Conference. Monterey Baxendale J, Carberry Day J. GreenJ E. A J. Replacemetnt of the flow smoothing screen in the ARA 9' x 8' TWT by a longcellhoneycomb. 1990. ARA Memo 339 Dougherty S Jr. Fisher F, Boundary N O layer transition on a 10 dog cone wind tunnel/flight datacorrelation, 1980.AiAA80-0154. Dougherty 5 Jr. Steinle W Jr. Transition N F Reynldk numer corparlsons several in major transonic tunnels. 1974.AIAA 74-627 Murthy V. Effects compressihility S of and Freestream turbulence on boundary layer transition highsubsonic in and transonic flows. 1986,AIAA06-0764 LongD. An experilmental evaluation of test section noisein transonic wind tunnels, 1990. AIAA90-1419 MichelR, Coustois ArmsID, Transition E. calculations three-dimensional in flows. 1983.ONEATP 1985-7 CreenJ E, On the influence fren-strean of turbulence a turbulent on boundary layer, it as relates wind tunnel to testing subsonic at speeds. 1972.RAE lR 72201 Hrcock P E, Bradshaw The effectof freeP. stream turbulence turbulent on boundary layers. 1983,Transactions the ASME.Vol 105. of IMabey C, Flow unsteadiness and model D vibration wind tunnels subsonic in at and transonic apeeds. 1971,ARC CP 1!55 Mabey D C. A review somerecent of research on time-dependent aerodynamics, 1984. Journal RoyalAeronauticai Society CarnerN C, Rogers V L. ACum W E A. Matskell E E C. Subsonic wind tunnel corrections, 1966. ACARBograph 109. B~rndt S D, Wind :untnl interference due to lift for delta wings of small aspect ratio. 1950. KTi Sw-den Tech Note TN 19. EvansJ Y C, Corrections velocity wall to for constraint any 10 x 7 rectangular In ubsonlc wind tunnel. 1949.ARC R& 7662 ThomA, Blockage corrections a closed In highspeedtunnel. 1943.ARC RAM 2033 Thopson J S. Present methods applying of blockage corrections a closed In rectangular
igsh speed wi.sd tunnel 1948 ARC Report 11385

4 5 6 7 8 9

27 28

29 30

31 32

10

33 34 35

II

12

36 37 38

13 i4 IS

39 40 41 42 43 44 45
.

16 17 18

19

20

21

47

48

49

51 52

Kirby D A, Spence A, Low-speed-tunnI model tests on the flow structure behind a delta-win 8 aircraft and a 40 deg swept-wing aircraft at high incidences, 1955, ARC RON 3078. Baldwin B S et at, Wall interference in wind tunnels with slotted and porous boundaries at subsonic speeds, 1954, NACA Tech Note 3176. Wright R H, The effectiveness of the transonic wind tunnel as a device for minimlsing tunnelboundary Interference for model tests at transonic speeds, 1959, ACARD Report 294. Coethert B H, Traroconic wind tunnel testing. 1961, AARDograph 49, Pergamon Press Kassner R R, Subsonic flow over a body between porous walls, 1952, WADC Tech Report 52-9 Browne Miss C C, Bateman T E B, Pavi-t M, distributins measured I. flIlgh. and on a daies A B, A comparison of wing pr- sure winditunnelmodel of the Super VCIO, 1972, ARC R&M 3707. O'HlaraF, Squire L C, Raines A 11,An Investigation of Interferenceeffects on

Jelly A H, Mundell A R G, rlrkins A J. Wind tunnel model support Interferencecorrections for a three-strut mounting, 1989, RAE Tech Memo Aero 2154. 71 Hardy B C. A numerical study of the aerodynamic interference of a model support system used In the RAE 5 metre wind tunnel. 1985. RAE Tech Memo Aero 2046. 72 Sala M, Quenard C. Airbus A310. essals dans ]a soufflerle Fl de I'ONERA. comparalson volsoufflerle. 1984, ACARD CP 238, Paper 22 73 Quemard C, WillaumoeJ. Calculs des Interactions aerodynmiques d'un montage de type '3 mats, 1987, ONERA Rapport Technique No 283/1464 CG 74 Haines A B, Holder D W. Pearcey H H, Scale effects at high subsonic and transonic speeds. and methodsIn model experiments, 1957, ARC R&M t ransition for fixing boundary-layer 75 3012 Rraslow A L, Knoxcritical height method for determlnation of E C. Simplifiedof distributed roughness particles for boundary-layer transition at Mach numbers from 0 to 5, 1985, NACA YN 4363. Van Driest E R, Blumer C B, Effect of roughness on transto on In supersonic flow, 1960, ACARD Report 255. Evans J Y C, Transition fixing techniques and the interpretationof boundary layer conditions on slender wings In supersonic flow, 1964, RAE IN Acre 2946 totter L tleith,Whilfield Jack D. Effects of unit Reynolds nuater, nose bluntness, and roughness on boundary layer transition, 1960, AEDC-TR 605 Nash J F. Bradthaw P, The mogniflation of roughness drag by pressure gradients, 1967, RAeS Aeronautical Journal, Vol 71, pp 44-46 Stallings R L Jr, Lamb M, Effect of roughness size on the position of bou,'ary.layer transition on the aerodynamic characteristics of a 55' delta wing at supersonic speeds, 1969. NASA TP 1027 Ashill P R. Fulker J L. Weels D J. The airinjectin,, method of fixing bounda-y-layer :ransitilol and investigating scale effects. 2 1997, RAeS Aeronautical Journaal,pp i4-2 2 4 Cartenberg E, Johnson W C Jr. Johnson C B, Carraway 1 L, Wright R E, Transitl'n detection studies in the cryogenic *nvironsent. In)0. AIAA-90-3024.CP Pearcey N H, Osborne J. lialnes A Ii, The Interaction between local effects at the shock and rear separation - a source of significant scale effects in sled tunnel tests on aerofolls and siege, 1968, ACARD CP 35. Paper II AshtillP R, Wood R F, WacLu 8 J, As Improved, se*l-inverse version of tht,viscous Carahedias and Korn method iVCK), 1987, RAE Technical keport 87002 Elyunaur A, EUperiences with transition fixing In the high speed regime at NLR, Crenzsclicht Steuerung dutch Iranaltlnfixierun, 1964. I*VIR Mitteilung 94-17 Haines A B, Experience In the use of a viscous sinulation mthodology for tests in transonic tnnels, 1990. AIAA-90-1414 fulLer J L, Ahill P R A model of the flow ocer swept wings ith shock induced separation, 193, RAE OR 8300fk Loc R C. Prediction of viscous and wave drag at high subsonic speeds by viscous-inviscid Interaction tmchniques, 1985. ACARD R-725 A method of determining the wave drag and its spanelse distribution on a finite wing In transoni, flow. 1983. ESOII0O 83022 Melnik R E, Chow R R. Uead H R, An improved -fscid/inviscid interaction procedure for transonic flo- oer aerofolh;. 1985, NASA CE 3505

70

53

similar models of different size In various transoslc tunnels In the UK, 1959, RAE Tech Note Acre 2606, ARA Wind Tunnel Note 27. 54 Page W A. Experimental study of the equivalence of transonic flow about slender cone-cylinders of circular and elliptic cross section, 1958, NACA TN 4233 55 Berndt S B. lbeoretic.- aspects of the calibration of transonic test sections, 1957, rPA Report 74 56 Berndt S B, Sorensen H, Flow properties of slotted walls for transonic test sections, 1975, ACARD CP 174, Paper 17 57 Sickles W. Ericisor J C Jr, Wall Interference corrections for two dimensional transonic flows. 1990. AIAA-90-1408 50 Smith J, A nothod for determining 2D wall interference on an aerofoil fro. - .asured pressure distrIbutions near the wall and on the model. 1981. NLR tR 810160 59 Cupeller C. Chevallier J r. Bounol F, Nouvelle 1dthode de correction des effets de prols en courant plan, 1978, La Recherche A;rospatiale.No I, pp I-I 60 Keep W B Jr. Yoard the correctableinterference transonic wind tunnel, 1976, AIAA-76"1794CP 61 iurman I U. A correction motod foe iransonic wled tunnel wail Interference. 1979. AIAA-79-1533 62 Kraft r M, Ritter A, Laster i L. Advances at ArrC in treating transonic wind tunnel ail Interference, 1936, ICAS Proceedings. pp 748-769 60 Minect P Creen L. Wall interference assessment/correct ion (WIAC) for transonic airfoit data (row porous and shaped all test sections. 1990, AIAA-90-1406 04 Ash.li P R. Wees 0, A method for determining -Interference corrsctionu Is noild-wall usrela from easurements of Ststic pressure t te stails, 1982, ACARD CP 335 65 Ashill I'R, Keatlng R I A Calculation of tuel *all ,otorfereore from wall-presauro a-asr eoen 1980, PA.$ Journal, pp 36-53 66 Kiripatrick ft, vuraard 11, Priorities for iigh-iift treirg in the uture, 190 AIMA90-14, 67 Coetnert Bo,Wludaviorrecturen boi horen Unterachaiigeu in*lr 'lgctetterlnr Besonderer Berucks;chtigung des 'scIh;ossenmn reislanals '940, t utsche tuftfthrtforschung forschunguericht 1216, translated as MACA tech 14- 1300 68 Jacocks J 1, Aerodynamic characteristics of perforated salts for transonic tunnls. 1977 ALDC-Tb.T7.61 tAD.A040904) 69 ilnachH t Jr, R,.sthnll IOH, Atronautical 2 0 faclit requiretmoentInto Ohs 0 0. 1990 AIAA4J,.I'75

76

77

70

79

80

01

02

83

84

85

86

87

88

89

90

3 91 LockR C, Williams R. Viscous-inviscid 8 113firkerT, Procurement tes ofa1 scale and .itrActions in vxternal aerofoil aerodynamics, Advanced counterrotating topfn 1990, odel, 19853. Progress Aeronautical In Sciences. AIA-90 10o,. Perganon Press 114 ACARD cCoummittee, Engine-airplane 92 Elsenaar The wind tunnel a tool for A, as interfetence wali corrections and in laminar finwresearch, 1990.ICAS.90-6.t I transonic nih ael tests, 1971, 93 Soners StackJc' Hrvey V D, infisence S, P. ACARI3-AR-36-71. of ssrface static-pressure orifices on 115 CarterE C, Experimental determnaton of houndary-layer transition, 1984.N8ASA 84492. TMd Inletcharacteristics Inlet and and airplane 94 Roeroen J, El S senuar Half-model nests in A. interference. 1972,ACARD-LS-33, Paper3 the t'iR high speedwIndtassel HST. Its 116 CarterE C, Technical evalsation report on technique application. and 1983.ACARDCP 348. FluidDynamics Panelsymposioum alrfrane/ on 95 Decher Tegeler High accuracy R, 0. force rropulsion Interference, 1975. ACARD.AR-81. accounting procedures turhopowered for lI7 Leynuert Air Intake J, testing. 1991,Air ACARDWorking for Intakes highspeedvehicles. AIMA 75-1324. 1975, tenting. Imulator 96 PughC. Harris E, Estahlishment an A of Croup13 Report experi mentaltechnique provide to accurate 118 Stevens H, SpongE D, HNo ck M S, F15 Inlet/ C measorenens the Installed drag of a full of engine tentse.,riques distort and ion spanmodelwith turhine poweredengine methodologies stien. Vol I - technical simulators. 1981. ACARD-CP-301, Paper25 discussics. 1978,NASACRt144866. 97 Him$n A B, At- lynmic Interference - a 119 Mitchell J t;. Fluid dynamic aspects of turhine general overview, 1983, ACARD-R-712, Paper9 engine tenting. 1983, ACARDCP 348. 98 Iturgsmullor W, Akheinn E. Kool J W, Engine 120 Aulthia Scimit7D M, Intake F, swirland simulat with turho-posared ion ifoulators, 1990. slnplirt..d methodsfu: dynamic pressure 136W Colloquium, distort assessemot, ion 1985. WKI Lecture Series 99 KranzC, Engine/airframe lntcr*ronce, 1983. 198884 - Intake Aerodynamics. ACARD-R-712, Pape,14 121Coo R W, Seddon The swirlin an S-duct J, of 100 Eckert van DitsholsonC A. Nonniknoa 0, I B, typical air Intake proportions. 1983. lBurgsmsllerLowapeed engine slm.,ut on W, ion AeronauicFal Quarterly a large scale transport modelIs the tCiV, 1984, 122Carriere Aperfo quelques P. da prohlewes posespar lei prises uctt~els asrodynasiques 10 I CAS-84-2 1 101 Dsrgsm~iler W, Scodroch J, Benefits and costs d'airsupersosiques, 1972.Icr Symposi um of posored engine simulation low speeds, at International nor les trogres des Racttors 1985. AIAA-85-0381 d'Avion. Marselille. O6LRATP 1102, 102 Reacts C, Transonic wind tunnel testing of D 123 SeddonJ. Coldsslth L. Intake E aerodynamics. propellers. 107, Canadian Aeronautics and 1983,Collins ed SpaceInstitute Symposium Aircraft on 124 Nacirodi A. Schmidt DOH.Esperienotelle P Propulsion System,Toronto. Canada Lntersuchungen ewa Unterruapr an 103 8-1av, D C. Pouniak O0 M.facilities for the Rampeneinlauf tlherschAl,strnong hei developmnt propollers and propeller of 1986. Jahrhuch I der DELR

Installatitons at ARA. 1988. PROS International Confrence on Advanced Propollers and their ,4Installation 04Wood Nrs N E. The design And commissioning of as acoustic liner for propeller wisne testinmg In the ARA transonic wind tunnel1, 1989.1 Noth E Aerolech 959 105 Harris E, Render A PH4,Pozslak 0ON,%ood N E. Recent wind tunnel tes.iog experience of cotra-eotating propollers, 1998. ICA$-90.4 I I 106 Barber L J, Performance evaluat ion of fuli scalepropailers wind3 by tunnel I tst. 1914, ALARI-CP-3h6 l07 Netzger B Ptop-fan F design and test experiernce, Iv88. RAeSInternational Conferenc a., AdvancedPcvpovllers their Installation andS sonAircraft 108 Gromet. e F, Aeroacouss ice of advanced on Aircraft

125 Jarocka J L, Statist ical analysis of distortion factors. 1972. A]AA-72-l800 engine Clay.Inlet 126 SurberL E. Fsglinees coupatihtiity, 1988, WKI Lecture Series 1988-04 - Intake Aerodynamics

t27 Maroon3, Sediock An anelog D, enditing system fur inletdynamic flowdistortion, DYNADEC Past,present and future. 1980.A[AA-80-118 128 Lotter K, Nackrodt P A, Scherhasi R. Engine surgesimulation wind-tunne In modelInlet dacts,1988,16thICAS 129 Jaursama Enporlwentl dtermination of nozzlee F, characteristics and nozzle airframe Interference. 1972. ACARDL.3-53 130 PriceE A, Investigation F-16nozzleof afterhody forces At transonic Mach namb~ers with emphasis support system Interference, 1981, on
A[1X-TR-80-27 131 German R C. Strut support Interference on A

propellers

1990, ICAS-90-4 1 2

A[AA-97-2920 110 Dengeliok L, Doerzbachor P. krysytZly j R R A She developmant and calihrat ion of an acoustic sail transonic teatsection, 1986, Alok-86-0759-CP Ill WungPHW C, Mam a N. forsey R, BouciA J. C Single and contrarotation highs spoed propellers flow calculation and performance predirm i,..1988.ICAS-83.2 2 4 112 Ssefko L.. C Jeracki L., R Porous wind tunonel correct ionsfor counterrotas propeller ion tsss ig, 1988. AIAA-88-2055

109 V-rm- D F. Hiugh-eJ F. Aerodynamic Integrat ion St aftm id DUD propulsion system,. 1987,

cylindrical miodel withi boattall 4t Mach snmhers fighter modelIncorporating turbine poserod engine ulmlators, 1981, AIAA-81-2635 133 SmithS C, Determining compressor me~t airflow In the compact multilion aircraft propulsion simulators wind tunnelI In applicatiloss. 1983, AIAA-83-1231 134 Coapton B, An oxporimental H studyof jet unhauwisimullttion. CP 150-16 ACARD
frow 0 6 to I 4. 1976, AEDC-1TR-76-40 132 Halley R 0, hrac N. Hiley P, She design of A wind tunnel VSTO.

4-36

0'00 WD~ 0

-o

0010 Ii 0

01
FIG I

02

03

L06 0

01
Mi 2

0,2

03

04

05

CL 06

A WITHINS GIVIN TEST R[F EATAIIILITY STANDARD 2TNNEL, lEGQl REF 15) (ARA TRATISON-C

BMUREN TESTS REPEATABILITY STANIDARD FROMIREF 13) TUNSNEL, (ARA TISANSC*IIC

NWNW fo-~ pmt" 141,81

17.4 K 226KN08

4M 2Rs lb 1170 h

Side1 14011.0 Ao*Mo1(

4.181( K N.72

W eN 63S8 AbI

FIG 4

EAL SIRAIN CAM[I BAIAC

o O

25 ID

0 2

00 20

8011-, .) 010

5 it(~~~~~ f(A~ RM ~ ~ ~ M

")tO ltWAA

4-37

TIC 3.

AIRCRAFT MODEL WITH 2 AFT-MOUNTED NACELLES

000
AC 0 0 .

L0
__

.
_ _

I
06

WITH NACELLES WITHOUTNACFLLES

07

VC

0a

00011 Co02

0 001

CLj 1 0

01
O 07 IA oe

PG 3b

STING CORRECTIONS, WITH AD WITHOUT NACELLES FOR 4OEL IN FIC 3a

i/2
/ .

FIC 3c

AKECKrT HOVEL

|TH 4 AF-M-O10TEO NA.CLLLIS

4-38

2.0

co a~l D

Ado~e

,o '
0. RM

.2.0 * ' SIORtFOC REWiOUAL

A ILMO AXIAL FOACE UtA*) XI.'AXIAL

iAL

ANAAL

FOR

to
e AXIAL
FORCE

A E.L2ENORM FORCE

FORCE I &= 400JE GMAN &MOEX '

AXIA N FORCE N_'OA)

-2 0 2x0 APPLIED PITITNGLOMELhT

0025

FIC

BALANCE CAIIPRATION

ACCEPTABIU RESUITS

FiC 6b

AL NCE CALIRATIONLNACCEPTABLE I

SIOLTS

. ... . . ..

..
OALAUCE

R O CE GENE ATORS

LOAO)NOTREE

MESURING ACHINE

CI 7.

IRAIIIONAL BALANCE~ C.LIATI(,N

F1*.7b

III% lUA

ALA.NCECAL16MITICAN

4-39

12

0 0 04 08 1.2

16

FIG 8.

ERRORIN MACHNLMRER TO GIVE I DRAG COINT ERROR AFTERODY DRAG IN (FROMIREFS 29,30)

I-

IsDX u

--

I mo RX

MAXIMUM OfIANR .

MtAII fl*.CI*O

01 A

0 02

00 000 I

rE0t

101

2930

4-40

Hetght above
1100,

-. 2 0

-00

-02

0 C.oe.

02

04

-01

cglo. do

FIG 9

MFASURED WIND TUNNELCROSSFLOW 1 tliSTRIftIMON (NASA AMEAsIIx TUNNEL. FROM4 REF 1ii)

0 04

06 M-of -ob.

00

1o

Oo.Wtd tunoo 0&11-o,

02

000100,0p -047

-058C06

01

03 d-ll 0.01000

0'.

0 % Polo, from in c offs 00o1thod

MOooot Typi.Mt

. ;,pot.%' 01nc~g

02~

IT

-.O A

0
litIC

0203
Pot., from dmg polo, ,othoo 0 07 03 04 0 5 00
method

E,,o~ 10% o. t U.0000,0.0000Ca.0

CL

FIFFM S Of VARIAtILITY

IN M~oMl NtUBFXE I10 SIMY.-IL2tIAL .L.'1XL 'ROSS111 LIUt)MiNATION prvmo Pt itt

4-41

P rb

Measured Oata
-Correcte

Data Id

on CL

._

_ _'_...

_ _

460

470 (L- C?

80

TIME 520 seconds

530

540 CL-0 s

550 TIME sec,--s

TIC 12

Md 0 8 DATARECORDSFOR TIMIE VARIANTO

T"TINEI., FROMREF I5) (ARA TRANSON~IC

4-42 Tunel AEWC 1 16T Turn AEO, Tunnel {WallsTTred' AT 4T AEOC Tunnel (Wals Tape Screen) 4T with or ONERA 6 S.2 Modone 6x NASA/Ames I TWT NASAAnes TWI 11 (Walls Taped) NASAAmes TWT 1 NASAnAmes (Wolls 4 TWT Taped) Calspan 7WT B ARA Bedford Ltd 9x8 Sm 0 '
* * a

o 9 V r" 4 k o 20

j9

Tunnel NASA/As - PT RAEBedford x 8 SWT 8 NASA/t~oley 16T NASA/Longley I6TOT(Test NASA/Lonqt% 8 TPT Idwe NSR&OC7x 0T NASA/Lagley SPT 4 RAT Bedfordx 4 HSST 3 NASA/Aesx7 SWT 9 NASA/Langley 4 SUPWT No iTS 1l ;light Data Fig17

rFreon

Mean FairedLine

oe ist r

-oaaeBy Lao-tT

IBond.j

Overalliee

002

004.

006 0

ol

02

04

0-A

yB 10

20

30

40

100,per

-ent

FIG IS

FeceM TEATS WITH I0' CONE (FROM EE" 33) 7TBATSITIcNBr1fNOL[o N~IlBO DEDUCED

---

O'rl ARC q(urs - -- 30 rel t2 PT ci


ARfC 4 TWT - - ->4 __ _. RC 8TPT A FLfGHTDATA ENVELOPE

00

ATCll

TWT

A-

Open symbOls Filled sy,bo s

da Bus Tubuleuce (mass fluctuaton) Peesure Iluc'2tl0R data

60 5'0 40
""

-A A.I.R P. . 1 5 1 . .... 8 T b

, .,,
O 10 2 0 40

2 005

0
1Ppu S0 AC

02

RCprms,/

-I

A14

RN

4-43

-0-12 02
Acf
Doo

0 "I, Ha
0

0H

08 _o

Cf o
01 OtV,,,

-006
-0 04A /\, H

0 C : 47L T,=_'';

= 02 0

a
02

*20)01 (U1l 100/(e Ue 6995

ACf Cf
0

with UYUe: turbulence intensity


Leu = length scale of free turbulence (05< LeU/995< 5)

fIC 15

DEPENDENCL SKIN FRICTION AND S.APE FACTOR TI.NEL TURBULENCE OF ON

M 0$0

is>
F
-

01

~Nj
F,C 16 DEPENDENCE RMS LNSTLADY OF WING-ROOTSTMAIN ON SVRFACEOF UINNEL SLOTTEDWAJLS (FROM REF 41)

4-44,i

N
Mred C

CCO

Stollb.9'..

\C

FIG I?

DR M

ANALYSIS FOR A LIFTING WING Uncorrected Corrected

SIC -0,07
S/C -0-13

0
+

0'9

0'7 C L and

C
--

LLEL

CD
0'6

05----

:;1
0
lit IN

TI7
IS 20
(a (deg)

10

25

30

35
IR(, ill 42)

APIR AItON Of lIOAXACI CORICIONS AT H41(1Lill

p
4-45

I.1

UP'STREAM

DOW~NSTREAM .50

FIC 19

CIce AG. LONCITUODIA. DISTRIR UIO OF WALL TUNNEL IN A 2D PERFORATED REF 48) (FROAM

ABLOK

-0 iS

0,

oDip

FIG 20

LIlT TIhNEI CONSTRAINTFOR A SMALL MOISFLIN A PERFORATED CIRCULAR TINNEL (FROMR F 42)

.. \Ctostd Tul am

K 0

R% I --

// 0." Iev to.

Assumed

for

AI ptSst

PK..
K 0 5 .

/a
. .

AotiO1.,

o.0s',. 05VST . .

. to.k

UPSTREAI

-05

AM4 LWIIITIE

l 1(, 21

CORRCTICINS BUOYANCY BLOCKAGE FOR ARA IR ISONIC TIINNEL 1

FIC 22

OPEN-AREA RATIO REQUIRED FOR 0 OP INCIDENT SIIOCK CAMFLATIO WAVESIN TlL AKATIDLNEL

4-46

V
WALL PRESSURE MEASUREMENTS1
0

WALL PRESSURE MEASUREMENTS9 CLASSICAL METHODS 110

CLASSICALMETHODS

(o) -x
1816

(b)
0/cLo

()

-108

I4
12

/-

107 .....

-1

0
02.
0

7Z7
7

103
. 102[ *, ::

"Of0
6 10 14 18 22 Geometrc Ind0nct 26 (gsl 30 6 10 14 16 2 Geo etra tntdenI.( deg 26

FIG 23

COMPARISON 2-COPONENT WALLPRESSURE OF MEASUREMNTS AND CLASSICAL METHODS FOR PREDICTION OF TUNNEL INTERFERENCE (ROM REF 66)

0013 0012. 0012

~
0

546

1N0) LE TC

-_

2-co0po1tnt ,thod i .sc . L.n.r Theoty

of

02

03

04

05,N

06

FIG 24

COMPARISON PRLWICTIONS BLVEACL CORRECTION,20 MODEL. Of OF RAF 8 T R F1 T INNFVL, - 0 71 (PROMREF 64)

4-47

d~p/dg

bA

7.0 10.0 '40 A z0

0L125 01 A LI 0.125 015 7012 0125 3S 0250 W%4 .

55001142 1.1

FIG 25

PERFORMTED WALL CHARACTIRISTIC 60'

INCLINED HOLES (FROM REF 68)

0000Cll021 040 a MVS 0ll(tx(.

~COOSLClto
50410 a4 a 045$

toy C"

2f~

03

04251A,4310915 .7 "M. C5 25251S14AV4051041 *1 52111

VWIAC-ASMR-sl0ot 25252N I :3 I

0324f4 028

-'
DATA W200110251

028 024

11 DAT t-

152.4 A-

0201 30

to

CO

1.

(U)00b)
30 1.0 a0 5

FIC 20

C(#IPARISON OIF NAVIER-ST.ES AND EULE PREDICTIONS TRNU LIFE CC46sIAINT CORRECTIONS OF (A) P'RE TEST (b FOST TEST (FF602REF 37)

tOltS taOSe 5000lrots. oto 57\ 0,0 coiVo and

asofomty

lDotny

-gS

__

FIG 27.

REAR FUSELAGETO OBTAIN STING CORRECTIONS DETAILS OF BALANCED USING CURRENT TWIN STING RIG AT ARA (FROM4REF 15)

II

FIG 27b

IWI TYPICAL CLOSELYCOUPLED FUSLLACFLAYOUT RIJOIRINC TEST ON LSHANCED

STING RIG

/
w"

/ / /-I,

C/.-

///
/

/
Roltt,O Adaptr. Wllg
Ostol S.q 0000t'it

fIG 28

,LO CFNSLRALI*.IOKS OF i~OOFS iN FAHANCl 1AIN SINhG IG

~4.49

----

central strut gurd and dummy strut

FIG 29

(a) 3-STRUT ARRANCD4.NT SINCLE (b) STRUT ARRANGEEt T LAYOUTS STRUT FOR INTERFERENCE TESTSIN RAE5 METRE TUNtNEL

ifs

ac

'a

... . .

;.

0L

i.

.1

FIG 30

c(PARISott OFGUARD INTERFERENCE CORRECTIONS

() --.- Transition- fixed


--- Tronsition free

/
Duta valid betweeni Boundaries AA and 68

AA Shocki 015c off of transition trip approaching trip 86 Ditto with Shock fromrear as a result at shock Induced seporation.
-

drat
-

-- ,J/"

- A Alt . ovemont transition of is YmpathY extension with of supercrrtical flow

0-5

07~

PB 5 DragOcrep CLBoundary

Buffet-onet

dQ.0O17Strarltioafree 06 At nt tronatn.ni of o s"'Pathy wit 0 4 Ixtiruilori at 02 supericitlcul /o ' -2 0 2*

iIL*0-Itt tranusition fied. genin result

Limit o Subcritical Flow

()LitCurve Slope

Prior ToBuffet-Onset
FIG 31
EXAMPLES SPURIOUS Or RESULTS WITHFREETRANSITIONi FIC 32a LIMITS 04 RANCE VALIDDATA Or WITHA PARTICULAR TRA~SITION TRIP

-10 -0 6 -062C -02 /


A

Trips 0 25c

wit

6CL

trip at a 0t15c*0,25c CO ---Rorqess%&to Infterwloa laid ofl betetian i d a aO1*1110*1 islIeferoiq lhe wane dm;

/uiri a
-F-, CL

tinOesCndto

2Pointl TraosltioniSveepPossibI# InRone 2 3Point transition Sweep Possible

FIG 32bi MIETHOD [ICIERNININIC FOR LOWER CL LIMIT TORANCE VALIDDATA OF

t IG 32c, tSEFIUL RANGCES DIITERIIIT WITH TRANSITION4 TRIPS

4.)! T crtero. of

A02

A..

Tr~fbnftOW

CIA

0.02

Rw

nf

Lo 021-

--- R-4-

MEASURED DATA.TRANSITONSWEEP. FORWARD -ArT CNE$NFArTTRAI4S*TIONTORmipFORTEST R

(A) (8)

Renod -w (~t tVmAiton led" tdgt) new -s~n svoSftp (e~tropolatdto None' RH "sbhw)

EXTRAPOLATION Rtuom BASED TO ON , (I) UP M EASUREDTREND TORem PARALLEL COMAPUTED TO TREND FROM RCWtT Rot TO

FIC 33

EXMPLE OF TW0 SIMULATION APRACHVq

FIC 34

FIRST STEP IN EXTRAPOLATION PROCEDURE

RFLIENT

atit

FURTNEST AFT
A4TERMEDLAI!

afern

FRTHEST AFT
ITERMEDIATE

NEARLL

0-4

_____Lfto

(A AP.F XA1O WII

L)O

--

MAm.I~ - AI -PI

R4wUP (voublowwIwoKqwRQ

CV.IUbopI"m A SWELP

wet,

SIMULATION TETS EXTPOLATION ATER RA4QM FWJ.MOO(LWANEFLLOE R-K44C HVIAOWEA

SMLTO ExTRAPOI.ATFORAFTERTSS

RAHOE HALFMODEL

FIC 35a

SIMULATION SCENARIO3

FiC 35b

4 SII4X)ATIEN SCENIARIO

4 -52f
+ 17 V~~1
.-

RESULT$ _EAUREq

-CALCULATED,

CALCULATED, x ,* roos U1N 7- CALCULATE, 605 XI XTM'UO R 4W

4_-1-MI'X.0

F 03UCE

SCXMN o.to,* . 11 N

WoUUT

0U10'Xvk.UU

B-08

-.-

R.6.SXU'XmUU p

06

06-

0404

020 0o 0 22 o X 1 0 6 04 0 2
4

'.

-06

0 "

0-.04 I0.1 CL M . 0.735 06

COO M .073S

THEORY ALLOWIN STAOO VISCOU-IWD ITERACTIONS FIG 36b FOR FIG 36s APPLICABIILITY ZERO-LEVEL or CRITERION (FRop4 s6) "r CH04OOu, C1r ) SUOGESTS

APPLICABILITY OTECRITERION' Of CL

0 4

1~ ~
10

(0 1'.UOUI1I'0.&M 16,m

4.%I

XTm.U2$
0 1. 0

-( I-

10

00010U.1.3

09

It..

30I~ X

Xm
--

11

005

I 07 06 04U065

66

C720 68

025 2

020 72

02 74

Xvi ItgR

035 7$

OLD0 S

FG36(4)

CEOSEQIICE FAILURE SIMULATE 0Of TO UUPERCRITICAL EELOPMENT FLW~ (FROM EEF.86)

05

01

0'2

03

04

05

00

FIG 36c

APPLICABILITY 611i OP CRITERIONCL - 0 65

4-53

WAII
yLIS Whoolc OF

FIG 37

TYPE+SOF SIMULATORI (FROMIREF.9)

TANX SELINO&

?PC% )

MASP M LATO E S OII.4U FO

MSTFL
j(2O~~)

L F## OWWV
OT oZSLE

94'I

5 A A LCI SI

>PctL/Y

NACLE
TION SUCUL

HW FO

NAr E,

4 (JFO 0 0W

FI AAMCSLATO 9

AK

4.54

MEASURE ObR, DEFINE

CGMtEt

WFAW- WM

WTI 1

I. A

W.

V N

t co

O K W

,.

UCY

PTP. TP.AP.-PS

CDPv'

p.-

BL-G

XGF

am ""T

Xoq(f CTw. (PTF/k YiWFJVi)


CoimlnCo FrmMTBoxjd Thu PVA. TVA. AVA PTP, TP. PS,AP

CTF
%W

PAM
MVDMN

nowit

CDP P~o

XGP f (PTP. PS)*WPfI~I-CT RPM TI CRPIC-W-I (CRPMI)


W1.CIc

XGP GA TM CRPMI -1t Cwt


WIt

m""{Opm
r"
Pim

T. Ho. PSI. Al. goI

W12 0m iF WFAN Fm=L


XGF Fm
ISI

PTF. TF. AF. PS, E


XGF. f(PTFIPS)* CTF *Wjrff ORA M .W F* M*I T AF. NF

DRAM ow Pm
,U

mk NF A

UFT

N.XFGXP nLAF

NE

ric 39

BOOKMEIIC PROCEPME (tROI REr.9)

... IS

I.w

coa

ntc

an aW

Sa.ll

sct.tSIIsC 41tI~t FIG 4 MODW FOPTURcS A PERFOM I 7 CE

Ia

EST IPC

FRMRF.10

]I% OWIS Alntu


ta OWQ, 1 C
1

Iflt

& I

II

t7a'I/II

14

WWLLSOF~ FVOM =INF~4A1TI SC (FRMU RL.t)

4-56

I*TII
&TO AW

I'V4'

SIN

KI

00.

IOLA , BAI,

Kit IMEED "rN


$LItA00

SMALL

I PEBmynT ttLt~ IO\ SLM

TRNOI

TUNE

(FRO

RE(WUMOM

AW N
EU It M!
0 0

wt./-

A
A
w!o4*Q 11"It 00

0~1W h~llow.A '~ 1*0*RT

o 0
N

LI

WO 10

411

1 .A

I'U 100((0'0

-"lT- -

noam

mt*IOIOA

C3

Vow

M
A.OoT l0 04 UA1

FOR DESIGN CONJIRA-ROIATINC ric 42(e) BALjA(CE PROPELLER

FIG 42(d)

FOR TESTP'ROGRAMME MODEL TYPICAL PRFORMANCE PROPENIER TUROPROP (FROM REF.105)

4-57

-Denton S---Exprment 2:o 0 " ,/04 646ton Experimentxperimnt /'C / - - -T AI sti Net x 'BladefThrist / + Apparent Thrust // // -s

rp

,I,

//
o
+i

60

64
BLADE ANGLE

600
BLADE ANGLE

64"

FIC 43

PROPELLER pERFONIJTHC: EFFECTS THRUST or DEFINITION (FROMREF.111)

BIILLMOUIH LtP

PERWH.RAL

FAIRI

ACOUSTIC LINER

/SEAL

FIC 44

ELECIRIC-UOIOR DRIVEN lRCE SCALE PROPELLER IN ARAACOUSTIC RIG IRANSONIC TIUNEL(TFOM REF, 1OS)

4-58 ENGINE STATION FLOW SURVEY (I)

()
SCREENS
* * ..* *

.....

**.. *. *. * .

SWIRL PROBE RAKE 56 To1 hea itubes ........................................ lFivehole probes 8Duct tatics *ErdeT 3 1u~tSIat"~

MASS SONIC rLOW THROAT) -/ (FIXED MEASUREMENT (3) SUPERSONIC XD N RA (ADJUSTMENT FLOW MASS NOZZLE 4 (SONIC THROAT) (2) FIG 5(a) ISOLATED INTAKE MODEL INSTALLATION
t{

.
':

DYNAMIC DISTORTION RAKE


* 3

t
,

32Totalheacltubes 32Endevcodynamic tdxs


ErdICVCO stitic$ duct

INCIDENCE ROLL AND A04JUMENT BLEED FLOW ADJUSTMENT

(b) BLEEFLO * BLEED FLOW EASUREMENT ...............


* .

PERFORMANCE RAKE * * 88 TOt head tubes 8 Erdevo dynamic tdi's 8 Ductics 3Endevcoducltstalcs

.
*.

.*

ENWNIEFLOW SUR E

OW'J (SONICTHROAT)

ENGIE FLOW MASUREMENT

FIG 46

TYPICAL ENGINE FACE INSTRUMEIN4TATION

n.;. NGINEFLOWADJUSTMENTr

FlC 45(b)

INTECRATED INTAKE MODEL INSTALLATION

CRUCIFORM RAKE PITOT (ROTATABLE)

EXIT PLUG (TRANSLATABLE)

U FRWEIGNEDPART

(a)

STATIC HOLES -&p,

p.-AND

EXTERNAL

MOMENTUM

BALANCE

(b

ric

~ ~

ASFO

FUSELAGE ALE LWTFRMAUIG FORC 47D SRV NEASUEMENT

FIC 48

FOSSlEB SCIIL

FOR NTERNALFLOW URVE

~REPRESENTATIVE
AFLBOY NOZZLE

45

-TWIN NOZZLES

- --- - -- - --- -UNNEL --

O(=0 TO 6
M~O O 12CONFIGURATION

ALTERNATIVE

RESARC

SUPPORT FLOWCGWT~t4INGSTRUT/AIRFEED SYSTEM & MEASURING

TWT TUNNEL FLOOR_________

5 COMPONENT BALANCE /AIRFEEO

FEUE

COMP'RESSED AIRFEED

FIG 49

NEW RIC FORAFER1OOPUiFORMANCE TESTING

Blow Sting Stinort


FIG $O

Ving Tip Support

Blade Support

ALTE$WI 91CS FOR IVF AFIFUORDt PERFORMANCE TESTING (Flu). REF12)

4-60.

FIC SI

A'(ER300Y TESTRIGS (FROMREF.12)

FIG 52

CROSS-SECTION&L OFOMAP (FROM VIER REF. 133)

PANEL MEMODSORAERODYNAMIC ANALYSIS DESION AND


National Aerospace Laboratory N1 a Anthony Fokkerweg 2, 1039 CM Amsterdam. The Netherlands

HW.N. lineijeakerKa'

SUHMARY

of panel methodsused in the aerodynamic analysis and~deeign of aircraft oraircraft components. Panelmethods can provide the flow about complex configurstions and are routinely used,ln the analysls the aerodynamicsof realistic aircraft +of shapes. However., panel methods are based on a mathematical mode.in which,much of the fluid physis ictignored. The report discusses the capbilitios and limitations of panel msthcds. the basic concepts of panel methods, choices thxt can be made in the Implementation of the basic concepts,

An,overviev to presented of several aspects

It say be noted that in aircraft devecopmont projects the application of panel-mothodsis gredually shifting from the.final, designphese to.

wards the ptelimlinary design-phase andevencon.

&sawoll possiblectypes boundary as of condition.

ceptual design phAs o.,primarily du oo: - increased.demandson the accuracy of-predicted aerodynamic charcteristicsin~the earlier phAses,of the designprocess. the increase in computer capability (speed and memory),, decrease of it, costs and improvement of turn-around times, modern data handling techniques, - availability of graphic displays and work stations for visualizing geometry and flow aolu. tions. Several panel methods. e.g. Ref. 5, 6 and 7, have been developed and are in use in the er s. space industry that are variations on the approach d escribed in Ref.. 1 end 2. Other investigators extended the panel method to linearized supersonic flow. e.g. Refs. 8 and 9. 3ecausa most of these "first.generation "methods do have some repanel strictions concerning their geometric and asrody. nemic modeling capabilities and require improve. ment of their computational efficiency several of. forts have been undertaken to develop *second.

that can be utilized to creatively model subsonic and supersonic floe. The discussion also includes aspects of the accuracy of the approximations involved, consistent formulations, aspects of low. o and higher.ordar panel methods, etc. Also rder discussed are the computational aspects of panel methods and possible extensions to nonlirear compressible flows, coupling with viscous floe methods and application to other flow probleas.

generation, panelmethod. eg. Refs.


1.0 INTRODUCTION

10-li.

The computation of the aerodynamic chorac

teristics be f o re the of aircraft configurations has been v a panel methods since the mid 1960's. carried outaby il ab il ity following the pioneeringo f l a rge -sc a l e &dig ita l work of Hess Smith (Ref. 1) and Rubbert 4 Sasris (Ref. 2). But even
computers work vt done on surface singularity methods, notably in Cermany by Prager (1,. 3) and Martansan (Ref. 4). Panel methods are presently

c t SEPARATION FI I HT ENVELOPE
ONSET BOUNDARY - LAYER

the only computational aerodynamic tools that have of the subsonic and supersonic floe about real-C life, complex aircraft

been developed to an extent that they era routine, ly used in the aerospace industry for the analysis configuration.. Panel ath ode have also ben used for the analysis of the floe about propellers, automobiles, trains. sub. marines. shiphulls. sails, etc. Panel methods have been used so heavily *ncause of their ability to ,

C US
t MACH NUMBER

SUBSONIC TRANSPORT CONFIGURATION


Fig. 2.1 Domain of applicability of panel method
(adopted from Ref. 33)

provide for coeplex configuration. linear potentiputer resources required for running the computer
code as well as to the manhour cost involved in preparing the input, al flow solutions of engineering accuracy at rea. sonable expense. The latter applies to the com.

The relatively easy input requirement of panel methods, very important from a point ,ser's of view, is directly related to the circumstancs that a diseoretizationis required for the surface of the 3D configuration only. This Ia considered to be an order of magnitude saimpler than the vol. am. discrstization of the space around the con!ig. uration generally meded for finite-difference, finite-volume and finite-element methods,

methods Is War their range of opplicaility is restricted to linear potential flow. I.e. non. linear compressibility effects ate not accounted for and the important case of local regions of super-critical floe and shock waves cannot be treated. The latter occur in the high speed region in a large portion of the flow field and possibly at low speeds (at the larger incidences used i start and landing) in smaller portions of the flow field. To account for such regions of super.critical flow would require the use of a transonic finite-difference or finitevolume code (full-p-

The major drawback of the present panel

5-2 tential Euler) a spatial or on gridcovering the implies thata modelbasedon Euler's eouations, entire spacearound the configuration. However, up whichallowthe occurrence shockwavesas well of to the present timethe spatial gridgeneration as the convection stretching rotational and of problem has not been solved satisfactorily for flow.provides appealing an alternative. a loOn complex configurations configurations exas with cal scale. somekind of modelfor viscous-flow do. tended slatsand flaps. addition, computer In the minoted features willbe required. Thisspecifiresources required for present-day finite differeally points -e the flow leaves it wh the surface onceand finite-volumta codesare quitesubstenti(separates) vorticity generated subseand is and al. For the purpose preliminary of design, wherema quently convected into the flowfield(*Kutt& conlargenuiber configurations flowconditions' ditions*). of and Altho.'gh'the computer requirements of are to be considered, full-potential or-Euler -Euler codescan ,beearby the current generation methodsrequire too mucheffortin termsof far of supercomputers. routine practical application computer resources wellas in termsof manas of thesecodes(to relatively simple configure. now. rions) startsto emerge hou-. -Asecondidrawback'of current'panl methods is thatthe compujational eff~rt.'and cost,is proportional-toN , or-evenH , whereN is related to the number panalsoThis implies of that tht method becomes rapidlyimpractical.i.s. for currentmainframes N of the orderof 2000-5000, for whichtypically panelnumbers are required the for resolution required for thecoupling at panel methods with boundary-layer methods. Thu latteris relevant application the method for of during dotaildesign. For conceptual for preliminary and design studies, whereconsiderably detail less Is required-the number panelsis typically the of of orderof a few hundreds. However, Is alsoan it experience the practise applying of of panelmethods that thisnumber panels oftenquite of is easily "consued', so thatcompromises have to be sought regarding resolution accuracyThe and latter requires insight intothe flowsolution, i.e.aerodynamics, alsoa goodperception but of the numerics involved panelmethods, in in the presentlecture overview iven an is of several aspectsrelated the formulation to and use of panelmethod. and the possibilities exfor tending domain applicability Improving the oi and the computational efficiency. discussion The deals primarily with methds for the giuiiLrj. about three-dlmensional configurations subsonic in or supersonic flow.It Is basedon the literature on the subject and on past and current MIA research, The latter i3 aimed at the development of a higher-ordar panel method(AEROPAH) of a panel and method (PDAtN) to be used in preliminary design studies. The first pertains to "the IX panel method-, the panelmethod workhorse used in theocetical suLsonic aerodynamic analysis duringthe last two decades and the NIXAERO panelmethod for flow. sub- and supersonic If the shocksarenct too strongand if the rotational flow is confined compact to regions. the flowmay be modoled Rgtggntl5glgy with emas bedded freevortexsheets and vortexfilaments. Now the rotational flowregions'are -fittedexplicitly intothe solution, rather than'-captured" implicitly part of the solution is the case as as for aboveflowmodels. Flowseparatiin trailing at edgesand at otherlocationahasato'be modelled through Kuttaconditions. just likefor any other inviscid flowmodel.Although has-to one decidea priori the presence of vortexsheets on and cores ad generally topology the vortexsystem the of mustbe well-defined. *fitting' stillrequires thatbot the position and strength the vortex of sheets and coreshave to be determined as part of the potential-flov solution. The treatment vortexsheets and vortexfile. of sants,freely floating a fixedspatial in grid. posesconsiderable problems for finite.difference/volume methods solving the nonlinear fullpotential equation for compressible flow.The conputerrequirements full-potential of codesare relatively modest, but application general to aircraftconfigurations hampered the grid-gneis by ration problem. A special classof methods for compressible po-.ntialflow is formed the so-called by transonic perturbation (TSP)methods. Thesemethods are boesd on en Approximation of the full-potential equation with some of the nonlinear terms retained and. with to the sae order of approximation, the boundary condition applied on a planar reference surface rather thanon the truesurface. This eliminates necessity curvi-linear bodythe of conforming gridsand a mich simpler Cartesian grid can be used,e.g. Soppa(Ref.19). in case ehockvaves are Absentaltogether end the perturbation the free-stream due to Ch on presence the configuration is small. the potenof tial-flow modelIs further simplified lhnearlby stion to the ifnear-Dotnrial model,governflow ed by the ?randtl.Claaert equation. Note that for Incompressible the saall-perturbatlon atsumpflow tion is not required, the irrotationality conditiondirectly reduces Euler's equations to Laplace's equation. the linoar.potential-flow In modelthe flowand the position the vortex of sheets and filaments can be solved for by employIng a bomgauJnvinerA nypaof formulation. In thisapproach singularity distributions the on surface the configuration on the vortex of and sheets are employed simulate to the flow.The formulation requires discretization just the the of grid Is not rei.e.& spatial surfaces, bounding quired and In thatsensepossesses a "dimensionloweringe property. The linear.potential flow medel is the model underlying the classical panel method, it is emphasized at this point that though the governing equation in linear, the problem is still nonlinear because the position of the vortex

2.0 2.1

HATEMATICAL MODEL

Zkron

In the aerodynamics of aircraft configuretionsthe Reynolds-averaged XggzX gku aquationsmodelessentially flowdetails. all However. turbulence transition and need to be modeled a in manner appropriate to the flow considered The computer resources required for numerically solving the equations a mesh thatsufficiently on resolves boundary the layers, freeshearlayers (wakes), vortex cores,etc.,are stilland for some time to comewill continue be quiteexto cessive, For mea high-Reynolds-nmber flow of interest in aircraft aerodynamics viscous effects are confined to thin boundary layers, thin wakes and centers vortex cores, i.e. the global flow fa. of tures depend only weakly on Reynolds yumber. This

5-3
sheets appeaisnonlinearlyIn the, boundary ,condi-

tionon'the, solidsurficoa< wll,a. in.the boundary conditions on the vortex shiet itself. It can be argued that for mast configuration. In cruise condition the wakes rsinfairly~simple, i... do not roll up within one-wing span 'down-stream of t al artached flw model' in which the voergexa) sheet is chosen as some appropriate, user-speodfied xjgiW'surface, rendering the boundary conditions-and therewithW he reulting problem fully linear. 60p .

C
CNE ANGLE

th;Ing trailing edge. This .1.ads to .ths'conven-

50.

TRANSONIC FLOW

40.

L"

SUPERSONIC
FLOW

oAs far a forces and moentsis concerned--0


onlyNavier-Stokeenthoda are capable to p.edict 30-

the total drag. the other ethode will predict the induced (vortex) and the wave component; ofthe. dg.20-S
2.2 Do:siL of snnlicahlOitv of 1rnol sethods The preceding discussion will have made clear that a number of substantial assumptions had to be made to finally arrive at the framework in which the panel method may be applied soundly. The assumptions are summarized here as: - high.Reynolds-number, eseentially inviscid flow -no flow separation other than at the trailing

0 LARGE

ENTROPY EFFECTS M

DISTURBANCE
10 LINEAR
-

THEORY
0 1

edges
compact regions with vorticity. i.e. thin wakes Sincompressible flow; or suall-perturbation compressible flow without shocks.

MACH NUMBER
Fig. 2.2 Attached conical flow on a circular cone at zero incidence

The restriction to small-perturbationcompressible flow implies that the linearized potential flow panel method applies to configurations with eln. der bodies and thin wings at low angles of attack and sideslip. The perisseble non.slenderness, wing thickness. maximum angle of attack and side slip depend on the free-stram Mach number K.. Nto restriction for %4 - 0. severe restrictlown at high free-stream Mach numbers, In general it can be stated that the pinel

The nonlinear flow effects can be sub-divided into effects due to large disturbances and ones due to non-isentropic flow (e.g. bow shock, strong normal or oblique shocks). As far as the nonlinear effects in attached supersonic flow is concerned a guideline can be given by considering the supersonic flow over a circular cone at zero pressure coefficient, which for this case Is constant on the cone, computed by linear theory. full-potential theory and from a conical solution of Euler's equationa as can be found in shock tables. It Indicates the boundaries of appllcability of the flow models. it shows that linear theso. ry Is valid in at least some part of the attached flow regime. This has been expressed by Stager (Ref. 20) in an Interesting way. iars this Is reoexprssd as follows. An aerodynamic characteristic F of a configuratieo can be expressed in an asymptotic fashion as:

method provides detailed but stmplifted aerodyn aos for tolcx configurations. This is Illustrated firthar in Fig. 2.1. which shows, for a subsonic transport configuration, the K.,- plane. Curves indicato the first occurrence of flow fea. tures such as shock waves and the onset of boundary-lays: ",)aration. The shaded area indicates the domain of applicability of the convontional panel method, possibly coupled with a boundary. layer method. The method Is restricted to sub. critical attached flow. The shaded region Indicate where the method will give reasonable re. suits, while even for the indicated design cruise condition trends in the aerodynamic characteristics will bo predicted correctly to some degree. It will be clear that extension of the panel meth. od approach into regions with transonic flow, without sacrificing its ability to treat arbitrary configurations, would greatly enhance its value for the aircraft designer, Another area where there is still a gain to be realized is to Improve upon the treatment of the wakes. For configurations with extendad flops or for combat aircraft that operate at higher Incidences the rigid-waks approach adopted in most *first-generation" panel methods is rather moade. quate and ways have to be found to account for the effects of non-planer wakes Interacting with the flow about the configuration.

F - Fo(geometry. X., a....), Fl(nonlinear cospressibill.ty. viscous effects....) . where F. is the leading, 0(1). term and F the first-order. O(.). corction term. The 01) ter is provided by the pane.n-ethod solution, which can be obtained relatively easily for complex go. omtries since it requires the discretization of the surface of the object only. The 0(6) term Is provided by he finite-difference, finite-volume or finite.element method solution, which can not he obtained so easily for arbitrary configurations because of the necessity to discietie nt only the surface of the object hut also the entire space surrounding the objart. Of course one should realize that F1 . the first.order effect, might doinate F . This is for Instance the case for the spats flow about a sphere, the transonic flow with strong shocks, hypersonic flow with real-gas

effects, etc.

5-4 2 3 o white-the preossure~coefflclentfollowsfrom thelinearizabion ofE2lna -linp . )l . +o('2) . .. :(2.3c) 3c)

The fullpotential equation wi h is with u the total velocity vector and 4 the'totil w velocity ential, i ttena po

whnere

whee -

4(1 -

2/)

"

(2.1b)

(2.3d)

is the density, the ratioof the specific heats (- 1.4 for air),whileU. and p. are the magnitude of the free-stre velocity m and density, respectively.The pressure coefficient follows Cp fromthe isentropic formula: Cp S.q,. p (2.1c)

When in practise the perturbations the free to stream are not small, Eqs. (2.3cand d) may'attain nonphysical values, I.e.lowerthanthe vacuumor exceeding stagnation the values. most'omethods, In usingEqs. (2.3cand d) the computed valueis li.values: itedto vacuumand stagnation

I) (c P vacu -"
and Ip )stag- (i

(2.3o)

Inc 2 with C - 1 (2.d) S Ur .heA q, 2 hre U' and p and p are the localstatic pressure rihe free-stroa static a pressure, reapectively. denotes C the pressure coefficient In incompressible flow. In the caseof incompressible p a p.,. flow - 0 and Eq. (2.1a) reduces Laplace's to equation, i.e. with u (2.2s)

2,/(-)

11

(2.3f)

way to extend of linear natural flow methods the capability potential is the applyEqs. (2.3a.d) everywhere the flow field in wherethe perturbation velocity smal and the full-poten. is tial flowformulation. (2.1a-c). the to. Eqs. in aining smallisolated regions, coupled each to otherthrough the appropriate boundary conditions on the commonboun.'ary.
2.4 od *" odtiq

(i)

wherev is the perturbation velocity potential. one finds: 62 a, OY


5

e.- 0

(2.2b)

whilethe pressure coefficient follows from 0ernoulli's equation, equivalently Eq. or from (2.1c) for the leit of M. - 0: C _C P E P
n

_I "r

0 t

,12/U

(2.2c)

2a-tflLM1&rtiLb of the configuration (Fig. 2.3)the condition imposed is t.atthe nor. MAl component the velocity of either vanish. es (solid body)and is a stream surface or is prsctibed. The latter amongst Is others required to: a siulate the flow through inletfan an face; - simulate propeller slipotream effects simulate jet entrainment: incorporate designoption; I a account for the effectof the boundary. layerthrough transpiration the concept. etc..i.e.

In the case of c=xtLIhjjj = Eq. 12.1a)is linarined under the assumption thatVV Is 0(e). In casethe freestra is directed alongthe x-axis Eq. (2.1s) reduces the Prandtl-Clauert to equation:

SwW KE'

3 5 a. OY2 40 nsNote that this equation is elliptic subsonic for fras-strea. Hach numbers (X. < I) and hyperbolic Z for supersonic free-stresa Koch numbers (m., > 1), Withinthe scopeof the linearization may be it assumed thatfor mell antles incidence of and sideslip compressibility reains the -he axis y sss. i.e. the ..axlsrather than the direction of the free stream ./I0t. For Incompressible D flow the Prandtl.Clauerr equation reduces to Laplsc,"s equation. Further note thatexpanding Eqs. (2.1a. S c) in the Rsyleigh-Jomen expansion for small Hach number,i.e. 1,, 0(d) alsoresults LUplace's in Fig.2.3 Airplane configuration equation. To the same order4f approxi~ation usedfor Eq. as (0 Oe). (2.3a), Eq. (2.1b) educes withEq. (2.2a) to

(2.4a)

l l1H2(k. ,)/U + 0(,2))

(2.3b)

where; is the normalto the surface v and the givennormalvelocity. n

5-s"

1i

(ii)<Onthe waee'vortex-sheet S 1two conditions apply, the'streaa surfac*Eondltion4Eq, (2:4a),vith'v-- 0 and the-conditionlthat ths'static'pressure'is continuous'acrossthe vortex sheet, i.e.

sheets.jncase of *relaxed.wakes!boundary~condi tionEqs: 2.4a and b are~lineari.andnildlynon'linear in terms of 9, respectively. but~both-conditions are highly nonlinear in terms of the, also to be~solvod for, position of the vortexwakes.

C()w p SC C p S)

(2.b)

2.5

reoresenarson of the solution torserml The solution the potential of flowproblem

,which as follows'fromEq expressed'as


p

(2.1c) carnbe

(S%)

C (S%) -0 p '. -

(2.4c)

may be represented. through'Creen's$third identity, in terms of singularity distributions (source q and doublet p, see Fig. 2.4 for the definition of p) over the surface S of the config,.ration and the vortex sheets S in the form (see Ref. 21 for incompressible and lef. 22 for subsonic compress. ible flow):

(1O+ + which leads to u.W

4) - 0

(2.4d)

S)

- #') - 0

(2.4e)

ith ea the average velocity ccross the voke vortex'sheet S .. (iII) At the trailing edges of S the Kutssos*dijW is applied that the fRow leaves the surface *s oothly *. (iv)

A ininity upttrem the perturbation


vanishes. The free-stream velocity vector mnayconsist of (constant) components dua Fig. 2.4 Definition of doublet distribution

to Incidence and sideslip. but also ones due to (small) steady rates of pitch, yaw end roll (p,q,r). Sometimes a user.spocifid onset floe (end total.pressure increment) is addedto model for instance propeller slipeffects. Also other incremntal ye. strea

) +

(o were

" q(

(opo

.q(2o)

lcity components can he used to model the


floe about configurations which parts move with respect to each other (e.g. etors separation). So we have: k(;) - U i(xcOscos . ;ysig * stsircosp} V(;o) -

q(s)--M
Sb

IRI
1 (2.5c)

are the velocity potential induced by the source and the doublet distribution, respectively.

s*(x) D()
...

In Eq. (2.5)i - lB](;4 ).


(2.4f) mtrlx JB is defined as

di. is the normal

rected into the floe fiel3 and the compresstbhlity

where aoi the angle of a&tack,j the angle of aid* slip, p - p q. + re is the rotation of the configuratyon etout the axis of the reference coordinate system, at, (;) Is the onset flow due to the slipthe s1iptroaas. 1) Is the Onset within X to the o a propeller. be appledVelocity
5

0
10 0 where B [-1. .2 1

01
.J and

(2.5d)

1 I0

0
/0 1/J

due to the relative motion of a part of the configuration with respect to the global co. ordinate system. (v) No upstream influence in supersonic flow.

1.a (2.0)

Above boundary-value problem is a problem in whick the governing equation is linear but the boundary conditions Eqs. 2.4. and b are nonlinear. The resulting boundary value problem will be linear only in cass of the approximation in which the wake Sv is assumed to be a fixed surface in assumption) valid for high. L. Space (the anpect-ratio. lightly-loaded wings, the wake vor. tex sheet not interacting strongly with other coo. ponents of the configuration or with other vortex

The velocity p..ential as defined in Eq. (2.5) atisfies the Prandtl.Clauart equation Eq. (2.)a) exactly. They also satisfy the far-field boundary condition. lb. integrals in Eq (2.5) have a sin. gular Integrand, which results, for regular q and s. in a situation where the potential is regular everywhere, except that it has a Jump across S. i.e. t(s5 )

e(
V 6
0

l() ,S) ; 1 0

(2.f)

here the superscript P sanotes the so-called Principal-Value of the integral.

-A-

5-6 _.The.velocityfield induced by the singulaci. tydistributions,folluws Eq.,(2.5) from through 4differentition, mE
* -or~along

where .- H I( 3n ,.h~ocalle4So.nrsl The vector.mis,p*rallelito hnoainnae of Incoepreosiblej.low in csse,ncssnormai to &nd the ,x-ais-(- tomprssiibilitytaxIs-).

wh

ere ffi;+3S() (2.6b)

jupi acros t

hevictyptnia

ssleydet

Isxcoinceus azump e oubetdistribution

0 with 'U()- W-ff([BlI)

v0normal 0

Mlet distribution has a jumpjin tangential in and. direction, since x 6.;;Z ;.,); In the case of incompressible the velocity flow (2.6d) in tetngential component only. This feature has as consequence that-modeling vortex sheets (which only)In velocity tangential a jump In ti's bsible flow requires a composite singularity of srntp dasoredisirio~tion of strength a O q--

X 3( Sb+. 1Phave snS~wcompre


-

~).~X((Bjd7(;))
k in /n

(2.6.)

B. v * 2

n,-Ln 1

U(SI)-iJ(S) SOURCE DISTRIBUTION


Fig. 2.5

DULTDSRBTO

Jump conditiona across sourcea sn. doublet distributions made to the so-called (Iinerixd) ~.Tems-lxi odfnd& sI eie h ~sfu z In the literature sometimoes reference Is afl ve

Hers we used the equivalence between the velocity Induced by a doublet distribution p with that inducod by a surface vocticity distribution ;of 4 stegh o plus that Induced by a concen. trated vortex of strength r - p alongthe boundary ofS. Both the vorticity vector ndthe grain ofthe doublet diatrihution Vp are tangential to the sur face carrying the doublet distribution. Note that In using the equivalence property it is assumed that p Is continuous everywhere on Sb end Secpt possibly at its boundary as Temain sAvetage of using the dooblet distriw& ion otciydsrbto rta thn th ufc Is that Kelvin's vortex lews (eag.that vortex lines only begin or end on tha solid surface and are closed othersiss) are satisfied autoaticlly Acrossa the surface tha velocity is diutomtially. see Fig. 2.5. I. u(x iS-) u -o-~ s i -aced q; o f
m)

0
-

pijo)(.) (0 V)-

-2

a0)

& *(5)

(.b

The (Ilnearized) perturbation "ans-flux vector Is then dofined as the differente between the total 4ass-flux vector and Its Iestamvalue. v ,I* which like v - lp has order of meg. nituda e t hedrvtono*ql2.b ene Eq. (2.3b) to order c. Employing the jump relation given in Eq. (2.6f) it turns out that ) V6xe is I~ 011-m (2.7c)

(2.6f)

showing that the doublet distribution has associ. with it a jump In the tangential component only, while the source distribution causes a jump In the direction of the co-normal. i.e. In general I-s both tha normal and the tangential direction.

2.7

corrennondencietwO2'--

olo Ey4

cnS drn

th

nvrge.nd

the'diff eence,

can be I 'ons Ned wponotsoeelati cmputaiona that oudn~,; ul t fir t~e iced o reuce distribution q, :-sourc r6~:)1nuce by fiel )idcd by aovCticity n:. and th~qve' city 6 Ins-p~ct~on of the compoents ' f Eqs. dis-tribution. (2.6bY ajid (2:gd)-learns that one can'wite: (x ;qe(Bl (2.8) ku 1-~.)

IteF inerleqain th or t sid '("I o thr"'~eidtediftnsm~ua~y


ary conkditio~nsdcupeedcl'b seatly ccs In both h ul~ye in he l l 'su48C neof in t- placs syetry treatment. n th~at OS ole~ the &est i:eatl

lddenotes the unit vector In l'2" wheret. respectively. x. y an'z direction, Symer in caseswith nr~ies~hodsd and nmangg In bothSgoae'try the flow Computing the thast singularisayediy teaiixing timecan be, vilb identical ty to the' one onl ths'mtrboard'aide. tf .then the po2.8 induced at x bytSMAHCN tentAi'fand tha elct re'spectively. one' f~nds that the po2.and (x )', &&i velocity Induced at xo by the complete a. E onfiguration are

Mote general cases of, geometric and not neceasacily aerodynamic symmetry can be formulated, like with in upper/lower combination starboard/portaxi-symmetry,etc. N-lobe side symmetry,

DOMAIN OF DEPENDENCE OF PDOAN FORWARD


S,

INFLUENCE OF P
S5

an

(e)0V (IS

I6;.)

(2.90)AT

Ii [0 0
10 0

,
1J

(2.90)

8 Nx -x).(-M.)((y 8-y) +(Z-z))mO 0 Fig. 2.6

Definitiens In supersonic flow

15* with I

(S)I.

2.9

This Implies that only the starboard aide of the co bs disccetized and subjectc onfigurAtion nae conditions. ato the boundary Afinal point is that in case the geometry Is eybecauseof but matric. the flow isnot symmsetric con( e.g. aide-slipping conditions tha boundary a can be reduced by almost the figuration) problem f..cor of 2 by formulating the problem In termsof of thn singularity and the difference the average on distributions port and starboard side. In this caseoe can write for the velocity Inducad by for instance tha source distribution on the complete configurattonw (O~)

In supersonic flow the solution of the Prandtl.CGlauert equation can also ha written in on distribution terms of a source and a doublet see Ref. 22. che aurface of the configuration. of the However, here the hyperbolic characcer quation Is to be accounted for by resricting the in of surface Integration Eqs. (2.5a). (2.5b). (2.6b). (2.6d) and (2.6.) to the it*&withinthe Mach conefrom the pointx6, i.e. the forward potential Induced by a source distribuvelocity tion becomes P x)-j V, so q('X)~Id I (.l

jmaxion~saLUi

~z:%
* -

I)'*

IS X;q ) 0

where Stdenotee that pact of 5 faillrith domain of doche forward Mach Cone from sx (the 1) . g o p nd * " #af ; ) (s ee i . 2 .9)I n Eq. (2 . 1" A occurring In Eq. (2.5b). This Is comensurate with mas produced by& the circumstance that all tha source has to flow aft throughs the aft suporsonic Hach cone (i.e. domain of Influence) rather than throughs both the forward and tha aft Kach cone. In care performing the Integration in Eq. (2.114) shauld be taken to extract the proper -finit Is In the senseof liadamard. The latter conPart* nmcted with the Circumstance chat lit -

s q U( : )
with qq -0+qP)/2 and

.. -. (S ( I V?( 'lI;q ) 0 q d 4 q. 4P/ /

(2. 10) (2.10b)

where qasnd qPdenote the source distribution on In portaide,respectively. the starboard and the, case of flow symetry ild 0 and Eq. (2.10alireduces to Eq. (2.9b). Tor the mirror-imaged point, located At ISlx 0 one then tots:
6 56
-

I)(

~q)+

'oq)) "SI(IS'

I(X.)
1(c.X) a 2 1/2(21b 2 ,( 2 4B2 2lb (:0)I ((y o 0 -y)

.1 a-iux..q d) - (S) ((S Is:q d)') (2.10c)

wl~~idi~~itJ s 'Ineg rdn breie X0 was o saingular only In ''ase . the (; th

tei " Lgveman e

SUPERSONIC FLOW,

LEA5ING EDGE'
oin' ' see. Eq.

tialzInducedby the doublet distributio, (2.c) o()


-

I kfBd) (;)

(2.11c)

COMMUNICATION
Fig. 2.7b Supersonic leading edge

source and doublet distribution foundin a are sieilar way from Eq. (2.6b), (2.6d) and (2.6e), i.e. uq() q
-

For a subsonic leoding edge the flowcorners around'the edgeresulting into,a singularity in
the velocity field at the 'edge The latter leads

lf Yq().,

st

ds6( )

(2.lid)

()" -I

I ;,dS(;)

(2.11.)

to thleading-'dge suctlwfi ci whichcounter. acts the dragforce.Thli'also impllisthat'there Is comunication between the upperand the I'ver wing surface, which means Amongst others thor at a subsonic trailing edgea Kuttacondition re. is quired ruleout expansion the flowaound to of the edge and forcethe flowto separate the at edge. At a supsonic leading edgethereis no cosmunication possible between side.the pressure finite,upperand lover is though diacontinu. sue,at the edgewhileconsequently leading. the edgesuction forceis lost.At a suporsonic trail. Ing edge alsoa discontinuity the pressure in may existbecause upstream influence not possible. Is Thisimplies thatat a supersonic tsxiling edgea finite jump in the pressure possible Is and also thata Kuttacondition not to be applied is at sucha trailing edge.

Sv(X v

I- , as

x ({ajd7( )) i0l'

(2.llf)

ioti that it &boe,, expressions whichhas an b. i~agifry! valuefor H > 1. only occurs as B - 1 -.2 whichis negative. It can be derivedthatthe Jumpsin the velocity potential the velocity and across the source and the doublet distribution identical the ones are to givenin Eqs.(2.5f) and (2.6f). In supersonic flowone distinguishes so. calledsubsonic supersonic and leading trail, and , Ing edges. For a subsonic edge the coeponan of the fre.-stream Mach numser normal to the edge Is smaller than1.0.for a supersonic edge thiscon. porent exceedsthe sonic valueof 1.0, i.e.for a subsonic leading/trailing the edge is awept edge beyond the Machco"e,for a supersonic edge the edge Is sweptforward the Mach cons.As illusof tratedIn Fig.2 7 the behaviour the flow near of a subsonic edge is completely different from the one near a supersonic edge.

2.10 githert transformarton In section 2.5 the solution the Prandtl. of Clauert equation, Eq. (2.3&). givendirectly was in termsof elementary solutions the Prandtl. of Clauert equation. solution also be formu. The can laceddifferently firsttransforming by the Prandtl.Clauert equation the Laplace to equations. The transformation [SIX.i.e. -.

C-

by. r - 81

(2.12&)

2 transforms (2.3.) Eq. with0 - #/B Into-

, so that4

sl I r

O1

(22. (2.12&)

SUPERSONIC FLOW UPPER-LOWER SIDE COMMUNICATION MACH CONE


. C +z

19. I a dy ao A6

LEADING EDGE

This implies that one has to solve laplace's aqua. ton for the so-called "analogous configuration" in the ) space. The analogous configure. tion' is thinner, has a smaller spanand a higher sweep, thanthe trueconfiguration the (x.y.z) in

space.

Fig. 2.7a Subsonic leading edge

Denoting the configuration physical in spaceby F(x.y.0-, it follows the', normal the vector can be expressed as:

5-9

aa lgu ay, oir

.{ ~
2f~

as'

ln. I.tes- In the derivation it has bcon as- i sued that the totalintegrated source-strength

~2'

1/2

From;Eq.

(2.5)

i folow

tha't

tit.r dimihsioua

-(Bn.

ny. n.)Tf

r.2 s
2

r21/,2
+ S 2

(2.13b)

)'dl the contour of the +nWe whrek-ne plus i e and~ wAke ",he i'zplane. The odinsional'ake'is dIstribution ou 'a descliption -ofthe potential.' and due' distribution to the'source The velocity a can doublet'distributionbe obtained'in sisilar resulting and (2.6e), (2.6d) way fromEq*. (2.6b),

where f -U1
2

1/2 'configuration

/2doublet Ue +'Y Z JSingle-valued the thatapplying noral-volocit'y it follows in n-BJnxthetransfomed planed condition . inu results

t() condition which iff r fr the trueboundary physical opae. .1, if the m.parprbion asosnption is made in whichf - I + 0(9')and the first term on tht lot of Eq. (2.14) is 0(2). or the arfor H. - 0. does one recover to0 condition, rectboundary pointof view is froma theoretical An advantage Cthart rule known'as that for aboveformulation, 1, the uniqueness of the solution can he generally Thereare whichis not the caseotherwise. proved, trans. Prondtl-Glauert-C~thert alternative further in basedon differences the possible, formsatons v between and #. scalings rule kno formulation, n as CGthert An alternativo V 11.but stillemploying - p/B . is to solveEq. incomin (2.12b) the transformed spaceemploying to distributions, and doublet source pressible transform the velocity components to physical and to applythe Eq. employing (2.12c) space space. in condition physical boundary withthe direct It appearsthatworking 2.5.2.7 in like formulation, described sections
and 2.9. in which compressible source and doublet

q()

(2.

C,

li:dl

i.e. perpendialongthe y-axis. with Y directed cularto te x-z plane,one finds:


-

(31 C+w 4 d ) - 1-1-d(~

I2d

( 21b

alongthe due For the contribution to the vortex it distribution can be derived edgeof the doublet that " " ;4 ) . p( for sb , (2,1(c) by induced a (compressible) whichis the velocity on located the edge of the distribution. vortex due flow In two-dimensional the disturbances to die distributions out less and doublet the source than with distance fromthe distribution rapidly in three.dimsnsional flow, io.. versus 1 12 versus /IEI e R2 dll 2 d uII, 'nlRdl etc. in with the differences character This corresponds flow, of two- and three-dimensional the two-dihen. atonal flow being forced to remain within parallel planes and not being allowed to escape sideways. from the threeit Finally is notedthatstarting dimensional formulation other, (quait-) wo-dimenatonal formulations can be drived. Theeo include fnow where the geomtY Scales the case of conical withx, I.*. linearly
-

of on are distributions employed tha surface the space,and the true in configuration Vhysical is condition imposed, has a slight pref. boundary trance. This certainly applies to the case of integration wherethe finite-part flow supersonic Is to be parformed. 2 11 Uo-dimentsonal flow is paperthe emphasis on In the present three-dimensionl flow applications. In casethe panel method is to be applied to a two-dimensional configuration mostly a three-dimnsional configuratio with a largespan-chord rationIs specified a of (typically order100).Alternativelytvo-diension l formulation can be developed directly starting fromthe two.divensional Laplace or or. equation, as a very InstrucPrandtl-Glauert tive exercise, from the three-dimnsionel integral distribution the singularity by formulation taking and direction integrating in constant spanvioe approach is The y(-,,..). latter over the interval here. taken In thisway one findsfrom Eq. (2.5b): Pq(;.) where o -

-(y/Kx.z

))

withK somecon-nt, while and q - q(y/'x.z/Kx) is id(y/x.z/kx).

y fq(2)lnji, 24 dC(R)
5(0
"E x + c'os

(2.15a)

02d

00

, X'6 + Z;s.

5-10

3. _IN ^~,
3.z

~os
S
EMA

N b
CONDIt(O
KUTTA CONDITION " E: So: DOUBLET

?,J ........ .............. toI I hn" . .. manner On the olid surface So f the configuraion'

impose this condition Is to substitute into. the gral representationslEqa* (2.6) directly into Eq. (2.4a)'. 'Hovever, 'bforethis iin~b;,done onezo to fix a remaining degree of freedom In the formulation, I.e.thereare two singularity ditributionsbutjust one bodry, ondtion on,b-l.e the' floe Iearbitroiy: liido the,volumieenclosed ., en Theafictitio flow within'by" .ro. with th; aegree of freedom, is'fixea by pefying S some relatIon betein th t wo slngulm.rity distrlbuttons or by choosing one of them.,In the *first. generation' penalmethods following the possibilitieshavebeen implemented:

DISTRIBUTION (MODE FUNCTION)

S SOURCE I~S,: S DISTRIBUTION (UNKNOWN)

IES I

Unknown Unknown

None or for LCO 'mode function"

Fig. 3.1 Example formulation first of of generation panelmethod Lift-cerry.over occurs when the wing inter. sectsthe fuselage, wing intersects tip the the tank, the horizontal tailintersects fuselage, the the vertical tail lntexsects fuselage, pylon the a intersects storeor a tip tank.etc. In not all a of thesecasesis the choice the oppropriate of way to handle lift.carry.over triviel problem as a as for a ssp.e wing-frselage intersection a of geometrically starboard/port-eide symmetric aircraftin symeetric flight. Fig. 3.2 provides example whic,the wing an in intersects fuselage the tip tank. If the the and wingdoublet distribution wouldend at the inter. sections therewouldhavebeen a discrete vortex alongthe wing.fusage junction and alongthe junction the wing and the ip tank. intro. of By ducing LCOsoent, carrying in doublet A a dis. tributon whichis constant spanwise in direction and which in chordwiee direction the se diehas aLter tribution the wing doubler distribution as has alongthe section. the vortex alongthe wing.fuse. lage Junction and the intersection of the es with the fuselage displaced the fuselage is to centerline.Here is willbe cancelled the vor. by tex fromthe port-side segment, lest for LO at symmetric flowconditions. vortex The tip-,tank/wing junction displaced alongthe is w) the lineof the tip tank whereit truthfully center sisulates the tip flowaroundthe tank,and downstream of the tip tank,the ring-tip vortex. This indicates that the 1O segments are ueed to position the diecrete vortices, associated with doublet distri. butionsthatare non-zero their at bounding edges, at physically correct locations.

itO: Lift-Cerry.Over from WINOS to BODIES

Here"ODIES' denote pdrtsof the configuration thathavevolume bt no clearly definedtrailing edgeand do not have a wakesurface associated with them.The flow over such& pert in isolation wouldnot geerate any liftforce. "WINCS" denote parts of the configuration have volume no that or volume, but in any casehave a trailing edge and an associated wake surface, whiletheseparts do generee lift,both in isolation and when partof a complete In tiched a configuration. cese a *ODY" to zero to -WINGthe lift does not fall is at. at the junction the bodyand the wing: the body of carries soamelift also,whichis oftenreferred to as lift-carry-over (LCO). Nest firstgeneration panelmethods started as methods that could not 'epresont the generation of lit forces These methods only a source had die5 tribution the surface the configuration on of b and did not feature wake surfaces S the methods whereextended caseswitil lift in. to by corporating doublet distributions wakes) in (and somerelatively simple fashion, mostlyin the form of what can be described a "Hodefunctionas ap. proach The 'Nodsfunction" (given shape, unknown tud) doublet distribution aituated eapli. is on the wing surfaceitself on som auxiliary or surfece(often the camber surface partof the camber or surface) Inside the wing. ee Fig. 3.1.At the trailing edge the doublet distribution on the ouxiliary surface is continued onto S"One of the probies encountered over and over again in the Application of thesefirstgeneration panel methods that is the doublet distribution (vortex system) of the wing has to be continuad intoor onto appended partswithouttrailing edgesuch as fuselages, tip tanks, etc. This is necessary to avoidin a fev eituations, to position a physically or in correct manner in most situations, the concentrated vortex aasocattd withthe second term in Eq. (2.6.), or in otherWordsto properly account for the lift. carry.over.

_-

__._FUSELAGE,

TiPTANk.

LIFTING SURFACE

'

MOVE VORTEX

rO CENTERLINE

LCO: Lift-Cafry-Over POS: Plane.of-Symmetry


Fig. 3.2 (t.00) egmsents Exampleof the use of lift.curry-over

VORTEX TO POS

The formulation of the problem isaucL. using the direct Implementation of the Neumanncondition. to the following Fredhol. Integral aqw, ton q distribution kind for the source of the second + +q $b
-

The exasple giver in Fig. 3.1 Is Just one of several ways in which , ift can be added to a basic Psumasn form-lation Other examples are a linearly on distribution Sb. againdetervarying doublet edge; at minedby a Kutt&condition the trailing on distribution the camber sur. a defining doublet that the conditions flow. face and applying for the internalthis surface is a trem surfuce The Fredholm integral equation givenIn Eq. (3.1) provides sound a basisfor a vell.condition. of Once ed discretization. th- solution Eq. (3.1a) velocity romputed i: is obtained the tangential of from the evaluation the integrM represent&. (2.6d) and (2.6s) tot subsonic tion, Eqs. (2.6b). for (2.11.) and (2.11f) flowand Eqs.(2.lld). flow, supersonic

0)

oda f. Vn.A . (o) . (3.1a)

for all Ib

b.

from Eq. R the kernel follows In Eq. (3.1a) flowa~l frol Sq. (2.lid)for for subsonic (2.6b) flow,e.g.14 - i/4*'RI for subsonic supersonic doublet distri. Thiterm due to the modsfunction as bution can be expressed Um=oda.(o " "l.(I6IS( (; )dS(;) o.;

(3.1b)

i where Andi

and (2.6d e) for tqs. follows from flow Xnd free Eq.(2.11* and f) for suoubsoni,

-. 1.
.V mnx(Un).vW-ViI I

Eq. flow, R personic i.e. -f - it.In (3.1b) 2


S, denotesthe auxifiary surf ce carrying the and mA-tunction doubletdistribution the wake is ontowhichthe doubletdistribution continued (e.g. see Fig.3.1).The downstream to Infinity distribution Is chordvise shap*of the doublet variation the am. of usually given,the spanwiss Is of pltitude th modefunction foundby supple. on Eq. condition (3.1m) Sb "enting Neumann the edge. elongthe trailing with .Kut&acondition

on condition the perturbation Fig. 3.3 Dirichlot potential

5-12 S3.2 Dirlchlet uofdwtlon ons

ROt'O" theasecond .generatlon" panel ath-f ods offeran . ac-iave alternative the.direct to above. Utilizinthe Juap' properties across the applicat ion of the'iumann condition desciib~jd singularitydlstributiona Neumi he r'condlt!6 for the flowexternal the volume enclosed to thesurface S can be converted Vb Dlrchle- by ihtoa
b

U (o.:
'

Vh A enresise Ion condit for the flow Inside whchmthe sertura in Fig. 3.3.is the formulation bation velocity potential is s'j v equsl to zero for all pointsn t S4 . This implies' tht. - 0 everywhere inside and therefore Vb alothat - 0 for x ' Sb It then follows fIroEj. (2.6f) and'thelleusann condition x c Sb for that0

. ,q/B2 (M.P,/(n.;). 0o.a


0 mv~/ne eAnd
+

S
SW
Fig. 3.4 conditionior in caseof Dirichlet

(3.2a)

forc s o op for o b

q/b + (I.

p))/(n e)

.v (3.2b)

OnceEq. (3.3)is solvedfor the dublet distribu. tion.the tangential velocity Sb can b obtain on ed as follows. The Dirichler boundary condition on respectively. Subtraction thesetw^ equations of S lopliop thatthe mean (I.e.*Pricipal) value then yiel'sthe following relation between the becomeso in) . . see Eq. (2Sf). Prom this source distribution and the doublet q distribution sassequatio8 folows also that it P on Sb: o(XoCS+) "s(Xo) (3.4a) n q - 82(n.;)(vn.Us.) B2(.vP) (.2c) and from Eq. (2.6f). selving froi uP we find (oS
0(0o

condi. of formulation the boundary The equivalent tionthat S should a streA surface be also leads to a Fredhol integral equation the second of kind,nov for p. The integral equation follows directly f-.oa Eq. (2.5)as:

e ',b.o oh

q/ n/;.). (3.4b)
3

ith Eq. (3.2c)this yields (;oSb) - ;kx) + vn;

ffsio) (x
Sbv

( -)

.;) .d

) b

(q(

o .z)di) (33)

wherefor subsonic flow9 and g follow from Eq. (2 5c) and (2 5b), respoctlvely.qfor supersonic flow fromEqs (2 lic)and (2.lA), respectively, In case of a bodywith a trailing edgethe wake attached it at the trailing to edgewili carrya doublet distribution. iuttacondition The requires thatthe doubletdistribution at leastcontino. is ous in function value, otherwise wouldintro, we due a discrete vortexat the edgeand consequently an Infinita velccity at the edge, This isplies that at the trailing edge thereis a discontinuity the doublet in distribution. Since usuallytrailing edgesare "nsturalbreaksin the surface the configuration discontinuity of the in the doublet distribution not cauaeadditional will difficulties the discretization the formula. in of Ths Dirichiet forsuletion does not require ficti. tiousauxiliary internal Lift-Carry-Over surfaces However. note that at the intersection of the waks of a wingwith a body the doublet distribution on the bodyhas A Jup (equal to wakedoublet to strength) (Pig.3.4) It thenmeansthatthe intersection the waks with the fuaelage of should be treated explicitly a breakin the description as of the cross-section the fuselage, of

This lastexpression does not involve evalue. an tinnof any Integral representation, the gra. just dientof the doublet distribution to be deter. has mined on the surface the configuration of together with the tangential component the free.tream of velocity (andotheronsetflows) and the user. specified outflow, The latter circumstance a clearadvantage is of employing Dirichlet the condition. further A ad. vantageIs thatnow a scalar function,. is to be considered rather thana vector-like quantity, u. Thismeantthatthe storage problem bly reduced. willbe shownlateris considers. As on computing timerequired evaluate to the integral representston is not changed substantially by the approach. in aboveformosiaion choke was med. to the set the grt velocity potential equal to zero In the interior Vb. This of results an in. In ternal flowfieldthatis identical fre to stress. alternative An formlation is to set the tota velocity potential 4() - p(5) 0 i .X equal to zero in the interior Vb eaW Pig. .q. of

~q

5-13

gi+

S w
is. (tOjj2

U*o.s Uo U1=O-

ob-

U U.. )~.-j~
Fig. 3,5 Dirichlet condition on*

P. .0

UixU.n*

rV

U*~fl~j

Fig, 3.6

Dirichlet condition onp a nd 0

Is Simt thon the totalvelocity zeroin the Into. rior. Eq. (2.6f) yields for thenormal component of the velocity on the Inner side of Sb:0O
.(.),

A coubination of the two Dirichiet conditions Is also possible (e.g. see Fig. 3.6). In this cae. conflow.The boundary ti flowIn V2 Is stagnant are applied ditions n~.doS 2
.love

iii 1~/A
- J52

(;.p~l(;.)

-on

0.(3.5a)

the Intersection Si Of the two volumes we then

i.e. for xot S and for the normal 5 coeponent of the velocity on tho outer side of b'

?6)l~ iO~+
1 . for 's,

n (3.5b) Eqs. (3.2 a and h). , whichrepluc. S

~q/B 2 + (.pnd(.

Ile*-0o n as Carrying out the analysis ahove leads to the situation In which the source distribution on Sbl are given by Eq. (3,.2c ) and (3.5c), ro. i onS, the jump pactv~ly.Furthermore, applying rather of relations leadsto the specification p., than q, as j's ) - k.03.* i 3.8a

thatnow It From thesetwo equation&. follows


2

B2(MO) g q (nm~v 0

(3lc

nd Note that forzerooutflow e incompressible fl ow Eq. (3.5c) yieldsq - 0. I.e. for thatcase the potential flow solution can ha found without employing a source distrihution. The Integral equation to he used for the formulationof the Dirichlet condition termsof the In total potential Is

from an Inteiraleqaton hile .q(. ,S,) followe on th wholethe problem can he 9epresse s the following of minedtypeof Integral set equations to be solvedsimultaneously: fo yo Oo , l

1";)-10 9 f() Sb~b Jq()t t bloh3

( .)dS('). ffq(')t (
q

.)dS(x)-

1,(0 jp;R(x.x)dS(x) Sbw


-ffq(x)t

(x .x)dS(;)-(x)tk( 0.)dS(;) Jf0 Si1


(3.8b)

( .x)dS(;) bqo

0 .x

(3.6)

for

~S;,:

again a FredholI equation of the second kind. dif. facing r"them intsgral equation Eq. (3.3) in the right-hand side only. Onc e thatl souinfEq (3.6) Isknwni followsx

ip(;)- ffp(x)t.( .o)dS() 3 bb2 0 S 4

ffq()tq(' .)dS(;)SI i'. so (3.8c)

b otJ.,a fr S fo o

u( eS* -v n - p(3.7b) oh n aideof Sb. while In on for points the exterior the Interior of Vb the flowIs at rest.

p) ('
Sb1.b2 0
-ffq(;)t

,)ds(,Z)ffq(')Z SI Si J

(;.)dS(*)-

(; *;)dS(x). fJ,(X)1t (.. )dS('X)


113.6d)

$bloh2 qo

5-14

the alternative to apply'both is Stilla further external Neunannand ,theinternal Dirichiei boundto sry condition. Thisdirect approach leads' a coupled.set integral of, equations for the unknown distribution. Moreover, source i d unknowndoublet thie,'ptentli have to be nov'both" the- velocity 'and in in incriase comipued resulting a substantial computaionaloffort. Thiacncerns rot as much the computing.time needed for the evaluation of the integrals specificall computing but the time involved in'havlng solvea system equations to of with twicethe dimensioncompared to the system resulting~from~thaforaulationsdiscussed above. is possible advantge of the directapproach that

of and a surface contains iear combination the firstderivatives Ax . The normal of velocity condition the trueupperand lowerwing suron faces is expanded termsof e is well.using Eq, in (3 9c). In theexpansionit~s'assumed chat,to velocity on
leadin4,order of &pproiiatlon, the

by may the truesurface be replaced 3 the'velocity where on Eke reference surface, see Eq. ( ,9c). u(x ) is'the perturbation velocity assumed be to of ordere. Addinrthe two expresslons resulting fromthe approximated noral,velocity boundary condition applied upperand lowerside,yields on

as. a the additional atorage required. The wel

,(r> u

<r,

'

. (

,+

in discretied'fonit sometimes more' accurate is than the indirect approach described abovein distribution computed is directly whichtheaource utilizing thatin the Inside the body the poof This tential'is zeroevery-where. Is apparently due to the circumstance thatupon discretization and setting potential the equalto zeroat a set of discrete pointson the interior sideof Sb result%in an interior potential field%hichis not exactly zeroevery-where zero to the orderof but approximation employed. case of coarseor ir. In regular panelling is imginable thatfor a it given number of panels the direct formulation in. per conditions panelmight volvingtwoboundary results a morn accurate in simulation the ex. of teror flow fieldthenthe Indirect formulation,

(3.10a) wherethe second tem on the right-hand side can in be rewritten termsof the slopeof the thick. nessdistribution the wing.T1.e of latter one is zero in caseAxu-a *, I.e. for avlingof infini. tesimal thickness. flow problem hand is The at solved a source by and a doublet distribution both situated the reference on surface, supplemented by the continuation thedoublet distribution of on the wake.Substitution thr Jump relation, of Eq. (2.6f), thenyieldsthe following relation q: for
2

"n

(.;),

vu +

)+ p..If,.l );, A n

B2(;.

3.3 Ljftin surfacealvroximation In the derivation the Prandtl.Clauert of equation it was implicitly assumed thatbodies are yieldsfiLtfne the two expressions whilewingsare relatively pointed and slender thin. Withinthe framevork linear of theory the u" v1u + j-'(+T)xO'nr + u; (so 5 strea...urface condition on the upperand lor wing surfaces be simplified a boundary concan to v n dition a reference on surface, e.g.the rasber (3.10c) surface any otherreference or surface sufficient. ly closeto the trueupperand lowerwingsurface The second term on the right-hand sidecorresponds 5 (Fig.3 ,. to the slopeof th comber distribution addedto the, not-necessarily planar, wing reference sur. fae. 1he add.on camber will be zero in case .U " fU -,;u, i.e.in the casethe reference surface X+ AX VU is chosen be the camber to surface the wing. of n It follows fro Eq. (3.10c). upon substitution of

(3.10b) wherethe last termon the right-hand side is zero flow for incompressible as well as in cospressiblv m to flow in case the co.normal is parallel the normaln.

s,

Eqs. (2.6dand e). thatthe lifting-surface approximation loadsto the following integral aqua. tion for ,:

WING REFERENCE SURFACE Sr

"

;r

_a~fJ 1 (;rxOP)R -J 4 (IBjdlhl dS 1 1 1

% r#

o. sr,vw
-

Fig. 3.7 Lifting surface approximation In th lifting-surface approxlation points the on wing surfaces defined are as . -r , ;u.1 (3.9a) .

(VnVn) S., n

r5)ffq9
S

dS q

(3,10d)

where 9 and 9 follow from Eqs. (2.6d and s) for subsonit flew vnd fromEq. (2.11o and f) for supersonic flow.Eq (3.10d) definitely a is not Fredholm integral equation the second of kind,it probably has to be classified en integral as equatienbelonging the classof Frsdholm to integral equations the firstkind.In aboveintegral of equation derivative p appears the of rither then while#ls the second the on value itself, term function the right-hand sidehaspers clear-cut a classification the integral integralThe genequation.equaoral experience of that Fradhole is tionsof the firstkind are no: as easy amonable to numrioal solution techniques those the as of no second kind.However, at present alternative approximation of formulation the lifting-surface Is available.

where r. u and I denote the reference (i1ifting') surface, upperwing surioe and lowerwingsurface respectively. Underthe assumptions that ' ,;u and its firstdarivatisee smallof order er* ens may write . ,
1

r . u.dr S

*0(2)

(3,Sb) (3.c) (3.9c)

u) - (r) ) + O (2) u 0 )

where ;r denotes the vector normal the refer. to wAss the + and the - referto the oncesurface, side,respectively, u s5 and lowec T - of order is tangential the reference s, to

5-15 -The',iftingsuirface approximation is.a thin-wing pp'roxiation with limltatioinein tleapplication toa1rbitsry wings.It mustalso bemenntioned that for'wings with asaharpsubsonic leading edge there In.formally no problem, though for est incidences at in there willbe a singularity the solution the loadingedge, i.e. 'x
-.

Sw S

STRAIGHT VORTEX LINES

I..

supersonic leading Fora wingwith i'sharp-edged but no edge there formally problem. here the 'i sloptof thethicknessdistribution at the leading sesliough to havea flowpattern edgemaust-be shockwave.For wingswitha withan attached blunt leading od&*'the wing-thickness source die. edge. infinite the leading at trLbution becomes invalidating liftingthe i.e. see Eq. (3.10b), surface approximation and leading to problems in Some Implementation. kind of a local the nuserical the proedgesmightrelieve of'blunt treatment blem. approxof The majoradvantage the lifting-surface of is imation thatnow instead the upperand the withone unjustone surface, lowerwing surface Is distribution, usedto reproknownsingularity the sent the flow abo4tthe wing. This reduces the of and therewith computational number panels effort. o conditions 1. 3.4 Agundarv conditions two On the wakesurface Soundary Eq. and apply:the stream-surface condition (2.4a) Eq. condition (2.4b). the zero-pressura-juap at Applying the streamsurface condition both see usingEq. (2.6f), sidoeof the wakeSw gives, alsoEq. (3.104): Mblet q(m.os)

CONS' W(M approximation rigid-wake Fig. 3.8 Classical The classical rigid-wake approximation is so atand linear it because renders.heproblem tractive on are conditions necessary the wake no boundary accusufficiently Veryoftenit yields surface. suchas for the caseof lightlyrat, results, strong wing. wings, for the caseswithout loaded etc. interaction. Also,in supersonic wake/tail because suffices flowveryoftenthe approximation influences. of the absence upstream of vortex with approximation itralght The rigid.waka linear. the linesfixedin spacerenders problem conthe average normal-veloclty Note thatthough on Eq. dition. 3.llb,is not satisfied S,, the across the is pressure to firstordercontinuous that the wake is forcefreeto that wake implying orderof magnituda. on used slight variation the A frequently wake Is to let the user specify classical planer alongwhichthe dou. lines.i.e.curves the vortex is (Fig.3.9), distribution takenas constant againequalto the valueof Wat the trailing *near edge.The vortexlineson the so formed linesonto the as waks"are continued straight 'farwake. In thiscase the vortexlinesare lineremains stillfixedIn spaceand the problem that the realize at. On the otherhand one should near wake eight not be force free, net even to firstorder. of The non.planer wake assumption, use for low. results applications, epee,high-angle-of-attack whichdoes not require problem also in a linear on conditions the wake surface. boundary

(3.lle)

to the (needed cancel distribution for the source jusp In normal direction due to the compressible doublet distribution), stream-suris condition the faverage) The second face condition: ( o u ).n - 0 (3.l1b)

It follows froa the linearized pressure formula thatwith Eq. ageinusingEq. (2.6f). Eq. (2.3c). (3.11b): AcF + 0(,2) - 0 2(l.p)/I (3.1c)

condition leadsto the classical rigtjako wake is directed Fig6 3.8) in the com(see whichthe approximationalong (also
This latter

Sw:USER - SPECIFIED VORTEX LINES NEAR WAKE'


SbLINES

pretsibility axis). On this suracc lines with - constant (withthe constant equal to the value of is t the trailing edge). whichcorrespond with ae directed in stre&swise di. the vortexlines, rection. Since nov ,p - 0 this wake does net disJust distribution, a doublet carrya source tribution.

Sw:STRAGHT VORTEX FA WAX LINEIFAR WAKE"

Fg CONST.

Fig, 3.9

Variation on rigid-wake approximation

5-16

VORTEX'LINE

SOLVE

i 0 FOR e S

WAKE: FORCE FREE'

CORRECT S5n USING U.


+ .VI - 0 FOR 6 S, -E

Fig. 3.11

Full wake relaxation

NEW So (7)
S t E iOVR N

ONVERGED
y
Fig.3.10 Wake relaxation procedure Although user-specified the near-*ake option does Improve modeling the capabilities the panel of be. method, therearm caseswherethe Interaction tween the wake of one component and the flow about another nearby component of the configuration is so strong thatthe fullnonlinear bomdary condtions have to be imposed. Examples j dolt wings k with leading.edgo vortexflow.the flowaroundthe sideedge of a deflected flapand the flow around the wing tip.The stream-surface condition leads to the two relations givenin Eqs. (3.11a and b) F&. the zero-pressure-Jump condition substitution of Eq. (2.6f) and Eqs. (3.11 and b) In Eq. (2.4s) results, without approximation into:

the local velocity, thusdefining improved an as. timato for the location of the wake surface. etc. However. caseswhet the wakeInteracts In very strongly with the flow abouta nearbycomponent of the configuration (eg.the flowaboutslender wingswith leading-edge vortexsheets) suchsimple hierarchical iteration schemefrequently diverges and the wakeboundaryconditionEqs. (3.lib end (see d) have to be solvedsimultaneously Fig. 311). pie two resulting integral equations p for and S (x) are highlynonlinear x. In termsof p in Eq. (3.lb) leadsto an integral equation not un. likea Fredholm integral equation the first of kind.similar the lifting-surface to integral equation (3,104). whileEq. (3.11d) nonlinear is (quadratic) the doublet in distribution. Is the shapeof An alternative approach to specify the *near wake* vortex sheet as in Fig. 3.9, but to allowthe vortex linesto move freely within this surface suchthaton S the zero-pressurejumpcondition, (3.lld) satisfied. reEq. is The suiting problem is only weakly nonlinear in p and a sieple quasi-Rowton iteration procedure be can used to solvefor the doublet distribution the on nearwake For thisformulation doublet the distribution on the near wakemeos to be discretized into panels. see Fig. 3.12.

iii,

-0

(3.lld)

Sw: USER-SPECIFIED SHAPE


NEAR WAKE I SN

implies that in combined with Eq, (3.l1b). which, Incompressible flow, as well as In linearized comprsssiblo flow, the vortex lines (lines p - conscant) are parallel the localaverage to velocity across the vortex sheet.WAKE can Eq. (3.11c) Note chetfroethis exactequation be obtained directly ratherthenvia the expres. sion for the linearized pressure coefficient. of To solvefor the position the wakeand the sin on distributions the body as wIel as on gularity the wake sialtaneously is a difficult tas',espocially for general configurations. main diffiThe cultyis thatoftenthe topology the wakevorof tex system not known.whilealso specifying is a successful initial guessto be inputin the procedure thet solves the system of nonlinear equations ia a non-trivial problem. frequently relies So one on a hierarchical "waksrelaxation* procedure (ose Fig.3.10)in whichone firstobtains the singula. ritydistribution a rigidwake (as in Fig. for 3.9)and subsequently alignsthe vortex lines -olth

Sw :STRAIGHT VORTEX

Cc

1i - UNKNOWN (S, to

Fig.3.12 Partial wake relaxation |

5-17 As a rpsultoi -the,:*partial.ake relaxationI U.-t4,, the- en-v itvn the vortex sheet, is perpenidicular~t oc~ p.ie-,the velocity vector Haies a~planse in norsal to,th~surface, bich, also contains thesurface vorzicity~vecrr;y -i' seeFig. 3l3_Ac iiowr'er ihine al~l~t.cn diin.ntiilfa, n are not not.parallel t.o eachoether..Aithin Implies that there atill ',is -&,force, tangential,,to the. surface, vorkin -onthe wake. This force' isproportiosal to lncase the wakevortex~sheet-is relaxed theliutta condition will ba satisfied-in the procees,of~the wakerelaxetion.-In this senseweke rolaxationis. '.he perfect-tuttacondition. requiring'n6-other input~thsn thel,,ocation' oftbe.separation line.In cs h aevre heti'o e~du chosen a fixed surface in space,-a different, as procedure msotbefollowed, seeFig.-3.14. There ate severaipossibilities in use in panelsenethsdsIn case ,the;doublet distribution thewake on Is relaxed ("partial wake relaxation') thepressurewill already continuous the trailing be at edgepoint on the wake vortex sheet. explicitly Is (11) In caseno condition Imposed on te wke shet omepanel methods orte foatore a nnlinear'Kutta condition in~which the condition is enforced thatat the trailing edge the pressure the upper wing surface equals the on pressure the lowerwingsurface. on Note that In 2D flowthis condition reduces a linear condito tion. Ispose the condition thatjust downstream of the trailing edgethe velocity vector directed Is alongthe chosen wake surface Sa. i.e. u.n - 0 at x-te.(3.12) The difficulty hire is that the direction of the wake vortex sheet at S,, Is directly irvolved In Kuttacondition. wingswith nerotrailing. 4or ar'gle end for lifting surfaces the direction is uniquely determined, namely tangential to the cusped pact of the wing, or tange6Ttisl to Sr. For wings with a finite trailing-*dgo angle the woke vortex sheet Is either tangential to the lower wing surface 2X to the upper wing surface 2Z. at enIsolated point on the trasiling edge, directed alongthe bisector (the so-called Mangler Smith 6 criteron, see Ref. 23).Wlhether wakevortex the surface depands the planfora the wing (the on of non-linearity creeping Intothe problem again). However, mst applications in Just one direction (lower upper)Is chosen. or

~(3

fj'(i)

VORTEX LINE
j.r-U. .VgeO0
*
-

Ilconstani

n
_L_+Up

U~

3(Mi)

FORC WITIN 5the FORC .edg WIHIN


Fig 3.13 Partial wake relaxation
.Force

on wake

*velocity

35 35 Krt& condition The Kut&acondition the condition Is that Is a smooth fashion. This Implies that tha wake vortex sheet ssoothly connects to the trailing edgeand thatthe velocity Is finite the edge. at In trms of the singularity distributitons this inmodiately Implies thatthe doubletdistribution is continuous the edge.sincesoy discontinuity at in p results in a discrete vortex, strength of equal to the jumpIn the doublet distribution, alongthe trailing edge whichwouldresult an Infinite in at the edge.

I
v. SPECIFIED O(U..)

TRAILING EDGEINE SUB-INCLINED CAP: V. SPECIFIED


Fig. 3.15 Inlet modeling In supersonic flow

3.3

FULL WAKE RELAXATION


Sb

3.

Sunsr.incltned surface

FIXE WAESRFC 2 WAE SRFAC St FIXD WIT:C 5 I-).CrX OR 3)


C, X-) -Co

Pr, OR
e

P1i 2 mj

LIX j 1 3

Fig,3.14inclined, ig3.4Kurt& conditions

In compressible subson ictor supersonic flow peturbtios ar asu to be small. This usully Implies that the trua surface Is only slightly Inclined with respact to the fre..streaa direction and generally will be sub-Inclinod with respect to the Mac% anglein supersnic flow. il410 . for instance at Inlet faces where an owever Inf low Is prescribed the surface will be superbut the perturbation say still be small. Thes* super-inclined portions the surface reof
th

quire, pecibundary~conditons; notgiven here, or alternatively specification the of~an'artifict.

pends'on,,the~preclia'focalationchoen, i.e',Neunnor'Dirchle,svhich of the, I

*1 sub-inclined cap'orsraap~c~ivering'rheinlr..On. -and.Dj'I unknownanfdf~whichtfofthese,,pa. lt the cap'or' 'he~specified~outflownshold rap cor.. raotersioreknown' a*nd'can be derived di. reipond to theirequired massflov'into, the, Inle.,, rectlyif66athejboindary conditiona, -&(see Fig.3.15). z I I II Also're~uired',iiae'ocalirepriientation However,'inlcase ofa~blunt-nooed fuselage~tho, for thegeosetry of,-thepainela; thisito,con. flow'ner the'supr-Inclined~porion ofthe'n~se' siient order, of &proxiitonwvith "respect' represents a'regionwheri the 'lineiarized-potential, toftherepriskn'ations for'tho singularity"ln'cae the'position'of vortextshets i the the nose for thiscase WillalleViat the problem. but'in-thenose regionthe flowsolution willbe simultaneously be solved to for,also'for x Inaccurate,' describing geometry the piala on,5 the of a localrepresentation required. Is The ltter'involves 380 further unknownparameters 3. CC SINT CovnlsnnciM~
2 NP'

SM u am'w 3NPtions Co-co" Sctce roeysisevc1 EQatorm, Nplrp a5 5 05005Hp .1 6 M .,~p~. 0 M.. AMoefficients 0,influence Co1",oi 0. &~ec 7NP

Approxisate, the required to orderof acturcthe Integrals overthe panelsufc corresponding the contribution the to of uin'gularity distributions the panelIn on the potential the velocity the N (coior at location) points where theboundary condiare to be Imposed.

Th* cosputational heavy (+. Nohip) task, constituted by sop 2 And 3, yields the so-called aerodynsic i.e. the velocity (AIC's), by potent induced at the point s~o k., -he sin. ial gular distribution expressed ity Is as:

pos~oessn~Aeccywucs~soand the velocity Induced at xas

hKWOSOW1.1kii' Fig. 4.1 1reakdoWnt taskswithin panel method of a

wh*r,* 1 .b nda.~svte I'.i q (4.1)Q. I-1).1aD * 1-1(1)111 the pa art raseters in the panel-viol' representatilona of'the source and doublet distribution, respectively. The AM's are a function of ;and of the gemet rc quantit;ies describing the panel surface. This ,.plies that the AMC'sare independent of the *etrodynsaic unknowns' (theQ 's and or the DI's). but may be A function the egeneetric unknownsof of the wake vortex sheets.

'

4.0

APPROXIATIO' ADil DISCRETIZATIONq

4.1 Cenralspygc In the approach taken In panel methods the fol. lowing tasks can be distinguished (soe Fig.4.1): . Subdivide Its wake Sinto of the configure. ,io,and he surface S.~ (#-IaIl) quadril&.

BODY

WING

WING

.ccoepliehod hierarchical mannot inA (Fig. 4.2) which iconfiurAtto. Is subdi. In the videdinto.s2cA.U part Intozsgrmnti each
(sometimes called gnstwork*s) aec s and t" sent Into A numberof uILig (rings) of, samenumberof panuels. Subsequently g, the metrical quantities (centroid. normal vector, curvature, twist. ae.) of ccci, panel are computed. 2. Replace the Integrals overthe surface by the Ia= of the Integrals over the NiP panels. Chooseon eachpanel,sufficiently accurate localrepresentaand mutually singularity distributions p tionse for theconsistent, and q. The local representations involve parataeters datersining the magnitude of the singularity distributions. Here Q, i-l()NQ denote the NQparamters In the source dietribution,. , i-l(I)iil denote the ND pareseters In the doublet distribution. It de-

AE F RNN. l

OY SG ET SEGMENTS

PANELE

STRIP

'"INTERNAL '

/IFT-CARRYOVER STRIP

"

Fig. 4.2 4.

Ecxaple configuration paneling of

Imposethe boundasryconditions at the collocation points In most aerodynaic ptal methods the collocation technique iosapplied

5-19
in which the boundary condition Ia applied at just one point per panel. Some other methods, not-discussed any further here' can be classified as Calerkin method..e. 'they involve the ourface integral over the panel of the product of the boundary condition with the local representatioa.. In most aerodynamic panel methods the number of collocation points is equal to the number of unknown parameters and of the order of + the that of off~the shouldbe'remiiki, alsonumbersomepanels. It -callid ilgher-ordir aehd~xlc l moe(btet odtlOns 'on the continuity of tsingulotity distributions across segment edges, which can considerably increase the number of al. gebraic equations to be solved,end inflate, the disension of the matrix-equation to be solved. S. solve the resulting non-sparse system of linear (non-linesr in case of partial or full-wake relaxation) equations for the unknown parameters in the local represent&tions for the singularity distributions (and geometry). for subsonic flow the matrix is fully populated, for K>)1 parts of the mtrix will be empty because of the forbidden upstream Influence in supersonic flow. Solution of the system of jinar equations requires of the order of H operations in ode do not have any built-in geometric pre-pro. ceasing capability and therefore fully rely on the availability of a CAD package to generate the 'defInition of the geometry, the sub-division Into parts. segments, etc. and to carry out the discretization (*paeling") of the surface of the cofiguration. Other methods a more stand-alone type of methods with geometric capabilities, incliding paneling options, definition of the goosetry through basic built-in shapes, etc., all with th to provide maximum+flexibility. and purposeof-minimizing the amunt of input data The dsip- and details of the remaining steps (216) will datrin the accur&cy~of and computer resources required for each appl'cation. As far as the'scnuruswyand-uong is concerned, the ai in the development of any panel method to be used in proliminary design should be to obtain, for lowest costs, the surface-velocity distribution to certain accuracy. i.e.

-h u (xeShf

+ O(h

for h

(4.2)

der of itxN operations in case an iterative solver is used, with it the number of iterations required for convergence. However, in both cases the coefficient m-lP tirlying H . with p - 2 or 3, is much smaller than the one multiplying MdU above. 6. Find to the required accuracy the velocity distribution on the surface Sb of the configuration.

a diret solver is used and of the or-

Here n denotes the -order* of the panel method. Host of the first-generatio panel methods are first-order methods, most of the "second.genera. ting methods are second-order methods. In the following we consider some aspects related to the formulation of a panel method of consistent

orderof approximation, discussion willbe reThe


stricted to first and second-order methods.

4.2 ZSall-JcIXaUre exeansion for velocity Consider the sxpression for the velocity induced by a sourcs distribution, Eq. (2.6b). In order to ait':ny aitters somovhat the discussion concerns the limiting case of incompressible flow. i.e.
Ox)) qoI -

7. Compute induced drag.Integrated forces moments,tht pressure, (lifting surface)and


edge-suction forces, surface streamlines. isobars. velocity and pressure at off-body points, stability derivatives. trimmedflight conditions, boundary.layer quantities. updated wake position, etc.. etc. The .,aerof the method will interface with step 1. where the geometric input to the program is digested and uith step 7. where the results of the flow simulation are generated. These two steps .iii detormine the ooMtric capability, post-proceasing power and also (of utmost importance) the user-frieondliness of the program. Some panel math.

ffq(;)-L- dSc) r 3

(4.3)

where o "(.t). I dnots the surface Of the i-tb panel and (.d is some surface coordinote system (Fig, 4.3). The intogrand in Eq. (4.3) is singular for xo coinciding with point.x(st) on S . In mest coase the integral itself is finite, rEough always discontiosous across Si. Therefore numerical quadrature Is not eppropriate. Hess (Ref. 25) proposed an expansion in which the char. acteristic singular behaviour of the integrand is conserved. The expansion results in a consistent

5-20

Fig. 4,3

Panel coordinate system

tha' possesses the 4fE.*4.3) aprximation coret behaviou ra crosses the surface S In the expansion nominator Is split Into I the l~or part and a part containing the curvature and twist of the pael, iLe.
-

'2

xxt~tx~)Ixt

The singularity distributions q and p on $I are ex.-sndod a similar In fashi 1I.e. on,
- V* ~~q+ asq *4p p-s As. + Atqo . tO(.d Pt 0 (.d

+ 0

dtxi a* p +
3

(4.4a)

-c

+ 0(4

(4.4b) t)

P-

with As & .a* and At - t.t*. with donoting the so-called expansion point.

AO

st.

0j

t'

(~ (4.4*)

the curlybrackets 1n Eq. (4.4) the termInside corresponds the lquatio' of the plane t ngenti. to a!to '. pno Iat X(s*.t*) and is defined the by vectors and nO tangential the surface x* to coot. diFt. dilectioni - coonstnt and y-constant. t respectively.
in the nee ft-*ed ofthe p-re.len* ore I(*,* in

Substitution Eqs. (4.4&.d) of intoEq. (4.3)and expansion termsof 6 and D yields In U,;)
-t0(K5i

The vector donats rho position the pointa of which the influence is computed relative to the tang nt plane. The panel curvatur/twist term ; 0.tt Is of order i2 where S'ise linear measure for the panel site in 0eithe daiecti n.

I3

"

'.dt'L d j4 I

L)
d dt

+ +

L*Od + Ji~lrt s d: t

and

-O(L I,

The tem" Io 5

. ( * 4 Ast;- * at; rder i ) the far field of the panel.In She following we

Is 3(r 76 2'6) Si, o + a 6O -ds 0(K5')

denote the order of magnIrude of D by

In term of the surface coordinate system the surfsoc element dS - (l,)ldedt Is expanded as: dS(X) where *xxJ and *, + a )) t'2the (S2

nef(af + Ad )-L ysd i

are epress ions containing the

Inhe te *ml "~r rof fld rd~rI

151

whee K Is a measure of the panel curvature and tWis All remaining irtegrals Involve en Intogrand with a quadratic expression In the nominator raised to the power 1/2. 3/2 and 3/2. For each of thee ntegrals closed-form expressions can be derived. ss

iurvthe and List terms evaluated at the enpas. sion point. SI 4

.,t

; a t

5-21 Itf;lowi, from the 'order of magnitude estimate Indtcated~for eich term'iG Eq, (4.5).-that for a, firxf-ordgr panalmethod-.in~the near-field AIC. c ottion a pasiel-wise conetant~reproentation, for q~nthe flatpanel~approximationsufficeos. For a first-orider'mthod-the-only-,trm~ofqimport. &nceCinthe'near-field' is thefirst tormingq, (4.1S); Aterm of-magnitdeo0(l), Acodr~oEq. (4;5)'a scnd-orde panal~method, requresa eprsenat~n~fo linar pielvis and panel craueadtithvtobacone , for-'snd all.1terma In Eq.-(4;5) bae~o included In the-computtion'of theAICts. However, note',that if the panel curvature-and twist areasuch that K6 2,, vhore4 Is-'the basic- lengthsxcale'- in~eaiscretization. e some average panel sine, the term in Eq, (4.;) due to the panel curvature and twist (the mast complex ones) are small of higher order and my be emitted. Reducion of loalby sub-paneling. in creasing but also with decreasing $. Is another K plxterma In Eq. (4.5). Applying the smll-curvature expanaion to Eqs. (2.6d and *). the velocity Induced by the doublet distribution, shows that a first-orde method requires a panoiwise linear re prsentation for u on a flat-panel fpproximation, while a secon-od toothod requires a poalwise quadratic representation for puwhile again panel curvature and twist have to be Included. The closad-rorm enpressione for each of the surface integrals In Eq. (4.5) contain the sae type of (4coeputationally expensive) transcendentalS functions (two logarithms, one square root and one Inverse cangent) multiplied by simple polynomial. type term and summed In a way depending on the specific expression. This Indicates that computetionally. carefully designed and programmed, if a higher-order formulation dossnet need to be very mh moreepensive ihan a lower-order formulation.However, will be clear that a third-.- or it evenhigher-order sothod. whichamongst others will Involve derivatives the curvature. of Is beyond practical limits.LE The smll-curvature expansion warrants that thIminuced velocity has the proper behaviour a4 the point xcrosses the surface t x - x(*-.t) and the Panel expansion point is usually choos as At the-edge of this vortex sheet-the doublet dietribution'Is indeed'oro (see Fig. 4.4).:-Since-it is rathe~r-imposiible to-fiscretito a-vortex sheet of infinte length t%-,tightly rolled-up, pert-ofthe vortex sheet-ii replaced by adiscrete vortexfilamnt, connected with the remainder of S, by u feeding sheet- (see Fig. 4.4). The vortex,wif of the correct strength and If positioned at the correct location. 'will provide- the proper flow field aay romthevortex filament but of course not narhecenter of the vortexcore where the yelocity wlbesingular In the numierical model.

FEIG HE
,.~..

+-,~)A
,,,'

VORTEX

5,,

FLMN~

S.

ROLLED - UP VORTEX SHEET


.

MODEL FOR ROLLED - UP VREXSHE

0
0 Fig. 4.4

------

Model for rolled-op vortex sheet

ONE HORSE-SHOE VORTEXPE PA

of location the of". ;

the collocation pointon the panel.Irrespective+

Although for arbitrary configurations cannot it proven formally, the sall-curvature expansionBObeBON given In Eq. (4.5) Is expected to have a coeposite error,2 the error do to summing over all panels, of 0(i ).I 2 In considering the smll-curvature exportso ::of,. I. thevelocity Induced by the doublet distribto tis otno b u m tat te orurlet ito 0(S) o It asotinou asuetast the bet odoubleto 0l6 )bfo higher-order formulations) across the panels* edges, actually already Implicitly used in the derivation Eq. (2.6e). If this Is net the case of apurious contributions In the velocity induced by a doublet distribution will appear. Similarly the doublet distribution should be xero at the bound, ryof I,*.to avoida discrete vortex that loat cation, however. in the description givenIn this paper we retained Eq. (2.6o) because the folof h-wing reason. Vortex wakes tend to roll up into co~centrated vortices (eg. near a wing tip). which are In tho vortex shaetmodel represented by a highly rolled.up vortex sheet of infinite length.

ttc
"

C COLL.
'A

T.E.

N VORTEX

SYMMETRY

/FREE.

SEGMENTED X
RE TRAILING LINE VORTEX

....... PANEL EDGE

VORTEX
+ COLLOCATION POINT
Fig. 4.5 Vortex lattice machod

5-22

scussedabove the smallr Under5thecond2tlon thatthe following expansion indicates curvature consistent. epproximatons',arpossible,for,the, evaluation thedn tegrsl of representation ofthe velocity induced sourceanddoublet distribu- . by fthe tions: P T

Note.that-thisrequires thatthevelocityie.to~be, computed two pointaper.panel, atthe~midpoint at of the 3/4-panel-chordllneangatthe-midpointof Furthermore note'that.F the,1/4.panel-chord~line. as givenlihEq (4,6),contains~both,thenoraland. tagetili force components For a specifir t ttio dimeiio nAl cai e 1 cin'be tsho vnlanalytic lly are that the computed iesultingforceand moment equal-to theirexactyalues,(Ref. 24)'; a remark-

Panel geometry

got Source q q*q,qq distribution (p5).p*,, Doublt distiu tin )

'* 't

lOtrder t

to2nd' order
'se

a 's's't

able~resultindeed-,
Furtherinote that in three-dimensional flow such a niceanalytic resultislnotavailable, -thoughit appears thataboveprocedure yields rather satisresults for overall, and moments. factory wingsaxial-force forces For~awept results-can be In. segments proved *unsweeping* thobound* vortex, by priorto applying Eq,(4.6).

, c) ss' st'ptt

Eq. (2.6e) to eccom. is A further reason retain to method(see-pruXd vortex-lattice moda e the so-called method.In Fig. 4.5),whichis &'lifting-surface the vortex-lattice method the doubletdistribution is panelwise constant. i.e, ) in 1U(o)G(x v(; ) the firstterm,given is in Eq. (2.6d). zero.i.e.the secondterm, givenin Eq. (2.6e), the onlyterm to be conis sidered. This mans thatwe haveto integrate have alongthe edgesof the panelonly,or rather to consider velocity the induced a vortex by of of strength alongthe perimeter the constant pael. at is distribution singular the Sincethe velocity vortices It is impossible derive consistent to a alongthe linesusedabove.In the approximation vortex-lattice method the pointat which the vs. locity to be computed is has to be chosen very carefully. turnsout thatIf the pointis It chosenas the midpoint the panel. of I.e.as far as the awayfrom the singularity possible, induced it In velocity to 0O()accurate. orderto avoid edge the havIng a vortexlocated at the leading vortices ara shifted away from the paneledges,or vortex a, most comonly referred to. the *bound' line at is located the 1/4-penel-chord of the panel, whilethe pointwherethe velocity com.is putedis the midpoint the 3/4.panel.chord of line. This actually means that the elements with conchordIn chordwise 1/4-panel stentp are shifted At edge the doubletdis. direction. the trailing tribution contin ontothe wakeas a stripof Is ed vor trailing to constant equivalent discrete 0. ticesalongthe sideedgesof the strip. at Similr to the shiftin the vortexlocation the to lading edge is the usualpractice off-setthe (trailing) vortex alongthe wing tip by 1/4-panel. s an . p the It mustbe notedthatIn satisfying stream surface condition the aid pointof the 3/4. at panel-chord line.thilprobably only pointon the the panelIs where u.n - 0. Furthermore. the ve. sincethe locity at thatpointwillbe continuous Jump In the velocity associated with a doublet to distribution related Op whichIs zero.This is Implies that In orderto havea Jump in the tangand ential velocity, therewith in the pressure and a contribution to the normal force. 4j has to be calculated from ose local non-constant representation for p. An aternative procedure compute to the contribution of the panelin the force on the theorem to lifting surface to applyblastua' Is compute the forceon the *bound' alongthe 4 vortex lineusing,see Fig. ,6z l/4.chord (4.6) for and a similar expression the moment, U

r
at -AS-6,

Fig.4.6 Forceon a vortexsegent 4.3 Far-field essns on for veioitrv the small-curvature expansion is Although expensive uniformly valid.it is computationally in whereIt and therefore onlyapplied the region of is really needed, i.e.In the "nearfield* the (Ix. panel.In the 'far-field' -0l>6) of the ansl as the nominator Eq. (4.3)can be expressed of r-r o" -r, " r
*

0(1

(4.7)

where -X -x te 0(1) and - ; 4a* + AtX is in 0(5).Subeti~ution Eq (4.3 then ields:

u(x

Z x:;x-jjq.,- ,, Jdsdt Rol S . e (...) 4 f(oe or At)dsdt St

i0(2

+ 0(5 )1

(4.8)

The second term in Eq. (4,) Is zero if the expansionpoint(as.to) coincides with the panelmidpoint. The truncation errorIn Eq. (4.8) Is suffia errorof cietly smallto guarantee eomposite that for the far-field 0(i ). Eq. (4.3) indicates expansion both a first-order and a second-order representation method require panelwiee constant a for q on the flat-panel approximation, curvai.e. ture and twistdetails of the panelor higherdoof distribution of are rivatives the singularity For no significance the far.field for Influence. the doublet 41stribution can be shownthatone it requires paneIwise linear a representation the on flat-panel Approximation.

5-23

Ofe hrgion wheetenafil

ie ;s, exans,

namc

gt

luatt~on of for.,,he cnistent, .. cuay r, t3rode "*


*X.X,'

to second- indthid-odr In th;~n 'ai: teredite fildt~i~ saed tht, the jitigiid in Eq. (4et3'A onisten acurte Into rtlonwith a standard, , numerical iter tionpoeu;tai slightlyTondoer =so coapiek'than Eq. (4.7) but is not as~complx. Pouel* s Eq. (4.).el'ty t

~~

ii,,r

4.4,

forogotential Smail-curvature esnansion -The~exproion~for*the-yeIcity potentil. can1 doublet'distribution injduced-by.aaourceoand~a be expanded~in &,simiiar fashion jilustrated, ae eVAl2.ver. Linordir~to abov:for the velocity,' the velocity to potential laethe velocity 0(S has to be evaluated to 0(S ). As an exasple consider,the small-curvature expanby aon for the velocity potential induced a From source distribution. Eq.,(2.5b) it foiiovs (incompressible flow):
-paneiwime

distributionat Doublet P*p a5 t ditribto

(*,*p,*.
-

55sa-t

p Ott

of so thatupon differentiationv one gets the yelocity up to,first and second order accuracy. respectively. Someof the second-generation panel ethods empioy the Dirichiet condition~in a o-tcailed low-,order formlation-in which the doublet distribution Is constant and the source distribution, If I alI panolwise constant. Within the fr;;wrko~fthesmal-crvaureexpansion this only the 101,implie for the veiocity that potential 5 ledin term treating from the contribution p of isretained, ota formally the method wouidbe O(S) In the velocity potentiai and 0(i) in the velocdt Itelf. However, It appears i"ty thatupon evalus~ins tangential the velocity tha surface at (. nx( xn) - %A).vmployinga higher-order reprsontation for p to evaluate Pp, Inmost cases ccito pressure distribution which Is as acsrte as the ane obtained from a first-generation &.~~t0( employing panelmethod the Neumntn boundary condi-

'qo

~used,

.3q*Jf~1~ Si,: q: ff AS-dsdt + q* f I Si

Q 0 OW

isi

t-st0
,,suits.

-5
Qjf qljS~II dt )

$11 2131 +
-(I S

(4.9) D

D(~

It follows from Eq. (4.9) and froma similar expression for the potantial Induced by a doublet distribution (taking continuity of doublet paneledgesInto account) chat for strength across the method employing or a first- second-order Dirichiot condition the ame type sf panel-wise representatins are required as for the corresponding machod with the Neuann condition. The condition under which the Isit two toe In Eq. is < S Comparison (4.9) say be neglected KS of Eqs. (4.5) and (4.9) indthafo chat'the same type of Integrals appear In both expressions, so have that also the same transcendental functions to be computed. This implies that coaputationelly the formulation In terms of a Dirichiet condition will net be cheaper than the formulation In terms conditions, but of course the Dirichlet of Neumtann formulation requires less storage, since a scalar matrix to be cum..jted is coefficient influence matrix, cuefficiont Influence thana vector rather se q.(4.1a and b). for thin Far-field expansions can also be derived potential Induced by source and doublet distributions, however, the details are not given here, It should be noted though that in the far-field exdistribupension of the potential doe t* a source tion use has to be made of the circumstance that the total source strength used In tha simulation Is of the flow about a configuration zero or a quantity of smallar order of magnitude.

o usan~fo Ifuosr~fnst The snall-curvature and far-field expansions for imapressible subsonic or compressible supersonic flow follow lines analogous to the ones given for Incompressible flow In the preceding sections. For subsonic flow the analytical evaluation of the resulting Integrals provides no real difficulty compared with the evaluation of theirincompressile caunterparts. Fot supersonic flow matters are much morecomplex because now the paneiwiso Intogracion is over chat part of the panel which Ies# forwithin the Inteeectlon of the panel with tits ward Math tone from x~ (see Fig. 2.6), while also of care must ha takentat the finite-pact the Intogral is extracted, see section 2.9. Here we conaider the suall-curvaturt near-field approximation first. 4.5 It turns out. sea Refs. 8 and 9, that the following procedure is the one that leadsto the desired closed-forim expressions with the least effort: Start from the closed-forms expressiona for ccooapoestt i e. tlow the case of o5opi .ieth adb-where3 - IK) expressions amar to the onsa 3rivsd In sections 4. and 4.4 for the valocity and the potential. respectively; which ac(ii) Rsplta the factor I/4*by 1/2x. chatIn supercounta for the circumstanca sonic flow all the Influence of ,hesinguIn is distribution exercised downlarity stea direction only. i.e. In the aft Kach con*;: 21/2 and 62 2 by .11-KI (III) Esplace B by ill-Ill1 overyahere IT/he -closed-form otpressions. . The expressions are now where i-(-) (i)

(iy Etricie 'r'lXpr'o'~I~ti' ()slonw. The'ioi

c~iple*-'al'~d'xprosioo athrfrhanthe .

the where1the'Q arer 1,se thparaTrser in, rpre-, seta (j in4 noeWay ,r _!nhi ' a &rrInge g~i~~ apai "!I.,ti o 4lad'; h coficie ntosa a a- iiq(4l)dpn ojn'at typo -af.,0.'alo1keres. ttivncheoen'for,; q~p;~.,p * In ai i~.ah'n '~ t5o d nerA5f f prc.-' p Di. !rsQj~ad

Io'arm

of the Influence exerted by the singulrity sould dis1tribut1;ono In uproi'fw.It th'e'real part b~ oe h the foarb.sic transcenit a~functions,(oe ;tw loe, "al om arctanget'anda ntht to the rit)qaeroot;; from' onec os) 'onvrt suchthat thepror influence in the of influ.ence' of th singularity 'distribution, on the panel is obtained,

~iiatl~tng

~ad'p*,zar

Fromt'W Above itwill be clearthat'the closedfore expressions the influence for coefficients In supersonic, flow are muchmore complicated than the onesfor subsonic flow. In the coding of thes*expresoloos a rather compltx'loiLel' structure is' required of coniditional' braniches. see Ref. 9. to account the appropriate for behaviour different in regions In the supersonic flow field. As far Is the far-fLid expansion Is concerned It is remarked that 'the domain where the far-field enpressions may, beApp'iied is situatid within the aft Mach cone with the Apex at somepoint downstream of the trailing edgeof the panel. on the other hand It can be remarked that In most of the space around a pael the influence is nero, and the need for a far-field expansion for M.>l is lessurgentiy required than for subsonic freestream Machnumbers, For the case of supersonic flowmuchcomputer time -anbe Avdbdee inn whteorntac. Plote pact, segment. strip or ring of penal* Is located within the domain of dopenden.9 of the poit where the potential or velocity is coeputed. If the parr.etc. Is not within the forward Mach conefrom the point considered the AIC's can be set equalto zero directly, rather than that the program computes zeros on a panel-by-panel basic, 4.6 the Inte. In most'aerodynamic panelmethods gralequations solved using the collocation are method. iLe.the integral equation satisfied is at just one point per panel. Compared for Instance to a Galerbin method, which involves an additional integration the surface, over collocation less Is AIpensive. b~it collocation maybe more sensitive to the secific discretination chosen. In moot panel methods the panel midpoint or tentroid is chosen as the collocatiijo Roto. This sdA to the aimplest enproions In the far-field expansion, see Eq. (4.$). Wi1th this choice problems ere also avoided with (nearly) canceiling wsakly singular contributions associated with diecontinuities across panel edges in the gocetry and in the singularity distributions and their derivatives. 4.6.1N,,exc1l sheme In the foregoing the contribution In the vs. iotry or the pottnctl due to the singularity d istribution on a pael w~s enpressed In ceams of 0 quantities at the panelenpansion point. e.g. q . q-. q* The next choica'to be made concerns the i Itgiiisbc express these quantities In to terms of the parameter valus to be solved for. Therefore the quantities at the panel expansion pointhaue to be expressed In terms of the iperam' ettrx e, Di appearing In Eqs. (4.1a and b). nd For exsmple, couldexpress on panel(i.J) one q:
qS(st.tj) 819i-1.j a1 QIJ 5

Iegoercqatte C 3oan **0 o A atte~o.. x follow fro -te ivn descrfp try of'the-cowifiguretion,(end,the-,user'specified wake vortex shsets)' Only in~case'the'wake vortex sheet is fully'relaxed-Is li'nce sry to choose alsotsa numerical 3chemeforeipressing xO. ae. in termsof a set of geometric porameters.' 9i. say I-l(l)liG. With'the choice of the numerical scheme It Is now possible writethe result the'ssall-curvato of ture expansion and the far-field expansion'in terms of Eqs. (4.1a and b). The Integral.'equation from the Neunanncondition, Eq. (3.1), then yields using Eq. (4.1b), the follovlog system of linear equations: NQc - ki'n(xk)Qi i-L-1 ND 1k . (kD~I*

4.1 (.1 wherek-I(l)NQ. while;k- k-l(l)iP denotes the location the NP collocation of points, prinin Cipleone per paiel. In Almost all panal methods the numberof unhnown soirce parameters N02 equals the numberof panels NP. Eq. (4.11) ie to be sup. 14sented by as many (i.e. ND) iKutta corditiona 1 (soosection 3,S)as there are unknowndoublet distribution parameters D used to describe the mod-function type of doulet distribution. The NI Kutta conditions ro volt In NO liner equations In rase Eq. (3.12) Is imposed, or In ND mildly nonlinear equations In case the pressure across the trall.g-edge pointjie equated, i.e. -P Cpxx. In a similar fashion follows from the Dirichlet It condition. (0.3). usingEq, (4.1a), Eq. that: N iiD i iI b 0 Xai~qt 1'-%aQI (4.12)

for k-l(l)ND. where %,aro b. are evaluated at The the collocation pointmi. t k;.Il)NP.-ND. righthand side in Eq. (4.12) an pecific the source parameters QI)- i-(I)NQ are found from Eq. (3.2c) which spocifim the source distribution q In cae of the outflow V &a the free *tram onsetflow 0 -ie. q - b2(6. ;(V n The second term cn the left-hand side of Eq. (4.12) stes from the contributien In the source distribution due to the doublet dietribution. see Eq. (3 .2c). The*su, c ppcrametort Qt,11(11 r through q - - - (in: d) expressed In terms of the 0 unknowndoublet parameters i. Li..()ND, using eXpression like Eq. (4.10) to relate mid-point quantities to doublet parameters. Application of the Dirichlot condition on the total potentiel. which leads to the Integrel equstiongiven In Eq. (3.6) with the source distribution given In terms of the outflow And the gradient of the doublet distribution, se Eq. (3.5c), leadeto!

ImQLIoj (410

~E~D, s'

ND

N kIqj
-

.skq..k

(4.1)

5-25 for k l()ND.Qin'AnEq.: (4;13) ojlovsfro.the knownsourceidistrlbution q - BN,(n.)v,vhllenov thseecondlterm,onthe rlghtaccounte~or thefree-stream onsetflow.Qt whichfollovs again, fromthe source distribution- -B (;.V), is q subsequently expressed in terms of the unknown doublet parameters1 , i-l()MN: D surfaces the integral equationfor the On lifting doublet!,distribution. Eq. (3.10d), results with Eq. (4.1b), Into NQ
17 1

points. For Eq. (4.16)i t-has been assumed thatwe are dealing with linear boundary conditionsonly. i.e;withrigidwakevorteasheets consisting offixedvortex lines " Eq. (4A6) can be expressed inthe following equivaleniform: . (A](S) - (I?
.

71

(4.17)

'~fk)i 1-

wherefl denotos aNUII matrix and I) denotes a column vector with NU elements. InEq.'(4.17) the matrixA, the so-called *aerodynamic influence coefficient matril, depends or. known geometric quantities only. In case of *partial wakerelaxation- doubiet the distribution tha vks is relaxedand we have to on deal iththe linear equationsresultingfrom the -surface condition o the ,solid surfaces of
the configuration in obination with Eq. (4.15b)

knxk nn( u.v_)+ % .(u


n+ U.. r k)(4.14)

+ II)x-r(k -. k

which Is quadratic in system of-the unknown doublet paramters, Le., the terms of equ~ations Is now: i r i (4.8

fork-l(1)ND and with

denoting the cotocationI( :k

In Qj surfaces. Eq. (4.14) points the 11fting on follows from the knownpart of the expression in Eq. (3.lOb). whileQf is the sameas described above. For the boundary conditions the wake,Eqs. on that: uIng Eq. (4.1la), (3.11b and d), it follows ND ki ;O and ND ki."( )
i -l k ' ( k

of someiterative To solvethis system equation proctdure to be used,usually Newton-like is a method. Neton's methodthe solution itera. In at tion number it+l is obtained fromthe solution at iteration numberIt by solving the following eyerm of linear equations: itItt+ i r( lit (4,Ib)

NQ%-;(k)IN] i (4.15a) ; )Qt + %. ) + . ;k- _ -0 (4 )

The matrix the left-hand on side of Eq. (4.1b) is It matrix. will or the socalled gradient Jacobian be clear that for the linear equations contained withinEq. (4.18a) the corresponding eltments in the gradient matrixare independent S and are of identical to the ones in the aerodynamic influence coefficient mtrix of Eq. (4.17). quadratic The equations Eq. (4.15a) in resultin elements that do dependon S and need to be re.calculated at each stop of the iteration procedure. However. af, the expense of the rate of convery,*nce of the iteration procedure, could freezethe one Jacobian its initial at value,or restrict to it the linear tars in Eq. (4.15b). In caseof fullwake relaxation doublet tei distribution the position the wake are both and of of and we have to dealvith a system algereaxed braicequations that is linear or quadratic S in and highly nonlinear in C, the unknown parameters of the local description of the geometry of the wake vortex sheets. So now Se have (F(SC)l - (r) (4.19a)

for k-l(l)ND and with x , k-I(I)ND denoting the collocation points on tre wakevortexshaets. In Eq. (4.1Sb) is to be evaluated thc collocaVP at ton point xk on the vortex sheet, which as before can be expressed ters of the unknown in doublet distribution parameters D i-l(l)ND. Note that Eq. (4.15b) Is quadratic In terms of the DI', hile both Eqs. (4.15a and b) are highlynon. linear in termsof 0i . i-l(l)NO, the paramatar occurring the numerical schemesto express in x* x*, etc.At the panelmidpoints termsof the in nkrwn geometric pramters. In Abovediscussion system equations has the of been derived assuming that the configuration consistsof merely surfaces wherethe Naumanm, the Dirichlet, the lifting-surface or the wake conditioneare applied. The system equations a of for configuration built-up of a mix of surfaces out with Nauan, surfaces withDirichlet. surfaces with the lifting-surface surfaces and withwake boundary conditions easily is composed fromthe relations gIven in Eqs.(4.11) through(4,15). Here we write the resulting system of squations . i- a.s - ri. for i-I(I)NU (4.16)

whichleadsto the following iteration procedure:

[] siit S

[I]ItWI.lit- r.FSit.fit)) (4.19b)

In Eq. (4.19b) the firstmatrix the left-hand on where N is the number of unknown singularity paaide contains elements of the aerodynamic influ. roasters $S. J-l(l)NU and the aIIs denote the ence coefficient astrix, the eecond one can be aerodynami influence coefficients. The right.hand termed the 'geometric influence as coefficient side r i . i-I(I)NU of Eq. (4.16) contains the freemtrix*. The latter represents the response of the stream onset flow, the outflow distribution and boundary conditions to changes in the geometry of the contributions due to the known singularity the wake vortex sheets. Its computation neceassi. distributions, evaluated the KU collocation rates taking all at the derivatives, respect the with to

5-26
Sposition

vector. of~the~expreosiono derived in , . For Interpolatory Tsplines it odiensionisthe sections 4.; hrough'4;4,:thenormal vector, etc., , situationis simll.r thouh 'in-genoral more coma rather'laborious bothfromuthelvlew.point plicatedi. task, ncase, the'data'prescribedinvolves _ de. of the panel method developer asfd~froathevyiew. rivatLveo; --"' .. point of computational expense. Moreover, because of the nonlinearity withrespect~to G,,theevalua. z tionof the right-hand side of Eq. (4.19b) re quires the re-coeputation of the aerodynamic in. fluence coefficients each iteration, at vhilealso the strong nonlinearity termsof C usually in re. tq quiresrgua updtin ,.,.tsreguIar~updtln& of theJacobian during~the, a, iteration
process.* |

In s ta!on 4.2 through wediscussed 4.4 consistent pproximations leadto an accurate that discretization the integral equatlons. order of In to get also an accurate solution.of discretizthe od integral equations the panel method formulation chosen should clveant, ie.,the difference be betwen the solution of the continuous problem and that of the diacrttizd problem should decreese proportional sP, to for S- 0, vnure6 is a measure for the average panelsizeand p > 0. It is well. knownfromnumerical analysis thata consistent discretization a vsll.posed of problemthatis ale Is also convergent. Stability a panel of methodIs a property the system algebraic of of equations that-results the discretization from of the integral equations. Therefore stability has everything do withthe chosen to formulation and used in scheme like Eq. (4.10) withthe numerical the discretization. specifically the con and with the in nucber the matrixInvolved solving of di fion the system algebraic of (linear nonlinear) or equations, in the presentof the thismeans thatfor the th specific chice context panel colloation point numerical schemes for the localrepresentations of the unknown functions q. p end x and their derive. tives must be chosen such that the resulting math. od is stable. Proving stability of candidate num,. rical schemasfor dicrotizing the integral aqua. tions. givenin Eqs. (3.1). (3.3). (3.6),(3.$b and c). (3.l0d) (3.llb and and 3.ld) for the Neu. mans, Dirichlet on V. Dirichlst #, Dirichlet on on Sand 4. lifting.surface or wake boundary condi. tions.is a difficult task.Most investigators base their choice on arguments from what is known aboutthe stability interpolatory of splines, sup. plmentad with numerical experimentation. The splin.stabi ity arguments useful are because the observation of thatcoseonly ap. the plication the boundary of condition at a colloca. tion point is dominated by the function value. firstor second derivative the unknown of quantity evaluated tha collocation at point. Splitn stability refera the typeof data to be to described a certain at position warrant to a sooth nrn.osciliatory polynomial-typo of interpo. lant. In the table below this is indicated for one.dimensional splins of odd and even degree.

S -i s

e.g., q'=q, ; q" -. ,4.


etg.

/2AS

etc.

a) SEGMENT - WISE SURFACE COORDINATES

q' q

q4

~ ~.,.,

,.,.

b) LOCAL REPRESENTATIONS
Fig. 4.7 Possible numerical schemes

SUBLE s

S PLE

I
0

I
SUB TE
-

0 TESUPSUE \

ISplit.e

oth

First

Second

degree

dd Eve 0
Even

derivative be specified to at NOd Mdpoint Mo od N odso


idpo~nt Mode Midpoint

MACH CONE COLLOCATION POINT

EXTRA COLLOCATION POINT


fig. 4.8 Location of collocation points on a strip of a lifting surface in supersonic flow

A different choice of the location at which the data is prescribed does not always lead to a com.

plate loss of stability, for instance the condi.

tions at the end of the interval can have a dmp. ing effect.

5-27 ,Fo~tio~cnfiurtinswiththe lieumann. o ba fsn Iig.and withilthe source distrlbuwhich resulted-in the~integrel. tion'as unknOwn, equation of Eq. (3-1).q* isthe dominant contribution.,Therefore' astableechee6is aaeven-degree poiitsa'sunnonparmeei.For-,a&firt-order sethod'this issimaply 'aconttant-,sourcm-,distrbun rich' "Fthe Forme.thoseslyig hgerr r anta.. for. ,of ,the'noaericalschemecan tion '~rcso takemany form. Thse,scheme may~lhavobeendarived finitetdifference.;typeofeoxpressions on a, rtoo sgment-wloe-defined rectangular computational domainvolvinsgsurfoce coordinates. or from A, least-squares~fit,,ofthe,parmetera atimmediats neighbouringmidpo ints in terms of a local Cartesiancoordinatesyotes, etc. (seeFl!.4.7). Clencly the efficiency of's higher-order~panelmethod still depend strongly on the compactness of the n.mwrical scheme used, For thick configuratins with the Djjshjij bounayc ond3if and with the doublet io~f distribu. tionas unknown, whichresulted the Integral In p* cotriequation Eq. (3.3). Is the dominant of Is bution. stablescheme herealso the even, A function values. basedon mid-point deiree scheme there im'no iKutta condition to~beappliad'at ,the_ trailing edge'(there is nocorsuication'betwen upper andklower side),and an~additionalcollocation point-is chosen'jutuptressofthetrailing, edge.
- ,FPlr~ding'a stible~numerical scheme for the two coupled yak*bounds"' conditions'used',incase wake vortex sheet'-is fully relaxed is even are,,difftcult.-TSbeirst condition. Eq. 31b). Is,very similar to the lifting-surface condition, Above and is doainated by, the 'sitond derivatives of p. The second condition. Eq. (3.114), is quathe firctderivative. draticin-p and Involvos Both'conditionslare highly nonlinear Iiterme of the unknown geometric parameters. Usually some kind-of mixed central and directionally hissed numarical scheme,. found by intuition and nunei.ca. enperimentation is arrived at.

4.7

Some furthr

aorsoftecmntto

In the case of subsoi. c flow a disturbance decrease* magnitude with Increasing distance In fromits source, specificallyd'scontinuity (in a In Introduced the function vallue derivatives) or doesnot causeserict's numerical modelusually problems.

For IfiInfjfigsurfcs whichwe derived for *PANEL-CORNER POINTS the of Eq. (3.10d). choice the Integral equation numerical scheme is less trivial. It turns out o PANEL-EDGE MIDPOINTS that in subsonic flow the leading term of t04 equation Imposed ;(st.i) Is proportional at tc derivatives p at (st;,tt). that of so She second n ev 0dere for a #ocond-ordor formulatio scheme based on *id-point function values will provide stability, which results In a second-order accurate method. Stability for a method of lower z A, degrov of accuracy, which employs a linsar repro-sentlton for the doublet distribution, turns out pointto the to require a shi~ft tha collocation of paneledges. uith Its own problems related the to singlar behaviour the Induced of velocity at pael edges. Apparently here the diff*Lonce in the type of the underlying integral equation comesato surface. The--integral equation for lifting-surfaces being not a Frodhola integral equation of the second kxnidY making finding a et~ble discretization mere difficult. The table below aumerizes, for subsonic flow. thnA degra. of the local reprasentation for the singularity distributions loading to stable discretizations of first- and eecond-order panel methods. Dosiftint Htoe Cntri bution DIVJMi.E LIFTING ACCURACY

A, A,

A, 1. 1(1)4: PL.ANAR TRIANGULARS B ETA LNRCUDIAEA

let-order 2nd-order ' us Qunsart QuadraticDIRIHIAT Qadrtic uadrtic P* Me p* .Me Quadratic Quadratic S~lF~t asee e ti.&long ctwhere at

A Boundary condition applied panel midpoint at In supersonlc flow rho situation ie difficult again and the choice of chs collocation point doponds on whather the panel leading edge is subsonic or supersonic. If then panal leading edge Is subsonic the collocation point Is at the midpoint. if tho panel leading edge Is supersonic the colloration point Is Jnst downstream of the leading edge (sea Fig. 4.$). Fuirthermore, In case tho lifting surface has a supersonic trailing edge

i.49Sbdveo fannpea in supersonic flow disturbances, due to sa actual discontinutit but also due to the ones introduced by the diecretization. propagate undamped ilAch lines to larga distances downsstem of tho disturbance originated. This quite soriosly hapers tbhe application of low-order methods to complex multi. component Aircraft configuratio'ns. and a higher-order method Is almost oblisatory for such contigurations. In such a higherorder method the ale.elsrity distributions and apecially th, panelled geometry should ha As continuous as Isiblo in order to prevent the occurrenco of nec-co-aifficimnt degree of accuracy canculling spurious uayes from panal edges. In this respect the subdivision of a nonPlanar Panel Into planar sub.psalis not only avoids the necessity of sao of the higher-ardor terms die to panel curve-

5-28 tureand twist, aee Eqs., (4;5),and, (4.9),zbut also provdes for6lower-order methodageometriccontinulty} ndleisssevere~spurlows' wav e s A n ' of c e n . ', usedexaspleof- thesubdiviLio kofVa,-qudrilatera I panel into planar sub-panels is given in Fig. 4.9. The panel and each of its 5 sub-panels is planar and the sub-panels arecontiguouawithoeach other and with~thesub-panel of~neighbouring pa.ls, I I - I _ I Z Another even-more-serious probles constitutis the waves,,spurious.ones or ones fromtruly-represent ed breaks in-the geometry, chitpropagate into the interior the configurationand may give riseto of a sequence of-spurious internal reflections-that eventually'destroy the.accuracy of the solution. This isillustrated in Fig 4'.10'(Ref, which -9) shows'the effect'of refining (in axialdirection) the paneling on-the aft-cone a cone-cylinder. of cone configuration at H. - 2 and zero incidence. In the interior the aft conethe reflecting of Mach wavescausethe source distribution oscilto lateseverely, thatactually so the 'bestanswer In the-applicationof agiven panel mthbd the level'of;tbi Accuracy.obtained -dependon many factori.Here'we meintior: (i) 'Panel distribution. Depending on-the type the nme'ricl lchems'chosenfin the local~repre- of sentationa the panelin aybe irregular-to-a samaller-otto-a larger-extent. ostpanel methoda use interpolatingiclieisthat iccort.for the non-uniformity ofthe pameling. In-tlt case the paneling-shouldthave a basilpanel size,say , 1. areas where thereeanr large'changes in-ue geometry wherethe-singularity-distributions and are expected to vary smoothly; In areas where the curvature and/or twist of the surface the configuration of are,-larg*the panelIng must be refined, alsosection see 4.2.Also in aceaswherethe singularity distributions exare pected vary rapidly'a to finerpaneling requiris ed. Truly automatic, solution-adaptive, paneling procedures have not yet been described the litin ersature.

'Cp

0.4 0.3 -5 0.2


x ef

M-2.0

a .0 0

0z ref

MACH CONE '

11 4 0.60.8

150 PANELS .... 200 PANELS -------220 PANELS

0 0.2 0.4 0.6 0.8 1

0 0 -0.1 -I

-0.2 -0.3. -0.4

2
J!
I,

,i
Fig. 4 10 Result of panel sethod with Neumann boundary conditions for supersonic flow(Ref 9) is obtained the coarsest for paneling. Ref.26 In some of these problems could be alleviated by devising a special composite source-doublet (triplet) singularity distribution with the property that it hes a (partially) cancelling interior ve. locity field, However, for aultiple-component con. figurationa spurious reflections are still present, It appasre that a higher-orde: formulation employing the Dirichlet boundary condition, vhich results in a constant or stagnant flew in the in. tsrier of the configuration, is superior in avoiding much of the undsired spurious reflections d'scussed above, The computing timefor a specific configuration is generally less in supersonic flowthen in subsonic flow.This is achieved considering by exclusively the region dependence the point at which th* of of influence computed. the complete part,seg. is 1: sent (or strip) outnide the upstream is Mach cone from the point its influence Is zero and the mdividusl panel influences do not need to be consldered. the otherhand if the panelor a part On si the panelis withinthe domain of dependence mote logic is to be executed computs to its in. fluence,

(1i) Typeof formulation. is knownthat some It types of formulation are not suited for specific applications.well-known is the application A one of a method with the Neuann boundary condition ta the internal flow in tubes with a curved axis awl or varying cross-section, a wind tunnel. or Hero the leakelg through the tunnel walls can becom , rather excerslve. A further example provides the flow througjt a long flow-through Picelle for vhich some of the first.genera:on panel methods have shown to produce unecr-ptable results. A fital example is the flow about thii wings where wthods employing od-function doublet a dI.'trithtlon on an internal auxiliary surface and a sovree dietri. bution on the wing surface say run intoproblems. These problas are caused by the prescribed ahape of the mode function beingdifforent from the one that the solution tries to es ablish, This means that the source distribution to takeover part Ias of the taskof the doublet distribution, ex. for ample leading to a large value of the source strength on the upper wing surfase and an accompanying large negative valueof tht soure* strength (i.e. sink)on the lowersutface. The gradients associated with thisphenosannw will detrimentally affect the accuracy of the slution.

5-29 (iii),Fre-stram conditions. The truncation error of.,the method depends on the.gradients of the solution. Formally, meansthat.increasing the this incidence wouldrequire finerpaneling. ,a However, from spractical ;as wall as frams computational point of view this is not a desired situation and usuallya paneling set up onlyonceand used is for all flow conditions, Apartfromthe pointof view of numerical accuracy it should be kept in m.nd that the panel method is basedlona.rolativoly simple odl-of the real viscous and compressible flow. Thrrore at flow conditionswhore viscous effects become of importance; flow conditions for which strong shocks occur; flowconditions near the condition where a wing leading edge becomes sonic, especially in case of bluntleading edges; ate.. the correlation between prediction experimental data may turn and out to be unsatisfactory.
4.8 Solution of the system of euaion

overthe Iterative one.This.since. oncethe L-U decomposition has been accomplished, each new solution-requires one matrlx-vectorsmultiplijust cation-whilean Iterative method has to startall over again for eachnow right~hand side. As far as the implementation (super)conputers on is concerned it is remarked that both direct and Iterative solvers are vectorizable, but~that for largesystems equations larger of the number of I/0 operations required for the Iterative solver my become a bottle-neck. In the present report the iterative methods used to solvethe system of linear and nonlinear equations occurring for configurations with partly or fullyrelaxed wakesisnot described moredein tailthangivenin the preceding section. As far as the (linear) casewithout any formof wake relaxation concerned it should remarked is be
that in order to obtain a solution for a series of

'&he solution of the system of linear equa. tions, Eq. (4.17). be obtained various can in Ways. Here we mention direct methods (Gaussian elimination. decomposition, L.U etc.)and itera. tive methods (Jacobi. Causs-Sidsl. etc.) oper. &tingon lements of the matrLx A or on partitions (blocks) the atrix A. In the latter of case the blocks ore to be chosen carefully, exaple as for blocks containinr elements all associated with the influence a .trip the wing,or a ringof the of of body,on itself. and are methods more robust the In general direct a solution almost always obtained, also in case is of irregular paneling. Irregular ordering of neighbouring sesgmnts neovorks. etc. However, or (timerequired NU ) sy becomerather the CPlJ excessive scalarminfrasa compters or on on workstations. Iterative methods are lesscomputer 2 tise intensive ( itxlU), but in some casesthe iterative solution procedura converge my slowly or mightevendiverge. The rate of converge.4a depends on the choice of the iterative procedure more specifically and on has arthe way in whichthe user of the method ranged the sub.division the configuration of into parts. sefs.its. s:rips. etc.Failure converge to viMlrequire switching an alternative a to proemof dureor a re-paneling partor of the complete configuration, this for instance in order to get larger blocks in the iteration matrix. The rate of convergence of an Iterative method depends on the diagonal dominance the matrix A, which on its of turndepends the typeof the Integral on equation that is beingsolved.it appears thatthe systems of equations resulting fromdiscretization of Fredhola integral equation the second of kind. Eqs. (4.11). (4.12) and (4.13). givethe least problems. system equations The of resulting from discretication the lifting.surface of integral equation. Eq. (4.14). can causescalmore probleas d.ring execution an iterative of procedureIn the literature the subject solvingIteratively on of nearly il)-coe.tioned systass of equations means are discussed to improve the rate of convergence, but most of thesetechniques nt applyto the do matrix equations typical panelmthods, for For supersonic free-strem Mach smbsers is also it noteworthy remark to that.if the paneling ar. is ranged frm nose to tail, generally convergence of iterative methods is faster than for subsonic ach numbers. Finally, it ust be noted that in case, at fixed tath nuaber. many right-land side vectors are to be considered direct the method aay be preferabla

free-strem conditions at fixea b. mar use can be made of the superposition principle. can It be shown that for given outflow, given onset flow due to propeller slipstreasm, etc..the solution at any value of the angle of attack a, angleof sideslipP. steady rate of roll p. steady rate of pitchq and steadyrateof yaw r. can be obtained by combining basissolutions, six denoted here by the column vectors SP, i-11)6' namsely as: S - S(I + b1 ) + (S S1 )h + (S .S)h 3 + 2 2 3 + (5 -s1)(Ap .hll) + (S '- )(iq-h 1 ) * 4 5 1 1 + (S -S)(Ar.hlFl) where 6 Ap - (p.pl)/(p j, 2 -pl). pl/(p2"pl
)

(4.20)

aq - (q.qi)/(q 2 -ql): , " ql/(q 'q ) 2 l


ar - (r-rl)/(r,.r,) ; I - r,/(r2"r,)

h1 - (cosp( inAt 2AinS)/sint22 -0)f . g h2 . icoso/coso1 if " cstsdinl/sin(02 "0 1 )lf

h3 - g1 with

fl" in(ea 1 )/stn( 2 "o ): f 2 " sir,(0 2 "*)/sin(o *oi) 1 2 & g " sin(D.D 1 1 )/sin(0 01 ); 92- sin(A"D)/sin(P ) 2 2 2 "P 1 f " f+ f2 " I: g - g1
+

in Eq. (4.20) the basissolutions 1-1(1)6 ore Si. solutions pre-selected at combination# freeof streaa conditions, namely: S1 -S(Ol'Plpqlrl) 2 " 2S3 - S(al.D2,Plq,rI) S4 " S(sl'p 1 'P2 ,ql'r1 )
S5 -

S(a,.,plql~r

In the case of port/starbuard-eide symmetry 0 , p . 0 and q n 0 and just three bsis solutions are 5 required i $2 an 5). In thatcase Eq. (4.20) ( reduces to S -Slf 2 + S2 fI 4 (S5Sl)(aq-if) 1 (4.21)

moms=

5-30
rain, analytically, the-derivatiVes with respect stnce'first-gnerarion panel mthods produce non-

ued o o9 to 0, O;-p, q and r of the solution, and of thetforces and moments, ie; stabilitytherewith derive,

E .(4.0) l~cn aoe e E. (. r

tives likeaCL/8s, C1/ap, la/dq, etc.

eods whil e zero drag coifficient,. In'thre-dimensional o n flow with lift the,patiel method should provide a suffi, ciently accurate estimite of-th'induced~drag~and
for iupirsonl~flov also of the wave drag. For the integration of the pressure over lifting surfaces the integral over the leading edge and side edges where the pressure'distribution is sin. gular requires-specialcare. For an accurate prediction of the contribution of lifting,surfacesin the forces (and especially the Induced drag)snd the momehts the inclusion of these edge-suction forces is crucial.

tute dirag dtil !n iSn~

Using above procedure ve see hat 6nce for a given ' Mach number, the sLx(three) solutios~havsebeen obtained; all other solutions and their derive. tives'follow in O(NU) operations. Further basis solutions can be'obtained for e.g.: the mass-flow rate into an inlet characterized by one or more mode-function type parameters; the de. flection of control surfaces modeled employing a lifting-surfacetype of approximation about a mean position-of the control surface, i.e. with the paneled control surface at some fixed reference deflection and the boundary conditions on this ease surface accounting for the incrementalde.

FAR-FIELD

flection: etc.
4 9 Focsad oet

UPSTREAM U

S 0 S

FAR-FIELD

Once the singularity distributions are solved for, the velocity at the panel collocation points can be coaputed and therewith also the pressure on the panels. Forces and moments %ron then deterlned by integration of the pressure distribution on the surface of the configuration. e.g.: -

-$fp(x).(;)dS(x)
bh '.() (;-" Sb ,) (;)d-S (;)

(4.22a)
(4.22b)

which have to be made dimensionless by q.$sf and qSrof t .fmoment reference In Eq. (4.22b) C. de. notes a , respectively. center. The2 integrals in Eq. (4.22) can be approximated to O(0 ) by using the simple mid.point rule. e.g.. NP
-

TREFFTZ PLN
.

Fig. 4.11

lar-field momentum analysis

"

I-1

at

0(12) (4.23.)

where there are NP panels o the solid geometry of the configuration. The Integration in Eq. (4.23) should be over a closed surface. Not accounting for parts of a closed surface corresponds to as. suwing that p - 0 on the left-out part of the eur. face In most panel methods the pressure ceeffl. delnt C is integrated rather than the pressure NP

Most panel methods have an optice to compute the drag froa a so-celled 'far-tild' analysis in which the drag on the configuratlon Is obtained from the application of the 1lieof the coneerva. tion of linear momentum in a large volume sur. rounding the configuration ;,ssteady flow. see Fig. 4.11. Ihis approach automatically accounts for the edge-suction fores on lifting surfaces. Coweervtion of momentum applied to the fixed volume enclosed by I . S , S (Jhe Treffis plane) t Sb and both sides of the wske S gives p INgh he$t. S

,6

;)'ldS(;) St + St

(4.24)

S u-+ S

and a similar expression for the moment. Since in ease of a closed volume the surface integral of n(s) (and the one of A -(;)) vanishes, Eq, (4 23b) is Identical to Eq (4.23a). if ro part of a closed surface is left out in the integration it is equivalent with assuming that p - p on that part of the surface For a body with a bae thiappears to be a good first estimate of the pres. sure in the separated flo region. the experience of applying panel methd.s to co, figuratione with a finite number .f panels is that, ulng above integration procedure. most of the forces and momente can be Computed to a suffidlent degree of accuracy. However, this is gener. ally not true for the (induced or wave) drag force. It Is maost obvious In the two-dimensional case where in a potential flow about a closed con.

with -the nomal pointing into the volume. From it ni follows chat the force on the oef igura. tion can be expressed as, wing that the Integra. tion in Eq. (4.24a) is over a closed surface and that mass Is conserved:

pp) ." ).) (x) (4.2b) S whore denotes the force on the body, i.e is the eurfoce itetogral Pp. over S of t . In taoe S from the flow); S end S are at an infinite distance oof one ham (in sieonic nuratlon,

WN!"

5-31
Su S : P P " "
1 3

ysis. a result which does not depend on the freestream Mach number directly, only indirectly throughthe doublet distribution. It follows from Eq. (2.5f) that the jump in the perturbation velocity potential equals -p(t), so

p - P.-

while'to sufficient degree of accuracy S': p-. .nthat:

St'

n--

P - .+,Ut),.at.; 2p-p n x

,n)dt

(4.24f)

The latter implies that the wake surface $w is chosen to be approximately normal to the Trefftz plane St . This leads to:

(4,24c) + "U.%tjdyd. f$((P-p.).x -.


$t where 1 is the velocity induced in the Trefftz plane. 9ext employing the expression for t.* qua. dratic pressure coefficient. Eq. (2.3d), results into

In order to evaluate Eq. (4.24f) one needs the yelocity distribution 1t induced by the doublet distribution on C in the Trefft plane. This can be obtained by a YD panel method applied to a system of vortex sheets with given doublet distribution. It is general experience that obtaining the induced drag from Eq. (4.24f) (the x term) yields more accurate results than using Eq. (4.22a) resuiting from the direct integration of the pres. sure distribution. A Still an alternative way to compute the forces and double. monts acting on the configuration is to the singularity distribution. integrste Referring to Fig. 3.13 the contribution to the force of a vortex line is found from glasius' the. ores. leading to "P)x;" eb + - ff0(t + uP)qd $b

pJtl(

litl2)e,

UUtdydc
Ct

(4.24d)

Upon writing 03in ters of the pertucbation potential V. which to consistent order of magnitude satisfies Laplace's e uation in the Trafftz plan . and rsing Greenes theorem in this plane, one gets . ( . ex'ii)(+ t%)s 0
0

(4.25)

* Uxnd Umen)t (4.24,)

which in principle also includes all edge.suction forces, but is expensive computationally. 7n supsrsonic flow the situation on S i differ. ant. due to the circumstance that disturbances are propagated undamped from the configuration. Calculation of the force on the configuration from the conservation of linear momentum is much mors elaborate for this case and not considered here.

P-F1V(x+)

where C is the trace of the wake vortex sheet(s) S in t~e Trefftx plane St the normal and t I's arc length along Cw . iq. ?4.24e) is recognized as the classical reeult of the Trsfftx plan. n.l-

NO

5-32
Panel source Doublet B. C.

DUCIAS.NEUMAiN Flat (13, 1962/1972 BOEIN-TEA230 (23,1967 NLR 363, 1969 MBb (63, 1970 Flat

Cont. Mode function (external) Cont. Mode function (internal) Mode function (internal) Mode function (internal)

Nean

Resarkis M<I

Neumann

M<1 lothert rule M <I NLR comp rule H<1 M,<l

Flat

Const.

Neumann

Flat

Const.

Neuaann Neumann

HUNT-SEMPLE Flat 171, 1976 USSAERO 381, 1973 NLRAERO 39).1980 PAN AIR (101.1975 DOUGLAS 0. H 311),1980 ROBERTS (123.1975 MCAERO 313, 1980 SAA [141,1984 HI$SS 115).1984 VSAERO (163,1981 QUADPAN 1173, 198) Flat Flat

Mode function Const. (internal, opt.)

TS:Conat. Neumann LS:Linesr Non-polynomfal Linearized TB:Cont. LS:Lin*ar quadratic Quadratic Neuaann Linarized

M <l,>l M <1,>l

Flat Linear Sub-panels

Neumann/ M<I,>l Dlrichlet Neumann <l

Mode function Quadratic Linear (external. quad.) Cubic Cubic Mode function (Internal. bIrnb) Quadratic Quadratic Quadratic Conat

lNeumann

,-0

Linear Flat Sob-panels Quadratic Linear Flat Linear Sub.pannls Flat Const

Ditbchlet M.-O Dirlchlet M,-07 Dlichlet Dirichlt M<I,>1 IM-O() %rabe relax 1M<1

Flat

Conast

Conat

DlrIchlat

SC Boundary Condition TB, Thick odies LS LiftIng Surface presently use in TableI Partial listof methods or auxiliary surface the interior on the In sur. of surface thickwing. or by lifting All thesemethods are first-order ftces, esthode. ecnd-neatio ethods. Refs.10-I. all in accounting axe way for panelcrvature, and saploylng higber-order representations distributions, Some for the singularity methods stilluse the mode-function formula. solving for q through the Nsumann tlion, others employthe Dlriboundary condition, condition solvefor p. A and chletboundary number of such methods are under develop. sent.e.. Ref 18 and at NIA AEROPAN/PDAERO. advanced methods. moorder Refs.16 and 17. quit*successful methods Theseapparently employthe DIrichlet boundary condition on with the flat.panl approximation a constant source(if any) as well as a constant claiming hiher.ordar doublet distribution, accuracy.
+

3.0 51

EXISTINC PANELdET

.OS

-enra BoecrIotIon and oe enoral in At present there ara manypanel aethoda the potential u.3 capable computing linearized of (see flow about3D configuration. for a partial listTable1) Threecategories b distin. can guished. firsrenertlon methods, Refs.1-2.5.9, condiall with the directNauann boundary op. tionappliedand uslngthe flat-panel Lift a proximation. is genrated throug)h doubletdistribution an on mode-function

I. shos,! comparison (Ref. 27) of reo suits, oftofrt neraton-nethods (Refs. 5 and 7),viththehigheir.order (3rd orderi) method of Roberits-(Rof. 42). -.The cese'considered -is the In. compr~isibli'flow aboutthRl' i~ at 5,deg inciaenci forja pinel seheae of 12 strips panels each~. The left-hand side of the figure shows the comparison of the x. and the y-cospononr
-Fig.I*

~of..6O

5-33 function doubletdistribution becomes even more apporent..This jedemonstrated for the chorduse surface C distribution, shownfor the 2%-thick wing In,% plot at the right-hand side of Fig, 5.1. The advanced lower-order methods Refs. 16 and 17 employ the Dirichlet bounda~y condition. Eq.

VX~RA RAEWNG 4 . 0. a 5M-0.0 1. -OA 1-0.2

a=-5

VC-O02&

-0.2.
0.2 00.4
0.4I I It-*

1.

064 '

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0O

V Y
0.2 0.1-a-. - -- -X/C -7 -0. 0. 0. 0. 0. 0: -0.2 -0.3!
Fig.5.1 Comparison first-genorstion methods. of panel Ref. 27 of the chordula. velocity distribution a wing for thickness 5%. The s-component the velocity of of do*snot differ such for the threemethods conthe However. y-cosponent of the velocity ,sidered. computed by the method of Ref. 5. whichemploys as mode function an internal piecowio constant dou. blot dis'ribution (equivalent to a vortex lattice, see Eq. (2.6.)). of user-specified shape, shows large deviations from the *datuom solution of Roberts and from the sotu. tionof the methodof Ref. 7 which employs a more contin uousa 'optimized' Internal soda function. For even thinner wins the necessity of the mare advanced Internal mode function or some other formulation withtess severe constraints the modeon (3.3). with the apparent Inconsistent discretiza. tion of a panelwiso constant source and constant (instead of linear) doublet distribution the on flat-panel approximation. In order to find the surface velocity, (3.4b) to be evaluated. Eq. is Using at least first-order accurate nuerical ditferentiation obtainIi., Is required for s to as fully coneistent first order method. it iosauggested that In spite of this one still finds the surfate velocity with 0(h) accuracy. For the 2D case Oskaa (Ref. 28) investigated the accuracy of a method with the Internal ilirichtet condition on the total potential # - 0 (which leads to q -0 If the flow Is incompressibte and v. - tt. seeaE (3 5c)). LCIDI ILOPE -3

0o

ROBERTS (DATUM) [121 0 H-S (S4EETS) (60 X 12) [71 NLR (60X 12) [5)

-6i

KTO2PROFILE

L9I()

Fig 5.2

Accuracy of solution for 2D flow using Dirichtet boundary condition and second-order formulation (Ref. 28)

5-34
for quadratic localrepresentation In Eq.-(3;6), and distribution a curved-pmel ap' the'doublet proxigatioi wa used. 'while in Eq. "(3.7b)'a 4thorder, scurate numerical differentiation'wss eeto was applied a ployed evaluate Vp. The method to section, inThe Karman-Trefftz airfoil 12% thick method-3s vestitiori showgd that tha'resulting ratherthan-second-order'sacurato third-order from the as (Fig.5.2) one nighthaveexpected of results Refs.16 and 17. This is a puzzling this, situation ene and, :"y wonder what causes anomaly, is.the.awlysis basedon the small-curvature expsnsioperhaps too conservative, or Is there some hidden annihilation of error terms in p of the process obtaining fromEq. (3.6)and the surface velocity fromEq. (3.7b)? factor A that might have contributed is that the panel scheme in non-niform and adopted the usedwas highly wherethe curvature the gradient and nose region becomelarge. in the solution codtion versusWrichlet boundary Neumann In thissection consider we someof the and of strengths weaknesses method thatemploy the Neuaann boundary condition methods and that The boundary condition. disemploythe Dirichlet and cussionisbasedon the literature on experience gained at MLR. 5.2 conThe methods usingthe Neun boundary on distribution the dition solv fr the source Lift surface the configuration. Is generated of on distribution doublet through (modefunction) a of surface the interior the conin an artificial of surface the config. figuration on the actual or of components the Alternatively, lifting uration, as surfaces are configuration treated lifting for to a distribution b solved carrying doublet which of distribution knownstrength ends source accounts effects for due to the wingthickness
Some positive (+) and negative (.) assets are:

The meth6ds using the' pixichle bound ry conditloii solve -forcthe-dotblet 'distriburion on' the surfac' of2 thecofiguration. Aile fie source disirib'ution (necessary-intteformulaiion in terms of tho prturlitionpotiftial j)'follows from as algebrsiC-relation. -Forthis category of methods 'the follow'ing appllis: ' + for. accounted is implicitly Li'ft-carry-over configuratios.-the for a wing/body However, of compointersection the wakeof lifting components with non-lifting nenis(wings) as (bodies) has, to be identified the edge of distribua segment across whichthe doublet (Fig.3.4). tion Is discontinuous More accurate for the samecomputing cost, or for the sameaccuracy lesscomputing cost? flow. Better behaved supersonic in Lessstorage required. More sensitive irregular to paneling and gaps in the geometry. n

+ + +

In view of abovepoints It is quiteevident that boundary coidition with the Dirichlet the methods are considered be a definite to improvement over boundary conmethods with the Neumann the (older) condition is surface dition in whichthe stream much fashion. However. in imposed a more direct to methods pracin the latter expertise applying assuap. in ticalsituations, whichthe underlying locally, has beenbuilt tions are oftenviolated carry does not automatically up. Thisexperience over to the newermethods. methods higber-order versus 5 3 liper-order (for the Regarding matterof choosing the application) lower-order a end/or davelopment several argu. method, rather thana higher-order aents, pro or contra, can be put forward. a For
low-orJZder method it can be remarked that a lover-

for to The formultion appears be forgiving at paneling, leastfor subsonic Irregular flow(Ml), or through user-spocifled Lift-carry-over auxiliary surfaces generated Automatically to is subject somearbitrariness. in Thinwingsmy causeproblems casemode. functions t4 used (Fig.5.1),the lifting, for may surface approximation be inadequate thickvir. as well as for wingswith a s edge. bluntleading becauseof flowscannot be modeled Internal ,leakage. in Ku4-weve reflections the in. Spurious in terior the configuration, case of of supersonic flow.

ordermethod: and mainto program Is lesscomplex design, to is rain.Less Information required define have AIC expressions and less the geometry out. to be worked because higher-order no has more flexibility + or is continuity pre-assumod required. nor In the singular. in neither the geometri ity distributions. features the in non-physical can introduce which vortices flow fieldsuch as discrete numerical may giverise to spurious, afrects. flowunless for supersonic ti nor suited (R *triplets' ef. 26) or somekind of averaging(see 1sf. 29) Is introduced.

. THICK WING METHOD


Fig. 5.3

LIFTING - SURFACE METHOD

1l-pe of panel method (Courtesy Fokker Aircraft g.V,)

mm

5-35

THICK
0

-0- TEA,230

14LRPANEL
(L S.)
-&-PDAERO

P
-0.4 -0.2-

00 Pt0"6
-0.4. TEA2 PDAERO c=0.02 0: 0 ;.NLR PANEL -0.2.

THICK

0c

.t05 L..8 0.0-

.. .. 1.0" .

0.0
oX 0.2 .25

02 6113.
0.0.40.6"

HICK

( iRAE WOING -,c,

0.4/
0.6N NLR PANEL PDAERO 12 x60 12 x30 CPU-TIME 170 sec. 105 sec.

ql. 0.549

Fig. 5.4 Comparison between thickwing and lifting surface model Similarly. the hber-grdgr methods are possibly "ore accurate for the panelsize rendinr zero.i.e.in an asymptotic to sem. This does not Imply that for a specific(coase) paneling the solution of the higher-order methodis moreaccurate than the one obtained with the lover-order method using the samepaneling. more econeic whena fine paneling reis quiredfor, for example, subsequent a computation the boundary of layeron (partof) the surface the configuration for of or ces such as close-coupled lifting components. required for supersonic flow (K.>I). for and wake relaxation. lessflexible because more ordering re. is quirad in the specification the geoetry. of computationally expensive casecon. more in tinuity geometry of and singularity dietri. butionsacross eeant boundaries to be is enforced explicitly, leso attractive develop to because tho. a roughanalysis required minimize is to the computational effort, time-consuming code and maintain. to -os difficult vectorize. to 5,4 Whattwe of anel method to use The type of panelmethodto be uied depends strongly on the purpose of the application (Fig. 5.3).If the methodIs to be used during conceptual or preliminary designphase* an aircraft of project, whichmany possible in candidate configurations studied, are thereis no needto consider all the flow features greatdetail. in This mans thata method that provides six-component the forces and moments, stability derivativea and spanwise loaddistributions within certain to a not too demanding levelof accuracy will suffice. Also very oftenin thesedesign phasesthe designerwill lookfor trends ratherthanquntitatively accurate data.For thistype of application a relatively coarse paneling allowed whichwill is cut down on computer run time. Also,as demon. strated Fig.5.4. it will be allowed employ in to the lifting.surface approximation whichreduces the computer cost(thenumberof panelsrequired the lifting for components is halved) even further. is also possible It that for the purpose somedetailstudyisolated of partsof the conflguratf -- considered, ! a e.g.the wingflapsystem, utilizing finepaneling a for that part. but neglecting the interfernce due to other This partsof the configuration, typeof consider3tione may leadto the for application duringearly designphases desired situation wherethe *turn.around.timo is such that the method can be usedon a workstation Interactively a mainor on framecomputer. addition is required In it that the geometry ba handled can (defined, manipulated. etc.)easily whilealsopro-and post-processing can be carried out fast.In suchan envirorment the designer can Investigate rapidly effect the on the aerodynai charactaristics due to for example changing the position and tye of the propulsion system, changes wing-tail in lay.out, flapsettings, rollanSgls(formissiles). etc. As far as the lifting-surfee approximation ts concerned Pig.5.4 provides nsighWt in the accuracy an of the predicted pressure distribution. ahows that It for both wing thicknesses thick-wing the the and lifting-surface givecospsrable results, model except near the bluntleading edgewherethe lift-

It should be noted here that in manypracti. cal situations the panel scheme chosen, because of restrictions computing In budget computer or core memory,Is Just fine enough to resolve the relevant flowfeatures, that in thesecasesthere so is no advantage "ing the higher-order in method, Hoever, withcomputing cost decreseing core and sizesincreasing userwill tend to increase the the number panels of and the higher-order method will eventually become ore economic. should It alsobe realized thatat locations wherethe so. lutionis (nearly) singularly behaved, frose quently ocurs at the edgesof lifting surfaces, but alsoat sharptrailing edges, ate..higher. orderaccuracy formally is only attained the if singular partsin the solution extracted are and treated explicitly (e.g.Roe.30).This may be pursuedin 2D but is far too complicated be to extended the general framework. to 3D

---

5-36 . ang-surface approximation invalidated. t/c is For 0.02 the result the NLR panelmethod of deviates the result the lifting-surface of method. Thisdiscrepancy due to the failure mode function Is of to represent doublet the distji~ution corractly in the trailing-edge rejion, see also.Fig. 5.1. AIRfO-L4%CIRCULAIRARCBICOVEX L' 25,Di'.x.5 BODY PANELS C' , C
-'O 94

F'

-,

Fig.. givesan example~of 5.5 thi npplicationof a panelmethod (PDAERO) an Isolated~component to of complete configuration. show the inletreIt gion of a nacelle, the comparison computed nd of and measured pressures Alongtwo sections within the inlet region. This Is a type of configuration wherethe flowresembles internal an flowand application a first-goneration method-emof panel ploying the Neuseann boundasry condition resulted in the inunsatisfactory results. However, applying ternal Dirichlet boundary condition on the pirturbatIon potential resulted the rather in satisfactory correlation theory of and experlent, shownin Fig. 5.5. The paneling of the nacelle is shown in Fig. 6.1. Moredetailed investigations,a later at phasein an aircraft project, requiring accurate localcharacteristics as pressure such distribution. sparises axialloaddistributions, and hingemoments, root-bending moment. etc. Viii ask for a finerand alsofor the actual wing surfaces to be modeled, maybeevenfor wake relaxation. or Clearlythisrequires accurate, an reliable and comutationally efficient method.

.-. , -

(ALL OIMENSIONS ININCHES) MOMENT REFERENCE POINT: x re .1442 cL

flfuc
5OTsel

C(t n t.

s t= A'0 *, "
.

C p'C

X iw. Ap!
I
WASS4 4

-f,

pzE

A .... COMPUTATION I N E 0i,.5 D -xi~k lg nos engine h 05d COMPUTATION


-

--

.,

"

IV. R. BYPASSm

1.06

'

A Fig. 5.5

COMPUTATION EXPERIMENT

I V. R. BYPASS -0.09
.

. *

Detail study of the fiowinto an engine Intake

In caoss here the configuration operates at both subsonic supersonic and speeds (supersonic transport aircraft, combat aircraft, missiles) it willbe advantageous use a methodthat applies to to both subsonic and supersonic free-strem Mh numbers. The main benefit hare will be the saving in manhours prepare to justone input ratherthan the input for a subsonic flow method and a sepaps. racs onI for a supersonic flow method. Suchan example Application the iIRAERO of method Is givenin Fig.5.6. It Showsthe lift and drag coefficients function am of'angle attack of for two subsonic and for two supersonic Mach numbers, as have beenobtained a single run of the code. in

.a
*

Co5.4I Af o
5

7T S

esc

TrxsC.5.

Fig. 5.6 Aerodynamic characteristics wing-body configuration predicted by NIRAEO method for sub-and supersonic flow.

5-37
Abovediscussion leadsto the~conclusion a that general purpose panelmethod. probably secondorder, with severalaerodynamnc~modeling options as Neumnnand'Dirichlet boundary conditions,' thick wings and-lifting ,urfacee, default and use.'-specified nearwakeswith~and without partial relaxationinflow.and'out1ow'se ents;-optons to model (the effect of).propellermslipstreamsand due jet plumeoptions toaccount for~effects to a,bound. boundary ,layrs, 'option interface,with to ry-layer m.hod. subsonic--an4Wsupersonic-flow. capability. etc., as well asautomtic-(re)-panelIng options~is the 'aerodynamic tool! that is needed. A.prerequisite for-such 'building-block* a system(*toolbox*),is-that, inspite of the many optLons, the computationel method remains extendable, maintainable, economic and aboveall "uer. friendly'. Regarding the latter It must he kept in mind that In the application panelmethodsthe costs of required preparing for involved the manhours in th input and anslyzing the results of the compu. the bare computing tation generally exceed far costs. The firstfew item of abovelist,require that the user has access toa geoetry package for geometry miapulation..On the other hand. during preliminarydesign studies pirts ofthe onfigura. tion may, have a simple shape, e.g.,cylindrLcil.fu. wing, sage, constant-aitfoil-section zero-thickness fins, etc. ,Some of the panel methods avaLlable have an extensive geometry definition capebility,facilitating operation the methodin a of "stand-alone" fashion.

- i

F
"

ENGINEINLET
_______

DIVERTEt

(BY~, PASS),ETRA

6.0

PANELMMOD

vIRiI T

INTAKE Fokker modeling (Courtesy Fig.6.2 Geometry Aircraft 5.V.)

Is in As sketched Fig.6.1 the panelmethod 'bedded between pro- and post-processing. The is main taskof the e-processmin the generation the of the inputf')r method, which includes:

PRE - PROCESSING

FPANEL RESULT POST - PROCESSING


Fig.6.1 Panel ethod enviromoent

definition the geomtry, subdivision of into parts and seopnts. determination of inter. sections between wings and bodies. etc. paneling of the individual segments, or in case the method features automatic paneling features, specification the panelachem of
paraMeters

specification of auxiliary, non.configura. surfaces such as near-vko surfaces, Lion inlet face. lift-carry-over segments, internal surfaces carrying the %ode-function doublet distribution. etc. inspection of the paneled configuration (eg. Fig. 6.2) specification of the normal velocity compothat eimulates viscous nentvn distribution inletand outletflow,etc. effects, spocifIcation of slipstream data as velocity and total-pressure increment, other userspecified onset flows. etc. specification of free-streaom direction, Mach (roll. steadyratesof rotation number, pitch, yew), incresntal onset free-stream velocity due to motion of som part of the configuration. etc.

vectors (Courtesy velocity Fig.6.3 Surface Fokker Aircraft SV.) Inspection will involve the visualization of the paneled geoetry as a wire frame (Fig. 5.3). with or without 'hidden lines', as a "solid eodal'. as of to a -pincushion- checkon the direction the with workstations rather norals, etc. At present visualization ;ackages are sophisticated graphical widely available. Inspection of the paneling of corIox configurations using different types of visualization techniques is nowa much easier tabk of thanduring the earlydays of the application panel methods. woere printer output ,gas the only inspection tool available,

an||

5-38 The last couple of items on'above' list may require the interfacing with other methods 1like a' bound-, ary-layer calculation method, a method for wake relaxation, 'a'method'for is6lated propeller aerodynamics, etc As far a:the'icreaental onset frei-ream velocity is concerned W can be used to compute, ,ina quasi-steady approach, the separation of storeo from a parent aircraft. mnmnorlifting surface, condition; with or:without near wake,,etc. .Theovalue'are quoted for two co puter~systems, one scalatmatinfraef(rated 'at -2460 flops)-and one(on pr eoo)4sWperco. ... .. ter. From theseovaluos-:nindicatIon of required. CPtime fox~sr scaleroomputer.can be deduced from the differenceain thelprocnssorspeed in, frlao the oeffece::sdte~pie vr'much~ tensof'the floprati., For a suprcomputer,,the, Zoove of the coefficicuts dep'end very~aoch on~the

The min task of the post~yrocemgfnr'is the digestion of the output of the panel method. it

degree'of vectorization,' multi-tasking,!paralelli. zation,. etc. so ,that'.trasnslitlon to'other computer

generation visualization pressure and of plots. isoba8.urfaee (Fig.6.3)and free strealinWespolirsof'forces and moentsfor dif. ferent'Machknubers, etc-. - comparsii th data from other calculations or ifro experlints (weak-nteractlon) boundary-layer computation wake relaxation, archiving aerodynamic of date in a data-base sYtem. In the practical of panelmethods use the rapidand user-friendly visualization,of geometry and of flowresults advanced on graphical (color) workstations essential. is Panel methodsrun on workstations, smellto large minfrae computers on supercomputers. and The basiccharacteristics any panelmethod of are the following: - Number linesof the cods,Thiscan run from of a few thousand a vortex-lattice for mathodto more thanone hundred thousand for a generalpurpose higher-ordsr sathod. general code In the can be brokendowneasily alongthe main lines indicated Fig.6.1 and also to deeperlevels, in facilitating efficient segmented- 'capsuleor loading the object of code. - Nemory requirements 2 The memory requirement of panelmethodseis + O(N),whet N Is the num. aL ber of unknowns panels)and the value of A (or depends the method (Neuant or Dirichlet on or lifting-aurface boundary condition) varies but typically between3 and 7. This implies that depending N. out-of-core on mss-storage is required. Soe"pnal methods optimize the usageof main memory in order te cut down on I/0 to and from disk and therewith on turn-around time. This involves amongst other& the block-wise treetatnt of the AIC matrices. CPU-ties requirements. CPU-tle require. The sents of a panelmethod can be expressed as aic2N (a 1 or i 2) (1) wherethe coefficients depend very muchon the processor speedof the computer and for supercomputers the degree vectorization, on of multi. tasking and/or pasrallelination of

systemsis so difficult. Note thatabovetableindicates'that, oM sofor lution a scalar-computer; iterative on the solutionprocedure requires less computer tim than the direct solution procedurefor N's exceeding a value of 2.0 times the number of iterationsre. quired. which is almost always the case. For the vectorcomputer thisvalueIs evenlower. However, in both casesthe Iterative nothod willrequire sore 1/0 operations. As an example consider a SO0-panl case which will require less than5 minutes CPU timeon the 2megaflop ainfrae and less thanone houron a workstation with 1/10 of that processor speed. Finally is notedthatthe higher-order it methods PAMAIR requires, a specific on typeof computer system. substantially computer more tim thanindicsted the tableabove,thanothersecond-genin erAtion methods thanfirst-generation or methods (seeRef. 31). Also the high.order panelmethod ISSSrequires relatively muchcomputer time(e.g. Ref. 32).The reasons behindthisare not easily assessed, the elaborate in whichthe nubut way mericalschemes are set up (likein Fig.4.7b) mightbe an Important factor.

7.0

OPPORTUNITIES I4PROViENET FOR

Thereare several areas where(existing) panelmethods be improved (see alsoRsf. 33). may Referring Fig. 6.1,wherethe various to partsof a panelmethod are indicated, following the items are considered. =,f5: This partof the program handles the goometrc Input, 'stand-alonein panelcodesit also actsas pro-procesoor defineand subsequently to panelthe object considered, this part of the In foienan method automatic &I for a (curvature. even procedure generating or eoluton.)gdgovyAJnln, wouldrescitin an increased accuracy the noof mericalflow olulmation.

AU; In this partof the program the influence integrals svaluated. The operational are countof ). In panelmethods, In the table thatuse an belowsoamvaluesof the coefficients thispart Ia O(N for the system iterative solver of equations, this appearing Eq. (6.1) in are given, part of the program accounts moat of the total for C01 time Hereattention has to be paid to the . " sic -,u -it vectortz~tn of the code,such thatit runsefficlentlyon super.computers. an example the As of 4 . " Cyber 180-96) (2-8)xi0. 96 l.I0 3O 6 speed-up that Is obtained on super-computers NCbr SO. 62 (2-$)xl* the 2 340,9 x lO' tablein chapter givesthe coefficients the " 6 in NECSX-2 (3"7) _ 2.0xlO c<10_ CPU-time formula. Eq. (6.1), for running the (scalar) NIX AEROpAM/PDAERO method a scalar panel on computer and on a vector computer. "Infra The valuso refer to the CPUrequireaents of NU's DAERO/AEROPAM and as far *s the Cyber is code It shrvs that even for the unmodified code a subconcerned also of the sub/supersonic NLP&M code, stantial speed-up a factor 15-20is realitof of bothapplied a number testcases. to of The range ed. Not* thatbecause of differences coremeo. in of values given refers to different types of runs ry used (Cyber 962: 1 Kwordin a virtual memory eachas withor without symstry, DirichIet, Nou. environment, SX-2: 16 Mvord main nemry), part of

5-39

the reduction the CPU time is due to the of smaller amount-of activity'required.is ex'. I/O It can patted thit the' CPUztie 'required'for'AIC be reduced furtherbyre'srranging.theomutaton suchWthat,'s greaterpartoftheco for-cooputing ' the'lC'$ svctlziso(e'al-o Rif,'32) :. - " Areother of interertI~'isx area lgziof the 0(82) operaional' o tlf, e.g.' toO(NogN) , this uithoutzs-acrificing thieaccuricy of the:solution: Althoughljome studli have been-initlaitd in this area,e.g.Ref. 34, progresi has beenslow. 2Q: In thisepart of the program the-system of, equations solved, is either usingan iterative procedure a direct(L1)decosposition or procedure.The direct solufton~require ) opera. O(N cione, but~oan bar I to alargextent. see the tablein chapter'6 whichshows' aspeed up by a factor 750' of Thesiterative solution procedure,-mostly blockJacobi Causs-Seidel, or requires O(it*N) operations, with it'the number Iterations, but this of at the costof an increase the number I/O in of operations to'becarried with the matrix out during the iteration. mainproblem The with the iterative procedure Is'thatfor complex configura. tions, 'depending'on paneling, numberof the the Iterations beixcessive or-itmay occurthat may the procedure failsto convergence. for lift. Also ing surfaces in-subsonic flow (which did not resu.t in a Fredhola integral equation second of kind)the convergence rather Is slow (e.g.see Ref.9). Morerobustiteration procedures, con. as jugate.grsdient of methods, type are to be Investigated. For running panelmethod a workstation a on or lowor-end mainframe the availability an iteraof tive solver remains must.For a vector a computer the CPU timerequired for the voctorined LU-decoa. position remains relatively modest t3 higher up numbers panels, of but eventually N count the will takeover and an iterative solution procedure mightbe preferable. the itarative For solver one should alsotake intoaccount the increase in turn-around time due to the increased opera. 1/0 tions needed. RESU : In this pertof the programthe velocity and pressure distribution computed are and are used to computeloaddistributions, forces and sosents the complete on configuration on its and individual components, center pressure, of veloci. ty and pressure off-body at points(e.g.for streamline tracing). etc.Also a file is prepared for use duringpost.procossing.

whichthe flowfielddue to the propeller isoin lation superimposed the'free is on stream anas additional onsetflow (seeFig.8.1).The data for additional- onriet flow is obthe (time-averaged) rained from a propeller program basedon. for instance. blade-element momentum theo r

Fig.8.1 Simple modeling of effect proFeller of slipstream flowaboutthe wing on A panqlmethod can alsobe used to compute the steady flowaboutthe rotating isolated propeller in a blade-fixed co-rotating coordinate system, e.g.Ref 35. For thiscase in whichthe free. stress has to be directed alongthe propeller axis,one bladeof the propeller a segment and of the axially-sysmetric is discretized hub into panels. The influence the otherbladesis sc. of counted for using sulti-lobed axial symetry. The min difficulty the modelis the wage,whichis in a helicoidal vortexsheetliteracting sore strong. ly with the flowaboutthe propeller bladethan is the case in a conventional interaction a wake of with the flowaboutthe wing thatgenerated the wake.In the isolated-propeller methodit Is as. sumedthatat some distance downstream the pro. of peller disc the wake is fullyrolled and all up vorticiry contained is within a cylindrical vortex sheet whichformsthe far-wake modelof the slipstream. On the slipstream far wake the angle between the vortexlinesand the axisof the slip. stress is constant. From the Isolated-propeller solution the additional onsetflow in the configuration-fixed coordinate system obtained is from someaveraging procedure. It "spears thatabovesimple add-on onset-flow mode,. in whichit is assumed that the interaction is weak and the slipstream is not affected the in interaction. i not always adequate and an Inprored modeling required. is One possible Improved sod.lis to approximate the propeller an actuator as disc carrying doublet a distribution givenstrength be obtained of (to fromthe tise.averaged loading the propeller of blades). Dovn.stresm the propeller cylindri. of a cal vortexshet. whichrepresents vorticity the withinthe slipstream, trails from the edgeof the actuator disc.Both the actuator disc and the trailing vortexsheetare paneled, the strength of the doublet distribution the wake vortex on sheet is determined from conditions similar the ones to used In the partial relaxation conventional of wakes.In thisway at least som mutualinteracnlonof the slipstream the flowaboutthe con. and figuration accounted is for.

8.0

FURTHEt MODELING ASPECTS AND RECENT AREASOF INTEREST

Thereare numerous areaswherethe panel method technique been applied has successfully. Someareasthathaveattracted sos attention recently considered below: are 8,1 Pronulslon installaton effects. The renewed interest propeller in propulsion of transport aircraft has brought alongthe need to predictthe effects the slipstream on the of configuration aerodynmics. During preliminary design studies mostly simple a modalis used in

Ur

5 40

EXPERIMENT.
r P

o
NLR

PANELMETi-OD N AER . ,,

tion..Tis IillustratedinFig. 8.,wlich shows.thejift coefficient, isa fnction- ofangie,of


attack-foc a-sple wing-body configuration in, the measured values, -the values, following from the NIR panel mthod which employs the thick-wing, modeling end the.values from SLRAERO which employs, the lifting-surface approximatton.'The 'lifting- -

incopresible flow-Three results are presented:

CL 0.8 0.6
0.4 0.2
-

THIKLIFTING THICK i "-SURFACE


WING
.factory
,

-,'
'

10
-0.EXPERIMENT

00-

surface methodgivesthe best correlation with experimental data, However, the latter method yields a less satis-

representation ot the chordise pressureONSET BOUNDARY distribution. OF LAYERSEPARATIONIN

A next stepis to *pply,forthe lifting components


of a configuration relativalyslmple formulas for the development of the boundary layer on a-flat plate employing the computed Inviscid velocity or pressure distribution in a stripuisefashlon. This will give a first estimate of the skin friction and of theaboundary-layer displacement thickness. Subsequently the displacement,thicknesscan be used to model the effect of the boundary layeron the inviscid flow and specifically on the lift, The latter can be accomplished in either one of two ways (see Fig. 8.3). In the flrst one a new wing surface Is obtained by adding the boundary. layer displacement thickness to the solid wing surface. This approach is not very practical because it would require the definition and subsequent discrstization of a new geomtry, which is rather elaborate for a general threedimensionel configuration, while it also requires a costly recomputation of the AIC's. In the second, more practical, approach an outflow velocity distribu. tion vn is computed froe the displacement thicknose such that the actual surface transpires enough fluid to cause the resulting Inviscid flow field to be displaced by the same amount as in the viscous flow. Re-computing the pressure distribu. tion from the solution with specified v yields an improved estimate of the pressure distribution, lift, etc. in viscous flow. During the detail-aerodynamic design phase a more accurate procedure will be required. Now the pres. sure distribution re-computed by the panel method can be used to re-calculate the boundary layer on the surface of the configuration, etc. Under cruise conditions the flow will be attached. a weak Interaction may be assumed and the hie:archical procedure followed, iterating between the p0tential flow method and the boundary-layer method, willusually converge. Ho .ver.for configurations typical for take-off and landing conditions, which feature pressure distributions with high gradients, a strong invisdviscous flow coupling Is to be taken into account. Furthermore, for eg wing-flap and slat. wing configurations also the viscous ake modeling needs to be considered in more detail. An importent it*a in any coupling of a boundary.layer

-10

-5
-0.2 .-0.4

10

15

(deg)

Fig. 8.2

Correlation of computed and measured Wlitcoefficients.

8.2 Viscous effecsa 8.2.1. Liftings omoonents The most simpie way to account for viscous effects, in en englneeri ng fashion, Is to neglect the thickness of wings altogether, i.e. consider wings merely as lifting surfaces. In this ayproach use Is made of the general experience that wing thickness effects (which increase lift) are cancelled by viscous effects (which decrease lift through an effective de-cambring of the wing). This will yield a reasonable total lift and spanwise lift distribu-

.. '-IPLACEMENT

THICKNESS/ SOL!SURFAC 'NEW SURFACE NEW PANELING


-- W ...... " OUTFLOW~-, O O n
"

WAKE

a)SURFACE DISPLACEMENT (not to scale)

method with a panelmethod It that typically the

arbitrary-geometry capability is further developed for panel mathoda then for boundary-layer methoda oreover. boundary-layer calculations require a ..... Nch stronger coupling between configuration sagets than one is used to in panel methods. 8,2.2 Bodv.lik, roooner. As far as accounting for viscous effects on body-like components is concerned it will be clear that in cases, such as

Oi

SURFAC

WAKE
b) OUTFLOW FROM SURFACE NO NEW PANELING
Fig 8,3 Simulation of boundary-layer effects

wing.body configurations. the simple atripwise

flat-plate type of approximation is not valid. Carrying out boundary-layer method calculations for an isolated body at small incidence may be feasible employing a weak-interaction technique.

5-41 paneled. i.e. the-panel method is applied to an open-ended body, In-& panel-method that employs the Dirichlet bounarycondition the f'ctitious part oi coe body hbstihe included-in the model, becauoe the ethod applies to closed'bodies only.

However, athigherincidence-separatlon might occur ano Also for the simple body In isolation be Accounted for. Sb q-:0

-NOT-

CNEUMANN
--------- [-C

LThe

CUT PANELING

classical rigid.wake approximation of straights trailing vortex lins(Fig. 3.8) is coupled components as wing-flap-tail~configurafigurations (combataircraft and missiles 14.-c, at >1). The rigid-wake approximation with a use. specified near wake .(Fig. 3.9) will'improve the modling only if the vortes'linos p consi.) on this part of tha wake are to a sufficient degree Aligned with the streamlines. Apeasible partial alleviation this problem Is to fix the shapeof of the near wake and its (paneling to -telexbt the dult dis t ribution locatiLon of the vor tex lines) the near wake by imposing AC - 0 on the condition, (3.114). Eq. Thislatter condit~on Is quadratic Is. thatant In so Iterative procedure is required, whichhowever doesnot Involve re-toma putotion the AIC's. of see Eq. (4.15b). Because of the weak non-liseaity it Is usually not necessa ry to update the Jacobian Eq. (4.18b) in eithordurIng partial sate relaxation. Thejbng of the neat-woke, stillto ho specified the user, by mightbe obtained roe a methodthotsolvesthe f non-linear problem ^n approximate In framework. An example of thisIs the method whichcomputes, within the framework of the 2D tim-depandent *salogy and employing a 2D second-order panel method. the roll-up soreor less arbitrarily-strutred of vortex wakes (Ref. 36). In Fig.8.5aa typical to suit of thismethodIs presented, whichalso serves to demonstrate complexity the wake the of of configurations deflected with flaps.

Sb

PNLD
-*

DRCHET

BASE CAP -

- -HLE

CLOSED BODY
Fig. 8.4 Mlodeling separation froma body-like component Within the framework panel methods of strong viacooseffects, specifically the rearend of a at bodylke component, be accounted in an can for engineering fashion Is shown in Fig. 8.4. as In a panelmethodemploying Neumann the boundary c.-Aition the body Is extended from Its base to InfisIty downstrea. At net too high Incidenc* It may be assumed that the resulting source distributionon the fictitious part of the body will be relatively sm-l1,Thisthen leadsto the modalin whichthe fictitious part of the body Is omtted altogether and the base of the body Is n VO20T WA9l OA9tT"i 0GWN PA(Y RAP PAIT4SPAN 0

02 Z 001 02 Z 00 02 00 -020

VOA2OT

to-

VORSDA

1 GOER

5
0

PRESSURE COEFFICIENT

00 z
-. 2\
4

04I
X X TE. COS at

12
E1

-0 L 4
00 S0 to

001
00

__

05

i0*o- y',

Fig 8. Exmpl ofresults of panel methods for configurati,.s with frem vortex sheets; sppro.,mate frame, works, a) time-dependent analogy. b) aiender-body approximation

5-42
SIn

cases where the interaction of the wake and the solid geometry is stronger, eg. separation from flap side edges, wing tips or for ,slender wings with leading-edge.vortex~sheets,tho two wake boundary be solved simultaneously.1hTe (3.11d)) have to conditions ( (3fllb) and oqs. fully nonlinear 3D-wske relaxation-probleo-Isa to-igh problem. Here also methods formulated in an approximate framework, as slender-body theory, are used for preliminary studies oras preprocessor for constructing the initialguess for~the method for fully 3D flow. Fig. 8.5b presentsothe result of such a non-linear second-order parel method (Ref. 37), formulated-In the glender-body.approx i motion. Sho n is-the solutia for the~flou'about a thin delta,wing of unit'aspect ratio-. tequenc* of incidences, Subsequently-such a solution is used to construct an Initial guess for the~sethod for fully 3D flow, see Fig. 8.6 for a typical result. Mre results and details of the vortex * sheet relaxution methods using (second-order) panel methods are g iven in Ref. 38 .

9.0

EXTENSIOWOFDOMAIN OF APPLICABILITY

'The domain of applicability of.the'pael, a-thod for linearized Potential flow,slaisi.ted to sub-critical flow. However approachotoof ti e cap&bility o~f the panel method extensi n flows with regions In which nonlinear compressibility effects cannot be neglected is possible. In one'approach the integralrepresentation for thesolution of the Prandtl:Clauert equation. Eq. (2.5S), includes the contribution due to a source distribution e n the flow field surrounding the object, i.e. the solution of Eq. (2 3 a) is now expressed as 9(0o
)

" vq(xo

9",(;0

+ 9'e('o)

(9.14)

where po and and q ( o


)

are defined in Eqs. (2.5b andc) ' vf ( [ ( . b

2ldeg A - 65 deg (X- 20deg

linear coopressibiity effects and o(x) the spa(field-)source distribution. In the -fieldpanel* method the spatial source distribution Is found by satisfying the full-potential squation Eq.(2.1a) at the points within V( ). It has shown in Ref. 40 that for the 2D (Irnsonic been goall orturbatton) case modeling of super-critical flow with shock waves is possible. In Ref. 41 the 2D fiele.pane.epproach was extended to tho fullpotential equation, Eq. (2.1a), using established techniques of contemporary finite-volus. methods for transonic flow,

Fig. 8 6

Solution of nonlinear panel method leading-edge vortex separation

VORSEP for 3D flow about wings with

64 + 150 + 50 + 50 - 314 SURFACE PANELS


0.2

(I*1

YI
ZX
.. ,.

-0.2.
-0.4 GRID 40 BY 8

0.40.2 0 02040608101.2
, -"
FRONTVIEW

VORTICES TRAIIG OFF


TO DOWNSTREAM INFINITY

-20.TYORPANNIC11e
'15 10i '-p PANEL

TRANSOIC RASNC
"151
I

X1

HYDROPAN
HULL

300 PANELS

METHOD

Mr-THODI

PANEL

(FREE S
Fig. 8 7

WINGS'

.1106

.530

PANELS)

PANELS

.5;

Exoaple of panelling of hydrodynaical application (Ref. 39)

54
Fig. 9,1

5J=I

0 0.25 a.-14i 0,G 1.0

6.4

lcl4.

Application of field panel method to &-cooponentairfoil section (Ref. 41)

Panel mathods art also applied in hydrodynamics. For the flow ebout submerged or partly suberged objects the offsct due to the free surface say be substantial. At NL the NIR panel method (Rf 5) has best extended to hydrodyneeiCal problems by including the frse surface (Roe. 39) On the paneled free surface, which is appro. Yimated as a rigid surface, the linearized freesurface boundary conditions are applied. From the computed solution the wave resistance is deduced, Fig 8 7 shows th.e paneling for the application to a realistic complex keel configuration,

Fig. 9.1 shows the result of the application of this method to the flow about a 4-component air. foil section at 14 deg incidence at XC.- 0.2 and 0.25 It is observed that, aS K. increases from 0 2 to 0.25, a *sall region of super-critical flow, terminated by a shock, develops on the highly loaded slat. This example indicates that for configurations with separate compact regions of non-ngligible nonlinear compressibility effects the panel method can be applied with suc. reis, without sacrificing the capability of linear

5-43 panelnethods'to -treatrcomplex geometries. However. note, that the number field of panelsin. creases very rapidly the'extant the tranas of sonicflowregion(s) becomes larger. If the field-panel methodis to be applied successfully to 3D configurations computation. the alcosts for evaluating influence the integrals willhave to be reduced considerably, by2vectori. nation or preferably lowering O(N ) operaby the tional count. Another pointworthnoting that Is ideally regions the withnonlinear flowshould haveto be detected automatically the program by In somekind of iterative procedure. Ref. 42 describes an application of the fieldpanelconcept the compressible aboutdelta to flow wings with leading.edge.vorteo separation, using a nonlinear vortex-lattice methodto simulate the linear potential flow An alternative the approach fo. usinga fieldsource distribution the zonal(hybrid) is typeof approach. Here the full-potential equation Eq. (2.1a) solved the regions is in wherenonlinear compressibility effects ncn-negligible are and the Pratdtl-Glauert equation elsewhere. two The zonesare couplediteratively through boundary the conditions the interface on between zones the At Boeing(e.E Ref.43) the PAIN code is AIR to flowby superimposing beingextended transonic onto the arbitrary surftca-paneled configuration a spatial r.ctangular grid.Thu volumeintegrals on uniformgrid.wnichis not body-conforming. this are evaluated usingPastFourier Transforms Someotherdevelopents are described Sinclair by (Ref 44) who reports the development a on of field-panel methodfor three-dimensional ronfigurations withgeneral geometric capability, Otherinvestigators (Ref.28) suggest that there compressible flow are prospects that the nsnlin.ar in. in can problem be formulated termsof surface torrals only.though the latter have to be re.evaiuuied in an iterativa procedure

n
, Se

U _.

S :U.ffnz0:Cpf st
Sw:(U .iuP).i=O; (U +uPJ.V Fig. 10.1 The inverse problem 10.1 Thickvfrn, The problem be formulated follows can as (seeFig. 10.1). solution the Prandtl.Clauort A of equation 3a) is to be foundsubject the con(2 to ditions that the surface a streamsurface, Is Eq. (2 4e):
*

=O

p).i - 0

(10.1.) .

for x on sb and the condition that C (2t) - f(s.t)


o

(l0.1b)

alsolor ' n sb' FromEq (2.1c) follows it that thenu. whichis now tangential the surface to
satisfies:

lul /1].
whereu -

-2

*of(,.t)i?/( 2...L
2

"l)

10 0 INVEtSE PIOBLEM In the preceding chapters the boundary con. ditions referred simulating flowabouta to the givengeometry. e concerned so-called i the aIysisproblem Subsequently computed the surface pressure distribution integrated yieldthe was to forces and moments the configuration the on in design problemthe pressure distribution, and therewith forces the and momants, are specified and the geometry the configuration the of Is nought for solution Sometimes not the entiregoosetryis unknown, onlysomepart,eg. the body goetry is givenbut the singgeometry un. is knosw, the geometry giveneverywhere is exceptfor some part on the uppersurface the wing.the of geometry the wingbox is fixedwhilethe nose of and the trailing-edge region are to be modified, ett in the foIlowing somemethod. used in the design of the geomotry aircraft of configurations are discussed. Designmethods basedon someoptimiatien procedure whichpurely analysL method In an is used as driverto find for exaple feasible search directiona not considaredThe discusare sIon that is presented here is far from complete. it Is meantAs a firstintroduction the subInto jectof the inverse (panel) methods

(10 Ic) + Ov. -q. (10 Ib) is to be supple. mented son& additional by conditions suchas that the surface closed; is the planform given, Is or the position the trailing of edgeand the spanwise distribution the chordlength; someother of or conditions contraining the geometry, inverse The problen, whichcan againbe expressed termsof in singularity distributions, a nonlinear is problem and resembles wake relaxation the problem. However. "in difference thatnow a hon. the is intersecting closedsurface to be found,rather is thana single*.surface vortexsheet The general three-dimensional invers problem a complex is problem with suchrather difficult aspects that as the prescribed surfare pressure distribution can not he arbitrary, i.e. it should such thatthe be resulting flowsatisfies -oology rules A(stagnation separation , and ...also in connrctionwith the sign of m in Eq. (l0.lc). Moreover. the resulting goestry should closed be and should not intersect itself otherthanat the trailing edge Apartfromapects of uniqueness, a i whetheror not an non-intersectirg closed surface can be obtained. resulting the geometry should also be acceptable froma structural pointof view (vingthickness, trailing-edge angle.leading-edgo radius. etc In "at casesthe designproblem beens has formu. lted as a conatrained optimization problem in whichthe mw geometry is obtained minimizing by a functional like Pl ( t- 2,* -. gin - x f (bpW - CP(x))

IdS

(10 2)

5-44

supplenented certainconstraints In Eq. b . (10.2) and x are the "target' C pressure distributtonAd the -targetgeometry from whichthe geometry should notvdeviate too uch,whilew and V are weighting functions, Solving problemin the this way Is generally e successful mor thansolving simultaneously two coupled the nonlinear integral equations resulting from imposing Eqs. (lO.la and c) by a Newton-like iteration procedure, Inverse panelmethods havebeen developed two. for dimensional flow(e.g.Ref.45) whilealeso in PANAIR the tangential velocity distribution be can specified boundary as condition. Thereare several approaches solvethe inverse to problem. possibility the following: One is (i) Compute from the specified surface pressure distribution givenon the present (initial) iterate the geometry velocity of the potential (a non-trivial taskin 3D); valueproblem; (ii) Solvethe Dirichlet boundary (iLL)Compute velocity the component normal the to surface; (iv) Compute geometry the correction, taking all kindsof constraints intoaccount; (v) Determina next iterate the geometry. the of Thesestepsare repeated untilthe geometry correction sufficiently Is small.In thisapproach at is on the condition the pressure satisfied each step.the stresm-surface conditfon iterated is on. At NIR (Roef 46) a slightly different approach is followed whichcan be sketched follows. as (i) Define the target pressure distribution, the &s geometry well as and the initial target the weight factors; (if) Determine pressure the distribution the on present(initial) Iterate the geometry, of employing NIX panelmethod(thick the wings); (ii) Determine difference the betweenthe target pressure distribution the computed And pres. suredistribution; (iv) Solvethe inverse lifting surface problem, a linear problem(seesection 10.3), find to the correction thegeometry; on v) DOetermine next iterate the geometry. the of The iteas(i)-(v) are revisited untilin step (lit)the computed pressure distribution suffiis ciently Closeto the target pressure distribution. So. In this approach stream-surface the condition is satisfied each atop, at whilethe pressure dis. tribution iterated Is on. 10 2 Perur ton.ali. mrhod (MCAERO) at The methoddeveloped McDonnell has the possibility computefor a givenbase. to linegeometry not only the solution also the but derivative the solution of withrespect geometto ric erturbations This meansthat if *0- 1(X ) is the velocity potential somepointx on b at I-l(1NG. also the threemetrices IJb.m

for i-()NU, vere, 1 . 1-1(1)N denotes theloce-, tionof pointawon now.geoetry.,Althouh,this the requiresthe, computation three of additional,matrices, solutionsfor (partly) perturbed aircraft configurations readily are obtained (no - trlx equationto be solved), allowing extensive studies of for example different,wing thickness,andor camber-distributions, flap~settings, Also the, etc. effects aircraft of structural flexibility be can investigated rapidly. However. what orderof to accuracy the stream-surface condition satisfied is on the perturbed configurations not so clear. is The aboveperturbation analysis been extended has to the design problem, thiscase the derivative In of the pressure coefficient ;i with respect at to all perturbations to be determined has (likein Eq. (4.19b)). e.g. aC. NS k where*k" k-l(i)NS are the *'e withinthe 'stncil" (domain dependence) C at a givenpoint of of on the base-line geometry. t~s pressure Now coaf. ficient a perturbed on geometry becomes C C (xi.Yi'Zi+4z) ) (x) + Epi C C aO k j (I (1.a)

(10.4b) for i-l(1)NU. perturbation is determined The Az froma least-squares minlization of the differeocesbetweenthe target pressure distribution and " )Z i e E ) (10.b). given In Cp(X + With the 4zi, 1-1(1)h11, fourdthe pressure coeffi. cientcan be updated usinglq. (10.4b), etc. This procedure appears work rather to well for vingsas wellas for wing-body corflgurations. surprisingly also in case the dovistions the finalgeometry of from the base-line geometry no longer is small. Constraints mostlyusedare thatthe trailing edge is fixedin spaceand perturbations allowed are in vertical direction only.whileagainit is not clearto whatorderof accuracy the stream-surface condition satisfied the designed is on configura. tion 10 3 Liftin,surfaces As already mentioned. is a possibility it to consider the inverse problem the framework in of the lifting-surface apvroximation. follows It from Eq (2 6f). see alao Fig.2.3. thatacross the lifting-surface (assuming incompressible flow) r _ .r- ) _ 4r (10 5e)

with Art denoting pointon the upper(+) and a lover(-) side of the sing reference surface . Sr From the expression the linear for pressure coatficient, Eq (2.3c), follows it thenfrom the di. in pressure acrossthe lifting surface: -0 p p
r

!!,.

Jp()N j-( l

(10 e)

l (A p

' 2

Ar/u u/u o
/U.

O(s - 2 1 are computed, for the solution all shoutthe base(10 5b) for The solution a perturbed line configuration geometry is thenobtained linear by ccarapolation. from which,for givengeometry Sr the doublet of , I. distributionon Se can be deteined,. From the P sxigu of the pressure the upperand lover on NG aing reference surface one obtains from the ex. 6(ni) ( - xj)" (10 3b) pression for the linear pressure coefficient. the j(i dcj following integral equation for the source distri-

-5-45 bution-on the given wing reference surface: qISrf o S (10.Sc) shifting from the detailed aerodynamic design phases to the preliminary design phases. The extension of panel, methods into the tran. sonicflow regime and the improved handling of willhave a components wakesof closely-coupled direct implication the extension the doon of main of applicability the panelmethod. of - Thereexistpossibilities improving comfor the putational efficiency the panelmethod of by: vectorization or parallelization and on supercomputers .reduction operational of countfor the evaluation influence of integrals new formulations improved and numerics .better, more robust(andfaster) iterative procedures solving for large, non-sparse whichwill leadto a of systems equations further utilization panelmethods. of
-

2(C(~r+) + C(;r))

This is an integral equation resembling Fredholm a integral equation the firstkind and choosinga of stable numerical scheme express and its deto q rivatives the panel expansion pointIn termsof at a set of source parameters be solvedfor is a to non-trivial matter.In orderto automatically sotisfythe condition thatthe totalsource strength mustbe zero.it Is advantageous define to the source distribution termsof the gradient a in of source-doublet (doublet with its axis tangential $istribution. to the surface) Once the source distribution the doubletdisand tribution the wing reference on surface havebeen determined wing-thickness the distribution follows from Eq. (3.10b). againfor givenwingreference surface. Finally wing.camber the surface distribution follows fromEq. (3.10d). In abovesketch the Inverse of lifting.surface problem justconsidered lifting we the surface with unknown thickness distribution leftout and the presence any othercomponents the conof of figuration. Adding thesefixedgeometry components in the formulation provides realdifficulty no otherthan thatnow pert of the "atrix.equation to be solved stem fromthe inverse lifting.surface integral equation. (1O.Sc). rather Eq. than from the direct lifting-surface integral equation. Eq. (3.10d) The lifting.surfac formulation alsobe Used can in a *partial designoption. such an optionan In incremental caber is defined for example alby lowing the MAairfoil sections a segmentto roin teteabouta givenaxisby a yet unknownangleCJ J-l(l)HA. stillImposing boundary but the conditionon the fixedwing reference surface. This impliesthatthereere MA parametore the rightin hand sideof the lifting.surface integral equa. tion.Eq. (3.10d). Itsdiscretized or form Eq (4.14). This leadsto MA basissolutions. i.e NA S CS J Cj"l for i-lIl)MU (10 6)

Pr.-and post-processing, an essential are part of the Opansl-method enviroment. At all timesduring the application panel of methodsit should realized be thatpanelmethods are modeling the r-.L flowundera greatnumber of asa._ptions.

12.0 REFERECES 1. Hess.J.L..Smith. H.O..Calculation A of Non-Lifting Potential flowaboutArbitrary Three-dimensional bodiese, Douglas Aircraft Report Mo. E.S. 6062? (1962), J. of Ship Ro. f, Mo. 2. pp. 22.44(1964) Rubbert. P.C.,Searls. C.R.:A General Three-dimensional Potential Plow Method Applied V/STOL to Aerodynamics. Paper SAE 66004(1968). Prager. U.. Die DruckverteilungKirperin an miner PotentlelatrZsung Physik. Zeitschr., pp. 865-869 (1928). Martensen, We erechnung DruckverE.: der teilung dicken an Citterprofilen Hilfe sit von Fredholmachan Integralgleichungen zveitar Art,Mitt.Max-Planck-Inst. Stromuogs Porschung 23 (1959). No Labrujere. E.. Loeve. Th. U.. Slooff. V J An Approximate Method for the Calculation of the Pressre Distribution Ving-body on Co. binations. ACARDCP-71(1970) Kraus. W Des HMB.UFE Unterschall Panelverfahren Report MiB-UPE 633-70(1970) Hunt.B. Semple. G. The 1AC(iAD) U Program to Solvethe 3-D Lifting Subsonic Neumann Problem usingthe Plane PanelMethod Report ARC97 BAC(MAD) (1976) S ,odward, P A.. An Improved Method for the of Aerodynamic Analysis Wing-Body.Tail Con. figurations Subsonic Supersonic in or Plow Ct-2228 (1973). NASA 9 Hoeoimakers. H.5. A PanelMethodfor the Determination of the Aerodynamic Characteristics of Cosplex Configurations Linear. in ied Subsonic or Supersonic Plow Repoct NIX TR 80124(1980)

wher*S is the solution the singularity for para-tere for whichthe J-thairfoilsection set is at a unit incresental Incidence whileall the otherC Is are set equalto zero.The MA degrees of fre." can for instance used to prescribe be the sparoise liftdistribution, this in presence of the fuselage otherfixed-geoetry part&of and the configuration A further example utilizing linearized of the boundary conditions to have control-surface is do. flections. and possibly also engine-inlet paflow radeters propeller-disc loadparameters, s or degrees fredoo to accomplish of trimed-flight conditions

6 I

11 0 CONCLUDING Ri266
-

Panel metlod are important aerodynamic tools with powerful and flexible modeling capabillities.whichare eed heavilyin aircraft design projectsThe application the panelmethodis of

5-46
HgherOrderPanelM-hod for Predicting Subsonic Supersonic or LinearPotential FlowsaboutArbitrary Configuration .AIAA paper81-1255(1981). 11 Hess.J.L.:A Higher-Order~ftoel Methodfor Three-dimennional Potential Plow. ReportMDC J8519(1979). 12. Roberts, Rundle, Computation In. A..* K.: of compressible aboutBodies Flow Iod Thick Wingsusingthe Splice-Mode System. BAC(CAD) Rep.Asrm Ma 19 (1972). Bristow, D.R.: Development of Panel Methods for Subsonic Analysis sod Design. NASA CR 3234 (1980). Sea also NASA CR-3713 (1983). Lltstedt, P.: A Three.Dimensional Higher. Order Panel Method for Subsonic Flow Pro. blems- Description and Applications. SAAB. SCANIA Rep. L-.1~ RIGO (1984). Foranier. L.: HiSlS- A Higher-Order Sub. sonic/Supersonic Singularity Method for Cal. culating Linearized Potential Flow.AIM Paper84.1646(1984) Mashew. B.: Prediction of Subsonic Aeco. dynamic Characteristics. A Casefor LowOrderPanelMethods Journal Aircraft, of Vol 19. No. 2. pp 157-163 (1982). Youngren. . Bouchard, E E , Coopersaith, 8.8 R.M..Miranda. R : Comparison Panel L of MethodFormulations IesInfluence the sod on Development QUADPAN, Advanced Of An LowOrderMethod. AIM Paper83-1827 (1983) L;, T H.. Morcholano. , Ryan.J Y Tech. niques NuairiqusNouvelles danslIa Moth. odes do Singularitis I-Application pour i dos Configuration Tri-Dimenaionelee Coin. planes,Paper6, AGARD.CP.412 (1086). Boppe, C.., Stern. M.: Simulated Plowsfor Aircraft with Nacelles, Pylons and Wingieta AIM Paper80.130(1980) See alsoNASA CR. 3242 (1980) and NASA CR-4066 (1987) Stager. L.. Technical J Evaluation Report ACANDFDP Specialists, Meeting 'Aplis.. on tionof MashGeneration Complax Con. to 3.D fSurationse ACARD-AR-268 (1991). Kellogg. D Foundations Potential 0 of Theory,Dover(1954) Wlard, 4 Linearized C Theory Steady of HIgh-lpeed Plow, Cambridge University Press (1915) Mangler, U , imith, H a Behaviour K J of thc vortonSheetat the Trailing Edgeof a Lifting Wing RAE TB 69049(1969) James.i H On the Remarkable Accuracy of ti,. VortexLattice Disccetization Thin in VinE Theory Douglas Report DAC61211 (1969) Hess,J L Consistent 4sloeity Potan. and tial Expansions Higher for OrderSurface Singularity Methods Report HOC J691L (1975) 98 27. tc Triplet - A NewAerodynamic Panel Siogne. isrity withDirectional Properties. AIMA Journal, Vol. 18, No. 2. pp. 138-14i (1980). Sytsma, H.A.,* Hewitt, B.L . Rubbert, E.: A P comparison Panelmethods of for Subsonic Flo- Computation. ACARDograph 241 No. (1979).

2.

Oskam. Asymptotic B., Convergence Higher. of


OrderAccurate PanelMethods. of Air. J. craft, Vol.23, Ho.2,pp. 126-130 (1986).

13.

29.

14.

Hunt,B.. Hewitt. B.L.;The Indirect Bound. ury-Integral Formulation for Elliptic, Hy. perbolic and Non-Linear FluidPlows. Ch. 8 of 'Development in Boundary Element Moth. ods'. Vol. 4, Elsevier Applied Science Publishers (1086). Hooijuakers, H.WM.: Aspects of Second. and Third-Order Panel Methods Demonstrated for the Two-dimensional Flat Plate Problem. NLR M 81074 U (1981).

30.

15.

16

31. Margeson, R.J. Kjelgaard, S.O..Sellers 111.W.L..MorrisJr. *Chi.E.K.,Walkley. K.B..Shields. Eli.:Subsonic PanelMethods. A Comparison Several of Production Codes. AIM Paper85-0280 (1985). 32 Foroasior, DiEspiney, . Prediction L., P de. Dirive., Stabilit; do pour lee Missiles avon I!' Code do Singularites 81115s. Rech. LA Aerosp. No. 1989-4. pp. 33-47(1989). SlOoff. U. Requirements Developments J and Shaping NestGeneration Integral a of MothOda. PaperIMA Coof.on Hum.Meth.Aaron Fl. Dyn.,Reading (1981). NP 81007U. NIX1 Schippors, On the Evaluation Aerody. H.: of oasisInfluence Coefficients. 'Panel In Methods FluidMechanics in withEmphasis on Aorodyoamice. ed. J. Ballmann aI , Notes at on Numerical FluidMechanics. Vol. 21. Viewsg Verlag. pp. 210.219 (l9B7) Clark,R W , Valaraco, 0 Subsonic U Calcu. lotion Propeller/Vinp of Interference AIM Paper90-0031 (1990) Hosijmahkrs. W 8 An Approximoto H Method for Computing Inviscid Vortex Wiake roll-up. NRT 19U(95 Hosijoakers,11.M.An Approximate H Method far Computing Flow aboutSlender the Config. usttons withVsrtoxFlow Separatison. MIR TB 86011U (1916). Hoaijaaksre. U N. Computational N Aerody. namics Ocdered of Vortex Flow 81IX 88088 TB U (1989) van Book.C M. Pier., U.J . Slooff. U J Bondary Integral Methodfor the Computation of 0Pot.ntal Flowabout Ship Coof igurations with Lift and FreeSurface Effects NIXR TB 85142 U (1985) Piers, J , Slooff. J U Calculation of U Transonic Floeby Meansof a Shock-Capturing FieldPanelMethod AIM Paper79-1459 (1979)

17

33

18

34

19

35

20

36

21 22

)1

21

24

39

25

40

5-47

41.

Oska, B.: Transonic PanelMethod for tta Appliedto MultiFull Potential Equation AIA Journal, Vol. 23, Airfoils. component (1985). No 9, pp. 1327-1334
Kandil, O.A., Yates. E.C.: Transonic Vortex

44.

Sinclair. M.: A Three-Dimensional P Field. of Methodfor the Calculation TranIntegral Configurations sonicFlowon Complex Results. Aaron. J. Theoryand Preliminary
neuy19,p June/July 1988. pp 2-2. 235-241. LabruJire, Th.E.: MAD, a System for Computer of AidedAnalysis and Design Multi-Element Airfoils. TR 83136L (1983). MIR

42.

Flowpast Deltawings:Integrals Equation Approach AIAAJournal. Vol. 24, No. 11, pp. 1729-1736 (1986). 43. Ericksson, L.L.,Strande, S.M.:A Theoretical Basisfor Extending Surface-Paneling Methods Transonic to flow AIAAJournal, Vol.23. No. 12 (1985). 1860-1867. pp. See alsoAIA Paper87-0034(1987).

45.

46

Brandsa, F.J.,Fray.J.H.J.: System A for Transonic Wing Design withCeoetric Constraints Basedon an Inverse Method. ACARDCP.463, Paper7 (1989). also NL TP See 89179.

6-1

HIGH ANGLE OF ATTACK - AERODYNAMICS


John Lamar E
NASA-Langley Research Center Hampton, 23665-5225, VA. USA

SUMMARY
Theability to predict high angle-of.attack. nonlinear, aerodynatnic characteristics flight vehicles. of including aircraft, has made sigificant progress the last 25 years.usinga variety in of computat-,nal tools andinsightful analyses The key technologica, element which hasmadetheseanalyses possible is the ability to account the influence of the shedvortical for flow, prevalent this angle-of-attack range, geometrtes In on of interest,Using selected analysis techniques, applications havealsobeen madeto wing designinorder to improve include a These performance. maneuver their high-speed complete obtlained modifying the entire cambered wing by wing. anda wing whose modiflcation werefocused the on leading- trailing-edge flap regions. and with associated different levels of Various techniques. accuracy. txistmodelthis vortical flow influence Te to onesincluded inthis paper cover: suction-ana!ogy with free-vortex-fllaments. free-votex-shet modeling, extensions, and EulerandNavier-Stokes solutions Associated relevant features vortices arealsoaddressed, of including; thewing andflow conditions which cause vortex fornation. andhow the vortexstrength varieswith angleof attackandwing sweep Onc: this nonlinear vortical is present, stability of flow the the aircraft canchange rapidly with increases tn angleof at. tack. These changes needto be known earlyin the design process sincethe forces/moments being genrated canbe be)ond the ability of conventional controls to handlethem. Dipending on whether flow is still organired. the engineertog methods be ableto estimate aerodynamic may the offecls In general, whenthe flow beco disorganized or asymmetrical timedependent, best and the engineering tech. niqos areexpenmental. Furthermnoe, aircraft controllability may need befoundIonovel aefodynamic to devices or cngine-thrusl vectoriog. This is especially true at angles of attackbeyond which the onset large-scale of vortex bsrsting occurs over aircraft, Workin this is highlighted the area

CLL CLo CL. C,

longitudinal lift coefficient.

(OL/Ox)/qb
lift coefficient 0* at a lift coefficient curve slope rolling moment coefficient, rolling momciqS damping-in-roll parameter, OC1/8(pb/2U), perradian rolling moment to sideslip, due perdeg pitching moment coefficient. pitching moment/qS saled nonlinear pontio of pitching moment coefficient, C,,/(Asinocoso) normal force coefficient. normal fore/qS scaled nonoal forcecoefficient. CN/(Asin aceo) scaled nonlnar pottion of normal o) n forcecoefficient, Cp /(Asin o yawing moment coefficient. yawing moment/qS6 yawing moment to sideslip. due per deg pressure coefficiet.(p - Pgs)/q leading-edge suction forcecoefficient. 2 sucton-forceNS; (1/2)Kv.i asn o leading-edge thrustcoefficient. thauWqS sectionnormal force coefficient sectionprofile nose dragcoefficient section suction force/q sectionthrust coefficient chord reference chord cha-ateristic lengthin augnened vortex fctor chordat imgmidspan lift contribution from augmented vortex floW

1
C, C

C
CN C. Ci. C. C" CP C$ C C. CR c' c, c I

cO ACt

aC,
AC,.,

LIST OF" SYMBOLS


A
a ratio of wing aspect wherethe fractional chordlocation from constant changes chord loading valhe to linear varytng valuetoward
zero at traihnl edge

Ax dF,

P
P
9

coefficient. CP.1-C,,5 lifting pressure surface pressure in change upper coefficient from a - 0' distance alongtip chordto ccnmroid vortex lift of sidk-edge differential oige-soction-force unit edge vector forcevector n fotceerto
normal accelertion

& CA CD CL

aspan

Ix. Iz K

moments innia about X of the


andJ body axes. respctiiely Z

axial forcecoefficient. axial force/qS dragcoefficient. drag/qS lift coefficient. liftyqS

S)hv simlanty parameturs /etan(, i tan also.thousands fret alttude, and of

6-2 unsteadiness parameter defined by. Kp Kovta potential factor lift Ieding-edge-vortex-lift factor side-edge-vortex-lift fddctor nugmenited-vottex-tift factor reduced frequency; flc0/2V lift-to-drag ratio dittance along leading edge fromapex(see fig. 24); also,inboarddistance vortex to Subscipts. a 4n BD c ep d dyn I attainable avetage -bieditdowi erossfiow centerof pressure design dynamic designvarable index

K
Rd,,.

k LID

core fromlending edge. inches (seefig. 41) free stream Maclhnumber component Machnumber 2 to 2 of 2 normal 1 wing LE. McosA( + sin cta t& A) / 2 2 M (Akf -1)/ cOtA p staticpressure: roll rate.rudiec alto, q freesteam dymnic presunre fi. Oh velocity comnpooentindexnotation in R. Reynolds number r ttreamwise distanice from leading edgeto vorte acien point, identified with the rotatedsnctionforce;alto leading-edge radiannormali theedge to S wing reference area: also, leading-edge suction T leading-edge thrust f time: thickness U,V free stream velocity V. vertical velocity n'v perturbation backwash sidewasis. and respectively tvwi sum, induced of downuash andUn atan-Irad X, Y,Z localbody axess)stem(see fig 37) Xf, Y1,Z1 smienflap coordinate centered axes at t11V apexOfthe nlap(see fig. 55) X sector design of variables z streandose distance froesihe centroid of the areagiving augmented vontes lift to the reference poiot u/c fractional distance alongthe localchord of thecalledout surface x/c, fractional distance alongihe rootchord from the actualor theoireiic4.i apen 2/c, fractional distance from local leading edge in termsof chsord y - 0. at ru(l) s-coordit of vonen breakdown 4./c, fractional distuace along the tood cbord to the center pressure thenoot, of for liear portmi of tisenortmal force t,/(b/2) fraction of wingisentispan n/a fraction of local senruspari Y., z sealedl lateral tenitcal locations and of vortex core. respecively, - ttv/(f/2).,4/(b/2) I critical distance vortexcore to abuse uppersurfac. Inches. also, vtnical distance called out iurcdonde to over total chord wing camberelevation a/c

MA M.n MN

leadingedge lower tnrface maximum N normsal onset associated at at which significant with vortexshedding beings p potential flow contribution r root;alto residual a separation se tide edge t tip: alsotheoretical TE, Ic trailing edge lot total u upper surface 0 vortex Oine vcu Vie vortexflow contribution from the leading edge seeC vonex flow contribution fromthe sideedge to wing zz, Ytj zu second partialderivative with respect to xyz reipectively 0 vaue at C1, 0 1 first orderexpansiron perturbation in quantities 2 second orderespansion perturbation in quantities on freestreami iime derioative I utx Greelk Symbata a act_ ad am 0 anglte attack.deg of angleof attackat which CL~, occurs. deg wimgincidensce fuselage. on dog angleof asnack nortmal wing LE,. to tdu-'tnano/eosA). deg 2 angle sideslip, also.(1 - A1 )1/2 of deg, circulation or. esquivalent circulation associateduih leading-edge suction sealed circulation dueto LE nonescore. ['/buhia ) distributed biouvnd vorticit) ata point distributed trailing vonticttyat a point atlerondeflection angle. posrtis trailing edge do. n. deg soitex fiap deflection. poasitive leadingedge do n.de

LE, le

r
['

6A. Or 6. dLE

6-3 6L,
6
brE

69

C
17 A A A p 0. fl W

deflection, leading-edge flap streamwise positive LE down (inboard/outboard), deg rudder deflection. deg deflection, trailing-edge flap strearnwise deg positive TE down (inboard/outboard), deg tip rake angle, wing apexhalf angle,deg vorticiy vector surface

s1/b
dug angle, sweep leading-edge wing taperratio, ct/c,; also.2ndcoefficient of viscosity t i cone half angle, t (l/M), deg: Macth also. Ist coefficient of viscosity density of fluid velocity potential perturbation deg: also. sweep angle, trailing edge circular frequency cyclesec oscillation frequency,

wing configurations: as usedprimarily in thegeneralrespectively. the last aviationisport o 'transportindustries, as are two segments treated one.For configurations that sweep of aredelta-like with higher values leading-edge may the andrelatively sharpleading-edges, low-a range seg-eita be extremely small leaving only thelatter three a In to be of consequence. addition to establishing which segments specifically involved, the configuration also are flow, which forms in whether the separated determines as themoderate range,will ever develop into a significant. wake-likeat the or vortical-flow structure just become of higher values a is of Though the emphasis this paper high a. the flow and aroundconfigurations is the moderate higher a ranges part,because This alsobe considered. is done,in large will analytical to s usefulat high a have some the same of is a. application at moderate Th. higher a range most and it is of increasing frequently called thepost-stall range to interest orderto respond two aeronautical in research spins unrecoverable community need. The first is to prevent is from developing on aircraft. andthe second to enhance of theoperational effectiveness fighter aircraft, asdepicted project reported in the joint US.A.-Oerman X-31 research or byDeMeis (ref. I). Experimenal procedures techniques. are as which amalsoclassed engineering methods, currently in of the bestmeans obtaining the aircraft characteristies the and or due higher a range to the unorganized asymmemcal well to mructure of theflowfield not lending itself unsteady to tathemotical modeling. With respect the low-a range. and of fc. methods use in the analysis design engineenug in by are asreraft covered the other papers this AGARD (ref. 2). special course the which address: is This paper divided Into chapters prediction of vortical, separated flow, stability andcontrol flight characteistics. and in the high-c range, post-stall. on Thework presmed hereis mostly focused engineenng methods for predicting the aroyiarmie forces and rionsents. however. aspect this course, of with the analysis which deals of the som of the maenal addresses desig aspect the is course Thethrut of this paper on fighter configuttions. resultsof is to reader referred the published The interested over and lectures specialists meetings AGARD sponsored (ref. 4), (ref. 3) and 1983 in the years, particular in 1982 on infonmation high for additional details andsupplementary angle-of-attack aerodynamics

Abbreviations: DesignSynthesis Automated ADS code A thin layer Navler-Stokes CFL3D codefor FLO57GWB An Eulerequation wing.bodies generalized Free-Vortex-Shect FVS Edge. Trailing Edge Leading LE. TE VortexFlap Leading.Edge LEVF NF Normal Force Lattice Method Quasi-Vortex QVLM Suction Analogy SA SideEdge SE 2-D. 3-D Two-dimensional. three-dimensional Three-dimentional Euler/Navier-Stokes TEAM Aerodynamic Method Thin Layer Naviet-Stokes TLNS VortexLift VL VLM-SA Votex Lattice Method coupled with Suction Analogy VORCAM VORtex lift of CAMbered ings

INTRODUCTION methods usedfor on focuses engineering Sincethis paper angle-of-attack (a) arodynamucs, it as important to high distinguish. first of all. exactly whatis meantbyhigh a. Todo this. the a rangefrom 0* to 90 will bedivided of are the into four segments, bound% which determined as flow present, shown in figure I These bythe dominant flow attached dominates. are a segments low. where and there is a combhnation of attached moderate, where high, where separated separated or voeical flow present, the and or von al flow dominates, higher, where flow due lss becomes wellstractured to vortex breakdown ending a's massIV,Aall Thebeginning and (bursting) or on arc for a par iclar segment dependent the Mach number andthe aircraft gonmety which generates neighhoring the or unuwcpt swept flow-field Forconventonal round-edged

PREDICTION OF VORTICAL-SEPARATED FLOW contains a review of high-a vortical flows, then This chapter for presents analytical methods estimating the acrodynamac compafeffects of this flowfield - along with representative dsign opportunilies isons. andendswith some

6-4

VORTICAL FLOW REVIEW It is important to have a clear understanding of how vortical flows are manifested in aeronautical applications.,This section reviews some of the relevant background the reader may need. It is organized into three parts: the first discusses the pertinent local conditions necessary for vortical flow onset and formation; the second examines those factors affecting vortex growth; and the third does the same for vortex diminishment.

This lecture, however, will primarily focus on leading-edge vortices. Figure 4. taken from MeMillin et at. (ref. 7), shows regions where classical leading-edge vorical flow is to be expected in terms of the a nonmal to the leading edge (aN) and the resulting Mach number normal to the edge (MN) [This figure is the latest version of the information first quantified by Stanbrook and Squire in reference 8. These authors found it convenient to correlate the leeside, delta. wing flowfield with the quantities a and MN. the pimary ndepen&nt variables associated with 2-D flow.] in-"2h all

Vortex Onet and Formation Vorticily generation, which is simulated in inviscid flow solutions through the imposition of the TE Kutta conchton, is due in fact to the action of viscosity at the F. In addition. vorticity is introduced into an othrwi e invslid flow due to either the action of fluid viscosity along a solid boundary or behind a curve shock (see e g. Anderson, ref. S). with the focus of this paper being on the former. There the vonicity is contained within an attached-flow boundary layer and may lead to no other aerodynamic consequence than viscous airfoil. or wing-drig. If the airfoil boundary layer separatesnear the leading edge and then reattaches to form a recircutlation region, this is called a bubble separation, On a swept-wing, a bubble separatirn often leada to the formtion of a coherent, leading-edge vortex-system, due to the falling pressuresfrom toot-to-tip associated slt, voticity entrainmem and increased axial flow. The bubble vorticilty is now confined withir a small region called the core. which grows in size and vortex strength from apex to wing lip Core growth comes about due to the addition of shedvortacity.associated with the flow sansf, .g the Kutta conditon - a viscous sornteul - at the lesdIng edge. being introduced into the vonex system along a helical path (Sketches associated with these descripions are given in figure 2.) This general desenpiron allows for vortex ontr and formation to occur at a small a. but the exact manner in which it happens Is dependent on the wing camber. thickness, leading-edge radius. Mach sumber and planform. The plonform effect car be to powerful as to lead to novel flow situations Conssder figure 3. taken from Cunntngham (ref 6). where at oderate a this 65' swepihack trapezoidal wing has two leading edge vortex systems which merge into a single one ata slightly higher a happen at higher sweepback angles Those wings which do not generate a leading-edge sd vortex due to their plaufotm shape being rectangula or too low in s%eepstill develop a vortex system at the rip one way to descrihe the fonmstion of this system is to consider i to be produced by the flow moving from the pressure side 4 the wing to the suction side This well known tip.fiow causes the span loading to change from a 2-D rectangular type to that approxiumatingan ellipse ad provides the theoretical base for the traiing sortvs whith get wrapped up into the tip vortex Viscosityplays a role in the tip ,onex initiation and in the determination of the point along the tip at which the vortex actually sepaates from the wing This getnraliy does not

this data was obtained at supersonic free streams, III gure is heunstically Important is that it illustrates how the leeward flow changeswith increasing small to large values.
0

N (or a for fixed A) from

Vortex Growth This par examines the subject of vortex growth by focusing on two of its component pans.vortex strength and core location.

Ystexmgnglb It is well known that the strength of the


vortex system from a delta wing is a and sweep dependent. However. the manner in which they are related has only recently been established theoretically by Hlensch and Lucknng (ref. 9) when they used an analysis based on the Sychev sinalanty parareter. K(- tana/tante). [These authors are not the first to show some relationship between vorx strength and a parameter, for example, Smith (ref 10) used the parameter (a tan) 1 The Sychev parmerter is associated with slender bodies at irviscid hypersonic speeds but Ilemich (ref. 11) shows it not to be limited to that situation In particular. K has application to even moderately slender configueatiotn developing vortical flow at low spelds The result is that the vortex strength present at a delta-wing trailing-edge can be related to a and the .ing sweep by

r/U

C(tz.()03n1t o/c6 cx c..(sln'


2 r 2

0 2

a or

a/os o a)/(tanA)O

This . uotiuon shows fo a fixed sweep and cr tst 4

rJU

Increases rooloratly with o. which is an expected result It also shows for a fixed a and c, that ['AJis reduced with moreasing wing sweep, a result which may seem contradictory to expenence and therefore be unexpected The apparent contradiction is due to our intuitive understand. ing that in the ooderate to high a range, a 75* delta wing develops are %ontx lift than one with a sweep of 45* We associate the increase in lift with an Increase ii vortex strength, llowetr., the reason the 45* delta develops less vortex lift is not due to a loss in strength but to a loss in vortex cohence or stability kThis topic is coveredin the vortex dmiurment part)

6-5 Figure 5,'taken from Heinseh (ref. II). shows the importane of K as abasis of analysis for vortex streugth. In that solutions from the Free-Votex-Sheet (FVS) code for thre delta wings, each at K i 1, yield essentially the same nonthtensional value of vortex strength at the trailing edge and similar growths along the chordl.This can be thebasis of an engineering method, V:llgx Cre bcuion Changes in a and wing sweep affect not only vortex strength but the lateral vertical locaand 'in of the core. Combining a and sweep according to the parameter K. Herach (ref. I1) al" determined that engi. neering estimates could be made for the core location as well. Figure 5 also shows thtat K = I the vortex cores for thesethree deltas increase with distnce along the in a very similar manner. Two experimental examples of .ortex core/system growth areshown in figures 6 and 7. These figures (taken from LatMare al. (ref. 12) and Lansar and Johnson (ref. 13). respectively) illustrate the o effect for two aircraft, one U.Sv sod one Soviet, as determined from in-flight vapor, scren image& fNote that the vapor screen Images for the U.S.A. P-106B aircraft have been digitally enhanced after the 01ght Vortex Diminishment Thetopic of vortex diminshmest is larger than just the loss of vonex-sysem cohernce. It also Includes the loss-ofinfluence a coherent vorex system his on surface pressures. Each is discussed. Lns-of-,,or~cx =: 'The losn-of-votex.colerence is due to vonex-core breakdown or burst, which has as Its main contributors (1) 0 effects on core size or swirl angle, aMd (2) adverse pressure fielddisturbances. Whther the bent is of theswiulor bub,,letype - see Lambourne AndBrier (ref 14)- is not of concern here. bet when it does occur. the flow in thatregion becomes, unsteady and begins to rotate like s solid body with a larger radius than that of the corehereas, the flow ahead of the region may be steady and coherent, As bent begins to occur ahead of the trailing edge. Itcan lead to asymietrial flow situations which result in an mbWance in the Aerodynamc forces on the left andeight sides. especially lateral ones %brtexbreakdown occurs with hysteresis over a wing durng Apitching motion with a resulting lag. This is exaned later with respect to dynamic stall, The ablity to estimate under what conditions bnst will oc. cur for s configuraion of interest and how to conrol the retalting flow or aircraft are of particular interest to the designer working in thehigh a regime. Much of Outnill must be determined expenmentally Figure 8, developed frontthe basic data of Wentz and rIlman (ref 15). shows the exper. imental vanasn of the ckfor vortex breakdown at both the trailing edge and apexover a large. delta.wsng.sweep range As expected, aD-7Z < D0L , for a given wing. how. ..A ever. what is revealing is how vortex b eahdw at the tril. ing edge or forward affects the maximum lifting capability of the wing, as denoted by ac.m. In puticular, note that for A > 700 CL,., occurs at an a very near that for BD-TE. This shows the aerodynamic importance of vortex coherence. lowever, there is one analytically detenined piece of information, shown in figure 9 - taken from Lamar (ref. 16), thatmy be useful here, It is that the leading-edge suction dtstribution across the span for both delta- and mapped-delta wings have their peakvalue increase and oce r farther outboard with increasing sweep or tip chord. By itself this doesn't help. but if a corelation is made with the quantity aD - the a at which themeasured C, ist begins to fall below the sctio Analogy esmate a trend is noticeable. h is that aD increases when the s peak suctitonvalue increases and occurs more outboard. i e. becomes increasingly triangular. This correlation can he used as s tooi in trying to estinste which of several configurations will have the highest *D by simply examiing the respective suction dttrihutonts. iThe quantity crD it slmilar to aBD.TR except that it is applicable to wings for which a dama may not be available, and moreover it Implicitly takes into account the lost-of- intluence associted with vortex displacement.]

influence the surface flowis related to its strength and the distance to the surface; hence, the greater the distance the less influence the system has. From expenments. it is well known that the vortex-system vertical displacement (see fig. 7) and strength grow for slender wings over an a range, and continue to produce a strong influence on the surface However, after some a. and perhaps before loss of coherence, the vortexsystem is too far from the wing surface due either to a symnmessdcal asymmetrical displacement or - depending on the wing sweep and flight attitude and as indicated in the sketches in figure 10- and thereby loses its str ng influence.(Asymmetmcal displacement is often assoctated with vortex crowding on a very slender wing at higher Aiphas, shown in the right sketch, but it can also as occur for most any delta wing at sideslip) This lots-of-influence effect is apparet on a wing surface from either an oil-flow or a ireside. surface-pressure test. The effect is conspicuous from an oil-flow test by the patterns beconing mor spread out and not as sharp in surface detis. and from a pressure test by a reduction in suetion pressure a sore positiwe value or toJust maintain. to tag a coCUtanL with increasing a. Regardless of how value these changes in local surfaceconditions are detected, the influence of the changing coMiimons produce correspond. tag global ones. Thes include either a reduction in lift (in particular, for a fixed at a value >200 and with A tncrea. tag above 76*. as indicated by the basic am of Wentz and Kohiman. ref. 15)or nonzero values being developed for the acrodynuic lateral charactenstics. even beforc vortex burt occurs

L LtNWLa=.:: The abilty of the vortex system to

66
FOR USE IN ANALYSIS General' Various techniques, associated with differeni levels of accu. mcompl~hy~r~enera available to model the vortical flow are racand complexity, metants high a. The one included herein cover, suctionto free-vortex-filamnents. free-sheetwith analogy extensions, vortex modeling, and Euler aid Navier-Stokes solutions. two The latter ate known as Computational Fluid Dynamconsidered engineering sea (CFD)techniques and are not yet methods, but may hosoon. In order to demonstrate these techniques, at leastone comparison with experimental data is presented for each of them. All the techniques just listed are discussed in this section and are done so basically in the order of increasing complexity. Suction Analogy (SA) The leading-edge suction analogy of Poihamus (ref. 17) has was originally developed for delta wings and become known as one of the nore reliable engineering tools for estimating the aerodynamic affects of separated vortical flow at subsonic andsupersonic speeds Futhermore. it is also one of the most widely implemented techniques because the information it ieeds can bereadily obtained from analytical orcomputational solutions to the linearized potentil-flow equations Ibis section reviews the basic suction.analogy concept, cites representative methods in which it employed, details stmilanty use of it. and presents is extensions of it P= Polhmus (ref 17)postulated that the in-plane leading-edge suction [aforce) developed in attached flow was not lost when the flow separated around a sharp leadingedge of a swept wing, instead the force only became reorisystem This is ented the direction of the rotating vortex in called the leaing-edge sucton analogy (LESAor just SA). m that the extra normal developed as a result of the force shed vortex system is analogously just cqul to the in.plare force lost along each edge Flow sketches depictig this change are seen atthe right infigure taken from referII. ence 16 This figure also shows an application for a A - 75'delta wing at At = 0 TheSA allows one to use potential flow codes to compute the CN, and C S . which are then used potential in the lift and drag equations In particular, the flow pai of the lift curve is identified as Involvtg a factor called K, This factor is nothing more than the low as value of CL,, or CN. The vortex lift portion coies by knowing the factor Kj, which is deeresned by taking the 0(2 one-edge CS)/8(siaho) Figure 12. taken from reference 18. shows the Kp and K, k vanation for delta wings Sinular curves for arrowand diamonds wings are given by Pohamas inreference 17 Gt these factors are determined computationally or from curves, they are used in the formulas cted on the figure to obtaii total lift TIe sortex drag associated with a ,iog the ACD = CL tano. Note the having no edge force is just lift good correlation obtained with th Zero drag rm ioved. Polhanmus (ref, 17) also showed how using the Prandtlteansformations one could obtain other subsonic Glauert esu outindthteeadrm results ltro Polhanus in references 17 and IS demonstrated that the lift of SA was capable maisng and drag estimates to other pointed wings thandelta, i.e. diamonds, arrows and even wings with cranked leading edges. Moreover, Polhamus showed tn reference 18that this concept is not restricted in speed regire but only to the development of a leading-edge force. Since analytic solutions exists for thin delta wings with subsonic leading edges at supersonic speeds, Kp and Kwje can bedeternasned and they are reported by Polhamus to be.

I K

Kp

t=xA/(2E(k)),

where E(k) is the complete elliptic integral of the second / l, maid and k - (I - (OctA)h]t and Kj. - r([16 (A,7) ][A + 16)tl2/16E2(k). ligue 13 shows two examples of neasured and estimated lift at supersonic speed for an A a 1.147 delta, also taken to from reference 16 There are twomajor points be made here (I) the reduced amount of ,orex lift available at M - 28. as compared with that at At - 1.2; and (2) the not lift is well estimated at M.t 1 2 but as well at At - 2.8 = The poorer agreement at thehigher Mach is due to the lower surface of the Mach cone coming near the leading edge with increasing a which effectively moses the lower suiface separation line closer to the leading edge ad reduces the amount of flow turning around the edge Consequently. less vortex flow and lift are available Representative micthods: On:y two of the many methods which employ the SA are cited in this section. They are the Analogy (VLMVortex-Lattice-Method coupled with Sauction Othets are introduced in the SA). and DIGITAL DATCOM. later sections which deal with eutensions or design. VLM-SA.The NASA VLM code developed by Marga. son and Lamar (ref 19) contained the ability to calculate leadiug-edge suction Since then the code has been continually up-graded to include the SA affects, as indicated by the of the repo by Lamar and Gloss (ref. 20). and tiile that feature Is cunently embedded in the most recent VLM code release, refered toas Lamar and Ilebate (ref. 21) An of taken example the VLM-SA code is given in figure 14. from Snyder and Lansr (ref. 22). inwhich it was used to an predict he lonagitudinal-load- disinbuttois -CLL - for A - 1.147(A - 741)d, at three salnes of a. This work was doneto addless an initial con ei that the success obtaud wih estimating lift and drag using the SA would not he repeated when trying to estimate pitching tniotoL, The concern was based on the nowledge that to obtain reliable etitsates of C,. it necessary to have valid predictions was
2 2

6-7 of the longitudinal distributions for both potential andvortex not flow, something required by theSA. and Snyder Latsar obtained the potential lift cu.'vesby at a Sperfoiring spanwis integration of lifting pressures the a variety of longitudinal locations; whereas vortex camedirectly from theleading-edge-suction lift portion show thevortex lift distibution. Theresulting curves a contribution to become larger fraction of the total with predicted-total results a increasing andthe measured-and on over to be in fairly good agreement the a range Based is of thesuccess this early work, theSA concept now used for estimates Cm.routinely to provide reasonable using this methodare given later. results Other engineering DIGITAL DATCOM: One well documented for developed theUSAF by Williams andVukelich method, thatof DIGITAL DATCOM. It is a computer(ref. 23), is based systemfor obtaining "static stability, high lift and over derivative characteristics" a control, and dynarmic and Mach numbers o's The of range ancraft geometries. are computational methods primarily linear aerodynamics with the nonlinear lift portions forsome combinations of for being accounted using the planforms andMachnumbers only speeds straight-tapered, SA.In particular, at subsonic speeds and low-aspect-ratio wings arecovered: at supersonic or a wings with a subsonic-leading-edge only straight-tapered with supersomc-leading-edge an attachcd-flow-shock at zeroa are covered. TheSA is alsousedto estirt.te forebody the speedsabove a for at lift and pitching moment subsonic "ost of noneslift", K. Similaity. The useof the similanty parameter as re. beenpreviously introduced portedby Hcmsch (ref. II). has Ilere theemphasis is on discussing a link betwen a simi. to lanty parantcer (tuna/A), which reduces K/4 for delia To bodies. aswings, andSA for wings andsmoothslender again theCLL distibutions presist in this effort, consider of in sented figure14. Sincethe magmtude andshape each on on dependent a. theremaybe a scaling disibution arce o which would make eacherve likethatfrom anther delta to wing at soie othera. Ie, use a similarity paramevter son resultstogether. properly scaled the Ilemsch presents longitudinal-nornal-force distribution by a quantity similar to CLL - scaled (Asinoos) t for a family of thin gothic wings at Af - 0 and095 fora valueof thesimilanty paranet. (tana/A) - 0.3. These from theFreedistinbutions havebeenestimated scaled later Vortex-Shot (FVS) code- to be discussed - aid show potential and the to closeagreement exist among respective force vonex normal that total curves. Thismeans the scaled the contbetion between vanous wings musthavebeet also Figure I showsthe impactof applying inclose agreement force of scaling to FVS estinmatesthe total normal this same coefficient and to only the vortex flow porion asa function of K/4 for thin delta wings Iiersch atibutes the rationale have researchers found for doing this to the fact that"several conical thatthe %elcisyfieldof thepnrnaiy ones is nearly distbotion is not pressure eventhoughthewing surface sinlarity may that This suggests theSychiv slender-body at leastbe applicable to the portion of the slender wing or by body loading induced the vortical flowfeld," The result of to this action isthatthetotal C doesnot collapse a single the line, whereas vortex portion. Cv, does. What is also is interesting thatthe theory shown for theC,, curve is a law power fit" given by "best one-term C, 3.07(tanaIA)g,

w which isvery close sothe Pothamut ressit. wrtten is similarity form.as C,, = K*(t-*I/A),

thani, whereK ,,0 is known to become only slightly greaser range. over this sweep distanc to the cester This fige alsoshowi the fractional force as of of pressure the nonlinear portion of the normal that this term"doesnot notes Hemsch of a function K/4. range verywell", which "meansthatthe aspect-ratio scale subsonic to over which onecan expect scaleor extrapolate smaller than it would data is censiderably pitching moment " flows be forsupersonic mg Regarding smoothslende bodies,the resuln C. and the C estimate aredependent on M and on whether body is an elliptical coneor denved from a powerlaw The general resultsarcthat. C. cc (tano/A - (tuna/A)oti . elliptical conefor %f < IM > 1. also powerlawbody for At < I s or(tana/A) law power body for M > I am C, oc(tana/A - (tan/A)a..u, for elliptical cone, if < 1. Y > I; also law powet body for M < I o (tno/A)F -ir4 law power body foeM > I. gfljgig The LESA has spjaned a vanety of extensions An eatly onewasthe estimation of thevorex flow effect for wings with finite sipcbons Another dealtwith estimating theeffect of a vosex s)stem on a doAnsu.am areanot with associated its generation it is called an augmented more Ohers have addressed generalgeometnes. vortex ternm. non-zcreedgeradius, which include wingswith camber, thickns cifects. combariss of a wing with a canard. For flow or strake body. and asymnsetcnal situations. of wingsdeveloping %ortex flow. the concept she cambered of tortex action point is hscussed because its ,mpotane are lany of theother listed items detailed in thefollowing paragraphs
J). r ( )

6-9 the trn AOL with known terms, the Kuna- Joukowski relationship was used for tre differential leading-edge force, d&P. shown. result is that another factor is identified, as The 1R, which can contribute to the vortex flow aerodynamics based on quantities already known. and which uses the same trigonometic functions in its' computations for the forcelirent characteristics. as K,t and K.,... The only quantity not specifically known is Band it is used to provide a representative length onto which the forward shed vortex system acts. This quantity is a geometric term defined to be the streamwise distance from the tip leading edge to the apex of the trailing edge. and can positive or be negative depending on tie tip-chord length and the trailingedge sweep. For pitching moment estimation the streamwise distance from the centroid of the "affected atrea"to the reference point is needed, a quantity called i Figure 25. taken from reference 16. shows the relative size of the vortex lift factori and that the augmented factor is too largefor these wings tobe ignored. The cropped delta configuration atthe top right is the same as shown infigure 23. but the results for a lower Mach number. the am Note improved CL agreement up to the highest test a. For the cropped diamend at the left. the augmented factor iv the some size as die other vortex lift faciors andleads to good agreement with data, to the highest test a. again The other twowings, having notched tuiling edges, will not have such large values for the I?, ,. Fuithennore. both the A - 1 069 and 1.917wings have C,. rmolts which show a ift falloff by 24'and 19'. respectively. Even for a coherent leadingedge vortex system. notching the trailing edge reduces the room onto which flow reautachment can occur which will lead to a reduction as measured C,. Figure 26. taken from reference 28. shows the definition of to yield generally agreement for a less than 16*.yet good it leads to an underpredilcton of the CL in the moderate o range for this 45* cropped arrow wing Inthatrange the chord itself better tepresents th, length onto which tip the forward shed vortex system acts These points ate illustrated by oil-flow sketches show'a moderate and high at a valus Above a a 16' one of three things happens to this vonex system which cause CL to fall off. They am (1) the leading-edge sotex system bursts the basic delta - on breakdownthe TE would occur at a much lower angle at than 16*. (2) system gets large that its influence is the so dimiushed due to vertical displacement, and (3) the loss of reattachment area The points being madehere ar that there isan additional vortex flow factor beyond tlose of KIu, and Kv,. it a called the augmented lift term. and when used with judgnt~n about what the leadng-edge vortex system is doing can lead to good acrodynantc estuates A good example of this isfor the strake.wmg cimbination depicted tn figure 27, taken from Lamar and Campbell (ref 29) Ilere one can sce the nambet and sze of the lift augu'entation that regions can vary with a to reflect the actual flow. Cranked wings and wing-canards are other ronfigurational examples which may employ this type of flowomodeling a-variation. as relteted in the construction of both the k4,. and Ki.u terms. Sample resultsgiven in thenext chapter. are Wing camber Vortex flows and theseeffects on cambered wings are of increasing importance dueto emphasis on vortex flow control devices, such as leading-edge vortex flaps (LEVI). Extensions have been madeto heVLMSA code to account for theseeffects (ref. 21). Other SA methods which have also been extended, including one which uses the Quasi-Vortex Lattice Method (QVLM) as a potential flow base, as described by Lan in reference 30 Another one developed by Lan and Chang(refs. 31 and 32) ascalled VORCAM (VORtex lift of CAMbered wings) and ts derived from an improved versionthe of chord-plane acrodynamic-panel n sthed Woodward of (ref. 33), This code uses SA to calculate the vortex tie induced aodynamic effects on cambered wings, including those with vortex flaps, and is valid at those subsonic and superionic speeds where the lineariszed governing equations apply. An application of the VORCAM code to conically cama bered delta wing as M = 1.4 shownin figure 28. The is inclusion of the orte contnbution is seen to provide some improvement with the measured data over this restincted a range. VORCAM also has an option fot designmg a portion of a contiguous wing surface to represent an integral vortex flapinsm into the wing. An example of using VGRCAM in LEVF design given later. is Vortex action point. Lainand Chang (ref. 31) have also modeled the effect o5the center the vortex, called the vorof tex action point, moving boardl downstrean with a. This is physically correct not accounted for in the ongisal and suction analogy modeling schemes. In these e.iler schers. as previously noted, the vortex was assumed to reman small andalong the leading edge regardless ofthe a value. The vortex action point movensent produces no aerodynanmc effects for a planar win. but for a camherAwsed wing there will be differences hey ate associated with the local meancamber slope vatysng along the chord, thereby cawing the local contrtu,iois to lift and drag, calculated from the suction force. to differ from earlier results, flis can be under. stood by examining the nghithand sketch in figure 29, taken from Iref. 31) This concept was derived by comparing measured vertical velocties present near the wing leading edge with those associated with potential flow The difference is attibuted to the action of the vortex system and its magnitude based on data for wing at one value of a and was determined one to be V../2. Fron this cosept the strewanwise model flow was developed shitch has the chaactensiict outlhe in the left and center sketches on thins figre. The basic assumption is thatthe exchange lincar momentum into and out of a of control surfac of length 2r would be jus balanced by the section lading- edge sucuon force Physically. it says that the force required to keepthe control surface from moving

6-9 the term ACL with known terms, the Kutta- Joukowski relationship was used for the differential leading-edge force. dF,. as shown. The resultis that another factor is identified, k, which can contribute to the vortex flow aerodynamics, based on quantities already known. and which uses the same trigonometric functions in its' computations for the foroe/momnt characteristic. as Keu and Kv,, The only quantity not specifically known is antdit is used to provie a representative length onto which the forward shed vortex system sets. This quanuty is a geometris term defined to be the strearwise distance from the tip leading edge to the apex of the trailing edge, and cn be positive or negative depending on t'ie tip-chord length and the trailingedge sweep. For pitching moment estimation the streamnwise distance from the centroid of the "affected area"to the reference point is needed, a quantity called i Figure 25. taken from reference 16. shows the relative size of the vortex lift factors and that the augmented factor is too largefor then wings to be ignored. m'e cropped delta configuration at the top right is the same as shown in figure 23. but the results are for a lower Mach noumber. Note the improved C1, agreement up to the highest test a. For the cropped haroond at the left, the augmented factor it the sorte size as die other vortex lift facrors and leads to good agretment with data, againto the highest test a. Theother two wings, having notched tc,hng edges, will not have such large values for the k,. Furthernore. both the A - 1069 and 1.917 wings have C, results which show a lift falloff by 24' and 190.respectively. Even fo a coherent leadng.edge vortex system, notching the trailing edge reduces the room onto which flow reattachment can occur which will lead to a reduction in measured CL. Figure 26. taken from reference 28. shows the definition of Z to yield generally good agreeent for a less than 16, yet it leads to an underprediction of theCL sathe moderate osrange for this 45' cropped arow wing In that range the tip chord itself better representi the length oto which the forward shed vortex system acts. mese points are illustrated by oil-flow sketches thosn at moderate nd high is values Above a - 16' one of three things happens to this vortex system which cause C1 to fall off Theyare. (1) the leading-edge vortex system bursts - on the basic delta breakdown at the TE would occur at a much lower angle than IVl: (2) the system gets so large that its mfluence is diminished due to vertical displacement: and (3) the loss of reattachmnt urea. The points being made Iere am that there is an additional vortex flow factor beyond those of K,. and K., it a called the augmented lift term. and when used with judgment about what the leading-odge vorses system is doing can lead to good acrodynamtc estimates. A good example of this is for the strAe-wing cimbnaton depicted in figure 27. taken from Lamar and Campbell (ref 29). Hereone can see that the number and size of the lift auncnitation regions can vary with a to reflectthe atual flow. Crankedwings and wng-canards ate other configurational examples which may employ this type of flow-mdeling er-variation, as reflected in the construction of both the k.,rr and K,rr terms. Sample results are given in the next chapter. Wing camber: Vortex flows and their effects on cambered wings ace of increasing importance due to emphasis on vortex flow control devices, such as leading-edge vortex Raps(LEVF). Extensions have been mae to the VLMSA ce to acout for thes effects (ref. 21). Other SA methods which have also been extended. including one which uses the Quasi-Vortex Lattice Method (QVLM) as a potential flow hase, as described by Lan in reference 30 Another one developed by Lan and Chang (refs. 31 and 32) is called VORCAM (VORtex lift of CAMbered wings) and is derived from an improved version of the chord-plane aerodynatsmc-panel n nhod of Woodward (ref. 33). This code uses tio SA to calculate the vortex induced aerodynarmc effects on cambered wings. including those with vortex flaps, and is valid at those subsonic and supersonic speeds where the lineanzed governing equations apply. An application of the VORCAM code to a conically eainbrd delta wing at M = 1.4 is shown in figure 28. mhe inclusion of the ,orte. contnbution is ten to provide some improvement with the measured data over this restricted is rMnge. VORCAM also has an op-ion for designmg a portion of s contguous wing surface to represent an integral vortex flap in=x into the wing. An example of using VORCAM in LEVF design is given later. Vortex action point. Lanand Chang (ref. 31) have alto modeled the effect of the center of the vortex, called the vorrexaction point, moving inboard/ downstream with a. This is physically correct and not accounted for in the original suction analogy modeling schemes. In these earher schemes. as previously noted, thevortexwas asuned to remain small and along the leading edge regardless of the o value, mie vortex action point movement produces no aerodynanuc effrels for a planar win', but for a cambertAwisted wing there will be differences Theyare associated with the local meancamber slope varying along the chord, thereby cau~sing the local contrbuioss to lift and drag. calculated fromthe suction force, to differ from earlier results. lus can be under. stood by examlomog righthand sketch us figure 29. taken the from (ref 31) This concept was derived by comparing mcasurod vertical velonties present near the wing leading edge with those associated with potential flow,.me difference is atributed to the action of the vortex system and its magnitude based on data for one wing at one vaeueof a and was detenmined to be V./2. From tus concept the streamnwise flow iodel was developed which has the characteistics outlined in she left and center sketches on this figure, The basic assumption is thatthe exchange of Imear momentum into and out of a control surface of length 2r would be just balanced by the section leading- edge suction force. Ph)sically. it says that the force required to keep the coitrol surface from moving

6-10 exchange. awayfrom thewing, dueto the linear momentum by is just balanced an additional forceon the wing These by prewnt above forcesareboth caused a suctionpressure to thewing andleads r = c~e. Thesketch at the right shows of forcedueto themovement the reorientation of thesuction the vortex actionpoint. wing, not LE Round simulation: For a round leading-edge into all of thesuctionforceis converted an additional lift as Some of it remains a "residual" whentheflow separates. edge. suction forcethatactsin theplaneof the leading wings in the late1960's Polhamus showed for uncambered later with permission byKulfan in ref. 34) that (published to the therewasexperimental evidence demonstrate sumof leading-edge suction andvortex normal forceto the residual value the as be essentially same the theore'mcal of thethin. wing leading-edge suction. Figure 30 showsoneexample prepared Polhamus by from existing data for an A = I delta edgeandt/c - 0.10.which wing .h a roundleading on tuppons thepreceding statementThesketch the left at illustrates the forcesinvolved, andtheequation the otton 35, relationship. Henderion. in reference showsthe algebraic experimentally thattherewasevidence later demonstrated number this tre to render statement over a wide Reynolds range Thetwo mainmethods of simulating theeffectof round as leading edges thisPolhanus concept a basisfor use they detme-ting the vortex force, however, differ widely is and proceduresKulfan the basicassumptions computational procedure detemitning the for (ref 34) uses a theoretical in the vonex normal force,thenemploys equation figure
30 to find the residual suction, wherras. Carlson and lack (rcf 36) and Catlson and Walkley (ref 37) uv emprcal formulas to dterrmne the resideal suction at supersoni Wnd subsonic speeds, then employ this equation to detefrTn vonex normal force the

to ts This procedure extended Mach andReynolds numbers. theory. wings byusing simple sweep the 38 Unsteady flow: Lanin reference hasextended SA to QVLM met.A (ref. unsteady flow bycouphngtheUnsteady 39) with theidea of "vortex lag." Vortexlagis associated the lag with the phase anglethatexistsbetween wing motion at and thebuildup of thevortex strength the leading edge lag Lan deterined the phase angleandwiththatwasable the suctionforce.andhence vonex the to estimate unsteady flow aerodynamics is Others The reader referredto Lan(ref 40) and to Lamar to (ref. 29) for additional extensions the SA. andCampbell Free-Vortex.Filaments the Free-vostex.filamnt models are usedto represent shed vortex off edge using discrete by vortex system the leading similar to that shown in filaments which interact a manner countries have re.earchers insmany in figure 33. Various and methods based this nodel for both steady on developed unsteady flow. "Ilsy include MehrotraandLin (ref. 41). (ref. 43) for PacandLan(ref 42) andKandil andYates flow: and KatzandMaskew(ref 44) for unsueady steady of flow. The method Kandil andYale- highlighted Wcausc of its transonic apphration ipproach anda shock. This method uses an integral equation the of to capturing technique establish features transonic flow aboe the wing andin the ortex systemThese and includeshocklocation andthe dete ination of its shape
influence One interesting result repotted ifor a thin, A - 1 5 delta at o - 151 and M . 0,7) is that the captuted shock is cured, attached to the vortex sheetand doesn't extend to the leeward wing surface but towards the core This is sen at the right of figure 34 along witi a favorable compaison of pred.icted and experimental pfessa at r/c, - 0 80

Kulfa's prcedure. developed in a Boeing research effort, in in depicted in figure 31 and ",tich has been imptenrritemed the VLM.SA and other codes It is based on the assumptions that (I) the atrfoil nose section is parabolic and (2) tepara. lton begins wherver the local value of theoretial leading, edge s-iction exceeds the parabolic note drag value, el The sonex and its associated ionsal force are assumed to occur %hen the local a excreds that required for local separation, oa. an a I sn(a - o.) l stl(a - o.) Carlson assumesthat the thrst at the leading edge is hmtied by the amount of pressure thit can be attaunedthere An xam;A.i of how the hmting p ssure can reduce the leading-edge thrust to its "attainable" value is shown on fig. ure 32 llere the huoting pressure is sacuam and its unpact car be clearly seen with increasing a or the theoretical sec. lion thrust coefficient. cs (Note, if the luling pressure were on, cq,/cA would be I 00 for all vius of cit ) By analyzing airfoil dlat. Carlson correlated the experrmental residual thrust with no nl Mach number, airfoil thicksrs

The left sideof the figure shows three views of the leadingand traiing-edge free-vostex lines on the wing along with the invoisd LE and TE cores This method defines the core to be the centrmd of the enoss.sectiotll ara and Air the filament has made one revolution it is temmnated and its vOricity added to that of the core The fact that there are two cores may weemunusual but they result from different vontety shees The leading-edge shele produces a Counter.lockvse rotation about its core - when looking upstream. whereas the uamttng-dge sheet produces a clockwIse rotalton about its core on the right wing pariel due to the span loading not decreasmg nsoionically to cero under the influence of leading-edge votical flow Tr- e ,t.sectiosal mushloom.like in ihe trailing. shape of the system appears Odge region. Free.-Vtee.-St" C The fre-vOelex-shet (WS) code. deal-

and leading-edge radius ratios, andblmiting presswe The proper values for limitig press$re were determuned empically from an airfol data base that covered a wide range of

tned by Johnson Ct al (ref 45), swtisfies the Laplace equatins by using logher-ond-r panel technology to represent the loading on the wing and differs from attach-d flovwmthods

6-11 byvirtue of themorecomplicated boundary conditions. The chief difference associated the freesheet is with having to stmultaneonusly satisfy both the no-load andno-flow boundary conditions, this renders subject problem non'.-iear. A the second difference is associated with the near-wake bou.diycondition. This condition needs besatisfied second to to or. der accuracy orderto obtain correct tn results Figure 35, takenfrom Luckeing et (ref. 46). showsthesefeatures al. on a representative wing Solutions havebeen obtained with this codefor a variety of configurations andare catalogued the paper Luckrng in by et (ref. 47). That paper al. addresses solution procedures to beemployed order toovercome in convergence difficulties 'encountered with morecomplex configurations. The partial restartprocedure oneof those is developed. second The of the FVS examples two shown hereusesthis procedure. fke examples, cited in order of presentation. an A - I are delta wing andtheP-106Bconfiguration. Avolication - I delta: Figure36 (Lucknng ctal.. to A ref. 47) provides a comparison between measured the and predicted curveandspanwise lift pressure drstnbtion results for an A - I deltawing. [Tbe measured results haveconi from an experimental study reported Hunimel in reference by 48.1 GoodCL agreement notedover the a range to is up 30 After this a the flow physicsbeginto change from that asnmmed theFVS methodto thatwhich encompasses by vortex besedown. reference shows sinlasly good The also agreement dragandpitching moment with over theC range up to I.I. Regarding spanwise the pressure distrbotions. the peak suction pressure overestimated thedistrihution tiwat is and the leading edge misseddueto theFVS having no mean of accounting for influence the secondary the of vortex. The ma)or measurable effects thesecondary aw. of ,onex twofold firstly. displacing theprmaryvortex upward, therebyreducing the influence theprimary. andsecondly. of increasing suction pressure the nea theleading edge, to due the presence Ihesecondary of cnahics Atitn hlcat F-106B. partial to The restartprocedure a starting vortex- geometty., hich hascubiter sheet been specified bythe user or previously obtained ona similar (wing, wsg-fuwlage. etc.) configuration, to be applied to the currentone.heretheFPI06B This procedure been has usedin a three-step process, outlined at thetop of figure 37, to obtain converged a solution (ref 12) Ike first step is to acquire a converged free-vortex-sbect solution for flat. * deltawing Secondly. thatsheet at. a is tached to the iamewing mounted onto a geneti cylindrical fuselage with a conic-like forebody. Lastly. theconverged ortex-shect solution from previous stepbecomes the the inital guess the final configuration. 11e final contigurafor lion models actualairplane making threegeometnical the by chang Theyinclude changing wing plarom from ow s me with no imlmg-edge sw"ep to onewith S' of forwand sweep (going from delta to a diamond), introducing a closeapproximation to the actualwng camber, incorporating a more and realistic fuselage model In particular, theactualforebody. canopy.faired-over inlet arewell represented the afterand body is closely approximated Steps andtwo gaveno evidence convergence one of difficulties, however, someweremanifested when very precise a modeling of the actualwing camber, which exists outboard was of 80% local semispan. attemptedTocircumvent this problem an alternate method modeling the cambered-wing of was employedThis method wasanalytical andmadethe leading-edge coordinates of thecambered-and flat-wing to be the same. The effectof this wasto causethewing to be placedonthe fuselage a negative at incidence (approximately 1.25') as shown the sketch the in at bottom of figure 37. Figure38 showsthesurface panelrepresentation the of wing-fuselage theconverged and free-vortex-sheet solution at a - 19*andsubsonic speeds. Also shown the flight is location of a vapor-screen sheetandwhere intersects light It theconverged fre;*vortex-sheet. ILight sheet oriented is II 2 ahead of perpendicular the fuselage to centerline, as documented LamarandJohnson by (ref. 13).JSinceforce andpressure cotuparisons not been have madenor publithed for configuration - duein part thesmall number this to of a solutions made- only the vortex location will be core reported that.subsequently. and EulerandNavier-Stokes General Currently. methods Vhich model eitherthe inviscid Euleror the viscous Navitr-Stokes equations under fall theclassification of "expertcodes".This is due in large partto thespecialatiention required developing an in acceptable grid.andthe knowledge required stipulate to suitable parameters the flow solver Illo ,eer. this to situation is improving by the introduction of CFD codes that have suitable documentation sample arid cases assist to the newuserin their application. e g. theTEAM codeof Raj ct al (ref. 49).) adition. dueto the lnge requirenments In of computer memory time.these and codesaremostoften run to eitherassist understanding in experimental resultsor to helpguidethecxpennentalist m areas whereunanticipated phenomenon maybe presentAfter the Eulerand NavierStokes equations presented tensorform -summation are m convention iniphed by epeatd idcx, example solutions are given. itshould notedthat eachof thesesets of be threeequatons cuitains no bodyforceandincludes 5 and 7 widependent variables, respecti ely. To obtain closure, the equations continuity, energy,perfect gas of total equalim-of. statefor static pressure, Stokes hypothesis bulk viscosity for andSuthetland's law formolecularviscosity arealsoused. asneeded ] EulerEqos O(pW)/0(t) + (Pq) + i,,), -0

6-12 Navier-Stokes Eqns This figure alsoshowspredicted spanwise pressure distnbtons at two longitudinal locations Primary vortex capture 0 s noted thecomputational solutions, though the peak in valuesfor suctionareover-estimated. is duein pas to the Ihs inviscid Eulerequations having no mechanism modeling for the secondary tertiary vortices generated the surface. and on The secondary vortex is themore influential of these and two itseffects have already been detailed FVS section in the Application toF-106B Thereason representative a vortex
system is expected for Nls wing is that its leading-edge ra-

f,

+ +(pq,) p6,), -Aq,,j, -p(qj, +q.,,) + , -

Eu.k.rC. ,Q&

Three-dimensional Elder codes, like the

FLO57GVB code developedRajandBrennan by (ref. 50), have demonstrt-Z only capbihiy capture not the to shock
waves at transonic and supersonic speeds but can capture eralized geometres

regionsrotational thi" lower of flow at and speeds gen. on


Unlike potential flow methods. such as

model the FVS.Euler codesdo not requirean explicit ofthe vortices Instead. they appear automatically asa partof the solution.reasons twofold' firstly, geometnra and The arc secondly, computational. The geometrcal issueis an easy to understand one in thatfor real flow expects separated tohe a one a flow generated thewing leading-edge issharp, at iflit regardless of theReynolds number, to thenaturalenfoeemen' due of the Knit condition there.Thus.if the geometry be to ndeled hast sharpleading-edge, cmialsoexpect one this condition tobe enforced compusationally bnngs This in the kampurational issue. namely,thatthe 'discretized Euler equations diffusive near theleading-edge, to are due iruocation error added artificial viscosity". accordig soPowell arid Murxman reference Since the in 51 Kutta condition oves its existence the effects viscosity ora to of diffusive effect,it is understandable these that authors note "any computational model hasa diffusive effect atthe that leading-edge -: ,frnuc realdiffusive t effectsshould tugger separation. regard.,s of the magnitude of the diffusion" Thus,a method employing the nvoiscid Euler equations is ableto predict, in a faitly reliable manner. leading, the edgevorex.separation effectsevenfor thevery complex geometries, provided wing hasa sharpleading-edge the (SeeMurmara al. 52) et ref Two solutions using Euler equations given. oneis the are fora 74 delta ing(A . 1 147)usingthe TEAM code in its Eulernrode.and the other for theF-106B aircraft basically a * dcltawing with cotc-like eamiber using an earlier serston the 1TAM codecalled n.OS7GWB of lit maybe surprising to thereader thatsomeCFD codes .an accommodate configurations having a high degree of geomtrical generality, nloes andChadcrjian (ref seealso 53) andGhaffan elal (ref 54)1 Strictly speaking, this codeshouldonly beapplied geometies in which thereis to reason expect a leadig-edge vortex-sy~uem beformed to to Reference 12 shows the P-106Bairctaft ad motdels that meetthis crtcna eventhough they do not havea sharp leadng-edge Application to A = 1 147delta wing Raj Itef 551present .omparisons. sinular to thatof figure39. betworn measured andpredicted andsurface lift prestures M - 03, where at the predicted values havebeen obtained with a II 0 grid of 30 x 37 x on ponts Thefilt agreement good to is up a = 30" Abovc this converged a vscdy-state solutions were not obtained

radius. ratio dius small: strefuwise to-chord reported is ic. to he lessthan 0 2% across span(ref. 13). Furtherthe more. a discetired representation a radiuswill yieldan of acute angleatthis or any other edge Th FLO57GWB code was alsoselected because, thetime work was being at the done,this codewas both readily available andfairly straightforward to usein assessing flowfield on a geometry the as complex as theF-1068. ThebasicF-106B wasanalyzed Pao (ref. 56) using a by C-H grid of 129x 25x 25 with points clustered aroundthe wing section.Thesurface grid representation F-106B of wing-fuselage configuration anda typical C-Il grid around twing stationare shown figure40 The conditions of in interest M . 0.4 anda . 19" Lift andsurface are pressure results presented figure 41. andvortex corelocation are in at thesame position as measure from flight vapor-screen itmages given in figure 42 is Regarding figure41. agreement the betwen n.asured and predicted curveslops is go'od lift over the range a from 10' to20 This isonly important within thecontext that the Machnumber t-' experiment 0 2 andfor for was the computatiral solution was 04. Theprydi ngleads to theconclusion that thereis no igwificn Mach number effectat this lowsp-eu At o'sabovs20" een the lift cuie slopes begin to differ significantly l is known thatvotex breakdown occursat the trailing edgefor a sharp-edged 60 deltawing at a's greater than 14' flat (Wentz and Kohlnsan. 15) Therefore. thu F-106B ref for configurarion difteerees at the Ligher the lift alphasma) be duetotheinabltty of the cste to model the flOA physitca accurately.tih the number points available Regarding of tle.coxpttatisital solution ata 30. it known thatthis is onedid not converge a steady to statevalue Themaor difference noted in the to 20 a range 10 is theactuallevel CL This differenme alsob m may attisbutable therelatively few number tonmuational to of pointsavailable to resolve configuration and flow field the Exainng the pressure portion this figure, it should data of he notedthat there no measured are pressure aaiable data for comparison thecomputed with values. The resultsat a - 19' showtheexpected dropin suction peak gong in from z/c. 051 to 89 The relatively high suction of 0 prssutes micall near to.miton icaourg cage is atnotutea us par toa asieling difficulty associated thesolution of wtth

6-13 and at the Euler god points available. Vnumber ofequations a leading edge in partto thesmall of a Figure 42 provides comparison the vortex coreresult previously. The and that from theFVS solution discussed from theflight test is theoneassocichosen vortex system the atedwith thevortex nearest leading edge,andit appears shownat the photograph right in the enhanced at theextreme on was The top right in this figure. selection based the fact at to thatonly the outervortex extended the leading-edge the inboard progressive station had thenecessary and tight-sheet a. (SeeMcGregor, ref. 57.) with movement increasing On the left sideof figure 42 the theoretical core locations ae from the FVS andEuler codes shown The corefor the FVS is at the endof the free-vortex- sheet andis indicated of by the triangle. Forthe Euler codea crosssection the flow-field velocity vectors just behind the light sheetplaneis on shown.Supenmposed it is the corelocation, indicated by contours. To from staticpresssre thesquare. determined as plane,inear establish the corelocation alongthe light sheet interpolation wasused. are from thepreceding summarized at Thecore locations the lower right of figure 42 for M i 0 4 anda - 19 ' in flight was made dung a I-g Themeasured location of ,celeration maneuver at 25.00 feet. andhasan accuracy of + or - oneinch. A comparison the lateralcorepositions edge of showsthatthey ae all within 30 inches the leading the with the measured location slightly inboardandnearer of the FVSresult Regarding height thecore, ills scn to be by vcll predicted the Euler code. between of for In searching the causes the disagreement core and ti predicted measure loation. it is clearthatthe with are differences associated unmodeled and/orunresolved that effects This suggests the F-106Bconfiguration needs to he mn in a Navier-Stokes codewith a sufficiently fine all and the gnd in orderto resolve geometry subsequently flow appropriate features c TheNaie.r-Stoes (N-S)equations differ from theinviscid Euleroes in thatviscosity as these of regardless the numerical solution Ilence. inherent wingthe naturally admt and can resolve viscous. equations surface, flow-field This allows the previously nussuig and secondary tertiary vortices effectsto be estimated averaged Often.a thi layer approxitnation to the Reynolds problems is N-S equations suffi.ient for manyaeronautical effectsare that The termthin layermeans the vitscous us fouased nearthe solid boundaries, a manner tinular to codewhich An the bouncay layer approximastiOs example et doesthis is CFL3D. doeto Thomas al. (ref 58). of A compotational example this iode is for the aric A - I deltawing alummel. ref 48) previously exaied with the FVS code
=a~tt Th.i A - 1cta Api-m , A 0 3 and modeled with a Sd of 129x 65 x 65 al M flow The lift andsurface in la inair 0 9.5 a a

in resultsfor this wing are presented exssup This pressure thatgood absolute lift agreement figure43. to figure shows at a i33 andd atte ooper lfttrend is reproduced the over-estimate the of higher values a. Sincethe predtctions the offset between peakCL value,thereexists an apparent with two setsof lift results.The offset is associated a flow which modeling difficulty in the vortex breakdown regime exists a > 33 for at the Regarding suction pressures a - 20.5*. it is notedthat the the for the forward stations, overall level under primary closely with theexperimental vortices agrees andsecondary station are at results. The differences theaftmostpressure of attributed to the presence unmodeled turbulent flow. available the This figure illustrates possibleimprovements in deltawing results, both thetift carve over the previous when pressures, oneusesa codewhich incorpoand surface ratesapproximations to the N-Sequations. FOR USEIN DESIGN Complete Wing are methods highlighted which 0.Imr: Two example wing designin the for yield low dragsolutions a complete vortical flow Thefirst is associated presence leadhngedge of with an attachedwith Lamarel al. (ref. 59) andbegins and wing camber twist design Thenfts flow. complete on shape operated by a geometrical constraint andthe is VLM codecoupledwith theSuction Analogy (VLM-SA). the design.by-analysis mode,to achieve final in a mnanual with code(WINGDES2) associated shape.Thesecond or (ref. 60). is subsonic supersonic. and Carlson Darden bet with thefirst method the whole somefeatures shares 60. The is process moreautomnated. methodof reference has on alto based a onex-lattice representation, two design modes known as"whole wing" and"mission- adaptive" The of provision for the design lattermodewhich hasa special here flapswilt be presented wing Eachmethod is illustrated byin applicason. a cranked wing for Carlson. for Lamaranda swept trapezoidal of and Theassumptions designprocedure Lamaret g between at. tref 59), along with the resulting companson data predicons and measured for a Pi-SCAMP cranked wing follow basicto theuse Ihe Assumptions inherent assumption. analogyfor a cambered-wing, is thatthe of the suynton flow would promotereattached lcding-edge votex system edge.As is well known with theaddition nearthe leading lift of positive camberto a wing. the pomential-flow will increase a positive angle of attacL Tbis increase at is. howccr. toupled with a condition uswhich the flow is more edge The "alignent" does with the leading nearlyaligned vortex that two things in the realflow: I) the leading-edge nearit. but will not only reattach is fouraed theedge near zfcc . an 0 ai w rn , ui .. on-sag the suction or negatisedrag. and2 reduces lft associated exists a dichotomy which with vortexflow [Ience. there

6-14

mustbebalanced. related A assumption thatthis vortex is systemwill be small, notbe shedinboard but extendto the tip. begin to comeinto play only on the upper surface and as thedesignC, is approached. Therefore, the procedure to befollowed is based the flow beingnot far from the on smooth on-flow condition. Hence.an attached-flow solution for smooth ot-flow is obtained from a mean camber design Method VLM technique Lamar(ref. 61) is employed (the of herein) usedastheinitial warped and surface DesignProcedure Application: The design and conditions soughtfor thejoint NASA-General Dynarcs cranked wing wle CLd =0 5 andMId 0 9.In additin, a rooftop ACp distribution (a ft wasinitially specified 0.7) alongthe chord. It shouldbefurther notedthatthe resulting solution for span load from the VLM attached-flow designcodewaselliptical in keeping with minimum vortex-drag considerations The method employed uniformly 20 hosoeshoe vortices chordwise at eachof 10 equally spaced spanwise statios on a semspan Thispattern wasalsousedin theVLM.SA code Spreeding conditions led to thesmooth on.flow mcdence ditstributton shownin figure 44 for the"wing box". The term wing-box incidence refers the incidence of the to centerportion of the wing chord (for this study assumed to lie between and75%of the local wing chord) Theex15 tremec variation of the structural twist, depicted this box in figure, from the tide of the fuselage the tip required to for smooth On-flowwould be impractical for any realaircraft configuration In order to provide a morepractical design from structural aerodynatue standpoints, final box and the incidence distribution (restricted 12') used,asshown to was in this figure er the structural boxreMaie atan cssntinily constant incidence an is twisted only over the outermost 15%of the serntspan Lines consecting thewing-box leading Md tralng edges at four different spansttions for the final Incidence shown are lafigure45 Though z/c the and /c scales ate different, the relative incidence vanazon across the spantisdicermable. Associated earch these iUs, as well theother with of as stations across wing, is theinitial smooth on-flow camber the rotatedbythedifference of thetwoa, curvesin figure 44 andpassing through the trailing edge Thiscombination of incidence canber wasthen euaayre using the VLM. and SA procedure to dettffnsne drag. and the strength lift, of the suctionforce along theleading edgeard toprovide a refernce for successive rodifications. Thecamberahead of the wing box (the 15%chord) wasthenrepresented by five equalstenspan cambered leading-edge segmVnts flap whose deflc1tion angles wereadjusted pamitnesrically while monitorng theVLM-SA draglevel.Eventhoughthese levels wereoptimistic, they wereconsodered relable in estimating the proper trend of lowering dragwith flap deflection angle.Aftera Set angles obtained of was about the 15%chord le which produced mimmum dragvalue, a the reulting camberwasSmoothed the process and was repeated about the 2 5% chordhim. Theis Smoothed caniber linesare shown in figure 45 and labeled final designed the caniber It should be notedthatthe final cambershapesart

changed the direction of the initial incidence distribution in (shown figure 44). in In order to put this camber on the wing. two things were done.The first was to shift thelocal camber distibution vertically to provide a constant elevation alongthe wing nudchord. The second to matchthefuselage was incidence to the finalinboard wing incidence to provide an evenwing. fuselage Junctum (Note the that CL. occurs ana of at about9 4* ) Photographs the designed of wing mounted on an existing fuselage appear figure 46 in For thefinal camber, VLM-SA codeindicates Cd. the thai occurs ata wing etslightly larger thanrequiredfor smooth or, all flow across span If the flow features th which are indicated are largely realized, then this should enable large a amount the available Icautngedge to berecovered of suction at the design point. Dataobtained for the cranked wing. whose design jUy was dericld. arecompared the nextsectionwith theory, in an assessment thedesignprocedure madethereinand of is Lift The lift comparison presented in figure 46 showsthat the VLM-SA method (solid carve), obtained combining by the potential-flow results with the vortex and side edges, predicts the measured lift from theleading lift well over an as range 3o -12' Above a of 2. thereis a lOssm the amount votex lift realized. Of partially dueto thelack of flow reattachment the regionof the wing-tip trailing in edge a coisequenee the realflow having as of insufficient chord thereto peemt the finite.sted vortexto develop reattached flow ad full lift IFor wings with trailing. edgenotching this lift los is increased.) Regarding the comparison thesolution from potential theoryplus with 100%leading-edge suction. it Is apparent tpto that about a * 8 theeffectof the vortex flow ISto reduce thelift. imdicatiovereattachnisnt on the lower sutface Another of interesting feature obtatnng CL,,(. 0 5)with vortex of flow is thatin compartso with the potential-flow solution for this Same cambered wing anangleof attackof about 2'liss is required. Of e it is realizedthattis wing was notdesigned reach CA with potential flow. Still it to 4 is interesting to realize theorcically there an aqlethat is of-atack redulction possible vortex flow is presnt on the if slender cantbered wing. eipecially sincevoitices would tend to form naturally on sucha wing. With regard the0% suction with sonex andthe to no lift 100% adusg-edge suctionsolutions, s note-othy that it thepresence the potential flow leading-edge of suction on the highly cambered leadingedgeactually reduces C, over the the a rangeShownThis is, course. dueto the edgeforce of acting tangentially to the highly cansered leading edge. therebycreating negative force a lift Dragpolars,Figures46 to 48 present dragdata the and other datato aid in its miepretaton For example, figure 46 sho%s boththe planar cambered and wing dragpolars mncompaion with two theoetical curss Over mostof

6-15 the CL range the planar-wing data follow the upper or zero edge-force curve asexpected. The camttered-wing data are generally much lower than the planar-wing data and approach the lower bound polar in the CL range of about 0.35 - 0.5. even though the wing is thin (maximum thickness/chord ratio - 3 2%) and the leading edge is CL shatp. Furthermore. at the design the data reach a level leading-edge suction. This lrge, equivalent to 77% of full value of equivalent suction is remarkable for such a slender wing. particularly at tis high Mach number maneuver fraction of condition. The data further show that a larger leading-edge suction is realized at CL = 04. indicating that been fully optirmzed the wing mean camber surface has not at the design CL. Figure 47 displays the saue cambered-wing drag data but here in place of the planur-wig lower bound polar are two attached flow polars obtained from the VLM-SA code. One is for full edge force, 100%leading-edge suction and the other for no edge force. 0% leading-edge suction. It is well known that a planar wing of the sameshape will have more edge force than a cotresponding cambered wing under the sameconditions, because a potion of the suction available on the cambered wing is distributed cherdwise over the surface. Thus. the figure shows that die displacement between the full. and no-edge- force curves to be smaller than for the planar wing Petiher. the data are quite close to thefull-edge-force curve for CL values equal to or less than CL,. This is in keeping with the onginal idea of being onat an angle of attack slightly above that for smsoosh flow. in that at smooth on.flow full suction is realized but of is distnbuted over the cambered surface In terms the suction available, this cambered wing achieves a level of effective leading-edge suction of about 67%. Axial force Another wa toestablish %hen flow changes beyond examning the lift curve, is to %ing. occur on the is examine the axial force, since it a sentstive measure of the edge flow Figure 48 shows theaxial-force coefficient variation for the cranked cambered wing asa fuc-tion of 2 the n e.because both edge-force and vonex-flow terns i have this dependency. It interesting tonote the sharp change in the CA variation near a n V".because at this lift same a the data of figure 46 show a rapid change The faired straight lines in figure 48 have atsocsatcd with them labels describing the types of flow which am hypoth. esized be present. From the tses of planvicw oil piho to 10" clear that atboth a - 5* arid the flow on tographs, it is CA the upper surface appears to he attached even though the curve shows that some change in the data has occurred It needa to beremember here that, since this leading edge is highly cambered. the flow at the edge cannot easily be seen from the top. At a = I" there is a definite indication of vortex activity on the upper surface, which means thatthe fomtrW orbecome strong enough to vortex system has jutt be noticeable method of Carso Ohe -mission-adapive" design-mnode a Carlson of Darden (ref. 60). to be illustrated here. hat provide a twisted and cambered allows it "to feattr whicuh surface restricted to specified wing regions". The resulting the cambered regions near leading, and trailing-edges may then berepresented with flaps The basic premise most of the wing fixed, say dueto structural that is with constraints, there exists a particular combination of leadingand traling-edge flap deflection angles which will yield the lowest drag or highest effective suction. The concepts of attainable thrust, suction analogy and vortex action point are all employed m this design mode of Carlson's method the This method is applied to 60*- swept trapezoidal wing shown mounted on a body in figure 49 in the following way Firstly, the "whole-wing" design mode is employed provide values at specified of CL, M and R These results a camber surface, which the designer may find helpful in selection of "usssion adaptive" or flap system design areas consistent with structural or other considerations. The provides a design moment. C.i "whole-wing" solution also - -0 17. which in the absence of any other specific C" insure an effective constraint is used in the next step to the contribution of trailing-edge flaps to overall lifting efficiency. Secondly, the program is run in the "missionadaptive" mode with CL. M, & and C, specified along of with a definitiaon the design area in the form of spanwise The results of leading. and traling-edge chord distributtons second run provide a "mission-adaptve" wing camber this surface shown in figure 49. Superimposed on each of the mean. camber surface wing sections are the flap-hinge-line locations and the limits of the design area The code also provides for automation of a flap-fitting strategy, illustrated in figure 50.The idea is to replace the smooth program surface with straight linesegments to generated camber approximate the design camber surface and its loadings. The resulting schedule of leading- and trailing-edge flap deflections is referred to us "code" in the inset sketches of figure 49 Thirdly. the designer selects an appropriate flap is On segmentation plar. this figure it labeled "modified" and consists of four leading-edge and two trailing-edge flapsegments Fourthly. using thes segmented flaps a related analysis code SUBAERF2 may be used separate but toprovide an estimate of the actual flap system performance has (Much of the text in this paragraph been contnbuted by Carlson in a private communication ] No direct comparison with datais given in reference W howescr, an off-design situation is analyzed for a two segment leadig-and trailing-edge flapmodel and the results The analysis predictvthe masured shown at figare S1. values well. SUBAERF2 can also used in a design rFhe analysts code is mode. It done by varying the leading. and trailingedge flap angles systematically while recording the suction level achieved for each combuation. Thesuction levels and flap angles arethen use as basic data in developing a "thumbprtnt" or an "optimization" chart to help select the best combination ]

6-16 Leading-Edge Vortex Flap Genm: Designing a wing with leading-edge vortex flaps (LEVE) is distingished fromthatof Carlson's complete wing method in thatfor this designproblem thehinge line is assumed known bet the flap shape (leading-edge outer boundary) andits deflection arenot. The -eader rennded is thattheLEVF is just a special purpose leadmg.edge device, as discussed LamarandCampbell reference and by in 62 illustrated in figure52. Initially. the function of theseflaps wasenvisioned to be thatof dragreduction for slender wings while maneuvering moderate high o. Since at to then otherfunctions havebeen conceived, asdetailed in thecited reference. Regarding accomplishment the of theinitial function, it wasto occur as a resultof the flap previding appropriated capture andorientation for area theentire separation. induced shedvortex system above itself. Dueto the downward deflected orientation on the forward facing surface the flap, a substantial of suction force wasgenerated thethrustdirection to provide the drag in reduction Furthernore. the"captured" vortex alsofunctions asa rotating fluid cylinder to turn theflow around the leading edge onto the wing upper surface, thereby promoting a smooth transition to attached on thewing nearthe flow hingeline Two methods highlighted for designing planforin are the of the LEVF The first is thatof Fnnk (ref 63) given in concept focu. and she scnM isthat of Hueltct (ref 64) given in sonse detail, An example of eachis given for a r-I06B configuration. [For otherapplications theinterestoa reader mayreferto the three conference publication volumes dealing with Vortex Row Aerodynamics (reft 65 to 67) 1 is seenon this photograph be located ahead the to just of vertical tail. Details of this system alsoprovided in are reference 11 Huebric The designprocedure Huebner (ref 64) is of based theanalysis sethod VORCAM andshares on of with Carlson's wing method ideaof amving at an appropriate the amount of flap area anddeflection angle.However, as previously noted,this procedure differs from Carlson's methodin thattheshape ihe leading of edge a part of the is solution This procedure differs from thatof Fnk in also thatthe flap geonmetry extendbeyond wing leading can the edge. n. a "bolt-on" flap implementaton, andusesa i numerical optimizing procedure. prinmary of this The goal effort was to develop the vortex flap planforimn. deflection angle. wing angleof attackto maximize L/D andsatisfy and CLd at M = 15 for ther-106B This speed chosen was as typical of a supersonic maneuver atsadvanced for tactical fighter. TIe F-106Bwas chosen the hpphcation aircraft as sinceits 60* sweptleading edgeis capable generating of measurable amounts vortex flow. of Analytical FlapModeling: Figure55 shows modeling the of a typical flopwith its designvariables (geometrical features) flap coordtates. TheX axis corresponds in to thewing leading edge(hingeline) in global axes.andthe dimensions theflap havebeennormalized hase a range of to of ze-o to one. The Yf variabledetermnes the flap planform shape.Usingthe VP.D4 flap - developed Fnnk - as an by initial guess, procedure this models theplanform shape in threeregions Regins one and threearcparabolas which areuniquel definedbytheir two endpoints anda slope condition at the pointswheretheynivet with region two. which is a straight tine. The specific design variables needed to define flap are shown the figure X(I) and X(2) this in determise extentof the threeregions the X direction. the in X(3) through X(5) provideactualplanfonn chord length and ultimately planform shape. X(6) specifies flap deflection the angleandX(7) is the modelangleof attack It is worthy to rote a few things aboutthis method.The apexof theflap is shown be at the ongin of the flap axes to In reality, thechord lengthatthis point need be zem. but not it is not a designvauribte remains and constant throughout thedesignprocessThe valueof Xi can go to zero while thevalueof X(2) cango to one. Thus.thepossibility exists that a flip design olution could yield a constant chord, taper. ivere taperflap rurthemsore. the value X16) or of waschosen sucha way that it represents aresangent in the of theflapdeflection angle. Certain geomeivincal constraints arisebased on this rsethod. In orderto resnct the flap to a reasonable X(3) through size. X(5) areconstrained thattheir naxtnim valies are no such morethai 10% of the leiadig-edge hinge- line length Also. to avoid -maningless flap shapes, value of X(2) must be the greater o equal Xi1) than to Procedure, flap-destgn-upttizAation process given or The is outlined form in figure.56 with additional details provided in both luebner (ref. 64) andLa=sar (ref 68).

EnnkAs illustrated in figure 53 thisconceptwasoriginally validated in 1978 a highly cambered on Pre-SCAMP design, just discussedIt was further demonstrated duringthe same wind tunnel thatcomparable test levelsof moanuer peefornxce improvement could be achieved deflecting cerby tain continations of simple planarleadirg-and trailing-edge flapson a planar wing of the same planfonn. Thesimple flap results wereveryattactive from a practical designand fabrication standpoint warranted and further study.As reprented in figure53, maony additional experimental anlytand ical tudtes wereconducted the simple flap concept An on examnaton of theresultsfrom these oec' sthe saled and Fnnk in 1982 the evelopirsnt of a LEVF designproneto dare - published refrence 63 A typical resultis shown is on the lower right of iis figure
Another example isgiven byLamar c Al in reference 12 for the F-1061)arcraft. Therethea rodynamic design of the LEVF was developed through an iterative process that encompassed Frnk's procedure. nd-tunnel results, and practical considerations constraintsA flightphotograph ald of the LEV mounted theaircraft is showAn figure 54 on in Theassociated flight test program is -. ech pressure a and flight performam data andvapot-sreen imagt are recorded - isschaedled completion bythe spinne, 1991 [The fce of housingforthe rotating-light-slscet system- oneof three systems needed in order to obtai vaporscreen images -

6-17 Apphcation. An application for the F-106B is given at the design condition of M - 1 5 and CL - 0223. The initial geome rtc design variable values X(l) - X(5), associatedwith the supersontc application, were taken from the design solution of vortex flap VF.D4 at M - 0 3. along with X(6) which specifies flap deflection angle. The design variable X(7), which determines a was started at an arbitrary value corresponding to a = 40, Figure 57 shows the initial and final planform shapes and other pertinent results from this design study. The flap chord has decreased for most of the flap, designated VF.DOI. except near the flap tip where it increased slightly. Flap planform area decreased by 6.5%. The flap deflection angle converged at 18.470. which as quite close to the slope value at and perpendicular to the leading edge of the cambered wing Finally. the angle of attack converged at 5 060 A comparison of the competed aerdynamic performance of VF-DOI and VF-D4 on the F-106B is shown in figure 58 The VF-DOI design shows an tmpeovement in L D at CLA of 0 6, or 9% over VF-D4 at 10 deflection Further. the improved I.D values extend throughout the entire CL range, The mital design solution is also inclided to show the total performaunceimprovement from the beginning to the end of the design process. Figure 59 shows the aerodynamic chrmctnsttcs of these two flap designs at LE m 300 and Al - 0 3 lie purpose of this is to determine te crodynansc characteristics of flap VF-DOI at an off.design Mach number. Minor variations occur fo CL and C, versus i. however, a measurable improvement in L.JDjna be quite satisfactory at Af - 0 3 A vortex flap designed for the F-106B at subsonic speeds is also given by fluebrier in referemce64 Pointed wings, Figure 61, taken from reference 16. and fig. ures 62 and 63. taken from reference 28. present expenmen. tal and predicted CL data for a variety of pointed wings Regading figure 61. it should he noted that since these wings is noted Thus, this figure indicates that the flap optimized for Md - 15 would
6

In this section only predictions from SA methods will be showe since it is very general and has been widely applied. (Though SA only applies to the vortex flow contribution to force and moments computed by potential flow methods. in this chapter its usage is sometimes broaden, for reporting purposes, to include the potential flow conrribution.] The examples shown cover geometnes from isolated planfoms to interfenng wing surfaces and at speeds up to supersonic. The order of presentation will be (steady) longitudinal, including some configuratons for which only CL is presented, and then lateral characteristics, which include both a steady and an unsteady example [Other examples can be found in the cited references.] Longitudinal Simple

sm Compansons are presented here for six

pointed wings with round and sharp leading-edges at subsonic speeds lTe configurations range from arrow to dia. mond Delta wing with LE radias, Figure 60, taken from Lan and Hsu (ref. 69). shows an application of the QVLM-SA method to a 600 delta wing with a round leading edge at low speeds The SA predictive curve is labeled "thin- sharp" and is swen to estimate the measured CL and C,, results well up to 10* and t. respectively When the mod leading. edge effects are accounted for by using Kulfan's technique. Lan estimates a noticeable aerodynamic effect. This leads to an extensionof the a range for which the CL andC,, agreement is good. 6 and 20. respectively The lack of agreement beyond these a values means that there is still an unmodeled affect. It is obviously associated with vortex breakdown, which is known to commence at the TE on a thin-sharp delta wing of this sweep near a n 12*. Lan and Ilso (ref 69) developed a procedure for quantifying this affect with and when employed for this wing produces good agreement over the enrse a test range

STABILITY AND CONTROL IN HIGH-ALPHA RANGE


This chapter cxurmnes stability ard control both aalytically and experimentally in the a range up to high-c Ilie analytical reults preseeted are based on the analysis rocheis dosettbed in the preceding chapter. are focused moe on stability rather than control, and are compared th experiments in. The expenrmental stability.ai cortrol resurs presented are not restncted to those situations that can be predicted, ht dude those fkom devices range PREDICTIONS FROM ANALYSIS METODS General Selected longitudinal stability results have already been pitsensed with the introduction of the various analysis methodi htch are likely to be successful in

haveno tip chordthe K,. values all zem. however, this are
does not preclude there being an augmoeted. k.,. term In fact, suce the sign of the augmrsented term depends on the sign of Z it is rteresting to note that three of these wings hav. positi|e values and one has a negatree eatue Positive augmented values produce hft above that of SA. whereas the converse is also rire Note the improved agreement at f - 06 achieved when the augmented teros are inclued in the Ct estimate Figures 62 and 63 Present the CL and C. results for the top iwo wings in figure 61. bt Mach number a townr and higher subsonic These figures show that the ability to predict

providing either longitudinal and/ or lateral control in this a

the expenrmental CL is similarly improved at these Mach numbers Both figumr also show that this extension to the SA gives a tremendous improvement in the ability to predict the expenrrunetalC,. This is more tree for the diamond img than for the arrow. since the diamond wing effectively

6-18 addsarean theregionwhere reattachment occur, can whereas, arroweffectively removes the area. In addition, the trailing-edge wakefrom the arrow wing caninteractwith vortex system asto moveit farther from so the leading-edge the wing, therebyfurther decreasing influence aft its in the region. Geometncal combinations Comparisons presented are here for five combinations of wings They include a croppeddouble-arrow wing, a wing-canard, a strake-wing-body. a cambered-thick lifting-body andwing combinatton, anda cropped-delta-wing body. Thespeed range covered both is subsonic supersonic. and Cropped-double-arrow Thethin, sharp, winguncambered. complex configuration of figure64. takenfrom Lamar (ref. 70). provides good illustration of how the various a vortex-flow termscan be used to estimate longitudinal the aerodynamics low speed.(Notethat thefigure legend at groups theseterms moderate high c, which just by or recognizes the particular elements eachgroup differ that of according the local flow features, outlined in figure 27) to as In particular, at moderate thereare two Kvje andkAe a's terms(an inner outerpair associated eachvortex and with systemi a K,... and term At higher a's. two leadingthe edge vortex systems expected to mergeinto only one are which extends from the apexalong thewing leading edgeto thetip This system betepresentedsingle and can by a Kvj. k,,t term which is thenconibiied with pevious Kva tim term produce total vnoex to a flow effect Thedetermination of the Z term used in k._ andi used is estimating needs to befurtherdetailed for completeness Cm (Thereader mayfind it usefulto refer to figure 26 for the 2 vanation with i justification I In particular, at soielrate c's 6 for the inboard systemis just thesrraunwise distance from the leading edgeto trailing edge thespanlocation at where wing s.eep changes, for the outersystem is the and 6 thetip Lhord Each assoviated term it halfwa, distance x the alongsincethe areabeing represested rectangular. 6 is At tugher a's Z is the streamoisce distaice from the tip leading edge the trailing edge to ape., which for this wing is a smal pos tie number The associated is computed the %aie i way asbeforesincethearearepresentedalso tectangular is Consideisg nowthe predicted and teasured resultspeseted in figure64.itis clear the high a ieory gives that better overall agteement either te modete i or pothan tentialtheoriCs, expected. The C1 expenmeital data are as well predicted to nearly28 andtheC,, up to16' For up highera's vortes system the growsin size anditoves farther from the surface overall the wing. especially in the aft region This givesrise to thefocmud part the wing still of lifting well.whereas the aft portion responds the effcts of to vortex dimirshment and finally breakdown Ober cxantples of unsplex tigi are given inrefetens 70 Wing canard Figure65. taken iotmreference presents lb. appli ationsof theSA to a wing-forebody anda wing-said at lowspeedsOnly the wing CL resultsshown, onthe arc left in thepresence the foreody, andon the right in the of presence a high canard of (a/l = 0 185) The wing-forebody comparison showsa variation typical of wings with moderate sweep because areknown to havea low c departure they andvortex breakdown, which leadsto theSA overestimating theexpeimentsl results c's at above8* However, in the presence of the high canard, favorable a interfernce results, andevenwith the reduction CL, on thewing, dueto in canard downwash. predicted the amount vortex lift is of developed the on wing The measured resultswell are predicted over thec range reachhighe CL values and than those the wing in the presence theforebody for of Strake-wiug-body: Figures and67, 66 takenfrom Lamar (ref. 68). present companions between experimental and theoretical datafor a complete strake.wing-body andfor its components, strake-forebody andwing-afterbody. The theoretical results, called high- moderate-a and havealready been developed ,d outlined on figure 27. (Additional modeling details be foundin reference For the can 71.) complete configuration (fig 66) at M = 0 2 it seen is that upto CL, the measured datais betterpredicted C. by the high-ca method Above the corresponding neither a. theoryappropriately models flow It alsoseenthat the is the two theories generally bracket C, data.againup the to CL,, or vortex breakdownThe ability of these two simpletheones to this is encouraging, thattheyare do in ableto estimate collectively the genetal nonlinear versus C, CL,tot characteristics this class configuration It for of can benotedthatthemoderate-a theory may.in general, estimate betterthe C,, resultsthanthoseobtained the with high-c theoty This occurs because moderate-a the theory produces loadcenter a fartheraftat a particular valueof CLan eventhough valueis largerthanthe dataat the this sameangle atiacL The potenial-flow curveisadded of to the CL,a versus c plots for reference The wing afletliod) and strake-forebody longitudinal aerodynait data andthehigh-aandmoderare theories given are infigure 67 for Af - 0 2 Justas for thecomplete configuratian, individual datatopneers are g-ierally well the estimated the high-cv by thor) or a collecitie combination of theories to CL, or large-scale up votex breakdown Whatis particularly u lf thatthe individual C,, com. is ponentt aretightly bracketed the high cv and nooderate-cv by thecoresThe CLdatafor the strakeforebody are, in gnoral, reasonably estiniated by thetwoclosly spaced well theories until the strake vortex begins break to down on the stlaLe higher satues a. The spacing atthe of bet,-t the two theones larger(of the wing iftetbody. with the data is tending be geneally or above estimates to on the from the high-a theory This continues until the strake vortex be. ginsto breakdown ahead thesing trailhi edge.From of this figureits seenthat.in general, cunfiguration has this its arodilynamic compneos betier estimael by thehigh-a that of thesry Lastly, note at te higher angles attackthe wing-aftertody lift vanations follow thepo teclal curveeven the though flow ts closerin a Ileluholtz type Cambered thi lifting bodyandwing A proposed bypersonic research aircraft configuraion composed a of

6-19 cambred-tuck lifing body anda cambered wingis shown in figure 68 (takenfrom Lamar, rrf. 72). Several different ways of modehig the various wing, body andaugmented vortex-lift effectsdeveloped tis configuration on at M = 0 2 wereexamined using the VLM-SA code. The method which worked bestin estimating theexperimental data wasthe onewhich only mcluded thecambered-wing leading- uncambeed.wng side-edge and vortex-lift terms - addedto thepotential termsandis given in figure 69 bythe solid curve. TheVLM-SA computational model for this configuration included the body andwing mean-camber slopes, wing dihedral, but thickness not Tht representtion is seen to be sufficientto well predict te values CL,0 Ca. A comparison of the results of and also showsthatreasonably good overall lift anddragagreement is achieved. However, both the andCn areonly well CL estimated uptoan a of about16". For larger a's, the influence the vorex of system evidently getting smaller is over the aft portion, mostlikely dueto vertical displacement. which causes measured andC. to decrease the CL and become morenose-up than predicted, respectively. However, theoverall results quite encouraging are given the modeling of the configuration Cropped delta-wing body:Figure70. takenfrom reference 16, presents companison the aerodynamic a of characterastis obtained on a cropped-delta wing.body model tested at M 12 with the SA methodresults the wing alone for at thesameMach number.The comparison shows that inclusion of the leading-edge., side-edge., augnened. and vortex-lift effects leads improved agreement the CL, to with C, and CD measured resultsover this restricted range, a Thevonex flow contributions toC,, areobtained byhaving therespective actat their cmtroids. andbypertonrung lifts theanalytic surface integration. both insideandoutside of the tip co., of theproductof thepotential-flow lifting pressure its and chodwise positron (All C D curveshavethe expeimental valueof CD,0 added I in Lateral tedy The QVLM-SA method, allowance leadingwith for edge vortex breakdo-n, hasbeenapplied a cropped to delta wing byLanand Ils i re fetrnc 69 to deeraane the lateral acrodynarmc charactenstics. Figure 71 showsthe agreement the expeninental resultsto he remsarkable with Lanpoints out thatthe inclusion of the side-edge vortex termsis important to get this good agreement (The steps noted in the theory curves dueto the vonex are breakdown cnitena - developed this reference affecting eachsideof in the model differently at a with 13 Or. e. 9 dpendent upon d the local flowconditions present.) U .Figure /2 showsresults obtained by applysng the unsteady suctionanalogy Lan(ref 38) theprediction of to of oscillatory roll dampimg a gothic wing Shown are for experimental andtheoretical data for 1 versus for 0 , a reduced frequencies k - (w b/2U)- 0 75 and 1.20 Tbe of thmeretical results predict expenmental quite well the data primarilyto an appropriate due modeling of theVortexlag andconvective effects. Notethe reduction in roll damping which occursathigh a's to a vortex-mduced effect due This feattre offers a possible explanation of the wing rock phenomenon encountered slender by wings geometries operating at those attitudes. EFFECTIVENESS CONTROL DEVICES OF General Conventional control devices aircraft maybe effective in on this a rangeif they havebeenproperly integrated the into dominant flowfield of the configuration Altematively, nonconventional devices which work with vortical flowfields should proveeffective. A few selected devices examned are to illustrate these points Aileron The rolling momentgenerated a single trailig-edge by aileron tip-mounted onto a cropped delta wing is illustrated on theleft sideof figure 73, takenfrom reference as 68. a function a. This aileronis mersed in the vortiral of flowfield off the wing leading-and side-edges at a 5' and deflection its effeciveness at generating rolling moment is constant the highest a. Deflection angles to test higher than 5' were tested, the growth in rollmg moment but doesnot increase Imearly with deflection Therefore. effectiveness the is degraded the higher deflection settings at Vortex-Flow Roll-Control Device The vortex-flow roll-control device. in proof-of.concept form. is alsoshown figure73. This device in seeks to develop useful lateral characteristics generating by flow asynrsneaes through a planform geometry modification In particular, the intentof thegeometical change to alter is thesynmrical flow situation by modulating theleadingand/or side-edge vortex system theraked.lip sideand on byregulating the amountof areadownstream thetip of lending-edge on which voex system to act. The the has experimental results the middle of thefigure showthe in sigaificant andlinea, left.wing-down, rotling-monsent growth with a for a tip rakeangleof 5. Though ero rolling moment produced zero deflection. the is at middle .gure shows Ci value this deviceto first exceed the for that of the aileron, just discussed. a of 12' near Theeffectof recreating the tip rake angleat an a near 18, shown the right in figure73, is alsoto produce on an almostlinea increase rolling momni. The results in of the precedmig indicates a potentially usefsldevice, which would work best whenboth tips weredeployed in thesame direction -either right or left Another feature this device. of though shown,is thatit produces not e:ther a p~overseno-cr yawMgmonment. The potential theorycurvesshown hereweredetermined from combntitions of symmetrical model analyses, andare sen toaccountfor only a small portion of the measured rolling morent
.

6-20 Leading-Edge Vortex Flap Rao and Campbell in reference 73 discuss many vonical flow devices which can be used to manage this flowfield in a useful manner Among them are full-span or segmented leading-edge vortex flaps which operate on the lower or upper nrface. The lower surface have already been LEV discussed an the design section. sothe upper surface type, shown in figure 74 (taken from ref 73). will be considered here. This figure shows a full-span LEVF and how it functions at moderate and high a's. Rao reports that at moderate or's "a vortex forms inboard of the flap whose sucton generates drag on the flap, but also increases lift on the exposed wing area." (Rao actually used the term "low". butthe term "moderate" is employed here for textual consistency in this lecture I Thus. the LEVF behaves very similar to a thin, unflapped. highly-swept wing with a sharp leading edge is the sme a range Whereas, at a higho coniuon Rao notes that "a dominant vortex develops in front of the flap while the inboard vortex tends to weaken. the net effect being a thrust force," T s is a relatively new flow feature, and therefore one which has not been fully exploited. Since the local vortex flowfieids behave differently in these two a ranges. Rao suggests that this device has "potential applications in different flight regimes". Figure 75. also taken from reference 73, shows some poten, tal uses of a sgarted vercsion this devi at high o's of The top sketch indicates how it provide a drag reduccan nonby deploying all ou segments at a high lift condition in order to get the thrst benefit off the front surface of the flap Theieddle sketches shows how pitch-up or pitch-down canbemanaged by deploying only the rearward or forward par. respectively By manipulatng the flaps in this rae the lift is maximized ahead of or beund the center-of- grasity, respectively, giving rise to the associated morr.erts The bottom sketches display how roll and )aw control can be accomplished through deployment of the devices on one side lnly a coupling of theright'fro segment with the left. and back one, respectively. The right roll peoduced due the is dflecitod flap,onex-system being farther outboard than that for the aideflectd side. Th right yaw is associated wath the thrust on the deployed flaps being properly oriented with respect to the cter-of-gravty. These eamples tugight single degree-of-freedom miOnons, i is clear that with an but approprate-control-system ancd sdh properly sized segflap ments controlled maneuvers more thanone axis at a about time are possible PROBLEM STATEMENT Aircraft operating at post-stall-flight conditions ar susceptible to quickly occurring, unusual maosns - including turnbhing,spinning, coning. wing rock and nose slie. These motins caused by the flowfield around the vehicle be. are coming asymmetrical, unorganized and/or unsteady Assocated the change in flowfield is a degradation us aircraft with control, primarily lateral, even at zero sideslip. This is itustrated in figure 76, taken from Mum and Rao (ref. 74), as a loss in available yaw control at higher os when the rejust quirement for control is increasing The significance of the flow/control changes, with respect to current fighter aircraft. is better understood when oneconsiders the evolution is design which has occurred for this class of airplanes is the last 50 years. In partcular, these aircraft now have a subsmtanal portion of the vehicle ahead of the center of gravity, as reported by Chambers (ref. 75) and shown here as figure 77. The consequence of this is to make the aircraft very suscptible to differential changes is the lateral flowfield oer the forward part of the configuration. (T reader is referred to the two papers by Chambers (refs. 76 and 75) a discusfor Sionof high a effects and expenental solutions, in panr si. lar stall/spin, on fighter and general aviation aircraft.] Apart from lateral solutions suggested by novel flow control methods on an aircraft. it still possible to use design is crieria for the "prevention of directional departure due to either stability or control charauienutics". as woed by Chambers (ref. 76). In particular. nuliahy aircraft are more likely to be :.sstant to directional departure if both C%. and the Lateral Control Divergence Parameter (LCDP) ar > 0 These cnteria -defined in figure 7g. are not absolutes but should be viewed as a useful guide with sihlch to examine each new aircraft design because they are based on a large collection of conelated data Even If proposed design fails a these coteria. there me still altemarive solutions to address the post.stallfight problem. POSSIBLE SOLUTIONS Possible solutions to the problem of redlced aia raft lateral control in this a range do not lend themaelves to well analytical tirett with a resulting mathematicAl approach. Therefore, the engineering method to use is that of an expermetsal proces, like that depicted in figure 79 taken from Nguyen and Gilbert (ref 77) A successful pass through this process vonld be one as which potential problems are iKntified early on aW solutions verified This section. therefore presents some possible solutions to this problem through either novel aerodynamic or powered devices Aerodynamic Devices Several novel devices hase been stuied whlch offer the potential to increaw the lateral control at she higher a s The ones presented is ths section workto control the relatively small nose vortex the long forebody, and are called on nose devices Some of the devices discussed previously, e g.vorexlift roll contrl device andLEVF (plu others t

POST-STALL-FLIGIHT CHARACTERISTICS This chapter examines the problems of post.stall flight, or flight at hilte o's. offers some and potential solutirus. In addition tk blenfits associated with "dynamc stall" are itraolced midan engiering method given for estmatlng the effects, Latly. soice flight at the higher o's impacts not rely the aircraft aerodynamacs but the enguictnlel also flosfield, a brief discussion is given of vays to iniumize this effoct

6-21 for pesented in reference should alsobe considered 68), flightcontrol in this a regime. forebody include in actuated considered Thenosedevices shrake,Jet blowing, tangential slot blowing andjet suction, 77. asshownin figure 80, takenfrom reference Though to the fluid different in activation, all seek change local melchartcs in such a way thatthe nosevortex on oneside than on the of the forebody will be closer to the surface other. One canview this as removing vortex symmetay. The closer the vortex is to a surface more"suction force" the from it generates the larger the yawing moment that and 77, s.,e. Figure 81. takenfrom reference showstheresults from oneof these devices.Note how effective the actuatedon forebody strike deployed the left sideof a genericmodel a The is at generating noseright yawing moment. effect is with strake deflection andto peakbetween seento increase , 400 to 55* depending the deflection, on Thrust vectong on employed for many years the Thrustvectoring hasbeen Harrier aircraft to provide total lift at zero (andlow forward) during "ski-jump" takeoff and airspeed, enhancement lift can flight, andin-flight thrustreversalSincethis aircraft the to thereis a need augment aero"fly" at low airspeed, reaction control system dynamc controls with a dedicated for Without a requirement flight at zero air-speed. (DRCS). would no longer exist to havethe enginethrust theneed near vectored (deflected) theaircraft's center-of. gr,;.) is are nor for a DRCS In particular. if thenozzles located enas theregionof aircraft rearclosure, is a conventional this puts thrustin a logical locatuio the ginearrangement, 'oe its deflection ia order to provide effective longittaonl is control The peeceding the curent fi.us in and/or lateral , wherethedeflected thrus t to i, thrust-vectonng research aircraft control as o' pat sorne,if not all. of theneeded for which conventional aerodynamc controls are itre nt. in of Theimportance this focus is reflected the curem fliet programs utilizing ther-15 SMTD. F-18 IIARV research andX-31 aircraft 77. Figure82. takenfrom reference illustrates howa naral. deflcted 30. becomes increasingly ineffective at generatait, by as yawing moment a is increasedThis is caused the being rudder flowfield not experiencing freeair, butinstead separated of amounts "the low energy exposed to increasing as wake"which comesfrom the %mg andfuselage, notedby Ngu)en andGilbert Figure82 alsoshowstheconabution 10*yaw -t due to yawing toment to thrust vectoring at mavimum power.It is miietsting to notethat thiseffect on re fuel is only weakly dependent a Is order to .... to it bumed durng a maneuver. would be beaeficatl have the yaw-corirol- transition occur gradually from being fully dependent the rudder lower r's to beingfully on at on dependtent thrustvectorng at thehighera's -isn ianner acro-yawcontrol by simila to thatdepicted the idealized. curve Theblending of acrodynamic controls, deploynent of novel aerodynamic devics andenginethrustvectoring over a wide a range an active areaof controls research. is interested learning nre abouttheaerodyin [The reader for reasons using thrust vecnamicandnon-aerodynamic by to toring is referred thecomments Poisson-Quinton in 781 reference DYNAMIC STALL There another is aspect flight in thisflow regimethatcan of well havea positive benefit. It is called "dynamic stall" and with is assocsated a pitching motion, in which theslender a configuration reaches given a rapidly beforethevortical This can flowfield can change character breakdown. e.g has response occur because vortex system a hysteresis the during a pitching motion, asnotedby Lawson (ref. 79). andleads the phenomenon to known as vortex lag with it coherent a to attendant effect of keeping thevortex system higher o. All this can leadto aerodynatmic forces/moments of in excess the static values.The studyof "dynamic stall" effectshasreceived increasing emphasis recent in years by of not manyresearchers, only because its positive effectsbut of therear concerns its impacton higher-a stability and by the control of flight vehicles Sec.for example, papers (ref. 81) and and et Naumowica al (ref 80).Brandon Shah Nguyen (ref. 82). a Ashley et at. (ref. 83) haveput together simpleempirical the theory to estimate Cy andC,, effectsof sinusoidal 00 pitching - a motion of (-cosfit) between and900. are ratio wings The keyingredients to keep - low aspect they parameter physics andan uosteadiness tsvc, of the fluid and to cull . '-hich is related .,e circular frequency the Their analysis follows 'he wing's -n rat.sssir of do/ is distance torn-.. acodynamic forceper unit chordwise assunco ' consist of threeparts.(I) a portion determined as of of from the uste change crossflow momentum, in -cdt a.4cj theory, but with slabsof incompressible fluid at orteiQt normal to thewing surface 0. (2) a portion bastsbyrotating the leadingctc lr'd on a quasi.steady tige fiction through 900(Poihami.s, ref. 18) and(3) a Ive,. dominantat thehighera's andfound by cromsflowof drag corsuteratons. In therange a's whereit-ahhility is above the wing. the first two part. ie ...sumed to present of actonly ahead the point zBO The third partactsonly 2 here to behindthatstationandis proportional Co, sin a. w drag CD, is takento be the measured (ornormalforce) at a aboutan astfl.ed at .he two-tiurds - 0* For pitching ot) chordlin 1c,), thenoemal forceandmoment about thataxis are" gsven as-

C,

. 6

(6m) I-

tan -A

8cO/V)cesa

+ 16( C/V)(zBo/a)cono 2 + (6C/V=)[3(zaD/1O) - f ixln/MOli 2x -. / + + 8(ic.lV)[(zBD1co) - llam + (oc9/V) [3(xa1CO) + (8/3) - (16xBLj3.b)) + CD,[sm a + (t/l)(Oeo/V)
2 } j

6-22

and C. .(XBDCO) - 1sin2a taA2 + (& o/V)13(zBD/CO) - 4(XBDIOO) + (4/3)] cos + (a4/V )[(3/5)(ZBD/c) + (4XBDO/9c)]}
-

perform analytical or experimental work in this a range, it

ts imperative that the fundamental features of the dotmmnao

(o/) 21(XBD/CO)
2

flowfleld be well understood. Therefore, an extended discussion concerning how these flowfields form. grow and decay has been given. This was followed by a representative sampie of varioic aalysis methods used to predict the effects of vortical flow, along with examples of each. The methods presented include the suction-analogy with extensions, free-

- (xBD/C)

isin a ( I[(cBD1/) ',fisin\J A + (ac0fV)i3(cBp/c0) ' 2 - (16xBt/3A) + (g/)]nina


+ (aco/V)21-(16/9) + (16XBD/3C0) 6(XDD/cO) + (12/5)(xoD/A)J}
2

vortex-filaments. firee-votesele. and modeling of thc Euler and Navier-Stokes equations. Vortical flow can also be used in the design process using engineenng methods. Four examples have been given which include a complete wing, a portion of a wing and leadingedge vortex flap systems Each of these was subsequently analyzed or analyzed and wind-tunnel-tested Upon exam. m,ng the aerodynamic results, the conclusion is reached that the salient flow features have been captured in the design process Stability and control in this a range has been examined with methods which employ the suction-analogy with extensions The result of which is to establish that these methods are able to make reasonable and quick engineering estimates of both longitudinal aid lateral global charactensties as long as the vortex system is coherent. and sometimes even beyond overall coherency. A general role is that once large-scale vortex breakdown occurs, the experimental procedures with their emphasis on a sequential process ard/or redundancy should currently be viewed as the method of choice, provided all of the test parameters and lengthscales are ',ell understood Novel control devices, such as forebody nose devices which utilize vortnal flows aid which (as of yell have not been nosleled aalytically, have been shown to mke a significant coniution to the lateral characteristics in the higher-a range or post-stall flightregime Thrust sectonng is also shown to make a significant contibution in this a range where conventional latial contuol surfaces fail An ergineenng method, which also employs the stction analogy, is provuled to estimate the normal force and dynamic stall pitching moment of delta wings underlgoiPg witl Though exact agteenent is not achiemsed measured das, this inCtho does captur the qualitative behavior of the loadings

CD,(coy/)((9)(co0/V)sin a - (l/135)(aco/V) )

Figure 83, taken from Ashley et al (ref. 83), shows measured and predicted results at a value of K = 0 04 where the Cm has been corrected to be about 77% e. They noteThe fir agreem between coreponding plots is clear." (Only one shown here.] "One can peihaps conclude that this very elementary attempt reproduces quite well that qualitatine behavior of the aitloads and might serve as the basis of methods for prelinunary-design estimation on lifting surfaces of supeenaneuvenng aircraft" [INGINEfINLET In order to minimize the pressure or Mach number distortion across the face of an engine or inlet, a successful imegralion with the airframe is needed This is especially iroe for fighters and is done by taking into aconunt. eatly in th: dsign cycle, the flowfieldin which the inlet is to be imnmersed Peener (ref. 84) illustrates this process on the Mirage 200D %shorewith forebody reshaping alone flowfield distortions (ref. 85) acre reduced 30%. Beyond reshaping. Leytnaeri shows how the faselage.boundary-layer flofield can be kept out of the inlets through hvertets tas for the Rafal. F-I IA. etc ). splitter plates (as fot the Rafale,. or hieeds(as for the YF-17 or F/A-18) At high c's. minimizing the d-,tortions hecon'es more difficult because of flow separation which cn occur at the inlet lip Leynaett notes that for a fixed sharprinlt lip there areat least two basic solutions either add vortex geeratlor to control the sepatatd flow in the inlet duct or bleed the bondary layerat the lip. as per Concorde For nassable inlets, there is a potential weiglsttnechatism problem but also a significant benefit This benefit is documented in figure 84 for the auxthary, door device, along with two other types of o%able inlets Theroiatng cowl lip (or cowl lip dloop) has also been studied, as reported in a summary by Nguyen and Gilbert (ref 77). and the results shown in figure 85 This figure shows the measured improeienits in pcessure recoscry averagetutolence and distoftion oilagund by increasing the cowl lip droop angle at the higher is

REFERENCES
l1) DeMets. R X-31 will turan a dinse Aerospace Amet ica. October1990. pp. 2629 (2) Manyauthots Engineering Methodsin Aerodynamc Analysis and Design of Aircraft AGARD Report-R-783, 'Ay 1991 Many authors High Angle-of-Atiack Aerodynanucs AGARD-LS-I! March 1982
.3)

CONCLUDING REMARKS This lecture focuses on aircraft high angle-of-attak acrvl)namics with their attendant vortical flo.fiehis In order it,

6-23 (4) Many authors:Aerodynamics of Vortical Type Flows in ThreeDimensionsAGARD-CP-342, April 1983. (5) Anderson, D., Jr: Modem Compressible J. Flow with Historical Perspective McGraw-Hill Book Co, p. 166. 1982. (6) Cunningham. M., Jr.: Practical Problems: A. Airplanes. In. Unsteady Transonic Aerodynamics. Prog in Astronautics andAeronaitics. Vol. 120.p 92. 1989. (7) McMillin, S. N; andThomas. L: andMinaran, I- M.: I. Navier- Stokes Euler Solutions for Lee-Side Flows Over and DeltaWings, Correlation With Experiment A Supersonic NASA TP 3035.December 1990. 7 (8) Stanbrook. andSquire.L C.- Possible pes of Flow A: atSweptLeading Edges. Aeronaut Q., Vol. XV. P 1, Febroaiy 1964.pp 72-82. (9) Hemuch. J.:andLuckling. J. M.: Connection BeM. tween .eading- EdgeSweep. Vortex Lift, andVortex Strength DeltaWings.J. of Aircraft. Vol. 27.No 5. for May 1990.pp. 473-475 (10) Smith,J. H. B.: Calculations of the Flow over Thick. Separation Conical. Slender Wingswith Leading-Edge Aeronautical Research Counctil R&M 3694.March 1971. (It) Hemuch. J: Similanty for 8figh.Angle-of-Attack M Subsonir/rransonc Slender-Body Aerodynamics. J.of Aircraft. Vol 26.No. I. January 1989.pp.56-66 (12) Lan ar. J E.: Ilalhusy. J.B; Fnnk.N. T.: Smith. R. H.: Johnson. D.. Jr.: Pan. I -L: andGhaffan. F" Review of T Vortex Flow Flight Projects the F-lO16B. on AIAA Paper No 87-2346 August 1937. CP. (i3) Lamar.J.E, andJohnson D. Jr. Sensitivity of T. F-106B Leading.Edge-Vortex Images Flight andVapor. to Screen Parameters NASA TP 2818.June1988 (14) Lamboume. C. andBryer. D. W.: The Bursting of N Leading EdgeVortices. Some Observations Discussion and of the PheomenooAeronautical Research Council R&M 3282.1962. (15) Wentz.W I1. Jr. andKohlma. D L VortexBreak don on Slender Sharp-Edged Wings J of Aircraft. Vol 8. No 3. March 1971.pp. 156-161. (16) Lamar. E. Some I RecentApplications of theSuction Analogy to Vortex-Lift EstimatesIn Aerodynamic Analy. ses Requiring Advanced Computers. II. NASA SP-347. Pi pp 985-101. MArch1975. (17) Polamus. IL C A Concept ,1, Vmrex Lift of of Sharp-Edge Delta Wings Based a Lea'ng.Edg.Scticon on Analogy. NASA TN D-3767, December 1966. (18) Polhamus. C.: Predictions Votex-Lift CharacE. of teristics Basedon a Leading-Edge Suction Analogy. J.of Aircraft. Vol. 8, No. 4, April 1971.pp 193-199. (!9) Margason. J; andLamar. E.: Vortex-Lattice R. J. for Aerodynamric FORTRAN Program Estimating Subsonic of Characteristics Complex Plaifors. NASA TN D-6142. Februay 1971. (20) Lamar. J. E.; and Gloss. B.: Subsonic B. Aerodynamic Characteristics Interacting Lifting Surfaces Separated of with Flow Around Sharp EdgesPredicted a Vortex-Lansce by Method. NASA TN D-7921. September 1975. (21) Lamar. J. E; andHerbert. E.: Production Version H. of the Extended NASA-Langley Vortex Lattice FORTRAN Computer Program Volume User's Guide. NASA TMI 83303. April 1982. (22) Snyder. H., Jr,;andLamar, .. I: Application M. of theLeading-EdgeSuction AnAlogy Prediction of to Longitudinal Load Dismrbetin andPitching Moments for Sharp-Edged Delta Wings.NASA TN D-6994. October 1972. (23) William. J.E., and Vukelich, S. R.: '1HE2 USAF STABILITY AND CONTROL DIGITAL DATCOM. Volume II. Implementation ok'Datcom Methods AFFDL-TR-76-45. Vol II. November 1976. (24) Lamar.). E: Extension LeadMing-Edg-Suction of Analogy to Wingswith Separted I-low Around thq Side Speeds. NASA TR R-428.Octo.r 1974 Edges Subsonic at (25) Bradley R.0: Smith. C.W.. andBhateley. I. 1.: ,brtexLift Prediction for Complex Wing Planforms. I of Aircraft, Vot 10.No 6. June1973.pp 379-381. (26) L.an. E; and Mehroma. C An Improved WoodC S ward's Panel Methodfor Calculating Leading-Edge and Side-Edge Suction Forces Subsonic Supersonic at and Speeds NASA CR-3203. November 1979. (27) Lamar.I E. Prediction VortexFlow Characteristics of of Wings atSubsomc Supersonic and Speeds J of Aircraft, Vol 13.No 7. July 1976.pp 490-494. (28) Lamar. E. Recent Studhes Subsonic J. of VortexLift Including Parameters Affecting Stable Leadig-Edge Vortex Flow. J of Aircraft. Vol. 14.No. 12.December 1977.pp. 1205-1211. (29) Lamar. E., andCampbell. P Recent Studies J I at NASA- Langley of Vorwtcal flows Interactig with Neighboring SurfacesAGARD CP 342.PaperNo 10. 1983 (30) Lan. C. IL. A Qual-Vtirtex-Laitce Method in Thin Wing Theory J. of Aircraft. Vol. II. No. 9. September 1974.pp 518-527.

6-24
(31) Lan. C E; andChang,.J..F.* Calculation of VortexLift
Effect for Cambered Wingsby theSuctionAnalogy. NASA6 C'9.3449. july 1981. (32) Lan, C. a:; andChang, J.F.: VORCAM - A Computer Program Calculating for Voetex of Cambered Lift Wingsby theSuctionAnalogy. NASA CR.165800, November 1981 (33) WoodwardF. A;Tiaoco. E.N:and Larsen. I W: Analysis andDesign Supersonic of Wing-Body Combia. tiona. Including Flow Properties the NearField. PartI in Theory andApplication NASA CR-73 206.August1967. (34) Kulfan. R. :Wing Airfoi; Shape Effects the on Development Leading.Edge of Vortice-- AIAA PaperNo 79-1675. 2979 (33) Henderson. P.: Effects Wag leading-Edge W. of Radius Reynolds and Number '.onguudinal Aeroenamlc on Charaicterstics, flighty SweptWing-Bodly of :onigurv:ons at Subsonic SpeedsNASA TN 0.4361, December 1976 (36) Carlson. W1:andMack. R.J.. Studies Leadang. 12 of Edge Thims Phenomeina AIAA PaperNo 80-0325. 1980 (37) Carlson. W1:andWalkiry. K B A Computer If Program Wing Subsonic foe Aerodynamic Performince Esimateis Including AttainableThrust VortexLift andl Effects NASACR-3515. March 1982 (381Lan.C E. The Unsteady SuctionAnalogy andAppli. cations AIAA PaperNo 81-1875. August 1981 1391 Lan,C r TheUinAeady Quasi-Vortex LatticeMethod with Applications to Animal PropulsionJ of Fluid Mechan. ict. Vol 93. Pt. 4. 2979, 747 p 140)Lan,C a Applied Airfoil andWin3 Theory Cheng; ChungBook. Co.- R 0OC- pp 459.374.2988 (41) Mehrira. S. C. inald C a A Thoeccal Inses. Lars, ligation of the Acrodynxawics Lox-Aspecuslatio Wings of with Prt3J Leading-Edge SeparationNASA CR-14304 ;anuary1978 (421Pao, L, anld I Lan. C E A Vorsex-rilament Core Wd Iodelfor Wingswith EdgeVortext SeparationNASACR1638-47. February2982 (431Karidil, 0 A. andYates. C. Jr Transoruc C Vorex FlowsPastDella, Wings IntegralEqu~ation Approach AIAA J. Vol 24.No 11, Novetmber 2986. 2729.2736 pp (44) Katz.I.- andMaskesu. UnsteadLaw-Sped AeritB dynamic Modtel(f oeeplcte Ainvraft Configurations of 3 Aircraft. Vol 23. No 4. April 2989.pp 3G2-3 10 ( 5) Johunson. T. Lu.P, Tuxico. -N .and EptonI F M A An lmir.iod anetlMethodfor theSolLior of Three. Dimensiontal Lea/ar;- bige Vortex Plow Volume s I1Thory 8 Document NASACR-32 /F. Jul; 1980
4

(46) Lucksng. I M; Schoonover. E., Jr; andFrank.N W. T. Recent Advances Apptyirg Free in VortexSheet Theory for the Estimation c' :x Flow Aerodynamiucs. AIAA Paper 82-00953. No. -auary1982. (47) Luiclaing, M. Hoffler. K D.;andGrantz.A. C. J. RecentExtensions the Free-Vortex-Sheet to Thory for Fxipanded Convergence Capability NASA CP-24l6 pp.85. 114. 1986. (48) llummsel, Onthe Vortex D Foration Overa Slender Wingat LargeAngles IncidenceIn AOARD-CP.247. of Paper 13. January No. 2979. (49) Raj, P.Olling. C.R..Sikora. S.: Keen. M: Singer, J. J S W.: andBrennan. a- Three-dimensional J. EulerftjavicrStokes Aerodynamic Method(TEAM). Vol F Computation MethodandVerification APWAL- lR.87-3074. Jane1989. (501Rnj,P. andBrennan Improvementsan Ester J.. lo Aerodynamic Mltsod for Transonic Fuow Similation. AIAA PaperNo 87-0040. January 1987 (51) Powell,K G. andMunnan. E. M -A Comparison of Experimental Namenical arid Results DeltaWings with for VortexFlapsAIAA PaperNo 86.1840. June2986, (32) Mutenan. M: Powell. K. G,- and Miller. D S., E Comparison Computations Experinmental for of and Data Leading Edge Vortices Effectsof Yawarnd VortexFlaps AIAA Paper No 86439, January 2986 1531 Fles. J. anld Chadterjan, M The Numerical N Simulation of Transonic Separated Flaw aboutthe Complete F-16A AIAA Paper No 88-2306. June2998 141 (shaffits. F, Luckinog. M.. Thionsas. L, andBates. I1 J B L Navier. Stokes Solutions About theP/A-IS ForebodyLeading-Edgte Extensio Configuration, 3 of Aircraft, Vol 27 No. 9. September 2990.pp 737.748 1351 Rts). An EulerCedefor Nonlinear Aerodynamic P Analysis Asseusmwa Capabilities. of SAE Technical Paper Sei 881486. Osioter 2988 156 Pao. 1. wortcal nowx Analysis F-10611 1 for Configo. ialons A2AAPapeiNo 88-3'45-CP. Jul) 1988 1371 Mc~jiegur.I DiTe Varvia-Scrcen Methodo R2Fow Vi. sualotro I of Fluid Mechanics. Vol 11. Pt 4. December 196. pp 461-511 (01 T1homas. L. Krit. S T.-aid Anderson. K I W Navier-StoIkes Compuatios of Voisical Flows uver LawAspect-Ratio Wings AIAA J. Vol 2S,No 2. rebiusry 1990. 205-212 pp 1(91Lana..I E. Schememsly. R T. and Roddy.C. S Deselopmractof aVortxif DesignProcedureand Appl.stiuoe a Slender to Manseuver-Wing ConfigurationJ of Aircraft. Vol IS No 4 April 1981.pp 259-266

6-25 C. H. (60) Carlson. W.; andDarden. M Validation of a and for Code-, Estimation Optimization of Pair of Compater of Aerodynamic Perfonmance Simple Hinged-Flap Subsom6d SweptWingi. NASA 7? 2828.November for Syitems Thin 1988 (61)Lamar.J.E.: A Vortex-Lattice M-thod for theMean Flanfoims with Noncoplanar of Shapes Trimmed Camber Jane1976 Drag. NASA TN D-8090. Vortex Minimumn
3. J. (62) Lamar. E; andCampbell. F.- VortexFlaps AdAerospace Fighters. for Control Devices Supereroise vanced Ameisca. January 1984. pp 5-9

for Strokes YawControl at igh Angles Forebody Actuated PaperN~o.87-2557, August 1987. of Attack ALAA J (75) Chambers. R : High-Angle-of-Attack Aerodynamics June LearoedAIAA PaperNo. 86-1774-CF. 1986. Lessons J. (76) Chamrbers.R.. Overview of StallSpin Techology Angust1980. AIAA PaperNo 80-1580. L (77) Ngnyen, T.; and Gilbert. W. P.: Impactof Emerging Aircraft Agitity. AIAA 90Combat on Technologies Future May 1990. 1304, (78) Poissori-Quiton. P., Comments Propalsion/Airfranie on Capafor Integration Improving CombatAircraft Operational on Course Fundamersbiliries In, AGARD- P.740 Special ratsof FlghterAircraft D -sign, Februiary 1986. Flow on SlenderWings (79) Lowion. M V.. The Separated No & Motion Reparts Memoranda 3448. in Unsteady Council. London.U K. September Research Aeronautical 1963 R (80) Nanmowier. T, Jamb. M. A.. andMargason. L. of Aerodynamic Investirgation DeltaWings wita LargePitch August1988 No Amplitnde. AIAA Paper 88-4332, G J. (81) Brandon, M; andShah. If. Effectof Large Aerodynamics on Amplitude Pitching Morions the Unsteady of Characteristics Flat-Flair Wings AIAA PaperNo 884331,August1988 for Research Hfighly L (82) Nguyen. T. Flight Dynamics Septem-. Sers 892235. Paper Agile Aircraft. SAETechnical bee1989. M (83) Ashley. It - Katz. J. Jarrati. A; andVaried..T Loading of DeltaWingsfor Low Aeradynanuc Unsteady of andIfigh Angler of Attack Proceedings dieIntrnational ASME FEiDVol Fluid Dyniaracs. on Symposiuni Nonsieady Jane1990,pp 61-78 Canada. 92. Toronto. Optiniisaio for P, (8.4)PFencer.Techiqtues Performance In ans in Cruise i Mainoeuvrabiliry AGARD-R-740 Special Februof on Course Fundamnrtals FighterAircraft Design. ary l9tFi of 185)Leyriacrt. J Fundamental, FighterAircraft Designi arnd EngineIntake Afierbody In AGARD-R-740 Special Pebeuof on Course Fundamentals FighterAircraft Design. ary 1986

VortexFlap (63) Frink. N. T.: Concet(for Designing 1983. December Geometries(U) NASA WP-2233. Analysis andSuptisonic L (64) Huebnes. D Performance VortexFlapsfor the Convair Designof Wing Leadingt-Edge Universrty Washington The Thesis, George Masters F-106B. September Science. and of School Enginceertg Applied 1985. 16. Vol Flow Aerodynamiucs. I NASA CP-24 (65) Vortex 1986 Vol (66) VariesFlow Aerodyniamics. 11 NASA CP-2417. 1986 Vol (67) VortexFlowAerodynamic%. III NASA CP-2418, 1986 and at (681Lamar.J E.. Nonlinear Lift Control HfighSpeed Fow Iligh Angle of Attack UsingVortexn TechnologyIn. nf on AGARD-R-740 SpecialCourse Fundamentals Fightr (986 February Airft Design. of (69) Lan. C I-. aridIlsa. C It Effects VoriesBreak. I.ateral-Duicronal Aerodynansics downon Logitudinal and No Wingsbythe SuctionAnalogy. AIAA Paper of Slender Angust1982. 82-1385. J. (70) Lamar. I.. Analysis aridDesignof Suake-Wing 1980 Coofigurations J of Aircraft. Vol. 17.No.), Janurary pp 20-27 N. J. (7 Lamar. E.: andPrank. T.: Exprmental aridAna1) Aerodynamic Characterslytil Studyof the Lonsgitadinal Snalie-Wing ticsof Analytically andEmpirically Designed NASA TP 1803.Jane Configuaiorns at SubcriticaSpeeds. 1981 of RecentStudies of J. (72) Lanmar. E.: Summary Some Affecting TheILeading Utt Vortex andParameters Suhssonuc July 1976 re Stability. AIAA PaperNo. 76-414, F, . J (73)Rps. I) M. andCampbell, F VotclFlwMn !cnce ' V 24 TechniquesIn. Prog Aerospace agemeart pp 173-22-i.1987 of D (74) Mumn. G.-and Rao,D) M Exploratory Studies

ACKNOWLEDGEMENTS
iho The authsorirshesto thankall rallcagoes. coenbued for informnationsnovis thislecture,also.Mr. II W Carlson. Dr Dr li I Hiratch. SMr L T Nguyen. E. C. Yates. ;r.- aridmywife. Joyce. far their martyhelplul suggestions and the regarding inarpt. ws it eter P Jacobs Mrts prepiaration. Corsir Smithfar manuscript

6-26 Low ModerateHigh a a apost-stall Iligieraor 5 coU


0

00 4

ee~bbl 0 0,000. e be kd
.5000wSl obooood r-k0 Oot-d

ortexo7-1 laSpil

30

Sofi yroob~olso nynbfl W1.01 Open

10

<
O.SiO-0 ShoOkrd0
W oSMo bubbl k OOO5rb

M,

a.deg

Ref. 7.

Fig 1. Schematic of lift curve by asranges,

Fig 4. Classification of experimental data for sharpleading-edge delta wings. Lateral position of core

.... 75: Aiiached Flow LEBobbie Separaion


4~ Core strength

~65

8-

c2
0

Delta Wing

30 3Dri

3f
i
'z 0 2 4 6 /r 8 10

Vertical position
4 of core

LE shoatlr/c Voreo Fig Coredelta Fig 2 Fkc sketches

2 4 6 8 10 /

Le-Ading-edge vortex characteristics for thin

wings, %tf 0,K =1

f oderate as

Higher a

2-vortex system Fig 3 A = 65*

1-voltee system

Anlofatcs10Ref 12 Fig 6 Multiple sortices foind above wing of k-1061) during flight, all = 25, 000fi , AM 0,4, 1? =-3x0 =

Ref 8 Vurtex systems on high aspect ratio wing,

6-27 Lasersheet plane "x Cameraa Ogee wingg a=158' 0.77 C, a =30.4",L aa04 A Increasel inrea
6-s7

Ref. 13. Fig 7. Vapor screen applied to Soviet research aircraft

Fig. 10. Effect of a and delta-wing sweep on vortex system displacement.

80

cC '0CLmax
BD-Ape ,
x

c
Cs ,

deg 40

9'=8F 'v~c;d &

''

E.P

4 C,

40

50

70 60 A, deg

80

90

'

Fig II Original application of suction analogy

Ref. 16

Fig 8 Variation ofa for vortex breakdown and CL,ma. for delta wings

iA~

A= 1 13 20155 10
iS

40

A:

A 15

0 '7"7 ..

A:"a

.'W e.36 .

kKK,10

40in 7,9

kK

'_

32

X.03

Y , NA .

A.4

Rilef16 Fig 9 Leading-edge suction distributions 4iad a for departure At 0

Ref 18 Fig. 12 Variation of K, and K, wings. M = 0

i,

with A for delta

6-28
U

u
2"I 7 0 Exp

"OM fJ0, I L~e~adit-hrd' edgeo. force tg

or

force Norm al

1.2

M 1 eSIMe.e2ge -Kpol$ 1.68 22 Ci vIe 8K

M:2800 Ki4=128
0 6

force Wietlon N force Sl .edge suck i force Ref. 24

oK,,,.
C1 .40
SC

.2

a. deg

11

a, deg Ref. 16.

Fig. 16. Vortex-lift concept, suction analogy applied to LE and SE.

Fig. 13 Effect of supersonic Mach number on CL for A = 1.147 delta wing

5r

A.

to--

------

0 Experment Total Vt Puzenula to

0_-____

CLI~~~~~ _ ' "___


200

Kp

1461

~~

hit v-Vro~lLJKj.08 ortex teoms


cpeOO."

I0001ei601y (L-Sk

20

K,". .2126

eLL

8 10

"0

a. atQ.

Ref 22 Fig 14 Longitudinal load d delta, At ; 0

rib r, 1 = 1 147

Ref 24 Fig 17 At.OdYrcsAiIC characteriatICS of A gular wing, M 0

I rvctan-

, 0 65 075

A 187 107 CN

N.
16 12 vIa ,

i701 146 ,%80,


N 85'

u T.. ioryTotory
I

71
35

* ...

\1

,. I

3264r
.24

oo02 03 777t,

N~1
56-

Kv,~1/
ao,

0 1 2 3 4 5 6 K14

01

2 3 4 5 6 K14

8 16

24

32 0D

3 16
0

Z4

3.2

fief II Fig 15 Sim ilarity for computational solutions f r tbin Fig 18 Theiret delta icangs, M = 0 K14 tan /A ,= cropped delta Wings

Rtef 27 ecal of A . %alues 0

&utd Ar/e for

6-29
1 0 x
-P~2otIMthory

I.ESdO2 Pol~o5eOo , vr 0nn t (


,S

A =41S K6 = 1431 A = Kl = , 101 A.0 K,, CL 1r 08 2412

.4~A=2

A=45 0 1.0 oo

K6 =2279 K,,.=2037 KA,=0I0 -Pote

A=III 4 e

f 6r

6r.

in.:
-t M 00
0 A 3.

Lae
-.
0

A . 2.0

r
0 4

exi en

40

8 -:O-i16 24320
4 8 12 16 20 8 12 16 20 a. deg

8 16 24 32
a deg

27. Ref. 28. Fig. 22. Aerodynam,- characteristics of two sheared Fig. 19 Effect of vortex normal force on spanwise load rectangular wings, M ;-; 0. center; Mm 0

Ref.

s M 145 A.2 M4150 075

A = 0 8735

C.31

N=6 3 -04 K

2 C0 o,307

Cn -, C. we C. P Experiment

2Y

FO

Kvso . 1 466

0% LE sucion

120 AO .I ",
Cs

cpermet ,21

0 1. Pot .I
T --

. " 26

.
8

.2enA 0 764

~
I24 32 0

CLP, CL ) L 08 C

CL!t 2
.V.:

0 4

100 LE suction t

'"
.2
CL

I=-uo
';'
0
0

16 24
'deg

32

,d

22

16

,I

Ref 27 Fig 20 Vortex lift at supersonic speed

Ref 27 Fig 23 Aerodynamic characteristics ofa cropped delta wing, M = 0 6

4e4N4.C,,

Actual CL.W.e

CLv:e

631-

62 ' '

"

2,

8. 2g'

o "" (A -i

"oV

Phinar polnlisi , dt d-

4G h' a-

2,'

Kuta-Joukowski ielaitonship dFs dFs(I)= PWnet(/) rf(ldl 5 ' "Jk' /l"().teads to

K010

ig 21 "lheoretical vlues of h', and Az/e, for ,, wingo with subsonic leading edges and sonic trading edges

Ret. 28 Fig 24 A mned vortex lftdevelopment

6-30
12Z07~ Kp,1I279 K7 6*

i.

Ci..Oai -. L..,
. c t. I. .. "'CL A-3

-i
L-A .0.673Potential
0 . '6 "1

PotentialflowA C

.3t..t 2-

7
4

,
06

Section A-A

.04" Experiment ACD

16A. IM C1 ,.sn 0 a 1'

1
....

/A:1.St
X

'0466

-02 -2 0

8 10

-.1 0 .1 2 .3

.4

..

.... "-2

24

02

Ref. 31. Fig. 28. Longitudinal aerodynamic characteristics of


delta wing with conical LE flap; A = 1.333, M = 1.4,

Cleo

VORCAM.

Ref. 16 Fig. 25 Lift chr.racteristies of wings with streamwise tips, A 0.

/"i4. -

....... c
DL-

On~wg tO U.0

Vm

v-.L
.L-C

11 0 ,

*--,CV/2
4

s " ,..'
* -' 2*" Pl~ni,21
5

c'C
e oA AA

0%Lt[aOcn., Pe~Seebnon

"

q.

i 12

On. 16 20 24R Fig 29

Reffl31 %ortex action point concept, Lan

Ilg 26 Effect on

8 Ref 2 definition on estimated CL

Moderato a

ligh a

CNv

Round LE

f low sketch

Sucto

:anogy

E petelnot

ii

fEe s f

44

- CCr
.( .,

CT.,x )lCl

fnentat~

10

20

Ref 29 lig 2; Effect of a on vozt-x flow models for compley configorations, VLM-SA

Ref 29 Fig 30 Relationship between vortex normal force and residual thrust, A = 1 0, A = 76, f/c = 0 1, FFA 10.-5601 airfoil

6-31 (A) Compute ectonPaboC nosedrMg


fV. RP

(B)Compe wi somooendhV suction edge


=u

12
:I

00

l~ \Q

Experiment The0ory

Vortex

whenc, z c forms (C)Vortex

flow

Det k V M,=04 NA , 63A&

grows (0)Vortex

LEcore \-TE core

0 z1

y's 10 --

- - - ------- -- - Mach contours

0 24-

Free-vortex sheet

Captured shock

Ref. 34 Fig 31. Kulfan's method of predicting vortex normal


lorce.

Ref 43. Fig 34, Typical free vortex filament solution, A = 15


delta, Af = 0.7, a = 15".

(I - U!Je;. Vo 0,u. 0
Ouadifalk: dowh~ts Linea souces 10[rpva
Wit

F"h~

r
-

ae 0

8 --161 2 0 ,2
-

no
2 Lm/'/

.o octio

0e"$ssuI 3, Noma mach/ 4. Thickness atio i'o 5. LE radiusratio 4 Ct_. ... Note C l a* (Ct r eop 20 "

Trading/ .F

/ / she

05---

Ref, 46.

a dcg Ref 36 Fig 32 Carlson's method for predicting residual thrust, Af = 0 6, i/c = 0 12, r/c 0 0048

Fig. 35 Theoretical formulation - panel method- FVS

tvI o"taWv

2o0

,-

12-

0809 Wa~~e 0 x/' -' 875,an-et

/
.. . . . . . . . . . . . . .. .

'I

.Ref

47.

. 1 0..

tr', Free vortex fiament mothido 33

fig 36 FVS solution, A= 1 delta Al .'a0

6-32
20 ctO$

StepsE-

0to

Q.
i41t.4A

,0 We\
... Y CL

Secton A.AH

~
0

.-

- Flatwlng

FVS wing camber -----

.203040

'0 24

810

Ref. 12. Fig 37 FVS solution steps and camber representation

Ref. 56 Fig. 41. Euler solution, F-106B, M ; 0.3

Ref. 12. Fig 38. nee vortex sheet paneling for F-106B
0 Expe, tri * E~O~19
2

StiotA-A
Cor

right

20
05,1,6

I10

a
t/C

o -040

oEtervU."
4
C 12

A 0Ref. Fig

Eatet

"2

Z'K~

0 t02.304. (LE t

42

Measured and 12. calculated core locations;

Ref 65

02

e0

typical wing sectton

)JrJJIII,.],48

Ref 56 Fig 40 C-Il grid applcatin to F-106B for Euler code

Ref ,8 Fig 43 Thin layer N-S slution, A = t delta. A , 0.95 10W

0.

6-33 2 Side of fuselage 2416Initial: due to \smooth on flow Final: due to t..\ structural nts. "cst '' .8 .24 Vortex 766

8of, 0 de 0

1:

'

.20 No edge force .16 no vortex CO .12 .0

66.6-A,
Full edge force

jdeg

.4

'.6

/ .08 1 . , 2 , C .4 f 6 d No edge force probable vortex

shown) -8 .Note:Is Incidence box here for the wing -between 0.15c 0.75c & Initial chord load -16-24 ACp

n.04 ndpr

04 0

Note: Flat wing CD 0 .8 added to theory curves

X/c '1CL Ref. 59.

Ref. 59.

Fig. 44. Incidence distribution for cranked cambered


wing.

Fig 47 Edge force recovery on transonically cambered


wing; MA 0 85.

.02 05 xc0 50
04,
-

0
0 90

01 03

LEvortex on lower surface


.06 sIn a
2

or attached flow .09 76.6

. 2

Y=0.15.045

it 75 =0 ,

CA .03 04 "0 -.05 10 110 a, deg

0 .1 -

w s

neLE

66.6 vortex on upper surface

nC c IIl camber with final wing box incidence Final designed camber elevai0on 4 Same Ref. 59,

I 15 175

Fig 45. Cranked wing mean camber shapes; A = 1 383, = 0 9, a=07 CL, = 0 5,

Ref 59. Fig 48. Effect of flow type on transonically cambered wing, A = 0 85. - 9M

Data D, I 4

*CL o mfia at ++ 13,0 4+

04 8 121620 ' *P1&" j, Ctan 0 *Cd,


24

co ''01" .. ..o-

$13 b2

'0

Ca.snbd

ngdaa
.

0
50IL18 C,

3
E1

0 a.

Ref

59

Ref 60 Fig 46 Longitud.ual aerodynamic characteristics for Fig 19 Restricted-area-design camber surface for 60'designed transonicmaneuver wing, A = 1 383, sAept trapezoidal-wing fighter, CL.d = 0 73, = 050, Al4 = 090, ALE = 76 6'/66 6', CL, C,, = -0 17, M = 0 5, Rn = 2 9 x 106 VLM-SA

6-34

D-9.

L.hdo

___ 0A.0.~~al

c A~ytW

0001808

F g0Fit ing 5

ffap

s r ac st

w g ds n

sufa e

e.0 12.

Fig.ig

surfacesc FVtring

530

flap hisapy

aoewng-dyigamurace

,1 'L

~jE )
C, C,,,

OE

TEf

C C

3o

20~
1 -I5

Ta

lhpewt d101 aa

2 00 5 101i5n

04 4 20 2 A 6 8

Ref 60 Fig 51 Theoretical and experimental data for GO*- Fig 54. F-106B with leading-edge vortex flap in flight swept trapezoidal wing fighter; Al = 0 3, R= 19 X 10 6, 1L*_2-'l 6/20, = 15*/12*

, 1 Region Plain wing .Vortes suction utliue


NM3nsaisly for lift increase

Region 2

Region 3

Vortex flap -Provides a thrust componeift for drag reduction

X(3)
1

L
X

()

XX

wiiotz ex fla Id

V~lfowild$ with

ad comparing th.%vrtex

'-0

WIS*X flap

X~l
greatly exaggerated

X()8arcal angle of attack

tan(E

Ref 62 Fig 52 The vortex flap concept

Fig 55 Analytical Nortex flap msodel with design vari. ahles

6-35 1.0 design 6et metiaoes Update n variales desg 12

--

8 6

10 8

-0 A.nalss . Pertudeign 40.6 01

2FO1&L50

6LE

nals

ign

r Yes7 WV.01 do 506 dogu022304VO

CA

FlCalde26gn(VF.D4)

0.16804

114deg 102deg

2.o imes largerithan the horIzontal scale

1=

IiA~tCA t

.2 r
*o

It

Ref 64, Fig 57 lnitr and final vortex flap design results for the F-106B3; M* 1 5, CL,4 = 0 223
2. lage tha tlrne

.. ang

,
sang

Ref. 69 Fig. 60 Longitudinal aerodynamic charactertcs of a Fina12s delta w ith round LE; A = 2 31, A 0

3
/o"

7
2

~
2T 7

C-1.4..2

5oe
4

a t Is 6

./e

/le

14

'6D

u,
A
16

32

L/D 4

Baseline
--

3/----...VF-D4,

2 1
0

...

LE: 10~ VF-DO1, 8LE =18.47 VF'D4, LE =3O

-c~
I'

i.~

""
16j
.

c .5..o n,,oI
1557
*

5,.6 Ks./nIC11IS

.1 .2 .3 .4 .5 CL,d CL 1

.6 .7

ct.

:5z...

P'

hi

'.'

Ref 64 Fig 58. Performance of VP-DOI with VF-D4 on F-I06B at Al = 1 5

Ref 16 Fig 61 Liti characterscs of pointed

ags.

6-36 A -_.+
--- 0en --vo01 0 .Lo~tIf P ex01y K =2011 : K 0 2031 d Kv $.=0392 =0.,0 00

i$6

Po t (LE O 221 p

ug4.fK A 2.4A 443 , P; e Kh Xo=0.2 r 2am

12
C0

M..eO4

M.O

12 . ACLLse

12

0'~

i 4 W
0 0

~
1 24
M~ deg.

t,
0

.4

24 a.

30 0

24 fte

Re
Ref. 16.

8 16 odeg

24

Ref 28. Fig 62. Effect of Mach number on augmented vortex lift and p'Ich; diamond wing, A = 0 843, A = 74*

Fig. 65. Suction analogy applications to configurations; M =03

16
- -

P 1 1t~t4 -01.. ty *160 012066 PLE 1vonux o102h heIM1 o0y,

.A / :,

."

P10otal (LE Agleol.) UxP0vo Cop-Okit"v 8

.-

14....

ory

3L

04 C ,. .. .

24

20 U

16 22

.0

08

62

Ref 68

Ref 28, Fig 63 Effect of Mach nu mber on augmented wrtex lift and pitch. arrow wing, A = 1 463, A =74'
-c432 4

Fig 66. Complete longitudinal aerodynamic characteristics for strakewing-body, strake AD 24, A =44*, Af = 0 2

~~0122440

48.1

40

0 1

24

24

1521

.0

VI5

16 2D

Ref 70
F.g 64 Longitudinal ae8odynamc characteristics for , cranked wing, A = ac1 , A = on80n/65r amt 0, VLM-SA 4M

Ref 68
Fig 67 Component longitudinal aerod)namic characteristics for strake-wing-body, strak AD 24, A = 440 =0 2

6-37

S-- - ---

2C
C,

Experiment OVLM-SA
C;.

417c,-(f

76*

1 2 P01

8004
04 -04 0 10 20 30 40 o deg
L

I'

00
-008
0-0120 -

004
->

CC
-

016

0 20 30 40 .deg

0 10 20 30 40 .. deg

Ref 72. Fig 68 Cambered-thick lifting-body and wing combination.

Ref 69. Fig 71 Stability derivatives for cropped delta, A=0333,A=05,#=5, MmO.

..... po iy Eapeeet PoL. wepy g (iL.ESF)yoa, C

i. 04

'

- ,V

0K3

Experiment Unsteady SA Q07 , . O)b/2U

68%

1.2-

,body .08 .04 Iegi

/
...
.4

./f'&i *.0.75 >1.

-L

?,d u5,-~ -80* 16 24 32 8


4

..8
112

Lz
0 10 20 a, deg 30 40 Ref 38 Fig 72 Oscillatory A =075, A 0

Ref 72
Fig 69 Longitudinal aerodnamic characteristics for cambered thick.body. ing combination, f = 0 2

toll damping for gothic wing,

M '

A=2 v.45* 0333 (Biconvel O.eclion

Exp e me n t Potential flow contribuion Potenial , LE vortex , SEvortex .ugrnened flowcontribuions Kp= 3558, Kvje. 0 957, K,$,. 1 125 0 338 kv O1 .. 034k
0 2
-

- - (Po i

nul, n " ) li

.. 004r 0 0. Aiearn

.0.ci d Dp lp

B.Sia $p a8 t a. 00135

-012
_4C

Ci.

-2-

, 8 16 24 32-4 a. deg 0 4 80 6,. deg

4
.

deg

12

4 Ct

16 4 0 C,deg

Ief 16
Fg 70 Aerodynamic cha&taensti:s ofa cropped delta Fig 73

Ref 68
Roll-control device effecti%enss on cropped

-ing body, ,I = 1 2

delta wing, Af = 0 2

6-38

Moderate a Ref 75 Fig. 77 Fighter design evolution.

High a Ref 73 Fig 74 Upper vortex flap concept

Military Airplanes *Stability:

A
Pich up
nlghl /R'

A
al high 1,1\
PRch-dOw f 98

C Cn

Pdyn

cos (I- (IzIl x ) Cl

sin

>0

Control

l Aileron alone.

cR

hl

LCDP- Cn 1

CIII

,a

,oilVa

A;leron + rudder proporliona! to aileron. Ref 73 LCDP K2 Cnill cr 'Sa


- C IO

Cn6a on

K2C16

>0

Fig 75 Ilypothesized application of segmented up. per sortex flaps for drag reduction and aerodynamic control at high angles of attack, filled segment denotes deployed

Ref 76 Fig 78 Criteria for elimination of directional diver. genre for fighter airplanes, militar) airplanes

r Available
lRequired

Yaw control

Maximum lilt
Angle of attack
Ref 74 Fig 76 Typical yaw control requirements for maneuvermg

LE

Ref 77 Fig 79 lligh. aero deselopment process

6-39 Actuated strake CN empirral theory Simple Experment -----Jet blow ,ng 25 r --Cm theory empincal Simple Experment

Tangential slot blowing

ON C N. 10 or 0 Cm 5

Jet suction 77

0 tO1 20 30 40 50 60 70 90 90 otdeg Ref. 83 Fig. 83 Aerodynamic characteristics during pitching motions, A = I delta, Al so 0, K = 0 04.

Ref Fig. 80. Forebody flow control concepts

0'/
20 15l Fomebody strake 60..,Rotating Forward Intake (F-15)

M-5
Rotating Cowl Lip

Yawing

moment 05

to
0

L Rudder
-recovery

5S.9

to,
105 ........ MPressure9

Ope

"-.-5-'t- L J- J- -- L 0 10 20 30 40 50 60700 805078 Angle of attack, deg Ref 77

Presur
Auxiliary Door Ref 85

Close a. deg

Fig 81 Generic model results

Fig. 81. high a special inet devices

100

~Fr-v

. veio"' 80xis

-30

CONTROL EFFEC OR rudder 1-0--10 vector g maximum power yaw

,t

92 -of

04e 6A Yawing moment coefficient I% dealized aero yaw control y oto .1~ 0 0,l

0 hocOoopV bmsbee 040" p xr'

0 10 20 30 40 50 Angle of attack deg

20

60

100120 r0

Ref 77
Low-speed )aw 82 Fig alt = 20, 000't , M = 0 3 control effectiveness, Ref 77 Fig 85 Cowl lip droop concept

7-1

AIRCRAFT DRAG ANALYSIS METHODS by Charles W. Boppe Grumman Corporation Aircraft Systems Division Bethpage, NY 11714 ABSTRACT and analysis methods A collection of aircraft computatonal drag for AGARD been drag reduction techniques has preparea the on Special Course "Engineeng Methods Fluid Dynamics Panel Design of Aircraft "Pressure. in Aerodynamic Analysis and skin friction (viscous), wave (compressibility). lift-induced interference (multiple coponents, multiple flow (vortex), mm plume), and exhaust fields), throttle-dependent (inlet and predictions amincluded Background information source drag is empirical data o complrmenucty handbook shemes and prediction acomputational drag to provided The need establish and is experience base emphasized illustrated Project type drag in describedwhich these prediction tools applicatons are The reduction processes paper have been implemented for drag by by concludessarnsaiing therole played computerized drag in predicton methodsaircraft design programs NOMENCLATURE Vt/Vo P q Xl. X2 MFR Vol R Log In p At. An A A eff p o/l -

Required Inlet Velocity Ratio Force Plate Dynamic Pressure Length Scales Inlet Mass Flow Ratio Volume Correction Factor Logarithm Natural Logarthm Density Founer Coefficients Wing Sveep Angle Wing Taper Ratio Effective Non-Dimensional Axial Location Boat Tail Angle H1eight National Advisory Council on Aeronautics Single Line Over.Relaxation Boundary Layer Displacement Thickness Thickness-to-chord ratio irag Free-stream velocity Local Chord Length Chord Length Average Profile Drag Coefficient Leading Edge Shockwave Angle Ratio of Specific Ileats Surface Local Orientation Angle

10

Crc
V d Cp I' C y L )1 C 1 a M Pcn Re Ct) CL C CD, Ss AR &tE GAW I/d NPR CDo C T A.S L T b V V 0 V2/V0
-

Cl,

lncompreasible Frncuon Coefficient Compressible Friction Coeffcient Velocity Denvuve Pressure Coeffic-ent Circulation Load Distribution (CCUICavgl. Chord Span Position Lift Induced Drag Local Coefficient Lift al oftack , Anglof-aVelci Downyash Velocity Mach Number Critical Pressure Reynolds Number Coefficient Drag Lift Coefficient Lift-Induced Drag Efficiency Facto Coefficient Lift-Induced Drag Suction Parameter Aspect Ratio Edge L~eading Flap Deflection flap Trsting Edge Deflection Deflection Angle Canard General Aviation Wing Length/Djameter Ratio Ratio Nozzle Pressure Zcro-Lift Drag Coefficient Chord Centirline Temperatue Area Length Location y/W2 Non Dimensional Span Wing Span Calculated Open Inlet Velocity Ratio Calculated Closed-Inlet Velocity Rato

1 NACA SLOR 8. t/c D U. CLOC CAVO CDp LE 0 e

IN'RODUCTION Aircraft design has evolved over thepast century into aprocess requiring increasing levels of sophitstication to meet rcquirements for expanded speed/alutude envelopes and combat flexibility with improved cruise efficiency and portion of this peniod. Over maneuvering perfoemance a large engineering practice designers combined good inventors and testing with sub-scale prirans to develop avehicle for fulland flight scale evaluation Aviation history reveals successes for most pan deterined bythe quality of failures were the that the vehicle engineering design effort. Following close behind baiic stability, control, and handling qualities, performance derived by naxiin gthrust and a or minimraing aerodynamic drag often makes breaksnew a forces poses concept Prediction of aerodynarnic drag design foridable challenge Elemental flow physics driving viscous and pressure resistance components canbe quite complex. In akhton. thre is anextraordinary number of ways in which the total combine to produce the drag can elements interact and I-I forceReferencesI provide useful insights into the Full-scale aircraft drag predicton problems complexity of drag this predction errors of 10-20% have occured in thepast, is needed for success within therange often not

-L

7-2 One problem facedby the designer is thatthere isa large gap between conceptual/prelmnary design "handbook" methods typically usedto rough out a newconfiguration, andthe subthe s~ale model testing typically usedtogenerate first true performance esurtes This gapexists in theform of modeling fidelity andtime. so it is usually beneficial if certain key configuraton characteristcs are locked-in before suitable not tesnngandinterpretation efforts arecompleted It is %ell known, however, thatsub-scale testing in wind tunnels can occasionally mislead thedesigner In thesecases, problem canusually be the traced back to testing anomalies caused wall interference, by flaws in simulatng viscous effects, model geometry fidelity, test procedure errors. support inierferencelfoulingPT and computenzed flow simulation methods Over the past20 years. haveevolved Applications on aircraft programs havefor the mostpart focused on the prediction of pressureAclocity and characteristcs, lifting forces andmoments, boundary layer parameters thesemethods canalso provide drag force But predictons If the applications engineer is careful in modeling, andcantake advantage of a computational prediction expenence possible On onelevel, computational base, multiple benefits are beteen simplistic handbook predictions serveto bridge the gap testing results Whenanomalies methods and tal model between sub- andfull-scale testing ace observed. computational predictions can beusedto judge which is correct (a third source)Finally. computational methods provide a means for the design engineer to better understand flow mchanisms that the generate drag forcesThis is particularly valuable for appliations thatradically departfrom past designexperience This paper describes thecurrent practice usedincomputational dragprediction for different types aircraft dragsourcesThe of examples included should pro.de a foundation or experience base could prove useful ir thai lure applications BACKGROUND .Jndetstansding various sores of aircraft dragbecomes the important sincethedesigner orapplications engineer must select the proper computational tool for thejob at hand At present, no single method is capable of simultaneouslv treating all drag components thataretypically of interest In addition, no single methodis capable of treating a complete aucraft configuration with sufficient accuracy for all drag analyses mighi bc that required In view of this. it is reasonable expect thata number to of methods%illbe implemented with theresultsbeingconbined using a component build up approach Toensure there no that is rmsunderstanding aboutthecharacter different drag sources, of a portion of the Background sectionstll be setaside for descriptions anddefinitions Drag Sources Viscous on skin friction drag isdenied from theflo, field sheang stresses a region of reduced in velocity neartheaircraft surface Theresulting zone slocity impaurent, or boundary of layer. hasbeen sketched fig I Total pressure in losses ariatbuted th,s mechanism can be measured thewake to drag in downstreama from thecomponent of interest Viscous drag can beaffected byaltentig theboundar)laer flow characteristics this ts usually accomplished by o.ampulanng theexternal flb. field pressure gradients The presure environment ought also be amenable propagating a luar boundary layer thatis thinner to reduction vGos than aturbulent layer with an attendant drag Also. turbulent boundary layer viscous draglevels might bereduced evenapproach zero local regions if the on in boundary layer surface velocity andvelocity gradicnt are very sull (CF - 0). but this is strictly trueonly fo two-dimensional flows It is possible for the boundary layer velocity to be negiuve When this happens, resulting separated regiu, the 1kw drag It (Fig 1) gives rise to another drag component pressure the should berecognized thatviscous drag is unique in thatit is only drag sourcefor which the force-generating mechanism acts are angental thesurface All other drag sources derived via to the integrated effect of normal pressures fressure o form drag is a "normal-pressure" type drag - the origins of which can be tracedback to multiple sourcesIn the simplest case,pressure drag (or thrust) is generated time the any normal pressure integration is non-zero Figure 2-A depicts the mostcommon occurrence using symmetmrc a body-like surface
-

, Fig 1 Boundary Layer Velocity Characterlstlcs with Wake Total Pressure

-" +y

DRAG INTEGRAND +IO A ) (ATTACHED FLOW)

DA DR + I() LOSS SYMMETRY TO OF DUE AFTFLOW SEPARATION k"14124_


_ _

Fig. 2 Body Pressure Field With Drag Integrand

7-3 in a uniform onsetflow. The resulting syemnetmc pressure distribution (noseto tail) is shown below the body. A third vectors acting on thesurface. illustration shows the pressure Below that, the integrand to be sunmed to quantify thedrag or the force can be seenIn this symmetric flow case, pressure form drag foree is ero because forward and aft eotponents the exactly cancel eachother Aircraft flows of this type do not exist a for most applications. Typically, for body wing shapes and flow separaion region will develop aft, asdepicted in Fig. 2-B. and symmety is compromised The result isthat pressure afterbody pressure recovery levels arewe,. ,.ed with an attendant drag increase. Form dragbcorne more complex at highspeeds at lifting conditions asother drag mechanisms and field. interact andaffect the normal pressure Lift.induced drag is denved from theproduction of lifting forces, predominantly on thewing andtailcanard surfacesAny surface with positive lift (including body forms) will be Wake vortical flow alters or induces flow velocities on the lifting surface Most important, an upwash/downwash field can be identified (Fig 4) When fully integrated, a net downwash exists thatcombines with thefree-stream velocity. The resultant onset flow that thelifting surface "sees" is rotated, andthe lift vector rotates with it The component of the lift vector facing aft forms the induced drag force Swirling flow carries energy downstream is the lifting surface wake.

--

on characterized bylower pressurestheupper surface than on


thelower surface. As a result. lower surface flow tendsto move outboard toward the surface rip, while upper surface flow moves inboard toward the centerline (Fig. 3). This flow mechanism ts thesimple result of flow migrating from a high.pressure region crossto a low.pressure region. At theendof the surface, these flow velocities from the upper andlower surfaces combine with flow that is particularly thefrestrcam flow to form a vortical strong nearthe surface tips or outboard regions Isis well known thatthe vortical flow character is dependent the lifting on surface load distribotion

L r " i
-

Tjw

Fig. 4 Induced UpwaslDownwash Field wilt Rotated Uft Vector

00

Wave dragdevelops as a result of differences in the flows A oneand compressibttly of air in subsonic supersonic dimensional isentropic flow model reveals that a streanituhe flow contraction will accelerate moving atsubsonic speeds, while atsupersonic speeds flow wil bedecelerated The the
opposite is truiefor a ssreamiiibe expansion, ie , a subsonic flow

up sill slow, ,hitlea supersonic flow wili speed in anexpanding sucamube Airflows about aircraft wing andbody components form streuantubes with combiations of contractions and

expansions
At high subsosie speeds (beyond M - 0 7 0 8. depending on theconfiguration thickness andlift level), flow expansions might create supersonic flow "bubble' embedded in the a subsonic flow If thebubble flow gains sufficient supersonic speed, flow discontinuity (or shock wave) will permit flow a parameters return to free-strean values. Through a shock to velocity vector, density, and wave. flow properties (pressure. abruptly with an accompanying loss of total temperature) change a pressure downstream (Fig 5-A) At supersonic flight speeds, wing or body component will exhibit an ay of shockwaves and expansion or Mach waves (Fig 5.B) Shock wavesandthe

r(A

can aircrafi wake extend farinto the flow resultant effect.on the

Fig 6 swhere field about theaucraftTis extent is depicted us vapor condensation about an F.14 flying at supersonic speeds can be seenFigure 7 reveals thatshock wave flow discoinnuites ranbe large enough to disturs a water surface some distance from the ircraft The dragmechanisms discussed up to this point am i undanental in 'he sense theycannot be broken down into simple that elementsA number of drag sourcis, however, are derived from wribination of fundamental dragmechanisms lnterfer,,-e throtledependent. andtrm dragareexamples

Y Fig. 3 Lifting Surface Uppor!Lower Flow Pattern with Load Distribution

7-4 PTOT

M 0 75, CoS- 0 009 p'[


S OC K

M 080, C0 s 0020 5

C P

M 0 86, 0s' 0 130 C DEVELOPMENT WAVE & (A) BUBBLE SHOCK SPEEDS ATSUBSONIC Fig. 6 Extent of F-14 Shock Wave Pattern at Supersonic Speeds

//

EXPA.NSION ,, Y WAVES, WAVE A

.05 CP0

-05 (B) COMPRESSION (SHOCK) & EXPANSION (MACH) WAVES AT SUPERSONIC SPEEDS Fig. 5 Sources of Wave Drag at Subsonic & Supersonic Speeds (Ret. 16)

lInerference dragtakes myriad forms The natureof the interference might beof a component or flow field type. but in all eases. flow velocity disturbance is responsible for thedrag a force An example of a pureflow feld dstutxnce would be that of a wing immcrsed slipstream of a propeller. Dynanuc in the prwsure would be increased theslipstrcam wake. In addition, In a swirling flow generting, n upwash on onesideof the hub axis ad adownwash on the oter alters winp loading Ai veryhigh subsonic speeds propfass), this interference effect wdl (i.e dratmaiucally change the wing shock wavepattern It can be appreci ted byusing this example that interference drag can easily haveviscous, prsssUe, Ifi-indaced. andwavedrag components, this greatly complicates the prediction process

M. Fig. 7 Interactlen of Aircraft Shock Wave with Ocean Surface

7-5 More common examples of interference drag canbe found wi h interfering multiple body or wing and body components. Figure 8 shows the viscous flow near a component surface where a pod a has other body form tmposed flow disturbance or some gradients and composed of both fa,iable unfavorable pressure a The effect is weakened viscous layer thatincreases the probability of flow separation. Here, both viscous andpressure drag levels wll boaffected. Wave andlift-induced drag variations will occur if this type of interference exists on a lifting surface at high speeds Figure 8 provides another example of interference drag. Aurcraft a problems because boundary component junctures often present layer flow along onesurface, suchasa fuselage, is often poorly velocity) that conditioned to dealwith a stagnation point (zero surfac, juncture (e g. a wing or might beimposed bya second tail surface leading edge) Atthis secor surface stagnation if point, theflow is liktly to separate proper fairings havenot patch is onepossibility, but beenimplementeJ A separation is a juncture vortex, Interference another potennal phenomenon the drag levels areoften reduced byproper fairings or fillets but designer must minimie the useof thesesurfaces considering th, with additonal wettedarea drag penalty associated by generated disturbances Throtle-dependent drag is nozzles. predominantly nearthe inlet faceandexhaust the Considering inlets,engine flow ratewill deterne the level establishes the level of suction of inlet spillage. This inturn forces on the islet face which may not benegligible for thick inlets with largenose-radii High levels of spillage might induce This inlet flow separation. problem is aggravated %hen design requirements dictate very small inlet eading-edge radii In addition, spilage flow will interfere with wing circulation if the nore inlet faceis neara wing sunfe Nozzle flow fields am complicated Figure 9 is a schematic sho, Iag theelements of a typical nozzle flow Key hereisthe plume entrament region andpotential flow separation region at theboaral trailing edge and Tl'sust variations will alter the external flow entramnment a flow possibly the separated region At high speeds, shock region with throttle changes wave may exist on the boattail caustng theshock to mgrate forward andaft Mechanisms that of levels are important because the alter afterbody pressure relatively large surface vector component tn theaxial (or drag) direction to need keep theaircraft in Trm dragevolves from the eqtblirium during cruise andmaneuvering flight It should be a that apparent a timming surface will always generate at component of lift-induced drag,but high-speed conditions, For a conventional tailwavedrag penalties might also appear a generates downto-mm design (Fig 10-A), thetail surface a load requiring the wing to produce addinonal up-load for given a total lift level The increased lift might result in measurable conditions if the untrimmed wavedrag increment at tcansonic isolated wing wasdesigned to be optimum at the total lift level above design lift level the Trimmed, thewing must now operate a For a canard-to-trm configuration (Fig 10-B), positive load to the tmm might eliminate this penalty, but designer must ensure not thatthe canard downwash field does impair loading on the does, mmnung drag the main lifting surface downstream. If it might include both lift-indaced andwave drag components Computational Prediction Problems are It was notedin the Introduction that there a number of to testing is factors thatmust tt accounted for if sub-scale

SHOCK WAVE PRESSURE GRADIENTS

BOUNDARY LAYER

..---

li

NI

A'AINMENT

JETPLUME 8OUNDARY SEPARATED REGION JETEXHAUST

PROFILE

Fig. 9 Elements of Complex Nozzle/Plume Interactlon Flow Fields

VORTICES NE,;KL&CE

SEPASSATION PO-NT

TAIL (A)AJFT CONFIGURATION DIVIDING

STRSEAMLINE

STRABELET
wuiui.

CONFIGURATION (8)CANARD Conventilonal Aft-Tall & Fig. 10 Balance & Trim for Canard Configurations

Effects Fig. 8 Examples of Component Interference

7-6 provide useful dragmeasurements performance estimation for The test engineer must also monitor power input durtng powered tests because drag measurement errors combine propulsion-type with power input measirement errors to yield a total uncertainty in resultant drag forces Flight testing presents similar problems in thattheengineer intent on establishing vehicle drag levels as often haslesscontrol over flight condition variables such angle-of-attack, and true engine thrust level thanhe speed, would in the wind-tunnel Flight condition parameter accuracy rs another important issue.Like sub- andfull-scale testing, the determination of drag forces via computational methods presents a number of difficulties. Soe of the more important problems that are described in the paragraphs follow Recalling Fig 2-A, it canbe appreciated that drag forces are predominantly established by summing the effects of normal pressure fields The nose and tail portions of the body or wing component contribute disproportionately to drag because of normal vector onentation. That is tosay, pressure anomalies in the mid-section often register hide effect on drag, while disturbances forward and aft can have surprisingly large effects But numerically, this characteristic magnifies another problem Considering the mtegiand shown in Fig 2-A. the drag force could be charicterized asa relatively small paraenter computed by taking difference between two larger paruraeters, the the integrated force dominating either end of the configuration is considerable room for error in this process because the There high gradient nose andtail regions are often compromised by modeling resolution constraints inherent in the computational scheme employed In addition.complex physical flow phenomenology characterizing these regions is approximated, to some extent, by flow simulation methods in use toda) Finally, small flaws grd orsurface in modeling the at nose tail or can generate numerical anomalies that register a s..able error in the drag level while revealing no apparent discrepancy in lift and moment characteristics It be appreciated why the can developers of computational nethods rarely describe the data correlation of computational drag predictions with test Table I Typical Excrescence Drag ANTENNAS (EXTERIOR) I BLADE (APR.27) 10.32 2 IN TACAN) IN2 44 (ANIAPX73) AS1918'AR 2 BLADE 2 I BLADE (F-111) IN I- 301 32 DECM (F.14) POD I ALO-XXX 2 PDS 4 BLADE 8 IN EACH ALJWING PODS(F-111) 2 ECM LIGHTS PROBES & 2 PILOT STATIC PROBES TEMP PROBES 2 TOTAL 1 A O-ATRANSMITTER 24 STATIC DISCHARGE PROBES 24 BAL CNOSCLHAG PROBES I NAVIGATION LIGHT I ANTI COLLISION LIGHT MISCELLANEOUS 1 WINDSHIELD REMOVAL RAIN ACCESS DOORHINGES I ARRESTING HOOK OPENINGS POD -INC IN DECH DUMP I FUEL t BlLEEDDRAINS - 4 5IN VALVEIN 2 I ENGINE 8WATER/FUEL DRAINS - 51 IN DIA 1,6 5 FUEL CELLVENTS (FUSEL XGE) 2 REFUELING SUMP DRAINS 2 ECS GROUND COOLING LOUVERS @ 2 OI.BREATHERS 14HOLES 3 COCKPIT SAFETY, GAS- GASPURGE GUN I AMMO VENT, i-COCKPIT EXH 2 OIL COOLER, 2 ECS EXH 2 HYDOIL COOLER SCOOPS 2 ENGINE IDGOILCOOLER & & LOUVER I EPUINTAKE EXH s I APUINTAKEEXH LOUVER EXCHANGER AIR 2 BLEED HEAT --- ----I,

Another problem associated with computational drag prediction deals with extraordinary differences in scale thai characterize the aircraft components Table I summarizes what often identified is as elcmants of a configuration's excrescence drag These numerous small vents, drains, pobes and antennas %ould require computational modeling resolution oeders-of-tmagntaue smaller than that currently in practice tday Simultanouisly modelng the global and detailed elements of complete aircraft would beimptactical given current computer technolog) and charging algorithms Compeunding this issue is the fact that all rmco physical phenomenon responsible for drag are not fully understood In view of this and the point made earlier (that no single computational method is currendy capable of treaning all drag components for a complete configuration) it becoties apparent why the applications engincer is often able topredict drag increments of decrements using computational methods, but the determination of absolute drag levels is not possible

Aircraft Deelopns'nt Process & Computational Method T)prs The aircraft development process (Fig II) involves a sequence of steps or design phases that is initiated by a customer specification for mission requirements and point design goals In Conceptual Design, the configuration is "roughed out" using relatively simple tools techniques The emphasis here is and usually on rapid tz-n-around Ilandbok methods, experience bases, and the simplest analysis methods are typically brought to bear Far Preliminary Design, the engineering team must firm up the aircraft external contours This stage requires both windtunnel testing and computationl simulations as flow eades in the opoimization process. The computations and sub-scale testing (both simulations) attempt to reproduce what will eventually happen in flight The objective is to mino ize design risk because anomalies idenutfied during a flight test program are often quite cxpensive to resolve The next stage, Detailed Design, is characterized by the engineering team packing the aircraft interior spaces with actuatoi, h)draulics, crew station, propulsots, etc In the final stage, one or two aircraft are manufactured for flight esaluaoons At this point in time, computational flow methods can play an important role in resolving flight test anomalies because it often impractical to is re enter the winrd-tunnel on sety short nooce The engineers task of selecting the proper method for a part.ula r stage of development and a specfi design objcuve is exaaoirdinarty complex lie must have understavding of an

This probleni of scale and current computing hardware limitations was underscored by NASA Aims researchers (Ref 12) who computed the smallest eddies found in a turbulent channel flow ata Reynolds number of 10.000 using ANavier Stokes formulation Fifty billion gad points %tcr required for an analysis that reached a steady state after une steps This 2(0 can be compared to a typical Reynolds-Averagd Navicr-Stokes analysis currently applied for aircraft appltcaonu where thegrad system point count might range between ,000 300.00 and points The time step count might be 5W to 100

7-7

PART 1: DRAG ANALYSIS METHODS


MDISCUSSION
_____

DRAG ANALYSIS METHODS

An engieering perspectve of computauonal drag prediction methods is now described with an attempt to identify the earliest known progress in the field andtransition to techniques usedon aircraft projects This composition is not intended to represent nor is it COMPUTATIONAL all of themethods or opimum techniques intended tob represent the vey currently available, best that oight nted the implenr Instead, methods are,for themost pan. those had with whit h the author has someproject-type experience and nih noapiaiiy into some canprovide isight apphcabihty. [ Drag Prediction Pioneers Drag Equations

CONCETUAL DESIGN (WEIGiIT iNTEGRATED SIZING' D WDOW SM E PW

TSIMULATIONS

&cnpoiesm CA/CM DETAILED DESIGN CMANUFACTURING & MCOMPUTATIONAL FLIGHT TESTING-FLOW

as Many well-known investigators, going asfar back Newton (1642-1727). have madecontribuions to knowledge that is fondational to curr'ent dragprediction tecl'niqucs (Ref 13) particularly noteworthy for engineering Two individuals who arc applications are Smeaton and Oswald FIg. 11 Aircraft Development ProcessaplctosremeisndOwd -SIMULATIONS

~TION

addition to to thephysical flow that is besimulated in of appreciasng thestrengths weaknesses thecandidate and identifird andthen Table 2)is category (see methodsThe best code within that category mustbe properly theproper implemnted Experinendirecaics thattheslection of a more sophisticated code within a category or theselection of a code in improve theflow simulation obtained a higher category maynot Trhis beatirbeted to algort'hm fortulation characteristics can that to andmodeling constraints Caution is necessary ensure more prediction accidents do not occur As codesbecome complex, probability of having a prediction accident the and increases thetrouble free "usage range'typically &=ases Computer resource requirements might alsobea factor in the seven levels of methodselection piecessTable2 illustuites analysis complexity andexpenseRelatve computng costs between top andbottom entries might range bcten I and the 14,000

work ohn Snicaton wasan English c':perimentalist whose focused on improving the efficieny of windmill andwaterwheel at blades technique His involved rotrg tess specimens theend of a 6-foot armandmeasuring theresultant forces For a flat plate oriented perpendicular to the onsetflow, Smeaton s 1759 testing resulted thefollowing formula in P.00049 V S
2

(I)

in where wasthe plate force. "V"was theairspeed nuiles per "P" feet,andthe factor hour. "S"wasthe plate surface aea in square (0 0049) came ho known "Smeatonis fficient" is later to as Co literature references. With a simple cigonometnic relation, both lift anddrift (theoriginal termnnology for drag useduntil theend for of the19th century) ould be computed glider wingsan propeller blades IIP'-T2P con ina 0< a< 10deg Iv sm2c, Dl -2P t~n a 2 I smn (DRA )
2 s

(2) (3)

Table 2 Computational Method Formulation Types CATEGORY FORMULATION TYPE I 2 3

O<e i

<10deg

CHARACTERISTICS

s Smeaton work is imporiant sinceit marks the initial effori to actually compute drag forces Also, historically, we find the firsi useof an empirical factor into which all te world's uncertainties still andunknowns coul be grould, an approach thatis in practice today advantage of Smeaton s Ldiienthars work with gliders took Coefficient Formula. bit ilienthal believed thatSnaton's a haveavalue of 0 0055 The Wright brothers were bit should moremeticulous in their work theydeduced aSmeaton Coefficient of 0 0033 from wind-tunnel testing andglider flights. This value wasusedfor designing the Wright fyer is wing, canard, andpropeller blades It now well known that the drag on a flat plateoriented normal the flow will d--rd on the Reynolds number At high speeds. Eiffel conclud- CD that 128 wascorrect, andusing thefelation D-q Cd S (4)

NEWTONIAN PRESSURE "POINr PRESSURE LAW EQUATION LAPLACE'S EOUATI LIN/EAR TRANSONIC SMALL PERTURBATION EOUATION NON LINEAR (PIAUAR C) B

AS WTH EXTENDED TRANSONIC SAME ABOVE SWEPT SHOCK WAVE SMALL PERTURBATION EQUATION MODELING FULL POTENTIAL EQUATION NON-LINEAL NONPLANAR C 8 (TYPICALLY REQUIRES CONFORMAL GRID) ALL ABOVE PLUS (NO VORTICITY FLOW POTENTIAL ASSUMPTION) COMPLETE

EOUAT*NS EULERS

a moe exact coefficient of 0 00327 is obtained, underscoring the WVrights skil yewts.Smeaton's Coefficient varied Over a period of 150 sem large. between 0 00550 and0 00327 - a rang-that nught complex moders applications, variations on this in but some I el can still ocrur whentheparameter interest drag e of is

HAVIER-STOKES REPRESENTATION OF EQUATIONS q REYNOLDS AVERAGEPHYSICAL FLOW VISCOSITY INCLUDING NS &TURBULENCE N PARABOLIZEDS

7-S

A similar evolution for the prediction of induced drag can be Prandl sketched theconnibuton of both Lanchester and noting
that But it is most important to recognize the engineering

greatly simplified byNACA prediction of induced drag was


engineer W. Bailey Oswald Focusing his work on providing a to means estimate atrcraft performance in thelate 1929rs, Oswald (Ref 14) established thedrag polar relation that is used to this day CD- Cti, + (/ AR e) (5)

"CF" where istheturbulent flat-plate skin fnction coefficient, "L" the airfoil location for maximum thickness, "tic" is the is wing section thickness ratio, "F"isa lifting surface correction areas "S' factor (table look-up), and .epresents surface
usedto build up drag t~umatcs for Similar expressions canbe bodies, wing-body combinations, andwing-body-tail configurations with power effects andcontrol surface handbook techniques is deflections The value of these comparabl. to mathemaucal estimatng techniques that are applied when using a desk calculator. It is useful to have some in approximation of theparameter order to pick out errors Ipeatgmore complex orautomated schemes in

This "airplane efficiencey (e)apphiedinduced is factor" to drag ts drag comparable to Smeatons Coefficient for pressure
These early pressure lift-induced dragrelatons might be and complemented with a cimpacable conmbuticnfor wave drag To thatend. a useful conceptual rele:on developed byR T. Jones(Ref 15)s' DRAG qSC = qn0b
SVol
2

i Inthe preceding paragraphs,importance of anempirical and the


handbook prediction experience hasbeenstressed base A sirila. situation e-tsts for computuons in there sa need that -4 establish a computAtional drag prediction experience base With this in hand.anengineer will know whento trast the computational tools for absolute drag predictions, when incremental dragpredicuris should be used. when to select and means alternate The subsection%that follow illustrate elements of a computational drag prediction experience base Skin Friction & Pressure Drag Fncton drag is computed bya number of cotnputanonal methods lit i mportant thatbasicformulas are: handto put but in computationsproper in perspectiveIn the USA. theKattunSchoenhere formula (Ref 18) (en approved for by has use NASA, the Navy, andthe Air Forcebased agreement on with test results. The average incompressible turbulent skin friction coefficient relation is. 0242 - IM logi0 (CFi Res) (8)

M +

I) L

2nq X2t

(6) where 'X arid "X2"are length scales calculated using the supersonic rule This supersonic reinon establishes area the contribution ofwavedrag dueto volume andwave dragdueto lift consistent with friction andlift-induel components As such. it allows the designer to esLablist reiatn the importaunc of various paramneters Experimental Experience Base& landbook Methods for A keyto proper application of conpxtatonal rnethos drag prediction problems is theprojec, engineers ability to recognize thefluid mechanical features characterizing thetaskat hand To thisend. it is advantageous someexperience "real to gain with to This world' aircraft ptoJoit problemas basisaccrues, some extent. in every organization astimegoes on But the interual or organization applications engir"r can enhance specific experiences with thoseof available reports and handbooks dloere's fluid Dbyamic Diag hook(Ref 16 isa compendme donated byexperimen'.aly detetrotid dragsourcis thatcover as ,Atraordinary range applications Theemp.ncism found in of .his sourcecanplayan important role in an project application becausesomedragsourcescontrbutingto Do,transonic drag rise, andinterference drag defy prediction nypurely computational means Another sourne thatproves valuable istheUSAF Stability & Control Datcom (Ref 17) This compendium summarizes prediction methods ratherthan testresults While not specifically createdfor dragprediction, thevolume doesidenify hanbook-yp estimating techniques thatwould beapplicable in the conceptual design phase many aircraft development of

Results using this formula are besttabulated for application In and reference, Lhishasbeer donein Table 3. Europe. the expresslon (Ref 19) hasgainedmore Prandd-Schlichtiung relation is accptuc Thit CFi 04S 0 (logtoRes )2I (9)

Comnprexstble friction coefficients canbegenerated from the incompressible Karian-Schocnherr coefficients byusiig the n.cthod of Ref 20coupled with thechartsfound s Ref 21 F'gur 12 is a graph that is convenient for project use One lirmittion of these relations is ais not that possible to perform configutition-specific prediction taskswherethe viscous drag level depends detailed surface shaping Tbis is on wherecomputational methods can complement thedrag prediction p.ocess Eppkes method (Ref 22) can byusedto design andanalyze two dimenional airfoil shapes when conpressibilty effects are small This formulation is vell-suited to applications characterized bymixed laminar andtuiiulent flow An airfoil section can besynthesized using Eppler's conforial mapping procedure by specifying regional pressure distribution

projects As anexample, therelation below illustrates the technique used estimate wing CDo to

characterisncs. The resultant shape can beanalyzed with then Epplees distrbuted surface singulaity scheme since is it coupled with anintgrial montentu/ienegy equation boundary la)er ixd But aie'tdot designed using Epplers methol will
ihe as only be good as method 5abihty to predict dragfoes It is herethatexisting works in theliterature do not provide

Co

- CF [I - L(s-c) I00wo]

(7) 1WS

can Epplef s ,method be sufficient information To fill this void. applied to establish acomputational prediction experience base using airfoil catalogs sucn asRefs23and Ican seen that 24 be

7-9 Table 3 Karman-Schoenherr Average Turbulent Friction Coefficients' (Incompressible; M = 0, Insulated Case, Smooth Flat Plate) REYNOLDS NUMBER 105 xl 2 3 4 5 6 7 8 9 10 x1 2 3 4 6 6 7 8 9 107 xl 2 3 4 5 6 7 8 9 108 xl 2 3 4 5 6 7 8 9 109 x 1 2 3 4 5 6 8
6

00 7179 6137 5623 5294 5057 4875 4727 4605 4500 4409 3872 3600 3423 3294 3193 3.112 3044 2985 2534 2629 2470 2365 2289 2229 2150 2138 2103 2072 1884 1784 1719 1670 1632 1600 1574 1551 1531 1408 1342 1299 12 1240 1219 1201 1186 1172

010 7022 6072 5584 5267 5037 4859 4714 4594 4490 4330 3838 3579 3408 3283 3184 3104 3037 2900 2689 2609 2457 2357 2282 2223 2175 2135 2100 2045 1871 1777 1713 1666 1628 1598 1571 1549 1513 1400 1337 1295 12 1238 217 1 199 1184 1168

020 6883 6011 5547 5241 5017 4343 4.701 4582 4 481 4258 3806 3559 3394 3272 3176 3097 3031 2974 2840 2589 2446 2348 2276 2218 2171 2131 2096 2020 1860 1769 1708 1662 1625 1595 1565 1547 1497 1392 1332 1291 1260 1236 1215 1 198 1183

030 6758 5953 5511 5216 4998 4827 4688 4572 4472 4194 3775 3540 3380 3261 3167 3090 3025 2969 2813 2572 2434 2340 2269 2213 2166 2127 2093 1998 1848 1762 1703 1658 1622 1 592 1967 1 545 1482 1385

040 6645 5899 5477 5191 4979 4812 4676 4561 4462 4136 3746 3521 3367 3251 3 159 3 083 3019 2964 2780 2555 P 423 2332 2263 2208 2162 2124 2090 1977 1838 1 755 1698 1654 1618 1 589 1 564 1543 1469 1378 1323 1285 255

050 6843 5847 5444 5167 4961 4797 4663 4550 4453 4083 3719 3503 3354 3241 3151 3076 3013 2959 2749 2539 2413 2324 2257 2203 2158 2120 2087 1959 1828 1749 1693 1650 1615 1586 1562 1 91 1457 1371 1319 1281 1252

060 6449 5798 S412 5144 4943 4783 4651 4540 4444 4035 3693 3486 3341 3231 3143 3070 3008 2954 2721 2 524 2403 2317 2251 2'98 2154 2116 2084 1941 1819 1742 1658 1646 1612 1584 1560 1539 1446 1365 1314 1278 1250 1227 1208 1 192 1177 114

070 6362 5751 5381 5122 4925 4768 4639 4530 4435 3990 3 C-8 34:0 3329 3221 3135 3063 3002 2949 2696 2509 2393 2310 2245 2193 21 ) 2113 2081 1925 1810 1736 1683 1642 16W0 1 581 1 558 1 537 1435 1359 1310 1275 1247 1225 1206 1 190 '176 1107

080 6282 5706 5351 5100 4908 4754 4628 4520 4427 3948 3 644 3453 3317 3212 3127 3056 2996 2944 2672 2496 2383 2302 2240 2189 2146 2110 2078 1011 1601 1 730 1679 1639 1606 1579 1555 1535 1426 1353 1306 1272 1245 1223 1204 1 189 1175 100

090 6207 5664 5322 5078 4891 4741 4616 4510 4418 3909 3622 3438 3305 3302 3119 3050 2991 2939 2649 2482 2374 2295 2234 2194 2142 21C6 2075 1897 I 72 1724 1674 1635 1603 1 576 1553 153 1416 1348 1302 1269 1242

100 6137 5623 5294 5057 4875 4727 4605 4500 4409 3872 3600 3423 3294 3193 3112 3 044 1985 2934 2628 2470 2365 2289 2239 2180 2138 2 103 2072 1884 1784 1719 1670 1632 1600 1574 1 551 1531 1408 1342 1 299 1266 1240

1328

1288 1258 1233 1213 1 196 1 181 1139

7 9
101 k
0

1149 Mult 'yTasaldVatiues by 10.3 T0O a5C F

1231 1229 1212 1210 1 195 1193 1 180 11 179 1130 122 1

1221 1219 1203 1201 1187 1 186 1_173 1 172 1 094 18

ainoi Lhicki.css in thss colkcuon will vary bctwren 6% and 24%chord Leading edge radius range 0 2% to over 6% is rhord Flap leniths of 17%to 30% chord are piesernt up to with 10degrees deflccuon angle The Reynolds number rangeiv 0 7 to miliotn Laminar flow extent vanesbetwen 0% and6D% chord Consider a problemn involving an aicraft component sout, vertical tad, or an~cnna blade which a syneem, low drag for airfoil must bed.sige The Figures that follow illustrate a potion of the aiornnv.coed eexernce bas thai wouldprove useful forthis type of applic.ion In all cases bede ubetd, to the results weregenerated using Focler (reetranisiton option This allows the transtion po- bersru lan"unar arid ',buleni flow to bedeteesined as pin of the soluv>, ptoces Ft example. uncreas inciden Reynolds n,-mber will vause ng e or

the ranstion point to moor forward on bleairfoil with an aucilnt increase in do, The transition point is "f-," analytically is the sense it need not be known or fixed a thut pno. Figure 13shows compmted/expenrental comparisons for NACA 65 series aurfoils Airfoils tith 6%, 15%.and21% thickness ae included Here,the aufoil typets fixed andthe cnomputatonal method must predict thelanuuar turbulent drag r'nditng These comparisons indcaat thatthe meshed capable is of predicting drag polar break point as thicknesk iclteases. Thee is soni error noed i tinimu Irtunar drag levels for lowReynolds numbers, is aggravated by inreases in this thickness In view of this. computed results for a n-w airfoil within this range wouid haveto beproperly adjusted io a.otni for observed simulation trnng discrepancies

7-10

O0005.

05

0004

01110003
Fiue1 eit eodsrisTi ie ifi

'o*
oefcetTe

is

Fig.12 Krme.SconerCmresbeAeaeSl"Fito hcnssi ope

utpesrsnflwmxn

ixesnee o

IaFtariih b rpign cnheientifC mrsileAeaeSi iedrg1 or aigreil dicthascond i es hi te a ir i thciguri th fixedandis cange NAC 4-Diii g y. he seies p 66 beo neeSvrairfoil h fye geicue s eturins fapdlton havitfecs bee includedion i STlustarfn thicknes of noe i a ithus 30%ntey flapdefl ete n oladegr Thew chor seasd atolis sim gin getem h ength b qut0goSm ichrepasty sin is fatgures tdturbulen flow frak-orn isontid fpAgremn a b5% irn aefo ma llu s reioknes aitewid nc wher flaur that~ mtos sdlngacies are ca vrtcl t al plcaininesfl it ipl lp

Fito

ofiiedt

The

cmpex m6ilhte tpe fof ra gprestineee r h acc can ueiedarfi fos wit multi plen iflaplcications etnd ie vdegre tideni eideve I 0 da b rmeto prderstrte hight resltwil lito oplr'st mo theelmnga dit la sowed casaetler laif t l scan rsse b onsui and. cothe oumge tot availatouti-ceenhghlf but vf in ssvrl iie nefrnefe idtne aaa ver highli ghvts difiuty odrgh ile ctianauy ta expeteciptsoa Thi imar mte enti abfilt tablishio A r comee inulat7on14 esection biase Dlrfaple 4 mplentened onerso compu deter Sltions'tkeoit3 minuides ihlf eutwieEpesnchdpeit h ii flpsoecaetlwrlitevsThsoagons eu. Intsoe vplmidtas conipreb effecuti-cncnt beg i-tr Eppl and etos sevensl lmiehds inotih-e ppiabl Ounel aaa oa coutsaionta o tha protiedr ratisaaby el fr ot ah ariesnohds numbeirsn rissoc airfil gasestcabiharn yrng~t iscusationvsi nterions iase fD aadCe ta

~~~~~~~~~~

~a

beharactenas fovraanfoewlyfdevelopedfairfoillmightnbeha lias and pckific cbteencedor copuionalh simuatifo de can shend rois aihrfoilp docases Onl hnd ewTh perma butte flcored aendt evaluateodiTh a setion famuetiy e fo o a2%cor lpAgemn hinelig:gap my otetia fow heomea b casig nn ueinth' thedelig Epthat ith o peaytoa me canali e. ouin ciieol ts o or

proachi is 26~e 27)ce atvnalss and=This andpuderg cmpues. oudeeapaetso pitese xpesnsiv ompesletont tha metod charcterstis fo anynewy deelopd ufoilouconseidperbtment uiatadtonlfft descric iesofrbusmeppiao drgpitions orndra cmght remulire Somedsnpectdfo reducng oe srfac optrarauon airoilsthat mla-eemen and oa tpobl ads pro mgnituempaly er for with ao adoi "stes mehdisarie methd is thanoplx 3nDumbea. neil w higsylito upplinbonh tpe s would beuthode frore itragsctions sta fDi n ie bysedoninvsthusWave dscus elmnt moelig sod tren tangl gu-sine. ai visos &M) includeddeigppochi26e howver does airol Soptes. eRefs

16NACA

650066

NACA66O 012 1

12

11

.-

08 ______ _____

8____

04~~~

ID

RF2)
6

PPLER, =3TS

04TEST

(REF 23) OEPPLER

Re E0x 106 --Re .90e10 00

0
R.0

0
0

00

16
NAA Is~. NACAi

012 1

12

-12.'

*u

00

004

00'

_0

NACA F4'021

ACAOOI2

00~m

00

00

000

0006

0012

0016

F004

0004
CD)-DA

0008

01"

0016

CD)DRAG COEFFICIE.IT

COFIIN

Fig 13 2.0 Airfoil Drag Correlation-Thickness Effects

Fi 42DAirfoil Drag Correlation Laminar Flow Extent Effects

16 F

FX 67-F-141 FLAP 30 OWLER WITH

NUMBER VARIABLE NOTEREYNOLDOS

CLLFT COFIINT

O TEST(REF 24 04 0

EtER,

FI~o 0.0-l

05
16~~ 10 I

0i

TET(E4) 6 371 -

ANALYSIS STEVENS (EXTENDED) EPPLER(STOWED) .X7

12

~0
-

L~ ,
U5tW.

002

003

0064

CODRAG COEFFICIENT

808

Fig. 16 Eppler & Stevens Method Prediction of AirfoilFowler Flap (Extended & Slowed) Drag Polara

04

10 R-0 Flat. 0#. 10, 00 0Re . 20 - 10'


--

Lift-tndusced D~rag in Btackgrouind sectin. lift induced dragregistes Asnoted she on the aircraft surfaceasa form of pressure drag Computational msethsods integrate compoted rnormsal cass pressures predict liftso insduced drag.but in nsuflyapplicuasos where absolute drag levels are ir,;*ncant, nnsatisfactory results areobtined Thssis leadssg edgesuctiosn forces(a the case because lifling surface component of lIfi-inducced drag) ame resolved so degreethat a depends thle on computationat method's panlle gVd resolution or As mssodlsog elemsent gnd denity is increased. lift or the snduced tevetwill decrease drag asymptotically approaching she

16 F. L V 52 K26

exact trsietevel wouild achiteved infinste or that be with resolutsons (Sketch A) Inoa project application. pressure
12 FULLY TURBULENT FLO integration resultsroustbeusedcautiously wishshe engineer ensurng shut compused she difference between Iwo configuratuons is aerinjynaric, socharacter not numerical and

0B
I )

IS

*0
Ithe 00 Reo.07.106 F&V . U Dea
0012 0016

A isose reliable approach (Ref 29)to computing liftinsduced drag wing-donuitted configuraions is that for proposed by Glaucn (Ref 30) It involves integrasig theload wake) (or circulation wosild beequivalent soshe spanload by given -C,(10)
Usingsthe notation of Ashley asd Landahs(Ref 31).the litfs&W

04

) 00 0004 0000

CD DRAG COEFFICIENT Fig. 15 2-D Foil Dreg Correlation-Tab Dellection.


Length Effecta

indaced drag bewritsen can L -PU'.


452

r(y)dy
Il

7-13then

2 Me (19)

a A~icBy

expanding the full aircraft loaddistribution out to 0 = 2n: (Sketch B). the distribution can be Fourier analyzed to solve for the coefficients The resulting prediction method proves vety fast requiring only seconds on common personal computers r 6on

60 40 20 CHORD PANEL WISE COUJNT SPACING) (EVEN Effect of Chordwlse Panel Dennity on Computed L't-nduced Drag Level

( ZY

Sketch A

Di

~v'u ~

so

,-(YOE dyidy dyt (y-yi)

(2 0 -yRelation for Fourier Analysis

Sketch B and the circulation ts represented bya Fourier sineseries

I - U~h Ithis yield-,

E A. tn n0G
a

(13)

Atecrft configuration spanise load distributions cso vary considerably depending on the general arrangemein and flight conditions A number example cases for bosh symmetric and of unsynuttec loadings have been computed using the Fourier analysts previously described These cases illustrate the typical range of lift-induced drag 'efficiency" factors that might be encountered during project applications 17 illustrates eight cases of symetriseac wing/aircraft loadings and. as such, only half of the loading is displayed All 10 cases represent a commron lift level at CL I Note that the slope of the loading distributionais "0' at the centerline or syri-uy M!pIn (is I,idSn n Wv, eityic 103d a. distriution results in inimumt lift induced drag This load is seen in Fig 17-A along with the compiled V factor of 1 0 Elliptic loading is miosteasily attained with ra isolated wing arrangement, since body Orfuselage components are very inefficient in generating lifi This causes a deficit or depression in the loading curve at zhe body locati In mny aircraft design applications, including fighats and transports. aerodynamtcists attempt to achieve elliptic loading In somec applications. howevee, the resulting combination of induced, wave, and frictton/pprtssure drag sources may not represent an optimum. although the conventional lift induced or vorten drag has been immnized Anober load distribution is that generated by nezzsection properties (Fig 17-ll)on atispenoidul planform isolated wing For this lo":iwg. drag creep (that might be

LAiFigure l~ P

~. b
(l4)

ol 12h2

Di -a

~AI(IS) ~j~-

and in coefficient form

I A, L 2 'p 2 S s 2pl

(16)

C,~-

D 2

AZSVE A2. 1puisconstant (17)

ro compiute a spanload efficiency factor "c on thebasis of

generated bysomec airfoil sections developing wave drag local prior to others) is nmmired Allwing airfoil sections approach drag divergence stmultaneoasly For this example. e' drops to
0976

Cq IARCIs
IlM

Loading on a fuselage is largely the result of wing carry-over (Fig 17-C) Often little can bedone to compensate for the load deficit that will =ost) bedetermined by the percentage of the wing span blanketed bythe fuselage surface Figure 17D shows

7-14

16

E
ELIPTIAL WING) 12 -P{(ISOLATED FUSELAGE& WING MOUNTED EGINE/POD INTERFERENCE

08"

00
-

------

04

e=0.921

TRAPEZOIDALWING CONSTANTSECTION

FUSELAGE &WING MOUNTED PYLON INTERFERENCE - - --

PROPERTIES

04

0 16--

0'.!

C
FUSELAGE INTERFERENCE

ENGINE/PYLON INTERFERENCE

FUSELAGE SWING MOINTED

04

J-/
ii WINGFUSELAGE WITH HIGH LIFTSYSTEM DEPLOYED 1

12

EXECUTIVETRANSPORT WITH FUSELAGE MOUNTEDNACELLE INTERFERENCE 04

e-0
0 02 04 06

92

a = 0549~

08

10 0

02

04

06

06

10

SPANPOSITION --

With Lift-Induced Drag Efficiency Factor Fig. 17 Examples of Aircraft Symmetric Loading Distributions

7-15 the loading that rmght exist on many executive ransports (bsiness jets) where nacelles are mounted aft on the fuselage vta a short pylon The nacelle/pylon combination inhibits wing circulanon near the wing-fuselage juncture In this case. e" might drop to a level of 0 925 Engines are often mounted in or about the wing curface using pylons For embedded engines or pods, an effect similar to that for the fuselage can be identified Pylon surfaces are ditferent There is typically an "end-plating" effect that increases loading inboard of the pylon station and reduces loading outboard This is the case because the inboard portion behaves more twohile dimensional in character % the outboard segment functions as a lower aspect ratio wing segment Figures 17-E/F/G illustrate these possibilities. The worst case is for wing loading featiring fuselage, pylon, and engine nacelle interference Perhaps the largest influence on wing loading dsrimbution will occur when effective high lift system! are deployed for landing The spanload efficiency factor can be extraorinarly low, but. fortunately, there is little concern about drag forces during this brief segment of any flight Unsymrincric load distrbutions are generated in sideslip and when control surfaces are deflected to roll the aircraft Figure 1b A illustrates the type of load distribution thatcan xist for any swept-wing aircraft in sideslip Asymmetry is created oy variations in liftng efficiency bctwecn two wing haives that now havedifferent sweep angles In this example, the starboard wing effective swep is greati than the physical sweep angle, while that for the port wing is less The rculting degradation in 'c' could be subtle (0 975) Fighter asymnmunc loadings can be quite severe because a premium is placed on roll effcciti, encss in air to-air combat Lifrindaed drag increases as the roll maneuser is niated Many aircraft designs use ailerons for rollcontrol While the rolling moment can be very large with control surfaces positioned near the wing tips, the resultant reduced drag penalty can be high for the same reason (Fig 18 B) In other words, it is apparent from these example cases that the lift induced drag level is affected to a lunger extent when a loading anomaly exists at the wing tp than when it is positioned inboard Another form of roll control can be derived bydifferential tad deflection Figure 18 C shows the combined lifting surface load distrbution that ought result for th.s type of asymmetric configuraton Figure 18.D is a stular plot illustrating use of wing mounted spoiler deflections to get-crate ichllaigm ments It should be tecognized that the cases highlighted in Fig 17 And b8 n, specific to any particular aircrafi The true deiAled are loadings for an airraft applicaticin will be a fNrnun of the gencnc coufiguranons geometry and design (ft Ie.l Threuc examples, however, should prove raefl fvi cit.bhshtng trends linked to wing loading whether transonic or supersonic, there is an incentive to achieve shock wave surfaces that are oblique to the flow direction This mintrmzes wave drag losses because the largest drag penalties are generated by flow through normal shock waves Many computational methods have been developed for highspeed aircraft applicanons, particularly at tuansoni- speeds But the character of complex three-dimensional mixed (subsonic/supersonic) flows presents a c ansiderable challenge for algorithm developers At present, ccmputanonally predicted transport cruise drag level accuracy might bc on the order of 1030 counts At the upper end of this range, the project requirement might demand errors that are an order-of-magnitude less Fighter applicatioius reveal larger prediction discrepancies den'ed from a higher level of three-dimensionality and the complexities linked to naxed (attached, separated. vortical) flows Computational drag prediction discrepancies greater than 100 counts are possible Three-dimensional computational methods have, however, demonstrated an abrl ty to predict surface shock wave patterns This allow s the designer to develop shapes that generate weak oblique shock waves. But perhaps more important, it provides a basis for applying simpler, less expensive iwo-dimensional conputatonal methods in a drag build-up process with potential for higher predction accuracies Figure 19 provides examples of transonic shock wave Positioning for three-dimensional configuranons Simple S.eep Theory (Ref. 32) establishes a means for relating two dimensional airfoil characteristics to three dimensional wing performance Similarly, Sweep Thory can be used to translate wing pcrformance requirements into a set of specifications suitable for two-dmensional airfoil design The cosine relations linking two and three-dimensional parameters are listed below M2 D M3.D x cos Aeff CL3.D/nOs Aeff t3-D/os Aeff CP 3 D /coC2 Aeff CD3-D /cs
3 2

(20) (21) (22) (23) (24)

CL2.D /c2 D

CP2.D CD2.D -

ACff

The proper sweep angle must be identified to implement these relations Ior an infinite sheared wing panel (Fig 20 A), there as only one possibility the panel sceep angle This represents the original embodiment of Sweep Theory Consideng a finite upered wing planfom (Fig 20-B). tw dunensiona! simulations of three-dimensional wing pressure fieldnat subsonic speeds reveal that the quarter.chord sweep angle serves ,ell as an

Transonic &Supersonic Wase Drag Wave drag losses aregener'ated by flowaboxt asraft the passing through shock waves As noted in the Background secoon, shock wanes can form at subsonic speeds if wing or fuselage surfaces accelerate the free-stream flow to sufficiently lunge supersonic velocities. Mixed flow regions featuring an embedded supersonic flow region within a subsomc external flow (separated by a shock wave at the aft boundary) are classified as =r.nsoi Transonic flows al.o exist at low supersonic speeds when small subsonic flow regions are embedded in an external supersonic flow le .at the nose of a f.selage or leading edge of a wing %here a stagnatiuon point generates the reduced velocity 'island') Inall high speed cases,

'effective" sweep angle forthe rioe formulas hsed above The most complicated situation exists for tansonic conditions where shock waxes ar present on the wing suthace Engusenng studies (Ref 32)performed daring the IhMAT (Ilighly ,Maneuverable Aircraft Technology) Program revealed that the local sweep angle of the shock wave provided the best effective sweep angle for Sweep Theory conversions This definition of effective sweep at transonic conditions may at frst bedifficult to undcrstind, however. it rught be made more apparent by consulering an example Figure 21 shows a set of wing pressuire disrsibutions extracted from the nud section of an aspect ratio 5 8,40 degree swept back wing with a taper ratio of 0 4 Thesymbols represent wind tunnel test measurements The solid line comparisons are generated using a transonc ic todimensional airfoil analysts mod"(Ref 331 with Mach and lift

7-16

-080
-.

~...

~.....1

WiNGi LOADING

CC

-6, o
04-

DF RETRIhALAILO FOR ROLL CONTROL. a 0.741

1222
7 04 1 ROLOTO a 079

N DEfLECTO SPOILER WING


FORtA=L CONTROLa. 0. 22

-08

-06

04

-02

02

04

06

08

10

Fig 18 Examples of Aircraft Unsymmetric Loading Distributions with Liftinduced Drag EfflcilerJey FsCor

7-17

Fig, 19 Shock Wave Unsweep Regions

(A) SHEARED PANEL INFINITE SPEEDS) (ALL


/Agreement

conditions specified using equations 20 and 21 The extracted a,rfoil shape scaled is using equation 22. taking soloaccount the gcometric local sweep hnes of the tapered wing (Ref 32) Simulation agreenctit excellent But to achieve this is agreement, variability in effective sv cep with local flow the conditions must be accounted for Table 4 illustrates this fact b) listing the wing shockwave position. sweep angle. and the derived 'effective" two-dimensional condtions usedfor anal)sis eases Note that as the wing shock wavestrngthens with moves aft, the Mach number n&d increasing free-stream effective swCep angl usedin equations 20-24 decreases cannotbeobtaned unless variation in effective this
sweepwith flow condition is taken into account

An aifod is now selected establsh wing drag prediction to potential using this build-up approach Reference provides 34 data for the RAE 2822airfoil The transonic design pressure

distribution isseen Fig 22 ITs in generate rie curve. adrag


(B) FINITE TAPERED WING (SUBSONIC SPEEDS) ,ot effot is requited Data thatis not all atthe sam lift level t.ust beadjusted Thn can bedone using Kom's relation (Ref 33)

M MCOt/c K

(25)

Dr David Kom of NYLUs Courta Institute determined thathe could design transonic aufols with parameters consistently that sumsed tocertain values depending on theIeel of technology (or aft loading). The design engineer can use Korn's relation to establish approximate trades between the airfoil Mach number, thickness. anddesign lift level 'K" levels for conventional aurfols arenear0.87 while 'K' for highly aft-loaded supcrcnocal sections can bena 0 95. Figure 23 shows an eerimental drag rise curve along with thecurve thatresults % points am adjusted to cosaion CL - 074 A design hen point for this section(at niximum MLAD) occur near M - 0 7

(C) FINITE TAPERED WING (TRANSONIC SPEEDS) Fig. 20 Effective Sweep Basis for Different Applications

7-18 DA-40, X.04 000 EXPERIMENT3 WITH APER THEORY (KORN-GARABEDIAN)SWEEP-i 2-0 ANALYSIS 0

M.-070 (A) 00 0 0 00 M .0 so 6(E)


--

00M0

00
I-

M.00
0-

M.-0 90

Fig. 21 Wing (3-D) Pressure Distribution Correlation using 2-D Airfoil Analysis &Sweep-Taper Thory

Table 4 Wing Effective Macn Number for 2.D Analysis


SLOCAL ILOCAL FREE0 200 EOUIV

12

CHORD
&

FIG I REDUCTION LC ! , 21A 21 25% 0 2 5%

K,~~
M, MI N 37 42" 0 70 3 7 42 32,30132 3V 130450 80 086 090 095 0 556 0 635 0727 0761 0819

SWEEP TREAM NORMAL SIMPLE A 02-D0AALYSIS MACH MACH THEORY MN 0 .1 Cp

-00 MSWEP )

8.
0

0 63 5 0683 04 0715 0 754 Fig. 22 RAE 2822 Airoil Pressure Distribution at M- = 0.725, a = 2.9

.!D 21-E 121F 1

70% .... 70% 85%

The resuleng "K"vluc is 0 89 Test da at S- 0 723 is closest to this point and there is no indication of atpprcmbl: rq, w separa despite a Nuon reaunbly sotg shock wae The shoci k wave is postoned at 55% chord A compendium of codelexpenment comparisons found in Ref 35 rc',cals that (consukcrig a large number of iltffer"rnt Computational rncthod ,j two-dimensionl computtonal drag pircdlictors vary by 2pproximately 5% At the rno extrtnrc M-0 74 caw a 25% variation be~tween methods can be identified Shock wave posoon vaned by a much a 5% chord fo .II0 72 whil the location spread for M-0.74 was about 101, chord "lTb, tmhnbqjc for i~nilating two.dimn'osions! drag andi shock w~zc location dlSCpv, s into tuc- znenssorl wing para, etters be tiustrated b, considering two '%rag can

7-19

f-

002
TEST POINTS

CO0 0 1

6 06

si 07 M

"

08

AR..8 X-042 LLE- SLE

AR .3 X-02

Fig. 24 Transport & Fighter Planforms Wing


ADJUSTED POINTS AVERAGE #CODE COSRELATION 20%

00i

toa purely three-dirs'-monal approach The exercise shows that CO AVEPAGE both drag prediction accuracy shock wave prediction and CODE CORRELATION accurat equally Important to properly determine transonic y are ,avedrag levels It should also be apparent fighter wing why prediction errors canbeconsiderably greater those than for
transpot wings, evenin applications where separated vortical or flo is notpresent.

07 08 M Fig. 23 RAE-2822 Airfoil Drag Divergence Curve Data &Adjusted Data (CL=0.74)

00 .I .. 06

I6

planforins one atransport and second afighter (Fig for a for are 24)Panform parameters listed below Wine Trnn t" AR - 8 3.-04 ALE - 25 deg Fighter Wingmany AR - 3 3.-02 ALE -40 deg

Prediction discrepancies identified in the preceding paragraphs represent average might beobtained byselecting a an that computational analysis atrandom and applying computed results by discrepancies can beconsiderably reduced directly These selecting amethod that performs better others onethat is than or mo accurate the for particular application it handA Computational Airfoil Cataog can provide die flow simulation by needed project apphcations engineers This expencce base type of catalog %oldhighlight thestrengths weaknesses and of provide sufficient test-venfication computatioanal tools and cases establish correction fActors awide range aafol to for of design conditons types and One relatively new techniqu that promise reducing shows for both airfoil drag prediction error and shock location error levels is of Drela ad Giles (Refs 26 & 27) The Drela and that Giles airfoil analysis/dcsitgn code is not like othersin that its
formulation includes an Euler soluton for theouter flow region thatis coupled with a two-equation integral boundary layer schemeThe setofequaonss solvedby aglobalNewton iterative process Comparison cases revealthatthe lamarltubulent boundary layer technique works well for a, stri'g Interaction cases potentially unmnng "adjustments thatmight beapplied higha Mach numbers where at drag prediction discrepancies aretypically 25% From Ref 35, should beapparent Navier Stokes solvers it Ot haveyet to demonstrate superority over more conventional schemes c.. Drefa's method) for drag prediction This is the (i case despite tN . ,i Navier Stokes methoomight require that to orders-of magnitude morecomputing resources current than coupled methods. Wave drag prediction at supersonic speeds often presents a simpler taskthanthat at transoeac speeds because the of applicability of supersnic. linear theory andthesupersonic area rule concept (Refs 36 and37) 1her are a few assumptions Oneis that flow disturbances propagate outward along Mach lnes andther is no dissipation with increasing distance Another is disturbances are furction only of thecross scionl a areadistribution, i . the "flow mitferfeuce" between

Using RAE 2822airfoil (with shock position at55% chord). the the effective sweep angle attransonic codition, forboth wing planforsms 19degreesThus, for a wing design Much number is of077. RAEda,at M -725(Eq 20) canbe usedThetwodimensional section generates counts (Cd 0 0107) of drag 107 at - 0 74 This translates in a 4 1/2 C count error t band friion/pressur/wave drag on the transport win& using a projected 5% spread prediction accuracy andEq 24 The in same 4-1/2 count error band would hold for the fighter wing assuming thatthedesign ift coefficients ae identical Now. at can usagued that5% chord discrepancy in shock wave te a location extsl. Considering the uanspon wing, a shock wave position at60% chord results in a h gser normal Match numbei tAeff - 187 deg.), 0 73.Airfoil drag level rises to 125 about counts, this effectively registers as a 6-count rune for ttre tansport wing. The fighter wing effective sweep for a 60% chrd shock location is 17 degrees. resultng two. The dimensiotal Ma.h number u 0 736 From Fig 23. theairfoil drag level rises 15? to counts yielding a wing drag level error of 34 counts - sincenow at the higher effocuve Mach number,a 25% prediction error applics Ernor generated this buiddup process by using both two- and hiee-dinensiooal techniques coupled via Sweep Theox), t wa result in errorsthat areconsiderably snaller thanthoseattributed

7-20

components to relative positoning is not modeled. Finally, due be configuration wave drag can predicted by it is asumed that coroputattons performed for an equivalent body-of-revolution Reference 38 provides von Karman's equation for the wave drag of a smooth, pointed body-of-revolution

( PLANES MACH ANGLE CUTTING (A)

Co

. 0

f6

da 1:121 n(x.)dx d', d


2

(26)

An equivalent body-of-revolution for an aircraft configuration is generated by selecting a number of longitudinal statons between the configuration nose and tail. At each station. curting planes inclined at an angle. u -$mi (I/N) (27)

i
(B) GRAPHIC PROCEDURE

relative tothe x-axis Fig 25-A) generate a planar area value (see that is associated with the s'ation " location This can bedone graphically asshown in Fig. 25-B. or the procedure mght be automated for on digital computers. This catting process is use performed repeatedly for a number of roll angles, asdepicted in Fig 25-C An effective drag for the equivalent body.ofrevolution at each roll angle is computed and these salues are then Integrated to arrve at the total configuration wave drag coefficient, This procedure, combined with von Kaman's relation 1Eq was automated by Boeing engineers and 26). documented with sample cases by Hams (Ref 39) The resulting computer program hasexperienced application throughout the .ircrafi industry since its inception in 1966 The Fr Field Wave Drag Program is very simplistic by any stAard Butthe flexibility and complexity potential forso modeling realistic aircraft shapes isentraordinary and its applicability range is quite large Predictions for shapes that appear to exceed the bounds of linear theory are often useful for engineering purposes Modeling flexibiliv, is Illustrated in Fig 26 This model t.x of NavyiGrumman P-14Tomcat wasgenerated tn the late 1960s (Ref 40) The aircraft ismodeled using a set of Aing and body type components Figure 27 shows how design engineer,, optimized placement of the various coenponeni to ratch as close as possible the optimum supersonic bldy areadistbution This -as achieved despite a number of cinstraints that included overall fiseness rano. nozzle exit area. and placement of internal elements Application of this technique is the primary reason for the F- 14being positioned in what is often called the i"ird Generation of Supersorc Aureaft" (Rcf and Fig, 28). It 8. is simply too expensive and there is insufficient time to perform the wave drag mintnrzauon process with this degree of integranon by expenimenation alone One facet of drag analysis that should be noted at this point is numerical opurization While the cngier cansequentially establish a shape modificaaon via a darect compuuonal analysis for evaiuation. modifiation and re-evaluation. etc. there is considerable Incentive to pcrform shape opurruzaon using the speed of modern digital computers In oiler words take the engineer out of the loop and speed the process Another up approach mght involve the use of imversnemethods These techniques synthe ic a shape based on specified flow characteristcs such as velocity or pressure fields A useful perspective on optamizatin methods can be found in Ref. 41 Tbe methods just described can provide valuable guidancc on ancraft design projents. but there are nIsuitons should be that recognized First. assnung that low-drag pressur/veloc-ty

STATIO 0 10 2030 i050 60 70

LINES ON TRANSPAENT PAPER

REOUIRED PROJECTED CROSS 40 SECTION STATION AT 3 3 5 5.

PLANE ROTATION FOP ONE ROLL ANGLE CUTTING (C)

, Fig. 25 Generation of Equivalent Body-ol-Revolution for Supersoic Area Rule (Refs 17 and 39)

fields are properly specified. it well known that an inverse is method as simply no( robust asita direct method counterpart as In view of this, during a project application, an engineer sequentially performing direct analyses with modifications will usually be able to surpasa the result generated by another engineer deugning by inverse methodology One way to circumvent this Im ation is described in Ref 42 Here, a type of

7-21

Fig. 26 F.14 Harris Wave Drag Program Model (Body & Wing Elements)

'modified duect' approach is described thatis in essenme an technique The key to this approach. how ver, is thatit inverse retains thestrengths of die parentdirect method Numerical opmization, in theclassical sense, constrained by is a number factors First.the optimization schene is of extrordinarily expensive implement due to the Lrge number to of analyses arcneeded to establish appropriate trending that Second. rcasonableness criteria aretypically notapplicable ard asa result some lengthy computational opunueanon processes convtge toan unrealistic or tmpractical shape. Finaly . the pacnt analyucal method s dragprediction fidelity is often impaired 'ls results in theopumizer procesaunt deficient

design applications and. assuch.requires lesscomputing than resources a methodsetup to treatgeneral optimizanon problems

MOEAL

ACTUAL A

performiansce informatison the with outcome being design a shape


thatis suspect Reference 43 describes an approach ionking to &-ound theseproblems In this scherme, aCONMIN optimizcr (Ref 44) is coupled with design variables thatare m nrated using conventiotal inverse: techniqu-s and ob)ctive funtons thatare "aerodynamics specific Theresulting numeruial 4
*an

27 n-n

optimizations as prixess essentially tuned h;LAlc acrvdjmnurna to

Fig. 27 F-14 Optimized Normal Area Distribution

-7-22

5 4..2 I 3 2 A 5 1 13 1 33OF 0_ ) 300 00 ry 19 s FIRST SECOND TID9F81.-3 I1F-81D 12 MIRAGE 3G


B-58

24

027

GENERATIONS SUPERSONIC AIRCRAFT

SYM AIRCRAFT 0

SYM AIRCRAFT 14F-102A I5 1 RAS5C F-105D 17 F-SA 19 F-106A X 20 F-14A I21 F-I111A 22 LFAX-4 2LR0 24 F-ISA 25 F-I6 I26 YF-17 27 F-I8
A

25 26

I2

I XF4-1 XF-91 3 XF.100A 4 XFY-1 6 X-3 7 F-100D 8 F-4E

I16

HAACK 13A a SEARS

I )MINIMUIM DRAGISFA BODY 06

C ______________________13 10 11I 8 9 7 FINENESS RATIO 1/d

13AB-S8vPOD

FIg. 28 Supersonic Wave Drag Trending (Ref. 8) Throtle-Dtpendent Drag


Aircrafi throttle settingwill affect bothltheinlet spillage level andthreexhaust nozzlepressure rato Theifleractios of the spillage flow on ile surfaces (Ref 45) and neighboring arny aircraft coriponents will generte a resultant dragor thrustforce Similarly, exhaust plume interaction variations might raiseor lower aircraft draglevels, depending on iheparsicular configuranonarrngemnt andflow charcteristics The with of flous associated this dragsourceguarantres compleuty thatos manyprojects the first throttle-depeodeut dragestimates titng is completed until will not bein hansd poweredsub-scale Ilowever. two exameples included hereto illuinratebo. are computational methods provide usefulinformation prior to can testing ustng a variety of can Inlet/uacelle surfaces bemodeled con',cutional methods subSonic. transonic, andsupersonic

results (afor of suepsiin The Firstite/IlaCelle modelas characterired by an open flow-through dUCt.The second or model's duct is closed A Ibid msodel identical t0 the first. but is theangle of-attack is srt 1090 degreesThe following relation
is usedto compute the local surface velocities

[~

- T-Cii V V V2

V. V 2

1
OE

JO*V
(28)

v1 , V' Vu os a T. V2 VOj0

The example Fig 29uses subsonic is the 'source" meil-ed of surface singularity panelmodelLan lie Ref 46 The appropriate be to identified A clever scheme outlined in Ref 47 Can used generate flow solut:ous for an)incidenice angleandany inlet flow rote bycomputing three inlt solutions andcombining dhe

distributions that arc Local selocinies convened to pressure are integrated toobtaindrag forcesFigure 30 revealstho this simple scheme provide accurate Lan pressure field detAIls over a large range shaiprs of incidence angles, flow rates A anid

7=_ +

& +

ZI

Fig. 29 Opets/Closod Models for NACA Inlet Casen

7-23 NACA 1-50-100 E55'60 000 000 AAA 000 ANALYSIS ViN.- -090 VVV' .068 ViN. -044 Vw "020
-

NACA 1-70030 EXPElS ANALYSIS 000 VVW.055 4 V4VV-0 3 ------000 - 032 A" V4VV 0 21 VV

....

-02'

24

I-.58
,d

06

- ---

-02-

--

06_

06

0X.O6

0402 0

cp O 2

1 -0 4

02

02

04

06

08

10o

NACA IO-050 EXPElSJ 000 00M A" YLS0 V" 096 -0 VN. 054 ----VV-.030 VViN 019 . V'08-

03

04

00
08

M1 64201291. 02 0A

06 .8x.8

08

10

-04022- 1 CP 0 Vs Z_

Fig. 30 Inlet Top Centerllne Pressure Distribution CorreOldtions (coni'd) typical relaton beteen Inlet flow raltecompuled drag levels and provfdea 31 As Inlet leading edge decrease, the nFig radii poteri tni- flow separation drag penaltes increases for In addition. aero-propulson bookkeeping rtequires that"additve drag (a furtonof inlet streartube geonetry) betncudd to
obtain total spillage drag levels When all of these components I the aiecsombined,trend shown in Fig 31 may bereversed,e, flow inlet nmy spillage drag torease with any reduction mass

-o8

08 -04 ,__t "0

ratio

A keyfeatureof this approach and the example descnhed is just in generating the that computational modeling is not adtered 10numerteal
-

.-C,
h 02 0 02

T
08 to

resitts In this %aly,the resultant drag levels are certain to bedenved from configuraton geometry andflow conditions, andnot from nunenca] discrepancies thatmight surface when thediscretized model is altered

04

06

Afteibodydragle.els.asapercentageofthetotal, canbequite

US 2E0hs2

__

ct. x't8 6,126012 W1

Fig. 30 Inlet Top Centerlne Pressure Distribution Correlations

for large fighter aucraft (Ref 48) at certain conditions This e level rmght behalf .,i total aircraft drag level (Fig 32) of auisiplume interacUons. an important Componentthetotal af erbody drag, can computed with patched be solutions (Fig 33)Two computer programs (one transonic speeds and for one for supersonic speeds) beer, have developed to provide this ty)p

7-24 SDRAG , TRecirculaung FE reat - Controlregions) separat in an Flowtachmaenltvolume analysis with SEPARATION NOTE' FLOW NO ORADDITIVE DRAG Supersonic Exhaust Plume -Salas fimte difference marhng scheme Plume Entrainment - Mixing profile to yield equivalent
-

displacement thickness

Figure 34 showsa typical stmulaton result for two boattal surfaces (Ref. 52) at supersonic speedsThe drag reduction trend thatoccurs with increasing nozzle pressure ratio is 0redicted well 02 04 06 08 Vfv INLET FLOW RATE SI FInterference
_ _ _ _ _

1.0 Interference Drag drag sources complex andlarge in number In are some cases computational methods can predict flow qualitic% thatwill aid in the drig minimization process In other cases. the methods can predict useful drag force information Oneexample of an interference drag sourcethat hasdefied flows. The lirmtanon appears beassociated with the inability to of three-dimensional viscous flow predictors to perform well whentheflow tn highly three-dimensional As a result, juncture fllets riught best be optimized with guidelines applied during sub. or full-scale testing Reference provides both design 53 guidelines andliterature references could prove useful in that shaping fillets

__

__,,__

Fig. 31 Effect of Inlet Spillage on Inlet LIp Drag

IERCN nI PERCENT Co t00

,It
Fig. 32 Fighter Afterbody Drag Levels at Subsonic Speeds, CL = 0 tLAG iNVISC PLUME ENTRAINMENT EMBEDDED SKELETON BOUNDARY SHOCK FULLY DEVELOPED LAYER SEPARATO STREAML. 4N E REGION MIXING

provide should berecognized that computational methods can


useful flow angularity andgradient information (seeFig 35 and Refs 54 and55) thatwill helpthedesigner conceive a shape that is quite good forthe first series experiments Testing of requirements, andtherefore cost,can be ummirzed A greater level of application success associated is with simulating both propeller androtor slipstream interactons The former is important because interest in highspeed,highof efficiency propellers (prop-fans or unducted fans)andthe latier isnaturally inked to aircraft concepts like the V-22 Figure 36 shows propfan tractor arrangement might exist a that inthefuture In order to matmie the benefit of the total propfan concept, propeller slipstream andnacelle interference must be mnimnuzed The slipstream flow is charactenned bya swirling motion wilh discreet vorticity sheets emratiag from blade trading edge. Immersed configuration components experience increments in Miachnumber andflow dynamic pressureAt high trasonic speeds. slipstream swirl effects will be dominant Figure 37 reveals a testsetupincorporating a

TRANSON
< I IIS T R I SMOOTHED REGION ".. 4R 1

i SUPERSONIC SMOOTHED REGION

~'-,--010
-

,,,..AYSIS

CONFIG TEST 8

MM 220 Fig. 33 Boattall Analysis Method Computational Regions of prediction across subsonic, Irsisonic. andsupersonic the speed regies (Ref . 4, 50, and51) Both methodsiteratively solve forfiveseparateflowfieldregtois foundin Fig 33 The flow regiOtnS Are. * External Inviscid Flow - Mapped region. conventional SLOR with rotated difference sche mn * Boundary Layer- Greens integral method for 6 * C 00
-

NUMBER1659 X t0 REYNOLDS

0 0 5 to Is 20 25 30 NOZZLE PRESSURE RATIO , i P; I,--., Fig. 34 Correlation of Nozzle Pressure Ratio Effects on Afterbody Drag Levels

7-25 VELOCITY SURFACE UPPER VELOCITY SURFACE LOWER .... ipropelles of propeller slipstream simulator positioned upstream a supercritical wing transport model. Wing section boundary the conditions weremodified via "twist" angles to represent wing swirl seven-degree velocity That is to say. the sections between propeller centerline andthe inboard radivs line weremodified to haveseven degrees more incidence the upwash. while wing sections between representing swirl propeller centerline andthe outboard radius line werealtered to haveseven degrees incidence representing swirl downwash less For reverse propeller rotaton. theupwash/downwash boundary conditions areinterchanged Clockwise andcounter-clockwise slipstream effects on thewing pressure field arccorrelated un Fig 38 using a transonic small disturbance method (Ref 56and 57) Note thatpressure field details arepredicted very well despite the wing's complex double shock wavesys.em The wing pressure distributions found usFig 38 can be integrated to generate spanwise load dsmbutions asvel as lift anddrag coefficients (seeFig 39) Loading plots reveal that the slipstream interaction will affect the hft-iduced draglevels, the alteredshock pattern suggests the wave drag is similarly that altered The lift level is predicted well as might beexpected (good pressure simulaton). but computational Orag the increment is greaterthanthat measured during the expenment This appears be an improvement over incompressible theory to but draglevels measured this particular experiment are in suspect because thepropeller slipstream simulator hardware of mounted upstream from the wing surface Rotor slipstream interactions aremore prevalent at subsonic thantransomc condiotn While little success has accrued in modeling thecomplex flow separation patIerns about helicopter fuselage shapes, someadvances havebeen madein simulating rotor download effects on winged vehicles (Ref 5S) is Figure 40depicts this problem. It known that the"download" or veritcal drag force penalty attribuutble to XV. 15rotor downash impinging on the wing surface vanes between 5% and 15% of the vehicles totd gross weight It becomes impoetant to refine configuration components to minunre the download magnitude Unlihke most aircraft prediction applications, this caseinvolves dragcoefficient levels that are very high (on the order I 0) of e plots in Fig 41dlustrate thatwing section drag uscrossflow vanes with te flapdeflection angle The agreement between testdam andthe computonal model is compromised bya shift unabsolute draglevel If the curves ar normalized .y the0-deflection dragvalues, it can beseenthatproper uend: ar ,nredieted This trending was obtained using an unsteady panel model (Ref 59) coupled with a free strearire representation of theseparated wake Useful design information can beexuacted eventhough absolute drag levels predicted are in error The speed with which computational models can be generated. coupled relatively low cost of analysis, pemuts a large with number of shapebonentation combinations to beexamined. In this environnent. out of.the.ordinary solutions can surface that mght not naturally evolve for testing based pastexperience. on The wing section download poblem is good example that a in thenummum drag or download doesnot occur at90 degrees of flap deflection, which would bethe intuitive cboice based on minimum area presented the flow to Engine nacelle interference phenomenon canexhibit a degree of complexity that defies treatment with current computational

------TOPVIEWOPWING

WING
-

NBCARD SIDE OFSTRUT (A) CALCULATED SURFACE FLOW PATTERN NEAR NACELLE STRUT WING INTERSECTION, BOEING 747 DEVELOPMENT MODEL 2 0 DISTANCE UPSTREAM ALONG BODY hi WITHU FAIRING A Dconditions BOUNDARY LAYER SEPARATION THEORY -. EXPERIMENT JUNCTURE WING BODY (I) Fig. Component Interference Effects (Refs54 & 55) 35

_,

Fig. 36 Proptan Tractor Arrangement

PROPELLER SUPSTREAM

SIMULATOR

miethiods opartis the Thiin result of whit might becalled 'hard


boundary" iteractions that impact lift-induced. wave, and pressure levels, But in drag addton, the nacellc iet &M exhaust components add a dege of dThrottie Dependent" drag de nibd u a preceding sub-section An example is now described that can beused to dlustrate boh of these effects and themannerus which they interact

m,

., Fig. 37 Propfan Slipstream Simulator

7-26

.0 0 08

Ci

*o.o

0
V

~
"0

04,03 080 121

it'-0

L~r,.N AAYSIS EXPERIMENT (RE 56) o000 SWIRL .7' XX 660 SWIRL.7 _

Fig. 38 Wing Pressure Distribution Correlation Illustrating Propfan Slipstream Interfence Effects (M =O.8 ca300g)
EXPERIMENT' PT 7 .1075 ANALYSIS P7 ,pT -150

000 SWIRL.r
*00 08 0 SWIRL-7' 0 V "'00

PROPELLER SI.IPSTRE' W SIMULATOR

06 (INCREM CCj 0 4 T.C0 02 DRA NT, L

0IF

6L DEG 12 SWIRL, IANALYSIS *0 EXPERIMENT

10

02

04

06

06 lk

500

TH4EORY NC SWIRLDEG

Fig. 39 Proptan Interference Effects - Wing Sponioad, Lit, & Drag Increments (M- 0.8 a =3 Deg)

rigiur

42shows supenmposed computed wing section upper jet surface pressure ditributionis for an esocuniveattransonic conditions Auirnsons smldistbancetheryanalysts (Ref 60)is usedOntheleft. wing-fuselage calculation reveals a a strong vsock wave running along length of the she wirg from the fuselage juncture iothe wing sipJuNt below this irc. dosensional Image,wing roottection cutisshownd4ong a with expentmental pressure datcorreluoion Thluirepresentsia 'nacelle-oft easeNest, acompartuo the with nacelle surface preseni beidentified The can nacelleismodeledaclosed! as surface thecomputatiornsl method. other in In words. thenacelle isaclosed formi would be ppropriate for afuel uuik as or avionics pod Note thaitheengine surface deceleratesflow the in front of the nacelle and acceleratesflow just below fte the spike. Wing pressure nacelle lip foenungpressure inlet a corrlastions confirm this type of character but level of the agreemtetnoti good that thenacelleff case This is as as for suggests part of thcnacelle itterfererce effeci is msodeled that

butapiece still missing In the comparison (Fig 42). the is third tue inlt rass flow ratio (0661ismodeled byspecifying appropriate values the 'low field potental atgrid points of that representinginlt face Now atcanbeteen agreemniet she with test bas data inproved considerably The flow. now charactered bymsore negative pressures, notstow down does asmnuch front of theinlet face.with less in spillage modeled. ihe flow acceleration or pressare astheinket is redaced spike lip Simulation agreement is diraacadly improved. Asnoted before. absolute Ichc predicied byathree. drag dimensional computational method will not yteld drag accuracy ievels suitable project applications, liut theinformaution for shown in ig 42canbyfoundational for cte application of simpler iodrilsia methods might brought to bear that be pars 7the using superpossition principles as of abuild upprocess basic for this approach %we described in the Trntionic, ideas and Sup'ersonic Drag' Wave tub-section

7-27

SLPTREAMBOUNDARY ,

0O -,A A
Ai.J

DRAG COEFFICIENT

0 EXPERIMENT A PANEL CALCULATION


/'(

I
CROSS SECTION THROUGH WING STATION

SEPARATION AT FIXED FLAPSHOULDER

0* 10

30* 0 90, FLAPDEFLECTION ANGLE A (ABSOLUTE)

I\ 9

BF

DRAG COEFFICIET 05

EXPERIME4T CODE A PANEL 0 SECTION A.A 0


-.-

30' v0
Fig. 40 Rotor Download Schematic (Ref. 58)

60.

,,,

90'

FLAP DEFLECTION ANGLE B(NORMALIZED) Fig. 41 MeasuredlCalculated Drag vs Flap Deflection Angle (Ref. 58)

.2,

CP -06 ,o o
I2

W.UGFUSCLAGE NACELLE OFF 48 ,04 0 0 .


0)

I
NACELLE ON

WING FUSELAGE NACELLE

MFR.066

NACELLE ON SPILLAGE INLET MFA.B0E6 EFFECT

Cp 04

CP00

08-

Fig 42 Superimposed Computed Pressure Distributions & Wing Root Pressure Correlation for G-Ill Configuration (M= 0 85, a . Deg)

7-28 In the following paragraphs, another interference drag source is highlighted. This source closely related Throttleis to Dependent Drag, but thecomplexities of aircraft afterbody shapes requirean additional level of modeling sophistication beyond what is described in thatsub-secxon The category to be examined now might be called "Integrated Afterbody Effects" andit can be thought of asan element interference drag of Theboattail analysis described in theThrottle-Dependent Drag section hasbeen implemented to treat a number of simple nozzle shapes Figures 43 and44 show r-sonic, drag prediction results for two boattasl geometries Applications engineers extend can the useof these axi-symnsemc body computational methods by implementing the equivalent body-of-revolution technique described in Ref. 61 This approach requires that various afterbody components (e g, multiple nozzles, inter-fairings, vertical surfaces, fatrings) be tail and sponsons, horizontal and combined into a single shape with anequivalent total area distrbuson A prediction generated using this technique can be seninFig 45 Dragrisecharacteristicsforaresearcha.rcraft o -. EXPERIMENT AFTERBODYISTING SUBSONICTRANSONIC ANALYSIS SUPERSONIC model havebeenpredicted well considering the afterbody complexity But this engineering approach far from foolis reveals in Fig 46 Here, test comparison case proof as a second results suggest a drag-producing flow separation region that mght exist at low speeds.While this experimental trending is unusual, may in fact be in error, this case be usedasan and can example to point out that the simplistic engineering method involving an unrefined equivalent body-of-revolution may not be suitable for all project apphcatons Some theshortcomings just described can be overconse of by integrating a boattal analysis method, anequivalent body-ofrevolution concept, andcorrection factors developed from test databases Figures 47A andB present schematic illustratig a this approach.Semi-empmcal corrections can be developed to account a number of aircraft features suchas empennage. for inter-fatrings, enginespacing, booms, base drag.andlifting surfaces. Figures 48 through 53 show examples of correction factors, configurations, models, andcorrelations that make up theappliedexperiencebase A sample analysis for an F-14 afterbody using this system(Ref 62) can befound in Fig 54 Trim Drag Over thepastten years, NASA hasdeveloped a computer program specifically for optimtzing multtple lifung surface. multi-control surface arrangements (Ref. 63) While many methods aght be usedto mninnze mm drag, this linearized lifting surfaceformulation is unique in thatthe solution is to as process opposed the simultaneous obtained byan itemraon equation solution process characterinzng methods A other special implementation of leading edgesingularity parameters makes it possible to rare accurately predict leading edgethrust levels (Rcf 64) Computational tme savings associated with this approach areimportant in view of the large number of surf ire deflection combinatios that must be ins estigaired to effectively minimize configuration nim drag, Perhaps of the one most imortant features method not of this as the it so much the factors thatchaacternze but rather extersive base experience completed bythe method's developers This provides theapplicatons engineer with critical inforiation needed assess methods ulity. An example configuration to the hasbeenmodeled is Fig 55 One useful feature the of conputanonal nethod is thatautomatic panelmodel generation is provided An input stick figure" (Fig 55-A) is thebasisfor thecomputer model (rig 55-B) The control effectors appear to be larger thanwhat might be expected from the inout nsodel boeaise eachflap-type control surfaceelement rve.iecs a deflection anglr boundary condition, butelements across the hisgelise haveapropnately reduced bound.y condition angles All AreshadoO, however, giving the impression thatthe deflected surfaces modeled are larger thanthe physical cointeparts A total suction pararter "Ss' is used a measure of howvell as the configurations drag polar marchesieidealgivenby"e' 10 (Eq 5) The Ss relation s

BOATTAIL GSOMETRY 012 0 0 0 0 08 - t o-. 006 TERM 1578 - " 765 025 0

004002 L 0 0o 07 , 0 0n 09 t0 i2 14 t6 t8 20 2 2 MACH NUMBER

Fig. 43 Prediction of TrI-Sonic Drag Characteristlic of Boattall-Sling Conflguratlons (Case No.1) EXPERIMENTAL DATA 3ACBOATTAIL COOE PRESZit978) NASA CA3020

- --

BOATTAIL G

OMETRY

10 08
05

CD

9 06

04
02

CL TN'4CdCtL I-ACD
CLTACti/CL4-

C/KAR

(29)

0
-

.2

5
MACH NO

,0o t

14

InFig 56.swil varybetseu,'O andAl n Ssvalueof'l" represents an idealdragpolar-hile represents zero 0 the suction polar where drag is simply the lift vector component in the downstrea direction

Fig. 44 Prodlcilon of Tri-Sonlc Drag Characteristics of Boattall-Stlng Configuration (Case NO.2)

7-29

EQUIVALENT OF REVOLUTION BODY (EBR) (HORIZONTALNVERTICALOFF) TAILS (EBR) BODY EQUIVALENT OFREVOLUTION (HORIZONTALVERTICAL OFF TAILS 05 RD 03 01 03 01 0 02 06 10 LID 14 10

0 4

0'

'2 1.6 2.0 24 U2


00

28

8
.

A AFTERBODY DRAG -TEST DATA 0016" * AFTERBODY DRAG MINUS z/ -EMPENNAGE 0012 EBR AFTERBOQY ANALYSIS 002 lA O00
*

A AFTERBODY DRAG -TESTDATA 0 AFTERBODY DRAG MINUS EMPENNAGE 0 016 EOR AFTERBODY ANALYSIS 0020 z U 0012 0008

06 1.JZom

07 08 09 1B MACH NUMBER

11

12 ,

0 4i,.-,0 (1

=
____________

00

06

07

08

09

10

11

12

Fig. 45 Equivalent Body Approach Simulating Research Model Afterbody Drag

MACH NUMBER

Fig. 46 Equivalent Body Approach Simulating F-18 Research Model Afterbody Drag

1GEOMETRICAL INPUTS

MB FLOWFIELD NOTE PROPERTIES M8

MB

MB

MACH NO M.0

METRIC BREAK

REY1POO NO pcDV FRICTV4 OORA DEL 3 THETA


06

OUTPUT)

NO

IUSMACH

NO

~(

PSNOZZLE. MACH NOZZLE EXIT

Fig. 47-A Equivalent Body of Revolution Methodology Using GAC-BOAT Code

GA-OTIPTPARAMETERS AXIOR2-D GEOMETRY CROSS. SECTIONAL AREA DISTRIBUTION

...... PARAMETERS NOZZLE INPUT FLW CN FLOW EOM T,. AS CO-V A2 SERN aETC... I"

M. Ro s*

GAC.BOAT

PROGRAM
----------------------------4NOZZLE PARAMETERS EXIT o -PS ,*MACHNO OUTPUT PROGRAM NAIP MODULE
* -

.SEMI-EMPIRICAL 3 D CORRECTIONS

IUS T NOZZLE GE LOSSES NOZZLE COEFFICIENTS MIXING _ LEAKAGES PERFORMANCE NOZZLE WEIGHTS CALCULATIONS

/
C 0

-a- COMPAEN DA DELTAS

FUNCTION NOZZLE iRLUST VECTORING THRUST REVERSING o NOZZLE C0

*WEIGHTS

BOAT GAC

3 EMPIRICALD CORRECTIONS SEMI


PPOGRAM

TEST DATA %CD SOLUTION ANALYTIC

CO

NOZZLE Cot EXITPARAMETERS *PS

NO ACH

MACN MAC

F. 4....e.... zeAfebd Isald elrronePedcinPiga

K
*0SID-229 V SID-165 EBRMODEL LI CORRELATION BOOM

7-31

AREA BOOM

MACH NUMBER

[,C

OEL

C F

Fg48ERB/Spacing Effects - Cruise Nozzle

Fig. 51 Boom Methodology

US/D.13 OSID-11 0SID. 229 05/D.215

a SO-2 29 X 5/0215 v 5/D .2 29 1-5/0.215

0 15

...........

00

+ a00

-0

0 (L BO 02

ILZ O NFI9

Z2R 0NP

05

2 1.5 NUMBER MACH

25

Fig. 52 Boom Correlation -Cruise Nozzle

Fig. 49 ERBIpacing Effects - Maximum AID Nozzle 015S TEST GAD SERIES GFAXI TEST FLUIDYNE SERIES COWJFGA

FiiCONFGF

'.

MB

Fig 50 Boom Configurations

Fig. 53 Boom Correlation - Maximum A/B Nozzle

7-32 NOW. E CID C 2AR /'1


7

< &
-C

L--MEASURED y0

'AFiBODY

FUSC EMP DNOZ

-r-J......
0

12TH SCAL.E METRICPRESS INTEG


-C *C 0

FRICTION &C CALC

Ss.0 S

C'AFT-ODy LEVEL 11 C R

ESR DEMp DINTERACTIVE -ESR DRAG COMPLETE OF AFTER80,Y AREA DISTRI

Ii
DRAG -, -I '

I (LEVEL d
IIIII I -~l Co, LItWO EDU AGCOE 0D FFENT

COEFFICIENT jl. /J4 UER COEFFICIENT E

IF

v'OCOOAC~TC7t

Fig. 56 Ss Parameter Definition 20 24

04

08

16 12 MACH NO

Fig. 54 F14 Afterbody Drag Prediction - Maximum A/B Nozzle

(A)INPUT -STICK MODEL-

(B) DtSCRETtZEDPANELMODEL

Fig. 55 Trim Optimization Code Modeling

iypsl.d setof computed resultsfor Awing Figure 57 showsa body canard configuration Thre surfaces (wing leading and 'aling edge flaps andthe canard) be deflected ix setto can mitxtttize drag across range tf. covering both cruiseand a of maneuveing conditions The control surface deflections, S$ parameter, andconfiguraton pitchtng ntiment are all irapped (F g 57) andshow the varatitons that ocir with different total lift levels At eachof five hft values, three plots cover canard deflctoxofn,tO. IOan 20degres Eachplot identfites th performance level possible ,ith %anous combinatuion uf wing lding and railing edge deflecoons flap

an By combining theInformiion in all of the plots. optimum consistent with trnmng theconfiguration defleclon schedule canbecoistricted (Fit 58) It tan seen thait more ideal be a polar can be generated theconfiguation doesnot have to be if timed These idealdeflection combtniions lead to a conposlite di.g polar Tihs picess is very tuneconsumintg and expeisnise if a complete surface dlhlcoiVoonentatun matnrx is performed in thewind tunnel References 63 d 6.Ian be sed to identfy dragpolar accur.c that mighi hexpected for a range f dfferext aircraft configutaion types

7-33J
2 . 005 20 020 .010 0 % -05 -005 0 0

401 1~'F
40 -01 .005S -00 415 -020 .00 .015 .020 -025

x rA0 -005

I
074

ooo
06

5020~

I. 00

0.1

.010

.05

111

41

405

00

20
IsI

T1
500

415

.20

060

20

4055

005

405

4065

410

405

40

420

I1

0U
002

I
070

5050

-0

56

05kA

20 1

20 % 1

005

a0

00a

40S8

015

-405

.270

41S

420

0 I0
0
. 70\

\, .

0.4
on8
so O

00

20

Is

072

6a0.0.

I6a lo* o anr o1gu na


elcin deFa Iednrri

Fig~~~~ Prdce 57 Lif-Inuce DrgCaaceitc (sCnors


Difrn Leel as a Functio ofCnr Lif

7-34

____________-

-PART

2: DRAG REDUCTION/MINIMIZATION

-8

TECHNIQUES
D SCUS ION DRAG REDUCTION /M INIM IZATION

Z 0

(A) EFFECT LIFTON OF OPTIMUM CAADDEFLECT1ON

were In Part 1,a number of engineering dragprediction methods

0*

<

affect drag reductions thatoptimire the configuration shape The of a collection of case studiesthatillustrate the useof for computational methods dragreduction/imiztation problems anddiagnostics The topics include (I) Wing Tip-Mounted Winglet Design

t~o -consists LEADING (B FFC IF NOPTIMUM O DEFLECTION (B FETEDGE TRIMMED UNTRIMMED

01

~5 1 1-'.a

I I(2) ..... 4

Transonic Transpot Wing Drag Minimization

(3) Natural Laminar Plow Airfoil Design


(4) Fgbter High.Speed Drag/Iluffet Reduction (5) SweptWing Tip Optimization (6) Hypersonic Drag SourceDiagnostics righter Transonic Maneuver Wing Optimization In eachcas, an effort is madeto identify thevalueaddedto the project byihe application of eogisenng draganalysis methods

01
0 (C) EFFECT LIFT OPTIMUM OF ON TRAILING EDGE DEFLECTION
-

I
-

[(7)

CC tn
7
-0

Wing Tip.Mnunted W~inglet Design 1 1 Wing np-mounted wingleis provide a reduction in airenaftlift induced draglevels thatcan be antributed end plateeffects A to

1 1

LIFTCOEFICIET CLspan The reaalti'g; wing root bending tmoment less is than ihat driv frm asimle ingtip extension sostructural ieight (D)EFFCT LFT N OTIMM F SUCTIN PARMETERpenalties are minaiied To be effetctise. however, the winglet lifi induced dragbeinefit mustnot beoffset bya large%etted dragpenalty This differetites nicts from friction OPTIUMOMPOITEsurface simple eod-plates. It also nnggents designdetais are thai UNTRUMEDImportant. Further, many wsngleiapplications requiredesign transonic speeds whereshe applications engineer musst ensure OP~tM COMPSIT thatinterferencl: effects andwavedragpenaties arenor again,design details areunpisnan Once LIFTIncurred POLAR

at

DRAG POLAR COMPARISON (E) DRAG


-____ ____

Fig 58 Optimum Drag Polar Shape Constructed Using Multiple "Lift" Polarns Generated with Control Sutface Deflections

empirical guidelines (Ref 65) developed by Dr. Whitcomab at NASA's Langley Reschb CenterKeycharacteristics ate sumnmaaired belou *Avoid rake thatwould reduce wsngtip region loading *Wingler spanshouldbecomiparable wing tip chord to length Winglet toe-out angleis comparable to wing tip twist angle Avoid merging of wing andwinglet supersonic regions Winglet cant angleshould be selected optimum for combination of induced dragandwang roo bending msolnent *Winglet juncture region should be treated a wing roo like juncture region Asoid trading edge load build up The final threeitems can beaccomaplished using both subsonic pasnel methodsanld transonic finite difference relaxation schemes. Figure 59 identifies typical winglel parameters

7-35 N=I: WNGLET SECTION UPPER SURFACE -ACESINBOARD

WNGLET H..C SWEEP

hZ

ZI

L7

7h
C CANT AN4GLE
..

ccZZIIIZz-z
G IIIWINGLET AIRFOIL. PERFORMANCE) HIGH FOR (MODIFIED IMPROVED SPEED Fig. 60 Originally Proposed & Final G-Ill Winglet Airfoil Section

ORIGINAL -TYPE GAW SECTION

UPPER SURPACE

ANGLE OUT TOE

SURFACE UPPER

SECTION WINGLET

-0

iin5~M25~7A

ODESIGN CL~

Fig. 59 Details of Wirigini Geometry

During the winglet design effert for theGulf'ucam, III aircraft. GAW type theseguidelines werefollowed and3 recommended airfoil %as Implemented The airfoil shape beseen Fig can in 60 During the first wind tunnel test,early dragrie %w0FLsH observed attributed to the arid GAW sectionshape ibis effect was verified usinga transonic two-dimensional tMaSottiC analysts(Ref 33)coupled with extended sweep theory (see -ransonie andSupersonic Wave Drag* section) Iterative anaiyses Performedre-contour the airfoil forward andafi were to regions. tnaking theshapemnore supercrtical in 0haravsecr The (lug 60) wasplaced backin the , inglet improved arfoil shape planforin for wind tunnelmodelfabrication All of the design wind iunnel testi goals-ere achiesed during the second (3uif'srram IIl winglet dragincrements shownin rig 61 are The benefit/penalty trade heidentified Testing can demonstrated a benefit that wasslightly greaterthanthat predicted by analysis Also. gains reduced wavedrug are b) lossesata higher-thant-dessgn number Stacb coupled with higher than-design lift levels Theaircraft performtasce benefitsTal5 thatresulted from this design effort are summarteein Table S dragrise) (to in Applying computational methods this cawe delay

COEFA'CIEPJT C WINDTUNNEL TEST 0 M 0750 TEST . M 5 IFLGTTEST M - 0 7S 2

0 0030

00020 00020) 011010 0 00020 AC DRAGCOEFFICIENT INCREMENT 0 Fig. 61 G-1llWInglet Drag Increment

al

oul fWige igo eulso

saved considerable model and time test expenses Further,is it judged thefinal wingiet drag that benefit would not been have us test high considering the constraints characterizing most time programs Designing byrepetitve testing often resnults in con'pronixed final configurations due liats onthenumber to of variables that be can assessed is particularly thecase I bas for aircraft with trautsonic, points design

DRAG

Experimental Validation 3 1%DRAG REDUCTION ATCRUISE CL

WEIGHT WEIGHT .112% AVERAGE OF CRUISE RNE . 7M(33% RAG I7M13% FUEL -6500 GAL/fR (BASED 600 HOUR ON USEI .I

7-36 Transonic Transport Wing Drag Minimization case study to The Gulfstream I executiveet is a good demonstrate computaional drag unirruzaon techniques applied to high-speed wings. This is bestaccomplished by examining the series Gulfstrcant designs from 1966 of (Gulfseanm 1) to 1984 experimentation was the(Gulfsiream IV) In high-speed wing meansto optimize 1966,design by shapes The Gulfstream i program fabricated eight separate wing shapes wind tunnel testing Including refinements to for theongtnal eight, twenty-four wing shapes wereevaluated and thebestwasselected the M=O cruise design point This for 75 may seem like a large number of wing shapes, for thatperiod but some aircraft development programs investigated hundreds of before settling onthe final one. wing shapes TheGulfsutrean configuration evolved dating 1979.An effort Ill wasinitaied to develop a new high-speed (M.0 78) wing by retaining theG-11 wing boxsinictrire andflap/aileron surfaces 62. and The G-11 G.11 wing planforms arecompared in Fig, Leading edge wing tip extensions increased anid wing areaby 15% Leading edgesweep Increased degrees aspect was 3 and ratio wasmaintained at6 0. The wing root chord extension decreased thickness from 12%to 10%.Equivalent twowing dimensional airfoil sections wereextracted from theoriginal GII wing at threespanlocations ( n - 0.182.0 351.and0 937) Theseshapes wererefined using the transonic airfoil analysis methodof Ref. 33. subject to theextent constraints identified in Fig 62 Refinements evolved using iterative direct analysis Figure 63 showsa wing pressure distrbution "before andafter" comparison whereshockwavelosses werereduced considerably Initial testing revealed thatG-Il drag creep chanctensncs wereeliminated andthis wasconsistent with the specific pre-test predicted wing section dragreductions included in Fig 64 - r 08 CP-04 0 04 . 387 NGSTATON IG-I1 0 937

.12 08 CP 04

WING STA]IOI146 5 1 GI0

I
04L

12[ 08CP 04

GT

Clearly aportion of the higier dragdivergence Mach benefit is


atti table to reductions is wing thickness ratio Equation 25 Shows that the inboard drag divergence Mach beneit of0 04 is greaterthan thatassociated thickness. 002 Figure 65 with qulatifies the total configuration drag reduction based on vcnrleatiom testing This

W/NG 7 IG II 0 182

Gi
"'

/ ,.

I!/

trinsonic wing design effort

underscored value of computaional methods The the performance enhancement so substantial thatG Ill wings wa werefitted to many existing G IIaircraft. These -e re called G IAs Pc most Intrguing, the hybrid aircraft nrw -, erc o; achieved with asingle wind tunnel test that verified computational predicuons

0"

~aps

a, .14s.

G iiiAMOILFig. 63 Airfill Shapeiwing PrOSSure Distribution


SKAP<00 o XG 0 004 I G.1 Refinement for Guilslream Ill

]
"/

(,02 STA AC0075 <2 0 s O2 STA145


0

'I

0 Fig. 62 Gulistream iVill Wing Planforms with G-Ill Airfoil Contour Modification Region
__________________________

t,

088

072

078

080

MACH NUMBiER

Fig. 64 Analytically Predicted Improvement In G11 I Wing Section Drag Divergence Mach Numbor

7.37-

shift loadinboard tocompensatewing load for suppression induced bythelarger engine nacelle transonic wing-body. A nacelle-winglet computational analysis method 60) was (Ref used develop appropriate wing contour changes the to in /Iunconstrained region (Fig 67) Care taken to inimize a was wing tip lower surface leading edge spik: that evolved with the EXPOISIMENTi twisting needed shift load inboard. This pressmse to spike AgoDI44COE Ngts required treatment avoid drag to creep would appear to that prior 241 W(1ST9T66)PUAIOA achieving the cruise design Mach number and level. Figure lift 68shows refinement implemented the DESIGN AERODYNAMIIC
G-1:1(1979)

1 WtNGNJNGLET TESTED 040

C' 0
u, t;

II
0 C20

/f

Is

15
I 1

G.IV design were goals achieved with testing immediately that followed the computational design program. tussel Wind

well pressure distributions agreed with ptc-test projectious The


wing pressure (Fig. showsdransutic field 69) a reduction in

outboard shock wave strength; thisraised the GIV cruise Mach


number toM.0 80. Perhaps interesting, from the most improvement shown in Fig 70where isa sall drag there benefit whser 0-1ll and the 0.lV wing-fuselage configurations arecompared the With nacelle installed, however, the perforntance iscrementlarger case is This illustrates improved component integration capabilities through computational
analysis

Q0 GLvnEA

IPOVEMEN7 tI~n n M~n l


MACH NO

024 1---4--4----460 64 68 72 76 so

I
84 60

Fig. 65 G-IIIG-1ll Comparison Drag Natural Laminar Flow Design Gulfstreara IV developmseni 66)was (Ref tiit, d in Mia ech 1983 Unlike the1979 effun., G.IV wing would re G-11 the redesigned structurall) feeueight re4.ictirn This nade it
possible to affect additional aerodynamic refinements could that reduc- dragand increase range Theprimary designchallenge centered geometric characteristics of the largefuselageont mounted enginet with 50% ore volume than bhe 11', lay G Spey engines Wing shape the 65-4 .h d location wasto be maintained .n past ant effort to preserve cotitri sitrices 2,d Ldies G-Il reduce deselopmencrt manufacturing costsFuselage arid stnixiure was

Another technique fom reduction/misnieiatton ir.olves drag


achieving thelowet levels of friction dral; ihut arederived from maintaining laminatr flow over aslarge asurfaceas possible. This might involve poweredcoincepts where suction isapplied to rerion. low-energy boundaiy layer flow, or shapes might be contoured to achieve favorable gradients thatdelay transtnon to turbulent flow, Thutsecon.ltact,called natural laminar flow, is only applicable to some aurcraft design concepi whereoperating cimrnstanaces manufactunsng/matmenance anad constraits make it feasible Wing sweep, example, mustbemoderate fee to inhibit instabilities in the lamna bosindar) flow If ihese layer requirements aremet, considerable advancespecrformance io can

leftuntzuched byconstraining thewing fuselage p...ture


contouts Coussicig thsese contasraints. mosteffective wing the design wiould reduceoutboard shockwavestrenghl (Fig 66) arid

beachieved applying modern computational drag by


minimizations tools,

Ca

WIJNDb 0 TUNE DAT

O FLIGHT DATA TEST Fig, 66 G-1ll Wing Pressure Distributioni Comparinon at Cruise Corillhon (Ma0 78, a 4*)

7-38 G IV BL450 SECT.ON MOD G.IVPLANFORM 29

~
6-Il LANFORM

G-IV AIRFOIl
MOD REGION_______________

ORIGINALCONTOUR

-----

CF P

ORGINAI. Cp

MODIFIED

Fig. 67 Gulfstreamn 11111V Pianforms %%lth Airfoil G-iV Contour Modification Region Fig. 68 G-lV Wing Tip Mod ta Eliminate Lower Surface Pressure Spike, M 0.78, a 3.5', 11 0.95

CF 0000rE

I
-.

C400

1200-

EXPEHIMENT

ANALYSIS
00

-04000000

--

0400

Fig. 69 Correlation of G.IV Wind Tunnel lest Pressure Jistributions with Pro-Test Code Prediction =4.0*) (M=0.78, ao

7-39

4
G-Il /

<

AIRFOI.

NACELLE ON

to

20 304050 0 708090100 (% X-ORDINATE CHORD)

Fig. 71 NACA 64 2 -015/Advanced Airfoil Shape Comparison

NACEILLE OFF

..L

08

R-

'
085

L 0 T-

6O FREETRANSITION 6 .TRu-36X-0 *TUR8ItN

6!
,
-0

5 -

MACI4NUMOER

0
s(A) o 5 0

070

075

080

', ..... '

01
08 ----

t 1
ADVANCED AIRFOIL

-4!

o0
'5

........ R..72X t0, FREE TRANSITION ARAA7.2X106


.

TURBUAON

04
/4

Fig. 70 OIV/G-Ilt Drag Reduction Showing Engine-Airframe Integrat ion Benefit Eppler s method (Ref 22) w;As discussed in the "Skin rnction
andltcssu~rcDnS'svb-sect:ton Because ofviscousflow

o2
0
-

0DD 0004

0006

0008

0010

0012

simulation conplexiiy. theseedfor a predictior experience base was stressed With anunderstanding of simulation formulation assumptions and lirhutations. theapplications engineer use can Epplu's methodto synthesize win& section shapes for new specific applications An application is inclufed hee io A the power of thi. engineering method NACA 6-Series wing sections %etc de eloped to exploit the perfon.nce gains pos ible with various extents of natural lamriar flow Many aircraft nowin production employ NACA 6-Scncs s.ct"'r Forthis case, NACA 6015 symmctric a sectionsees at a baselineEpplcr's methodis used togenerate an advanced seclion with unproved dragcharactensucs Figure 71 showsthe NACA and advasced Aifoil section shapes plotted to an expanded scalerevealing details thatare quite subde It.a be seen theimproved airfoil features a comparable nose that shape I is slightly thicker between 20% and60% chord while being somewhat thinner between 70% and90% chord Figures 72 A and72 B illustrate improved dragcharactensucs the achieved At two Reynolds numbcrs It might beexpectd thatthe largest dragreductions occur at the lower Reynolds number, and this is thecase It is Important to identify th theimproved pcformwc c e in several forms First. th extent of lamnar flow is increased evidenced bylower drag in thepolar as "buchke region Second, inci=ce rangeover which lairauna the flowcan bemaintained is broadened Third. oec transiiwn

DRAG COEFFIECIENT Co (0) NACA 642-015 AIRFOIL Fig. 72 Predicted Drag Polar Characleristics for NACA 64 2 -0IS/Advanced Airfoil Seclons

occurs (above CIL- 0 51turbulent flow drag level is Joer than that for the standard NACA section. Finally, all of these beneficial charascinetics are achieved with an airfoil shape havingsuwre thickness - a charctensm that vucId translate into a stncturt weight savings. It should beapparent considerable progress hasbeenmade that in airfoil design since theNACA s'ctions were conceived The advanced airfoil described here was Senerated with very simple parameter spocifications to tpple's method. Given sonic time. it is likely thatsecton shapes with lirger improvemints can be synthesized Perhapsiore important than performace in sonne applications. compulatuoal techniques of this typeallow the designer to tailor airfoil section charctcnstes for specific design problcm with unique goals andconstraints nmght This becontrasted the alternaive that involhes settling on an jurfoil to seiton thatfeatures someof thedesired attributes but its primary benefit is derved frrm the fact thatit already exists

7-40 Fighter High.Speed Drag/Buffet Reduction Flight testing revealed thatsubtle alterations to theF-14 nozzle, sponson, inter-famug, andcomposite region (Fig 73) required to accommodate new F110-GE-400 engines compromised the aircraft buffet boundary during acceleration at low altitudes Transonic buffeting occurred atlower Mach numbers when comparsons madeto theoriginal configuration In were addition, the intensity was higher. Many months flight test of diagnostics and"field fixes" resulted in no satisfactory solution of theproblem It wasalso unfortunate thatthethrast to find a solution assoonaspossible limited fluid mechanical phenomenon test;ng needed develop a good physical to understanding of theproblem source. The change buffet character was most noticeable bypilots at in the M-0.95 CFD was implemented because closest available sub-scale testdata wastakenat M=0 90 Initial flow simulations wereperformed using transonic smaUll-disturbance analyses Unfortunately, modeling constraints precluded any representation of the vertical M=0.95 analyses tails provided no insight into relevant buffet problem flow physics the Valuable diagnostic information (Refs.67 and68) was eventually obtained using the Navier-Stokes formulation PARC3D (Ref. 69) Figure 74 shows the surface gnddiug for the afterbdy region of the original andmodified F.14 configurations Modeling sophistication was increased by adding the vertcal tail surfaces. Comparisons with available M-0 90 wind tunnel testdata (Fig 75) proved that simulation fidelity wasquite good The only discrepancy noted canbe attributed to wake modeling limitations Computations performed at the Mach number of interest. M-0 95.eventually highlighted the

coM.)SITE IIN

SPNON /

problem area Figure 76 illustrates alow-pressure area


terminated bya shock wave on the nacelle afterbody It was conjectured that increased shockwave strength in this region

7)'

was source the the of buffet boundary shift Compnutations were

Wu.1-iswt

NOZZLE FLAPS Fig. 73 Modified Aflerbody Surfaces

it

tA

06 .04/

0 TESr DATA T) (W - PASC3 0 INS ANALYSIS) /

A B

~~-06
02 Cp -02
00 6 p.02 CPcci

t M

O4

F I40 "' Fig 74 Surface G3rlsldlng tNavler-Stokes Analysts for . ,4. ,

0.2 6000

7250 X(INOCHES)

" 600

FIg. 75 F-lnA Aflerbody Prosisuro Con-ilatlon 2t M : 0.90

-06 06 *05 M -04 0 i03

617-41
Mt L"132 ,ML' 126

---

F-14A F14D

~-01
.01

02

.n

V""

lbF-14

Fig. 76 Predicted F.14 Afterb-ndy Preacure Fieat M = 0.95 thenusedto identify mans to weaken shockwave One the inboard to back. solution involved deflecting therudder surfaces pressure region Thegoal wasto reduce this pressure expansions to thelevel found on theoriginal configuration Figure 77 provides a conipleit. three-dimsensional of the flow view indicated thata expansion/shock characteristics ComnpuratooxS degrees io comipeniate set for rudder deflection of 4 degrees(6 expansion pressures shown but as loading effects) would reduce Much nsf slr (4*11 in addition, the redaction in local maximsumr ar 78 and79) would alsoresultis a netdragreds.uxvn canbe in sees Fig 80 Flight iesing perfoermed August 1999l in confirmied computational predictions In this ease, the the buffeting problem wasnor idlentified bysub-scale testing Thec analysis lcedto a valuable engineering design solutin thatwav not identified during a flight testprogratri.Fursher. drag benefit. reductions wereidentified asa secondary Swnept Wing Tip Optimiration Over thepast' years. investigatirs haveexplored the possibility of rodamorg lift-inuced dragbyseverely sweeping taperng and wing ups The mrotivation for doing this isdersveiifroms observing the shapes bird wings ansifish fins recognizing that of :hese pianfoecss evolved naturally over mrillionsof yearstu have forms rmosi sent andcompetitive shapes survival tte effii for Cotriuwat~oxal inethods wereapplied investigate this effest in to thierud 19M~. in beidentified in Refs 70 and"1 IInitial ass pmedvvtvons 4drnified lift induced dragreductions on the ordet of 30% lBnt large benefit was theresult of a faulty drag this calculation scheme"naserscal drag"atzero-lift wasnot property remioved from the predictions tiade at incidence for When thedrag force wasadjusted this problem, thebenefit wascloserto 3% Perforitance gainsin this range havealso achieved hydrodynamiuc in applications (Ref 72) been Reveutly (Ref 73). NASA perforsed reststo gainmoiretnsoght into the lift induced dragreduton mechanism The miodels exanined hiavebeenincluded hee asFig 81 Testresultsverify A performance: improvemsent theorderof 3% (Fig 82) on [Drag redwvutis derived from swept bAck wingtipsippea u, be verysontroveruji in aocmautational senseOne soree kRef 74, ..laims th..i thereis little benefit to bederiv-d from shapes of this type,while another (Ref 75) believes thatthe benefit actually hisa&diffeteni hansctcrnd echansr. Other ariies iRefs 76 and77) suggest swepttips provide an effec~tive thai endplating benefit whi~h van bevisualized byiriaginig a vtew looking upstream a swept tip wing planform, thai is at some at ineidece angle That is to say. wing plainfoemi is planar at a that rero incidence may not beplanar or exhibit planar wing drag characterisu~at incidence Clearly thereistorevwork to be doneto fort outsa physical explanation foethe drag nseatanisnss involved. but oneaircraft designee dafided riot has to watt (Fig 831 Hlypersonic Drug Source Identification Interestin hyperionic flight hasgrown considerably daring the last desade Itis recognized thatas speed increases, aircraft dominated bywing shapes transition to vehicles that are dominated bybodyshapes (Fig 84) Underitaniding the Lharacirissses of body fomi at hypersonic speeds becomes important. not only because the body influence on propulsion of inegration. butalsohesause bodly the kthe largestcomponent) generates considerable pomoatn the total dragforce a of

P-tO WITH V'INIIOASO DE)IEC110N RUDDER

Fig. 77 Navier.Stoires Analysla for F.14 Buffet at M 0.95

7-42

-06 -08 -05 F-14D *5F.14DWITH6R .4' ML. 132 ML. 12

b 67f

-03

z 03

02
.1 Ci TvNNEL S.TING

CENTER

Fig. 76 Effect of Inboard Rudder Deflection on F-14


Afterbody Pressure Field at M = 0.95

PANEL
TUNL _ _ _ _ _ _ _ _ _ _ __STING

14I

I(Ref.
N F14A09$ <
-

Fig 81 SweP-BaC Wing.Tip Wind Tunnel Models 73)

13"

O F.14009 ,0o097 F '1A9

10

...

-8 .

-10

Fig. 79 Maximum Mach No. at8151 ------- -'---01 ,1CRESCE 0--ELLIPTIC

01

167

S
I 32 .4

40 CL

60

80

[
[0 08 10 12 14 MACH NO 16

Experiment Wing Tips (Rat. 73) Fig. 82 Swept-Back Drag Reduction due to

6406

18

20

Fig, 80 F-14D Delta CD for Rudder Deflection Angles

7-43 this study. Two d&ffrnt types of computational methods were applied. Onemethod (Ref. 79) was characterized bysimple Newtonian Theory. while theother (Ref 80) was a modern Navier-Stokes formulation Experience with these methods applied to body shapes (Fig 86 and87) indicated that reasonably accurate absolute drag predicnons could bemade By comparing the results from two vastly different computational formulations, the nature drag on forebody of shapes incidence could be better understood But the at computational predictions agreed closely, and theresults were not what wasexpected Close agreement indicated that there was probably not an appreciable amount of forebody drag attributable to voriex flows andcomplex viscous interactions This is thecasebecause Newtonian method useddid not the include tre viscous andvortical flow simulation capabilities The feature expected wasa dent or hole in thedrag polar not shape (Fig 88)0 SinceNewtonian Theory registers this effect, it might be appropriate to think in Newtoman Theory terms Netoman Theory is linked to perfect gas,compressible flow. obhque shock wave relations for M-C-Fig. 83 Aircraft Designs Featuring Swept-Back Wing Tips + sin 0 y,,1
5

(30)

Where for very high speeds. shock waveswith angles will lie 0 close to vehicle surfaces with local orientation angles e Also y

M.0 6

C,b

2s

um n

(31)

This "point pressure law" is applied breaking configuration by surfaces intorc and mo aets nd sumai ng oe 3 iu for nal windward elements while Cp- 0 is usedfor leeward elements

o MAt

__

Fig. 84 Dominance of Body Shapes as Speed Increases

the impact of In 1988.a projct was initiated to identify hypersonic forebody shaping on inlet flow quality The foundation for this work was reported on in Ref 78 The new study in luded flight conditions at incide c andsideslip angles A secoodary objectcv of this effort as to betterunderstand the nature forebody drag charactertstics at incdenre kRef 79) In of other words, there wasan interest usexamining forebody drag to Identify theelements of resistance c, friction, vetex, and (I wave diag) as variations occur with incidenc Figure 85 sho's the hy-personu vehile forebody shapes formed thebasisfor that

In Fig 89. a body form is represented a simple cone-cylinder by combination At zero incidcce, a pressure field is generated on the forwrd.facing surface. The cylindrical poon "egsters no effect As incidence increases. onset flow 'exposes' the additional body surfacearea nd o doing changes is the effective body shapetheshape generating forces Newtonian in Theory Finally. depending on te coneangle, an incidence angle is reached %here the effective body shape 'locked in" is (note thelast shape Fig 89) It is con)etured thattheirregular in hypersonic forehedy polar shape a direct result of wavedrag as lecls registering on an effective body shape changes that with angle-of-attack over a shallow incidence range 'Te analogy to this would be an arfoil or body shape ,.tscous flow %here the in effective shape changes with Reynolds number asthe bonridary layerdisplscnment thickness vanes

Iswasfound (Ref 81) that eprnentaliss weremeasuring this characters. in te past.and .jrve fits %ere usedto fair thedat sgnrmg the'dent" This oc.urred because, Most early Inm hypersom testing, data at large(five degree)Increments in Incidence scneusually taken This canbeseenin Fig 90 In this ase., computtional drag analysts served to highlight drag mneihaitssnand provide a basisto understand fluid the mchanmal foundations of observed charactenstics

7-44

Fig 85 Hypersonic Forebody Shapes for Drag-ueTolncidence Study

0.

08

-NEWTORIAU

TEST DATA-THEORY 0 01

-. -

BENT NOSEDICONIC

II

S04. 030 CL 02 0201

O0~
______________IN

BICONIC STRAIGHT 0 TEST O TEST SENT BICONIC *c-ARC31) STRAIGHT BICONIC *ARC 3D
010 012 014
SENT

BICONI BIB01 02'0

ANGLE ATTACK DEG OF~ c.

0l

Fig. 86 Hypersonic Body Drag Polar Correlation with Newtonian Theory

CD Fig. 87 Straight & Bent Biconic Drag Polar at Mach 6 0

12-

CDCLTANa D-2A.
.L

INC:OENCE-INOUCED DRAG FORM

08 0 CD 04

CD Fig. 88 Hypersonic Body Drag Polar "Dent'.

12

080 Cc 0

04

Fig.89 EfeciveIn ewtnianFlo odyShap

Newtnia Fig. 89Shape ingstn Body Efectivec

Flowt
M= 10 (Ret. 81)

X-29 (Fig 91) configuration development provides a basisfor thebenefits thatcomnputaional dragprediction understan~ding
methodscan havewhenapplied to bigo-speed ighter wing optitnization In this case, two-dimensional airfoil ainalyses idenitified in (applied a three dimensional wing designprocess) thai win& sectionload wasearneddifferently (on the apperand depending on whether thewing was swepi lower surfaces) shownin Fig 92 A andis forward or aft This haa been at attributed to decoupled upper/lower flow narchanisms supensonix espansion conditions (Ref 82) Reduced tr-ansonic shock%avestrengthandan ultimately resslts with reduced thai attendant redaction in wavedrag It can alsobeseen as moreseverewith inceases at incidence flow conditions become movesafi or Macb number, theforward swept wing shock wane eep into a region where shock wavesw angles are incri:Asc; just theopposite of theaft swept wing Situation Foraft swept to wings, a shockwavemoving back on the planfortt is fortced angles with an accompanying wavedragpenalty lower sweep (Fig 92 B) Dhiii.of -w se. is not to saythatwavedragactually decreases forward sweptwings with increasig flow sesent for 5 relief mecmhaism it that Instead, suggests thee isadrag that involved thatcan slow the rate wanedrag increases

Iligh-speed wing design wasaffected with a design procedure finite difference thatcoupled a two-dimensional trailsortm methodwitl a three-dittinstottal subsonic/snpersonit panel a method Tie Iwo-dimensional analysts generated streamuste wtng section. while thethree-dimensional methodidentified wing incidence or twtst distibutton along with any spantwechordse camber tnodifi~aaons (Refs 32 andS.) testsat effects %ere aerified bywind tunneli Theresultant sweep AEDC andeentiaily byflight testsaiNASA s Di)dcn Flight Reseasch Center. USAF study(Refs 83 va 84j provided key A insights into theimtpactof resalung perfonmance, levels Figure at Mach 93 compares polars two tra-ansonic numbersusing drag X 29 dataandthe dragpolar from anexiting light-weight that the reveals fighter and air superiority fighter Figure 9-4 an X 29 is characterized by lower thnist-io-wetght ratio but a into h~ghen nsium liftcoefficienit. All of this translates an during airin advantage dlytinti turnsthatmight beperformted to-air combat engagements 95) (Fig

7-46

Fig. 91 X-29 Configuration

((MAT WING PRESSURE K DAE.2 2.SR 0ANALYSIS SURFACE LOWER TAPER SWEEP ANAL.YSIS STNOR8WE TAE
04

x2 FIGHTVEGT X2

12

'z0

IR DRG

SUPERIORITY FIGHTER C

CUICEN

12
a

------............ LGHTWEIGHT FIHE

0 -C ...// UPPEP.1OWER DECOUPLED (A) EFFECTOF SURFACE ANALYSIS AIRFOILK FOR SHOCK LOCATION SWEEP n AnaOR MACH ATHIGHER LOCATIN WAVE SHOCK UjFit CONDITION AT a SWEEP DESIGN CONDITIONJ

AIRM SUPERIORITY FIGHTER DRAG COEFFICIENT CD

/
/
AFT SWEPT WING FORWARD SWEPT WING (B)SHOCK WAVE DRAG RELIEF EFFECT FOR TAPERED FORWARD SWEPT VOWS40

(a) M.12 Fig. 93 Drag Polar Comparisons (Ret.83)

Fig. 92 Mechanism for Reduced Drag at Transonic Speeds with Fcrward Sweep

7-47 AIR SUPERIORITY FIGHTER\ UGHT.WEIGHiT FIGHTER S2, UIGHT-WEIGHT FIGHTER

XX-29

MACH NUMBER O
Wi

AIR SUPERIORITY FIGHTER X-29 UGHT-WEIGHT FIGHTER LGTWIH

MACH NUMBER

FIG4TER X-29

Fig. 94 Maximum Lift &Thrust/Weight Comparison


(Ref. 83)

The X-29 represents breakthrough in advanced a manneuvering


technology. Not only were theoriginal conceptual ongins of modern fomrvadivcop technology identified via computational but drag predicuon methsodology. corsputational tools iscreR instrumniial in allowing thedesign to movesmoothly forward despite fact that noapplicable desga experience existed the base foe swtntfotward configurations which might haveenhanced confide~nce reduced lisk and levels. It is difficult to make compaions to existng fightems since. for example, the cxpetnenial configuration is required to perform all X-29 not the functions of a tactical fighter, Not withstanding this. it is winid-tunnel tost houts high-speed of that remArkable only 160 time wererequired to finalire theconfiguration concept (Fig t

POIIN XI (B) U . a.2


'"

96). thisisconsiderably less thattypically used fighter than for configuration optimization (Ref. 85)Y

Fig. 95 Aircraft Dynamic Turn Performiance at 35,000 ft (Ret. 83)

WINDTUNNEL TEST HOURS

X29

GENERAL, ARRANGEMENT FLIGHT CONTROL& PRESSURE LOADS GENERAL ARRANGEMEHNT WING PUANFORM,$=O CMBER.TVAIST

__

LE & FLAP TE SYS


STORE LOADS

NI

F-l6

DEVELOPMEN STRAKE CONTROL DEFLECTIONS PRESSURE LOADStHWET FLUTTERSTORE SEPARATION SPIWNSTAaLL SPILLAGENOZDRAG &
MISCELLANEOUS

.W
-

f__

I77=r
Fig. Wind Tunnel Test Comparison (Ret. 85) 96

7-48 CONCLUDING REMARKS A number of engieenng computational methods that can play a role in predictinglanalyzang drag components dunng aircraft described An attempt has havebeen development programs been madeto cover all drag components thatareof interest to for thedesign engineer andprovide somebas.s understanding whatmight be expected. r e. typical results andaccuracy levels. the A keyconstraint in any application, however, ts fact that quite complex Occasionally, this aircraft flows canbecome complexity is beyond thecapabilities of cutrent computational enginecring tools Further. it is recognized that no single drag method capable simultaneously treating all sources is of thatare important This requires thattheapplications engineer beclever andoccasionally creative Nothing, however, will replace need establish a computational drag prediction the to experience base -examplesof which wereincluded herein The role of computational engineenng methods can bequite vaned It should be apparent the drag prediction tools that described canandhaveplayed an important role in bridging the simple handbook methods andthe perforntance gapbetween establishing sub-scale teststhatcharacterize aircraft where development programs Further. therearemany instances these engineerng tools haveprovided thedesign teamwith key to :hz insights needed advance stateof-the art or soive problems withconsiderableresource savings This results in an element of creativity thatis derived from theability to inexpensively without thetime/expense constraints evaluate manyideas testing Finally. it shuldbe associated with sub-scale cases, there nay be discrepancies recognized thatin some between wind tunnel tsting andflight test results When this a happens, third source(coming from computational engineering tools) can prove valuable in thesensechata hird sourc of to ties information is often needed break Thercarc likely to be times, however. .hen the best computerized methods areinadequateA fall-back position s might involve the use of Smeaton original 1759equation tEq I) The applications engineer might also be advised to use Smeaton's coefficient (0 0049) which was shown tobe .onser1vatve - sincea good acrodynamicist knows that it is or important to keepa few counts of "drift" in his her back as pocket These counts areoften needed projects evolve (6) "Technical Status Review on Drag PrImction andAnalysis from Computational Fluid Dynamics Stateof the An," AGARD AR-256. June 1989. Drag." AGARD (7) "Aircraft Excrescence CP-264.1981 (8) Jobe. C E . "Prediction of Aerodynamic Drag." AFWALTM-84-203. July 1984 (9) Covert. E E. Thrust andDrag Its Prediction and Verification. Progress in Astronautics and Aeronauics. AIAA. NewYork. 1986 Volume98. (10) van der Vooren. J and Stooff. J W . "CFD-Based Drag NLR TP Prediction. State-of-the-Art. Theory. Prospects." for 90247L. Lecture Notes Prepared AIAA Professional on Development Series Drag Prediction andMeasurement. August 1990 Analyses and (11) Slooff. I W ."Computational Drag Minimization. Mission lmposstble'. AGARD R-723. May 1985 (12) Rogalo. R S andMon,P. "Numerical Simulation of Turbulent Flows," Annual Review ofFluid Mechanc, Iol 16. 1984 (11) lladcock. R , Aeronautcal Engineering Development, Part I 1759 1914.Manuscript to be Published. 1992. (14) Oswald. W B ."General Formulas andChars for the Calculation of Airplane Performance." NACA Report No 408. 1932 (15)Jones, R T, "Properies of Oblique Wing-Body Combinations for Low Supersonic Speeds.'NASA SP292, pp 389.407, 1971 (16) Iloemer. S F. Fluid-D)nauic Drag. Published by Author. 1965 al. t17) F[ik. R D. ct "U S A F Stability & Control Datcom.' Wright Research Development Center. [light Contkols & Division. April 1976 Definition of (18) Locke. F W. S..Jr.. "Recommutnendcd Turbulent Friction in Incompressible Fluids." NAVAIR 4 Report No 1 15.June. 1952 ( 191 Schlchtig. II . Boundary Layer Theory. McGraw Ihll. 1960 (20) Sommer. S C andShort. B J. "FreeFight Measurements of Turbulent Boundary Laser Skin Friction in thePresKnce of Severe Aerodynamic IHating at Mach Numbers rrom 2 87 0," NACA TN 3391 1055 (21) Peterson. . It. Comparison of Experimental and IB "A Compressible Turbulent Theoretical Results for the Nessure Boundary Layer Skin Friction with 7xro Gradient." NASA TN-D 1795.March 1963 A Computer Program for (22) Fppklr. R andSomters, M . D Airfoils, NASA the G-stgn andAnalysis of Low Speed T1 802 10. August 1980 (23) Abbott. i Aet al. Theory of Wing Secions. Dover. New York.1959 (24) Althaus.D et Stutga ter Profilkatafog1.Fnedr u. Vieg & Sohn.IBraunschwctg/Wiesbaden. 1981 W (251 Stevens. A et al. Nlathematical Model for Two Dimensional Muln Component Airfoils in Viscous Flow. NASA CR-1843. July 1983 26 Drela. M andGiles. M B. "Visous Inviscid Analysis of Transonic andLow Reynolds Number Arfoils." AIAA 86 1786.January 1987 (27) Drtla. NI andGiles. NI B. "ISES A Two-Duitensotnal Viscous Aerodynanic Design andAnalysts Code."AIAA 87-0424. January 1987 128) Bopp.C.W..csal. "STARS&STRIPES. Computatiooal Row Simulations for Ilydrodynanc Deaign." The Eighth Chesapeake Sailing Yacht Symposium Proceedings. March 19b7 (29) Ives.D. Private Communication, Pratt & Whitney. 1983 oj'Aerofod aud Aiatre (30) Glauer, II .T e Elements Theory, Ca uidge Universit) PressLondon, 1926

ACKNOWLEDGEMENTS Theauthor would like to thank thoseat Grumman who ha e of madesignificant contributions on a number drag analysis and R reduction projects The work ofP Ilavitz (G Ill). (X-29) and Iendrickson W Evans (F. 14).andG Spaeht to belongs to this group D Ies of Pratt & Whitney is be lift recognized for his formulation of thesynnnemtc induced dragFouner load integration scheme E Tinoco and1 M.Mauters (Boeing) provuled usefulinsights into thestate of receiving the an in fillet design Finally. the author appreciates permission o usephotographs of Scaled Composites. Inc design concepts

REFERENCES 'AicraftDragPredicton.*AGARDR 723.1985 Acrodynamic Drag." AGARD CP 124. 1973 CP 71 71. 1971 Aeronynamic Interferece. ' AGARD 'Drag Prediction andMeasurement." AlIAA Professional August. 1990 Development Series. and Aerosamaucs, kSt MiCornmick. B W , Aerod)nruamcs, FlgiAMechanics, JohnWiley & Sons.N Y 1979 if) (2) 3a (4)

7-49 and of Landishi, Aerodyna,,atcs Wings M.; (31) Ashley, H. and Bodies. Addison-Wesley Publishing Company, Inc, Reading. Mans 1965 (32) Boppe. C. W.; 'Computational Aerodynamic Design: X29. TheOulfstreain Series, and a Tactical Fighter.' SAE 851789. October 1985 (SAE Wright Brothers Medal Award Paper - 1986). a (33) Bauer, F., et al; " Supeical Wing Sections 11, Handbook," L-ecrure in Econonars and Notes No Spinnger-Verlag, 1975 Marhematical Sysrerns, 108, (34) Cook. P H.. et al, "Aeeofil RAE 2822 - Pressure Distributions, Boundary Layer, and Wake Measuremients,AGARD AR. 138, 1979 (35) Hoist. T. L,* "Viseous Transonic Airfoil Workshop Compendium Results," AIAA 87-1460, June 1987 (36) Whitcomb, R T., "Recent Results Pertaining is ihe Application of the Area-Rule'." NACA RM L53115a. 1953 (37) Whitcomb, R T.. "A Study of the Zero-Lift Drag Rise Characteristics of Wing-Body Combinations Neurthe Speed of Sound." NACA Report 1273. 1956 (38) von Karman. T.. "The Problem of Resistance in Compressible Fituds." Mensored Reale Accadema dtalia. Clause di Science Fisiche. Matematiche c Naturali. Vol XIII. pp2lO-265. 1935 (39) Harris. R V., Jr-,An Analysis and Correlation of Aircraft Wave Drag." NASA TM X-947, 1964 (40) hfendickson. R . Private Communication. Grummain Aurrft Systems Division. November 1990 (41) Ashley. II . "On MakingnTings the Besi - Aeronautical Uses of Optimization,' 1981Wrigbt Broibers Lecture, AIAA 81-1738. Augusi 1981 (42) Davis. W If Jr.. Technique for Developing Design Tools from the Analysis Methods of Computational Aerodynamics." A IAA79-1529R. also AIAAJournal.Vol 18. No 9, September 1980 (43) Davis. W 11. Jr. et al. "A Study to Develop Improved Meihods for the Design of Transonic Fighter Wings by the Use of Nunscrical Optimization," NASA CR-3995. August 1986 (4)Vaisderplaats. G3 N . 'CONMIN - A FORTRAN Program for Constrained Function Minimization Users Manual." NASA T1M X-62282. August 1973 (45) Tamvlell. and Tampnplm . "An Inlet System Installed R G Performance Prediction Program Using Simplified Modeling." AIAA 83 1167, June 1983 (46) Hems. L ArM J Smith. A MI0. "alculation of Non Lifu~ng Potential Plow About Arbtrary Three-Dimensional Bodies." Douglas Report 40622. Mlarch1962 (47) Sherman P NI and Lincoln. F W . "Ram Islet Systems for 1969 Wuteriet Propulsors." A IAA 69 418. M~ay (48) Henderson. W P. "Itopilsion integration for Militry Aircraft." SAE 892234. September 1989 (49) Grossim. B anWd Nllk. R . -Ihe Numerical Computation of the Transonic Plow Over Aftibodues Including the Effect of Jet Plume and Viscous Inieractions.' AIMA75-62. 1975 (50) Salas. MI D. 'The Numerical Computation of Invisui Plume Flow Fields,' AIAA 74.523. 1974 Over Afierbodies (51) Yaeger. L S . "Transonic 1Flow Includirg the Effects of Jet-Plume ard Viscous Interactions with Separ-aiion." AIAA 77 228. 1977 (52) Ckgu-pion. B.. "Jet Effects on the Drag of Conical W. Afterbodien ai Supetrsonic Speeds.' NASA TN D-6789. 1972 (53) Bushnell. D and Dorsaldson. C . "Cniri of Vortex Plows.' NASA TM 102693. June 1990 (54) Rubber. P arid Coldharnmer, NI. CID in Design An Airframe Perspective.' AIAA 89-0092. January 1989 G. (55) Rubbert. P andSaans, R., "Review andEvaluation of a Three-Danensional Lifting Potential Flow Analysis Method for Arbitrary Configurations," AIAA 72-188. January 1972. (56) Boppe. C. W.; '"Transonic Flow Field Analysis for WingFuselage Configurations,' NASA CR-3243. May 1980 (57) Boppe. C W. and Rosen. B S., 'Computatin of Prop-Fan Engine Installation Aerodynamics." Journal of Aircraft, Vol. 23. No 4. April 1986. (58) McCroslsey. W. J et al. "Airloads on Bluff Bodies, with Appbications to the Rotor Downloads on Tilt-Rotor Aircraft." Vertico. Vol. 9, No. 1, 1985. (59) Maskew. B.; "Predictions of Subsonic Aerodynamic Characteristics - A Case for Low-Order Panel Methods," AIAA 81-0252. January 1981. (60) Boppe. C W, "Aerodynamic Analysis for Aircraft %ith Nacelles, Pylons, and Winglels at Transonic Speeds." NASA CR-U066 April 1987 (61) Miller E and Delaney. F J. "Level II Nuzzc.'Afterbody Installed Performance Prediction Program, AFWAL TR88-3004, December 1988. (62) Tindell. R H; "Computational Plaid Dynamics Applications for Jet Propulsion System Integration," ASME 90-G 17-343. June 1990 al. (63) Carlson. I1. W, ore "Validation of a Compuier Code for Analysis of Subsonic Aerodynamic Performrance of Wings and Plaps in Combination with a Canard or Ilorironial Tail and an Application io Optimization." NASA TP-2961. January 1990 (6-I) Carlson, HIW and Walkley. K B . "ACompuier program for Wing Subsonic Aerodynamic Perfotmance Estimates Including Attainable Thrust and Vortex Lift Effects." NASA CR-3515. March 1982 (65) Whitcomb, R T. "A Design Approach and Selected Wind-Tunnel Results at high Subsonic Speeds for WingTip Mounted Wingleis.' NASA TN D-8260, July 1976 (66) Chandrasekaren. R MI, etal, 'Computational Aerodynamic Design of the Gulfstreamn IV Wing.' AIAA 85-0427, January 1985 (67) Davis. W I . *Applied Trarisonics at Grumman." Transonic Symposium - Theotny, Application and Experiment. NASA LaRC. Hampton. Virginia, April 1988 (68) Davis,.W Hf. -TseRole ofCFD Applied to High Performance: Aircraft." AIAA 9W-307 August 1990 1. (69) Cooper. G K, "The PARC Code Theory and Usage.' AEDC TR.87-24. October 1987 kC)) van Damn. P. "- Swept Wing-Tip Shapes for Lois Speed C Airplanes". SAE 851770. October 1985 (71) van Dam.C. P. "Drag Reduction Characteristics of AfiSisepi Wing TIPS."AIAA S&-1824.October 1986 (72) Boppe, C W ,"Sailboat Ilydrodynamic Drag Source Prediction and Performance Assessment.' The Tenth Chesapeake Sailing Yacht Symposium Proceedings, February 1991 (73) van Dam. C Pcit al. "Wind Tunnel Investigation on the Effect of she Crescent Plantorm on Lift aria Drag." Proposed Journal of Aircrafi uriicle.)Also - AIAA 90 0300), Mlarch1990 (74) Smith. S C and Kroo. I NI , "A Closer Look at the Induced Drag of Crescent Shaped Wings. AIAA9O 3063. August 1990 (75) Dafluari. M A . "Induced Dragof Wingu with Ilighl3 Swept and Tapered Wing Tips." AIAA 90 3062-CP. August 1990 C (76) B~urkett. W, 'Reductions in Induced Drag by the Use of Afi Swept Tips. Actunarwicil Joiurnal. December 1989 7 j 7i Ba.kcit. C W. Analysis of Cresceni Wings Using a Srbvonic Panel Method. ICAS90 3 62. September 19%j

~'irtl

7-50 (78) Wilson, G J. andDavis, W H .Jr., "Hypersonic Perforniance Sensitivities Based 3-D Equilibrium on Navier-Stokes Calculu.wsns," AIAA 87-2642, January 1987 (79) Boppe,C W andDavis, W H , Jr: "Hlypersonic Forebody Lift-Induced Diag." SAE 892345. September 1989 (SAE Wright Brothers Medal Award Paper1990) (80) Pulliama andSieger, "Imphoct T L.. Finite-Difference Simulations of 3-D Compressible Plow,- AIAA Journal, Vol 18, February 1980 k8l) Spencer. ; Hypersonic Aerodynamic Characteristics of B Minimum WaveDragBodies Having Variations in CrossSectional Shape,"NASA TN D-4079. Sepiember 1967 (82) Boppe.C W, "X-29 Aerodynamic Design and Performance." ALPAA Professional Development Series.Aerodynamic Analysis andDesign. October 1988 (83) Bursey. C H ;"Fighter ClassAircraft Performance Comparisons," AFWAL-TR-88.3081, November 1988 (84) Pinrof, M., 'X-29 Aerodynamc Specialists Meeting S. Repori," WRDC-TR-89.30..7. April 1989. (85) Bradley. R G . "CFDValidation Philosophy," AGARD CPP-437, May 1988.

REPORT DOCUMENTATION PAGE

AGARD-R-783 5. Originator

ISBN 92-835-0652-9

UNCLASSIFIED

Advisory Group for Ae. -pace Research and Development North Atlantic Treaty Organization 6.Tte 7 ruc Ancclle, 92200 Neuilly sur Seine, France 6.Tte SPECIAL COURSE ON ENGINEERING METHODS INAERODYNAMI1C ANALYSIS AND DESIGN OF AIRCRAFT

7. Presented as an AGARD Special Course at the Middle East Technical University, Ankara, Turkey, 6th- I10th May 199 1,at the von Ktirmlin Institute for Fluid Dynamics, Rhode-St-Gen~se, Belgium, 13th- I7th May 1991 and at the Universitad Polttccnica de Madnd, Spain 20th-24th May 1991.
8. Author(s)/Editor(s) 9.Date

Various 10. Author\%/Editor's Address Various 12. Distribution Statement

January 1992 11. Pages 248 T[his document is distributed in aceordane.. with AGARD policies and regulations, which are outlined ott the back covers of all AGARD publications.

%rcraft

Cost effectiveness,

Aeroynaic nalystanoDesign uAircraft" have beeii assembled in this report. Proven Ltigineermtg methods used duritigconceptual and preliininary design and development of nev. ,ureraft Loneepis arc lpreseted These methods focus on simple computational procedures%for Lonceptual and prelimitiary design, low-level analysis computer codes, and experimental techlniques for aircraft l)erftrinaiice predlictiotis ftic course waisaimed at helping train young engineers to appretiate atid sork with simple enigitirering tools itoenhance the art of cost-effective prelimi nary design oif new aircraft [he AUARI) fluid IDynatmics Paniel Spetha Course - for which this material was pr~epared \sas held 61h- 10th May 1991 at the Middle Eastl lechnital University, Ankara, Turkey, i 3thiI 7tthMay at thte son Kartoan Institute for Mlid D~ynamnics, RhOde-St-Gen6se. Belgium, and 20th -24th May at the Un~versitad Politctruca die Madrid. E lsl Aeroitauticos, Marind Spain. lilis cour5s 55 delpd .i(nd conduceted under the Lominted sponisorship of the AGARI) Fluid IDynaiitic Paitel. the AGARI) Conssultant and Lxchange: Programme. attd the Von Kdilidn Intiitute for Flid IDvtamiev

E0

< Eo Rz~

Q 20~ C6C.

Az

<

r,

~
0

~
0

S-1

Fj

<~2<

20

72

cc

00 > <
.<

0
cC 0-

'c 'O.C.
a

c 5 9E a0

A~

64

5.~
E,~~

j
x

~o~-'sg
&

j2

~12 ;l: ~~V)

g~

72

E b

r -

T-

~~

RUEANCELLE3ON~a:bSlE jM~(04738,57,00, T&I17:4

IRSODSUUAIN

TN AUX &;CNRSEIFSOA u dn;, e$ r 'do paSLAND * MANETV W e 4elin N rumtw it menahmfrmu de 6Pg mD'rectoro(Avjuuooc~eitinau' As.7t Weucntovldhae2au7t -' mtW "Mieykjia e~a k'* cftilc~n cu ,BLGQU mITALIEfm'rcpt~ e EasU ,iemtcdsp od Coordoe deeurfofftfluSUd> Aeronuuucas uarc Emt ajof, eFceden"Utcodc cct aoaealAAD

':VA

AR~hd-t~

-u

A..

j~jt;

:10el Defence Research

1"MiNestedaeds~elentet

cloA0ARD;

'PRt) oALV Nadond"ionsuuuandapaesdemaeiraso ;IfAGNpsNawW.trsrwLzorslrpNL

--toAORD

ini'S.# Amiisii~n

Gaiekoce knbhtwfoniainc ntr

~~D~diaA~r~ozcelise& arcalknhib AR
Anara
d Deec isiii
'

SnondoSHc f~ rts irr E~fraqpciea& ounotSplyDaso ,nc 'Emoea St= l0.ruisganotl;EX, o, - 1n-, 1 o'Aiu CollduflS- W~ We~lb~ s ckiir) 5 8 '7501S PailsA~un ig V, gn r ht u61 'aeoue-OR~crpsedmnc~ica Lcdi mkrolcbtodeho hade

92Sout

sou RhitdloRHost" Nido~aI~chir~e=

Nitoil Tc~c~l iforraio Siiie

ES

~ ~pou ilifw

pa c ione

chi

Iforitioh Llbrje

Fac.

-I-nt

oyiONS)UM -

Imp'nEperSptensd deAAD t ladenorninationAGARD~~~sinsiqmururn 'Jou lii L~lenougheayen qwlc~tect~atedeublictio~s

toiithiiksnUcedaffSdsi arCO tfm in&aousdseAt i esAIR-G33 slxi 372-Rnmolc~ AAR-RlO

firde

oommndekraf 6isAGR~tdcsrppoueg~s~ttibC-k

You might also like