19-1

Aspects of Wing Design for Transonic and Supersonic Combat Aircraft B. Probert, Aerodynamics Department (W31OP) British Aerospace, MA&A, Warton, Lancashire, PR4 1AX England
It should also be mentioned that the term balance relates to the mix of aerodynamic performance and the overall weapons system capability. The aim is to maximise aircraft agility, defined as efficiency of the weapon system as a whole (ref. 1). In this context aerodynamic efficiency plays only a part since The search for agility brings in many other disciplines. aerodynamic efficiency and controllability will form the main subject of this paper. The theme will be to review the fundamental features of high speed flows, the problems posed and the means of overcoming them by configuration choice and detail design. Such design work depends heavily on available methods and procedures and a section is devoted to briefly cover examples of CFD capability. The general problems which emerge will feature to various degrees depending on the precise operating requirement of a particular aircraft. Here this is illustrated by breaking the problems down to four flight regimes and considering the detail wing design of an aircraft in each. Finally. current and future trends which may strongly influence aerodynamic progress are mentioned. SOME FUNDAMENTAL FLOW FEATURES OF HIGH SPEED

SUMMARY This paper describes some aspects of high speed design, mainly wing design, for combat aircraft. This is done by fist reviewing the fundamental features and problems posed by high speed transonic and supersonic flow and the means of alleviating them. The resulting empirical / simple rules and methods allow an initial baseline configuration to be developed which is further optimised using the most appropriate design processes - involving the use of a number of CFD codes, which are described. Each process is then illustrated by briefly considering the design of four types of high speed aircraft each optimised for a different, but overlapping, flight envelope The current need for processes to treat the design of novel configurations which have low observabilty is mentioned and finally current ways of working are reviewed and suggestions made for future developments in the areas of design processes, aerodynamic improvements/fixes and the potential of new physical phenomena. INTRODUCTION This paper considers certain aspects of wing design related aerodynamics for subsonic, transonic and supersonic flow. The authors experience lies in the field of military aircraft design and the contents will obviously reflect this fact but some of the broad descriptions and conclusions will also relate to commercial aircraft applications. A major difference between military and civil aircraft design lies in the greater number of design requirements, viewed in the (MC) plane, to be achieved by the military configuration. It is not possible to concentrate too much effort on a single design point e.g. altitude cruise, at the expense of other important considerations e.g. turn rates, low altitude high speed dash etc. The design thus becomes a search for the best compromise - or better, the best balanced configuration which comes closest to meeting the overall objectives. At the extremes of the configuration range we have the unswept wing which is usually limited to subsonic flight and much of its characteristics can be described by appealing to two dimensional aerofoil properties. Years of experience with this type of aircraft both in flight and experiment means that flow features are well understood. At the other end, for sweeps beyond 60, the flow features though very different are still orderly with concentrated vertical flow being the dominant feature at low speed and swept shocks with vortex separation at high speed. At supersonic speeds available theories hold reasonably well for the attached flow condition. However this class of aircraft has poor airfield performance and high lift dependent drag. The above shortfalls i.e the subsonic performance of unswept wings and the poor manoeuvrability and field performance of highly swept wings means that the designer may need to look at the intermediate sweep region, 20 to 60, with all the associated mixed flow complexities in order to obtain the balance which he is seeking.

From the aircraft performance viewpoint it is worth considering the gross effect of various wing geometric parameters on the flow development, but first we look at the broad effect of supersonic flow on the configuration drag. A drag breakdown at zero lift for a typical combat aircraft at subsonic and supersonic speeds is shown in fig. 1.
Fig 1 - Typical Variation of Drag with Combat Aircraft Mach Number

Paper presented at the RTO AVT Course on Fluid Dynamics Research on Supersonic Aircraft, held in Rhode-Saint-Gent?se, Belgium, 25-29 May 1998, and published in RTO EN-4.

19-2 The dominant increase at supersonic speeds is due to the wave drag component consisting of fuselage, lifting surface and interference drag, collectively known as drag due to volume. At lifting conditions the lift dependent wave drag produces a further contribution. At subsonic and transonic speeds the lift dependent drag varies inversely with the square of the wing span, thus maximising span is desirable for manoeuvre performance whilst at supersonic speeds the lift dependent drag is strongly influenced by the longitudinal area distribution. Typically the wing contribution to zero lift wave drag on a combat aircraft is in the region of 25% and this contribution is dominated, for a given sweep, by the wing thickness as supersonically the drag varies with the square of the thickness to chord ratio. However an increase in thickness at low subsonic speeds is often beneficial, increasing maximum lift, but as speed increases eventually becomes detrimental as shock waves begin to form leading to drag rise, separation, buffet and eventually shock stall. Typical boundaries for the above are shown in fig. 2.
Fig 2 - Effect of Aerofoil Thickness Characteristics on Various Fig 3 - Lift Development in Subsonic Flows and Pure Supersonic

FOT the subsonic case the distribution has the classic LE suction type of distribution which results in a centre of pressure at 25% chord. For the supersonic LE condition this is absent and for the 2D case leads to a centre of pressure position at 50% chord. For a practical 3D configuration the a.c. shift lies in between these extremes.
1.5 c L 1.0

Fig 4 - Swept

Wing

Relationships

05

I0 5

0.5 10 15 20

L'C 9%

Another significant effect of increased speed is the change in pressure/ loading over the lifting surface leading to a rearward shift in the aerodynamic centre position. This is illustrated simply by looking at the 2D flat plate development in supersonic flow in fig. 3.

3D Ouantity

Value for Eauivalent 2D Aerofoil MC0S.A CL xc2 A

2D

M CL CP a %PRESSURE WC) tan-l

cp WC2 A (tan a set A)

cDPRESSURES~3A (t/c) set A

19-3 The effect of thickness can be obviated by the incorporation of wing sweep which works by reducing the component of Mach number normal to the leading edge, since it is the normal component which leads to shock development and problem regions. From this 2D to 3D simple sweep relationships can be derived as shown in fig. 4. These are often useful in the early stages of design. Combining thickness and sweep effects one can derive the boundaries shown in fig. 5 where it is interesting to note that to obtain a transonic corridor whereby the aircraft can fly from subsonic to supersonic speeds the maximum wing thickness must be less than 7 %.
Fig 5 - Combination of Sweep and Thickness

Ihe above examples are extremely well known and do form the basis of initial project considerations for a conceptual design. Starting the detail design from a realistic base is obviously the sensible approach. The first approach to detail design lies with the improvement of the longitudinal volume distribution using supersonic area rule type methods. A number of variants are available, the approach adopted in Fig. 7 is based on a transfer of area rule and has been in use at MA&A for many years. The basic concept is to minimise the streamwise slopes of the cross sectional area distribution and the method is based on far field integration of the wave drag. This approach has been very successful over the years and still forms the comer stone of initial project design work.
Fig 7- Shaping for Supersonic Flight and The Swersonic Area Transfer Rule

However sweep cannot be increased without eventual detrimental effect on other aerodynamic parameters, notably longitudinal stability. Increasing sweep increases the loading on the outer wing leading to flow separation, possible loss of lift in that region and eventually to pitch up. An useful guide to the limit on sweep as a function of wing aspect ratio is shown in fig 6. The designer transgresses this boundary at his peril in the early stages of design but, as the figure shows, a properly positioned tailplane and/or a powerful LE device may allow the nominal boundary to be exceeded. However both of these options will need detail development. Further the correlation was developed from a large data base of conventional wing planforms Its application to novel Lo configurations is not yet established but should still serve as an initial guide.
Fig 6 - Pitch-up
10 SHORTAL yAp1N 1093) .4lD YURLONG 8~.MC HUCiki (NASA RlYP 1339)

Boundaries

The author apologises that the above is rather simplistic but does serve to illustrate that the basic features of high speed flow need to be considered at the outset of a project design layout. What follows is a description of the various procedures for refining the design. This calls for reliable design methods and before describing each procedure it is worth considering the computational methods available to the designer. COMPUTATIONAL METHODS

The tremendous advances in CFD methods over the last 20 years have exerted a powerful influence on aircraft design capability. Instances of application are too numerous to mention so I will confine myself to specific illustrations to emphasise certain aspects of design. As aheady mentioned,

19-4

supersonic area rule methods, based on linearised theory have been available for over 40 years and still provide useful design guidelines. Linearised theory lifting surface methods have played, e.g. Refs. 2, 3, and still play, a large part in aircraft design with each firm having their own, possibly in house developed, favourite. Inverse versions of such methods, where the geometry is derived from a specified loading or pressure distribution, are easily derived. Such methods have been coupled with empirically derived limits on flow features e.g. limiting LE suction levels and limits on suction levels which could lead to shock formation and separation (Ref 4). Such methods do not consider thickness effects. So called panel methods, where thickness effects are included, are usually based on the linear Prandtl-Glauert equation with a correction for compressibility when required. These methods have been used at BAe to improve on the far field methods initial design to increase locally fuselage volume without increasing drag. The panel methods give useful information at low incidences through the Mach number range but all the above methods are not usually classed under the heading of CFD. This seems to be reserved for methods which solve the non linear flow equations admitting discontinuities i.e. shock waves. The earliest methods considered the small perturbation form of the transonic flow equations. A two dimensional version was coupled with an integral boundary layer method in Ref. 5 and was used for many years as a successful aerofoil design code at BAe. Later, and with a fair amount of ingenuity on the part of the code developers, 3D versions were produced which could tackle complex geometrys, including stores (Ref 6). Such methods, used with care and including special LE treatment, provided good quality results for a number of years. An example is shown for the configuration in fig 8A, a 42 sweep wing-body combination, which is designed to a high standard of wing design. The comparison of the TSP method with experiment is shown in fig 8B and is impressive.
Fig 8A - Swept Wing Body - 42 Sweep, Ak3.3

Fig 8B - Comparison of Theory and Experiment Transonic Small Perturbation Theory Wing + Body (Inviscid) - fk0.88, a=6.6

However progress elsewhere, coupled with the need to cover more highly swept planforms, led to the switch to methods which solved the full potential equations. These methods, in highly developed forms, are actively used today particularly for civil aircraft design, where wing sweeps are low and shock waves are weak. Thus the isentropic assumptions are not limiting. Such methods have also provided good results on military configurations (ref 7). Fig 9A shows the configuration and a comparison of the FP result with experiment is shown in fig 9B. at an incidence close to the manoeuvre design point, where we have significant shock strength on the outer wing. Experimental pressure tap data was only available on the wing upper surface. Comparison is very good.
Fig 9A - GA of Model 209114

19-5

Fig 9B - Theory v Experiment Mach 0.9 1 Wing

- Full Potential + Simple Body

Method

Fig 10B - Comparisons Experiment

Between Euler Theory - M = 0.9 - M2091/4

and

-o-

Experiment

More recently, during the last 10 years and in line with the need for the treatment of stronger shock waves for military aircraft, the development and increasing application of methods based on solutions of the Euler equations have been prominent. These methods have proved to be robust in use and have achieved good comparison with experimental data. Examples for the same configuration as in fig 9A. are shown in fig 10A.
Fig 10A - Theory Mach v Experiment - Euler 0.9 - Wing + Body Method

I.4 -0 04

Further comparisons of Euler at supersonic speeds also gave good results as shown for M=l.l and 2.0 in figs 1OC. and 10D.
Fig 1OC - Euler Theory v Experiment Wing + Body - Mach 1.10 a=6

Even at this higher incidence, with regions of flow separation, the Euler (Ref 8) code gives a reasonable comparison. The code used at the time could only cope with a simple representation of the body, and this led to a poor comparison of the pitching moment data, as seen by the comparison with experimental data in fig 10B. We will return to this point later.

19-6
Fig 10D - Supersonic Comparison M=2.0 I w.=4 - Euler v Experiment

high quality, possibly benchmark, results in many instances. Set up times have been a problem but current developments on fast response unstructured grid codes are now reaching the standard set by the structured codes - for inviscid flow. An example is given in fig. 12 for the wing-body combination shown in fig. 8A at one spanwise station. A number of methods are compared, including a structured grid Navier Stokes method (RANSMB, Ref 11) and an Euler method (with and without viscous coupling). If we add the incremental change in shock position, observed from the Euler results, to the unstructured Euler result (FLITE3D, Ref 12) it promises to give a good match with experiment. Pressure losses due numerical dissipation are also low from this program thus the method shows good potential for future applications.
Fig 12 - SMARM Configuration 68% Station - Mach 0.88 I a=6.6

R2 % SEMI SPAN

-I I,

The need for better treatment of complex geometries led to the development of Multiblock Euler codes for the prediction of vertical flows (Ref. 9). These provide a powerful diagnostic tool and have been used extensively on many aircraft projects. In addition to the complex geometry capability reasonable results have been obtained in the prediction of vertical flows on sharp edges. However care is needed when dealing with rounded leading edges due to grid dependency. More recent developments have resulted in methods based on the Reynolds time averaged equations with a variety of turbulence models. These have improved the prediction of vertical flows and have also shown promise in the prediction of shock induced separation. A comparison of an Euler and Navier Stokes (Ref 10) solution for a highly swept 65 planform is shown in Fig. 11 The latter solution captures the secondary separation line and moves the attachment line closer to experiment.
Fig 11 - Comparison and Experiment Mach of Euler, Navier Stokes - 66 Wing Sweep 0.85 I a=lO

cp !

FLJE!DhlcdiSLd All of the above types of CFD methods have been used at MA&A over the years with the bulk of detailed design completed using Euler codes with and without viscous coupling. Often the design emphasis is based on achieving a desirable form of upper surface pressure distribution which is likely to produce low drag or good stability characteristicspreferably both. However, increasingly the prediction of drag from CFD is improving so that reliance on the predicted force data is playing an increasing role in decision making between configurations. In some instances, normally civil configurations, drag accuracy can be improved by considering the far field components of drag-viscous,wave and vortex and thus avoid the notoriously unreliable near field pressure drag integration, plus skin friction. For the more closely integrated military configuration where the body and wing are often blended it is difficult to successfully achieve such a process and resource is often made to direct surface pressure integration. In many instances excellent agreement with experiment has been obtained as shown in fig. 13 for a complex wing - body - foreplane configuration. However the recommended way of proceeding is still to think in terms of using CFD to calculate drag increments due to design changes and applying these to an experimental baseline.

-3 Al
CP
LINE LHBEL EXFERIPlENT ELlER NAUIER-STOKES -0.6

A fundamental methods is the advances in configurations

necessity in obtaining good results with such provision of a suitable computing grid. Again, surface conforming grids on complex via the multi block approach have provided

19-7
Fig 13 - Example of CFD Euler Multiblock in Design

regime covered by the aircraft. For present purposes it is intended to illustrate the different aspects of wing and configuration design by considering four aircraft types 1 2 Wing Sweep < 35 35 < Wing Sweep < 50 50 < Wing Sweep < 60 Wing Sweep > 60 Subsonic aircraft Transonic biased aircraft with moderate supersonic capability Supersonic biased aircraft with very good transonic manoeuvre Supersonic aircraft

M=0.9 a=0 c n contour

3 4

0.01

0.0 0.0 0.6

0.2

0.4 L

10

lb

MAm

1.8

DESIGN

REQUIREMENTS

The above are very broad classifications and no doubt there will be some exceptions. It is obvious that a variable sweep wing configuration e.g. TORNADO fulfils many of these roles, in the aerodynamic sense, as was intended. However, this was achieved at the expense of weight and cost. The effect of wing loading on STR performance means that it is not classed as an Air Superiority aircraft. The basic design of each of the above will have many common features but the detailed design will be very different. For example, the simple 2D to 3D sweep relations will only play a large role in the design of type 1. and to some extent in type 2. Even in type 2. the role of the 2D section may only be to provide an equivalent form of upper surface pressure distribution. We now consider each design in turn. THE SUBSONIC surface flow. WING - but with supercritical upper

For a military aircraft these are numerous and a typical set are given in fig 14 which cover both a wide Mach number and lift coeff. range hence emphasising both subsonic and supersonic manoeuvre requirements. The same figure illustrates the type of flows expected and hence the challenge posed to any design method. It is evident that the major part of the fight envelope will have strongly separated flow regions hence calling for extensive low and high speed wind tunnel development in parallel with CFD design activity. Looking at this another way, only a small portion of the envelope is rigorously theoretically predictable.
Fig 14 - Design Requirements
TURN RATES

INSTANTANEOUS
l a l .

LARGE . 3

REGIONS

OF SEPARATION

The example chosen is the minimum sweep of the TORNADO wing (25) The multi role requirements demanded excellent field performance, good supersonic performance and reasonable manoeuvre performance. The variable sweep wing helps considerably in this respect from the aerodynamic viewpoint. The bulk of the aero design concentrated on wing sweeps between 25 and 45 with fully swept high supersonic cases mainly serving as fallout from the transonic area. IThe variable sweep greatly alleviates the wing design problem, as the dominant thickness problem is taken out by the sweep effect as shown in fig. 15. Of course the configuration layout reflects strongly the multi role capability of the aircraft. A principal factor is the outboard location of the wing pivot in order to minimise the aerodynamic centre shift as the wing sweeps - fig. 16. The resulting nib fairing acts like a strake improving high incidence performance at low main wing sweeps.
Fig 15 - Tornado - Variation of Thickness Chord Ratio With Wing Sweep CA,
12
10 8 6 _\_\\\ 12% @25% h In

to

l *
. SUBSONIC .

SUSTAINED

TURN

RATES OF SEPARATION FLOW

LOCAL

REGIONS OR ATTACHED

SUPERSONIC .

MACH
4.

6% @-A,,

Coupled with these requirements are the constraints set by the capabilities of the flight control system, the need to obtain a stiff low mass structure with adequate spar depth and consideration of installed store interference. These points will have different levels of importance depending on the flight

IiL
10

20

30 wk3

40 sweep

so

60

,\LB

19-8 Fig 16 - Effect

-1.c

SMALL INBOARD

GLOVE PIVOT

CP

STMPLE GEOMETRIC TRANSFORMATlON

PRESSURES

SMALL FULLY

SPAN, WING AREA SWEPT

__- ---

LARGE OUTBOARD , SMALL

GLOVE PIVOT AC.

Fig 18 - Aerodynamic L/D

Benefit

of Variable

Sweep

M=O.9

SHIFT

LARGE SPAN, WING FULLY SWEPT

AREA

The wing was sized for a manoeuvre design point and thus demanded powerful high lift flaps and slats for the demanding field performance. In addition high speed combat utilised the slat and some flap (Transonic slat design is still difficult currently even with the best CFD method available). A good standard of design was achieved by combining subcritical empiricism and much wind tunnel data. This theory, demanded a large slat extension at manoeuvre which resulted in a hockey stick track shape to cater for the landing case. As mentioned the low sweep design cases were based on 2D aerofoil data involving much wind tunnel testing and converting to 3D using geometric sweep concepts. Ihe level of agreement between actual 3D and 2D experimental pressures are shown in fig 17. Agreement is good w&dating the design approach. Currently such a design would be achieved without resource to 2D test data since CFD codes are adequate for the task. It is worth pointing out the aerodynamic, but not actual, benefit of the VG concept. Fig. 18 shows a comparison of L/D at a high subsonic Mach number compared with a fixed wing. The advantages of increased span are obvious.

c,

THE TRANSONIC

BIAS WING

In this example (Ref 13), improvements to the basic wing are obtained through the use of supercritical aerofoil technology, variable camber and aeroelastic tailoring. The wing and configuration considered is the 42 sweep case already shown in fig 8A. The requirements assume a high turn rate at M = 0.9 and a good subsonic dash speed. The configuration is fitted with a tailplane. The supercritical concept is an extension of the constant near sonic roof top aerofoils e.g. ESDU data sheets, but with the near sonic level replaced by a supercritical controlled region. The terminal shock wave is weak and so for a given lift distribution the resulting wave drag is reduced. This calls for careful design of the upper surface curvature but the gains over older conventional aerofotis are significant as shown in fig 19A. A reminder of typical design requirements is shown in fig. 19B where the main design point for supercritical development lies within the SIR region A.

19-9
Fig 19A/B - Main Supercritical Flow Features Wing
POINT

Fig 20 A/B - Off Design

Considerations

LOW LIFT
DRAG RISE

60S TECHNOLOGY ADVANCED TECHNOLOGY POINT A EQUAL -1 CP 0 CL CD I REDUCED WAVE DRAG REDUCTION OF

HIGHER LIFT
INCREASED -REAR LOADING

TYPICAL

DESIGN

REGIONS
Q SUSTAINED MANOEUVRE 1-g FLIGHT

cp I

POINTS

C AND

HIGH

SUCTIONS

ATTAINED

D Q . n

C LARGE

SHOCK

FALL IN USABLE BELOW DESIGN

LIFT

MACH

1.0 NUMBER

Off design conditions at low lift (fig ZOA) or lower Mach numberhigh lift (fig 20B) can pose problems for these aerofoils through lower surface shock formation for the fist and LE shock bubble separations for the second. However incorporation of variable camber with lower surface design can reduce the first of these effects where small up deflections of LE and TE flap greatly extends the potential flight Mach number at low lift coefficients.

This type of design can usually be achieved with the use of CFD codes alone since the aim is to achieve attached flow throughout. Fig 21A indicates a theoretical design exercise which designed the upper surface for a high g manoeuvre point and the lower surface shape for a low speed high speed dash requirement, with an intermediate design to serve as additional verification. The resulting experimental drag rise boundaries shown in fig. 21B vindicated the design approach. Additional LE down flap deflection will obviously improve the high lift performance at lower Mach numbers.
Fig 21A - Benefits of Variable Camber Theoretical Design - Equivalent 2D Aerofoil

Section

LE/TE Flap angles

0.0

I,,
0.4 0.6 Mach

, ,
. -3.zs3.5

~gI&$gs~;

0.8
Number

1.0

c -0.8-

-----

10

19-10 Fig 21 B - Benefits Experimental Verification of Variable Camber - Drag Rise Boundaries

[Link]
L \ \

Drag Rise

It is worth noting that the nose down twist induced streamwise on a swept back wing under positive g is in the right direction from the aerodynamic viewpoint fig 23A. If aeroelastic tailoring had been adopted using composites then a twist increment over and above that due to natural bending could have provided a further small advantage. The range of twist possibilities available with different composite ply orientations are shown in fig 23B In this example almost 213 of the required twist is achieved from the natural wing bending of a conventional structure.
Fig 23A/B - Aeroelastic Considerations

0 65

0.70

0.75

0.80

I [Link] MACH NO.

Good comparison between CFD and experiment for this case has already been shown in fig 8B. This type of design approach aims to achieve a smooth upper surface at the high g design point and thus the idealised geometry contains the aeroelastic deformation of the wing. To achieve a jig shape for manufacture this deformation must be determined and subtracted from the ideal high g shape. This is an iterative process between aerodynamics and structures. For the wing considered here the ideal twist requirement is close to lo, whilst the initial twist assumed for a wing build Og shape was 3.5, as seen in fig 22A. This implied that an aeroelastic increment of 6.5 was needed to achieve an optimum design. However detailed structural work using a conventional aluminium alloy skinned multi spar wing structure gave only approximately 4 of twist implying that extra twist would have to be built in to the wing Og shape. This was thought to be acceptable when combined with variable up flap deflection to alleviate potential lower surface problems near the wing tip. Fig 22B illustrates the benefit of reduced twist in the outer wing region.
Fig 22AB - Twist Effects

Aeroelastic

Tailoring ASPECTRAT103.3 TAPERRATIO 0.4

IO

P0SSIBI.F

This wing was also considered for manufacture (Ref 14) where the geometry was simplified, from an initial 8 control station wing definition to a three control definition. This was an iterative process involving CFD and aeroelastic and loading calculations as indicated in fig. 24. This resulted in only minor changes to its performance, as indicated in fig. 25 where L/D distribution are shown at moderate and high subsonic Mach numbers. The final results with this type of approach and careful design are impressive and shown in fig. 26 compared with a 60s technology design. Gains in L/D of 50% are evident at high manoeuvre lift conditions.
Fig 24 - Productionisation

-6.0 -8.0 Pressures Near the Wing Tip (h=0.87)

19-11 .Fig 25 - Experimental DESIGN SNAPS - .-- PRODCT,ONISED Results

Fig 27 - Areas Canard dignosis / design

of CFD Application Delta - EAP

Fig 26 - Effect of Advanced Variable Camber Wing Design on Lift I Drag Ratio

I /7) RATIO

HNlMUM DRAQ ENVELOPE Wrm [Link] LE U4D IE DROOP

The foreplane-wing layout has many advantages in this flight regime since it suffers from a smaller aerodynamic centre shift with Mach number than a tailed configuration, and hence gives benefits in trim drag. It also gives cleaner afterbody lines with no adverse tailplane I afterbody pressure drag interference. Coupled with relaxed longitudinal stability and a powerful flight control system this type of configuration works well over a wide Mach number range. A constraint on the wing design lies in the area of aeroelasticity. For the tailed configuration considered earlier benefits were obtained through wing twist variation. This was since the tail could be used for roll control allowing a flexible wing. Differential foreplane is ineffective for roll thus the wing must be stiffer to provide adequate roll via ailerons or flaperons. In addition on a delta wing the effective flexural axis is relatively unswept and any wing bending produces very little streamwise twist. Thus any wing twist required for aerodynamic performance has to be built in twist. An excessive amount will compromise the high speed dash case. However the swept LE wing can produce problems away from the supersonic region, principally in the formation of LE vortices and possible shock I vortex interactions. Also the presence of a control surface ahead of the wing and a complex fuselage shape means that we must consider an integrated approach from the outset and cannot think in terms of a wing in isolation design. Due to the complex nature of the problem it has been found that EULER codes, at least, are required to provide accurate estimates as incidence develops. These have been successful when used with empirical design rules. Example results have already been seen in fig. 1OA. Ihe main important parameters which the wing designer needs to consider or influence are mentioned below. It is surprising how basic configuration features which are not directly related to the wing can exert a strong influence on the design process. Ihe first of these is the complex fuselage shape and wing location which can lead to strong interference even at zero incidence. As we have seen in the Euler result shown in fig. 10B for a wing - body case the inadequate representation of the body has lead to a poor prediction of zero lift moment. To improve this capability a hybrid approach was adopted which combined the effect of the body, as calculated by a BAe panel method, and the Euler code for wing alone. This gave a much improved match with experiment for zero lift moment and stability as shown in fig. 28. this is an example where the designer makes use of, at the time, inadequate tools but improvises to obtain an effective design procedure.

LIFT COEFFIUENT

SUPERSONIC

BIAS

Here illustrated by EAP (Experimental Aircraft Programme) design and later EFA (European Fighter Aircraft - now EF2000). CFD was used for many parts of the airframe as indicated in fig. 27 but here we only look at the wing design problem. Obviously here a high LE sweep is called sufficiently such that the LE is subsonic leading edge suction at the main supersonic However there are many other important which affect and constrain the wing design. for, ideally swept to achieve some manoeuvre point. points to consider

19-12 Fig 29 - Comparison - Modified Mach Euler v Experiment 0.9 EuLm THKORY SIMPLE BODY Y EULERWING WITH PANELBODY * ExPERlMEm . Fig 30 - Evaluation 5u
PITCHING

of Hybrid
MOMENT

Method Data
AT ZERO

Against
INCIDENCE

Experimental
AND LIFT

cL
0.8

u=o I

1
/

9 d A
.&Y
Fig 31 - Design
C bf
0.03

04

OY/
0.2

MACH
Targets for Moment at Zero Lift

Application of this procedure to the EAP project showed the complexity of wing - body interference as shown in fig 29. It is apparent that the contribution from the wing alone is benign at zero lift and is dominated by interference effects. The procedure also compared well with experimental results as shown in fig. 30 and therefore was used to guide basic design work on the EFA in this low incidence region. A typical target zero lift pitching moment distribution through the Mach number range is shown in fig 31 where the transonic hump is limited by the loads on the foreplane. Such a basic requirement can thus limit the scope of the wing design as small variations in camber can produce large interference effects. The zero lift moment can also strongly influence drag at supersonic speeds through its effect on trim drag. This will be considered later.

0 07.

0.01

1:o

MM!H NO.

Fig 29 - Pitching

Moment

at Zero

Lift

The same procedure was also used to assess and design out trim drag problems. A typical prediction of the TE flap contribution to drag and moment, a prerequisite in the estimation of trim drag, compares well with experiment - fig. 32. Here again incremental predicted force data was used and applied to a known experimental baseline configuration to obtain trimmed aircraft drag estimates. It was also calculated, and confirmed experimentally, that use of TE flap to trim was more efficient than the use of the foreplane.
Fig 32 - Evaluation of

THEORY PANEL METHOD

AC,,, ACD Project

Procedure

AND

EULER

FOR WING

BAe

High

ACC, ACM
. --EXPERIMENT

Speed Wing Development Model Due to 5 Trailing Edge Flap - M-1.4

19-13

The leading edge separation was controlled using a combination of Euler CFD with an empirical LE separation criterion and wind tunnel data. Fig 33 shows that careful optimisation of the leading edge region, in particular near the wing apex can control the inboard vortex movement leading to regular force behaviour to high incidence. Additional control, particularly at high subsonic speeds is available with LE flap, where care is taken to maximise the upper surface radius in the region of the hinge. Effectively forming a slotless slat device. This device is scheduled with incidence and demands high actuation rates due to the high agility of the aircraft.
Fig 33 - Design Provide Vortex of Leading Edge to Control - Mach 0.90

should be as unstable as possible, allowing use of large flap angles to trim with consequent reductions in drag and supersonically, trim drag. However this must respect the need to recover the aircraft from high incidence and the FCS limits, set by a limit on the time to double amplitude fig. 35.
Fig 35 - Flight Control Systems Why is there a limit on Instability?

Intibllliy K MAC.

Y/S
0.2 0.4 0.6

0.8

SPAN
1.0

Examples of the effect of zero lift moment on trim drag are shown in fig. 36A. The effect is significant. Similarly the effect of basic instability level on lift dependent drag is significant - fig. 36B. Both examples are at the main BVR (Beyond Visual Range) manoeuvre Mach number of 1.6.
Fig 36A - Effect of Zero
TRIM DRAG

VORTEX MclvEhlENl

Lift Moment
SIX

on Trim

Drag

L.E. DROOP

SUPERSONlC

10

20

a
&OS oios
CC @ 33%

Cm0

The overall aircraft stability level, resulting from the capability of the Flight Control system, strongly influences the lift dependent drag of the configuration. A typical aerodynamic stability trend with Mach number is shown in fig. 34 where the capabilities of the FCS system are superimposed.
C

Fig 363

- Effect

of Stability on Drag hk1.6, CL=0.25

due to Lift

III 0.04

Fig 34 - Longitudinal

Stability Number

- Variation

with

Mach

Am-o Centre Position (Forward)

o.03i--optimisatlon by controlling the

Mach reducing overall the Irinr

No

pm@~.
stable

The two boundaries come together at a fairly high subsonic Mach number and at low altitude. Thus it is the low altitude subsonic Mach number region which sets the permissible stability level. From the performance viewpoint the aircraft

The search for a configuration with excellent supersonic performance coupled with good transonic and subsonic performance is difficult. It was found that the best balance was achieved by designing for the supersonic manoeuvre point, using a combination of linear and EULER methods with wind tunnel data, and achieving the sub / transonic requirements with LE and TE flap. The alternative of designing for the transonic cases and decambering to achieve the supersonic cases was not as successful. This is illustrated in fig. 37 for early work done on the Experimental Aircraft Programme. The

19-14

predicted drag values for different types of wing design each with scheduled LE and TE flap are plotted against pitching moment at constant lift coefficient. This approximates to comparisons at trimmed conditions. Transonic, mild transonic and supersonic designs were investigated but even the mild transonic option was found to give unacceptably large (4 YG) zero lift drag penalties. The final design selected was based on a refinement of a supersonic inverse design calculation. This was also the conclusion reached in ref 15.
Fig 37 - Supersonic k1.4, Wing and Drag - Lifting Surface Theory Body, CD v CM at Constant CL ^ 0 ^ <I

Fig 39 - Twist
Reduction in drag hue due to Twist and Camber

and

Camber

on 53 D Wing

20%

STR

10%

Showing variation due to lift with pitching

1
M=O.7 STR

1
M=0.9
n

STR M=l.

moment

.flOR

for combinations of L.E. and T.E. flaps

Increase in
Drag at

-6X

.(I06 CD 1

c L= 0.0

CD ,006

0.01

I
1

CL=0.05 due to Twist and Camber

005

n
M=1.6

lg

SUPERSONIC
. ,003

WING

. ,002

- .OOl

A typical exchange rate between mean wing camber levels, subsonic STR and supersonic lg drag was established early in the design work. This serves as an useful guide to permissible camber levels - see fig. 38. The advantages of incorporating detail shaping, twist and camber are shown in fig. 39. The final design provided large gains in the high subsonic flight regime compared with a planar wing with scheduled flaps, with only a small penalty in supersonic lift drag. For the supersonic STR point the design requirement
was met.

Here we have a more extreme form of the previous planform swept to 60 and beyond, possibly with a cranked subsonic leading edge planform. Extensive literature is available on the basic flow features (Ref. 16) and design methodology on such wings - the majority based on linearised theory with empirical improvements to account for finite levels of LE suction or limiting suction levels to avoid shock induced separations (Ref. 17). Excellent review papers have been written (Ref 18) on this subject. However the majority of designs based on linearised theory have concentrated on cruise conditions with the aim of achieving minimum drag and attached flow. Even for this case linearised theory on its own is not successful in achieving an optimum solution. Inherent singularities at the LE and the centre line results in over cambered regions. This results in the optimum performance being achieved at a much higher lift coefficient than desired (Ref. 19). Ihe designer thus needs to factor his design lift coefficient in an empirical way. For a combat aircraft, even without a supersonic cruise requirement there may well be a supersonic manoeuvre requirement, as seen in fig. 14. Linear&d theory even used in conjunction with experimentally based empirical limits may be inadequate for this task - though such rules do serve to give the designer a physical feel of what is possible to achieve. This is often not available from an inviscid CFD code. However, and in addition, a complex wing-body-control surface is unlikely to be adequately represented using a linearised approach. An interim hybrid development has been successfully used at BAe (MA&A) to partly overcome these deficiences. The rather tortuous procedure is illustrated in fig. 40, where the main aim is to design an efficient wing taking into account the effects of the rest of the configuration. Tiimmed conditions are assumed throughout and the process has been found to converge to adequate accuracy in three to four iterations.

Fig 38 - Exchange

Rate

0 004-

DESIGN

WING Supersonic /

I Subsonic Gain Camber DESIGN WING

19-15

However proper treatment of this area needs a fully non-linear code for the whole configuration but even these need empirical rules to limit the calculated pressures.

Fig 41 - Design

Modifications to the Spanwise Distributions

Pressure

--I

~~~~

gz////m smnEmlc WN General Arrangement Configuration NO-UWim MYGN WIG of the Q8513

Fig 42 - Datum

Wing

Ml .6 I a=7.3

UPPER SURFACE
-0.320

The approach advocated in refs 20, 21 concentrates on reducing the strength of the cross flow shock. However to apply the method a non linear e.g. Euler or FP code is required to predict the shock in the fust place, The geometry is then modified via a change (increase) in LE radius and spanwise camber to reduce the shock strength. Taken in isolation this method has produced large reductions in drag which was a benefit of using non-linear codes. However it is also important to consider the effect of such changes on the longitudinal stability variation with incidence and to check that the drag gains are not cancelled by trim drag changes. Such an exercise (Ref. 22) was undertaken in an attempt to improve the drag of the swept delta wing of the previous section. The basic wing body configuration and the design concept are shown in fig. 41. The calculations were done using an Euler code and the wing upper surface isobars at the main design point are shown in fig. 42. The cross flow development at a range of incidences are shown in fig. 43 where it is noticed that at the main design point C = 0.28 the shock strength is possibly just strong enough to promote a separation on the outer wing. A large number of designs were tried, in an attempt to improve on the supersonic and transonic performance of the DATUM twisted and cambered wing, by applying analytic forms to streamwise and spanwise camber, LE blunting, twist and the LE and TE flaps. Only three of these results are considered here.

-0.302 -0.2M -0.260 -0.240 XW20 -[Link] -[Link] -0.160 -0. I40 -0.120 -0.100

WING ISOBAR PATTERN AT THE SUPERSONIC MANOEUVRE DESIGN POINT: kkl.6, n=4g FINE GRID

19-16
Fig 43 - Wing Spanwise Pressure Distributions at M=1.6 Fig 45 - Effect of Redesigned Wing on Drag at Mach 0.9

co0 1

0.006. 0.004 0.002 Fig 44 shows basic wing drag (minus ideal vortex drag), pitching moment, and trimmed drag as a function of lift coeff. Ihe results shown are for the datum twisted and cambered wing (labelled DATUM), a wing with the twist and camber removed (labelled SYMMETRIC), and the final redesigned wing (labelled FINAL) which has 50% greater twist than the datum and less camber - especially over the inner wing. The upper fig. shows that the untrimmed drags of the datum and final wings are similar with the symmetric giving a drag increase. However the final wing has a more positive pitching moment which greatly reduces trim drag and gives a 0.004 (40 drag counts) reduction in total drag relative to the datum. Note that the symmetric wing moment is in a sense to increase trim drag. However the FINAL wing has much inferior performance at the transonic case M = 0.9 as shown in fig. 45. Thus, though the supersonic manoeuvre condition could be improved through this design approach, the negative inboard camber and increased twist would violate the &, constraint imposed on this design. In addition, as seen in fig. 45, the subsonic / transonic performance is much inferior than the DATUM. gThe balance required between sub / tram / supersonic must be considered at all times.
Fig 44 - Refinement of Wing to Optimise CD0 Trim
___ ---I

WING

0.1

0.2

0.3

0.4

0.5

The cross flow profiles for these cases at a key longitudinal station are shown in fig. 46 where an apparently near ideal profile is obtained with the symmetric wing and a far from ideal one with the final wing. This is though the basic drag of the FINAL wing is less than that of the SYMMETRIC case, as seen in the top figure of fig. 44. It is thus conjectured that the use of wing pressures alone at supersonic speeds to judge the implied drag benefits of different designs is not enough. The local flow development is different to the classic 2D supercritical flows on which the cross flow concept is based. Thus the overall force data should also be used and, of-course, trimmed conditions compared.
Fig 46 - Comparison -0. , of Various Manoeuvre
- 0.7

Wings

for Supersonic

M = 1 .cj lt.

FMAL P

-.--.syM*oRc I 0.0 001E---I== 00 -__--------___ ---_ --__ -- _-- ---------__ -----, --. -01 L 0.2 06 0 Od SPAN la
The above design problem also raises the vexed question of the use of blunt leading edges in supersonic flow. Various researchers have proposed different solutions, fig. 47, and much work has been done on aerofoils which show a benefit with blunting in supersonic flow. However it is difficult to maintain this benefit over a wide lift range. In practice detail design should investigate the possible options but it is worth noting even when the wing LE is subsonic the wing-root junction may unsweep the Mach lines and cause a local drag increase (Ref 26) which can cancel out any drag gains elsewhere along the span.

0.0

0.27

0.28

0.29

0.30

03co.32

19-17
Fig 47 - Leading
0.010

Edge

Radius

Along

Span

Rg

48a - Novel

Wings Experiment

- Comparison - 5oP Lambda

of Theory Wing

with

0.008

X * + Q

DhlVAWNQ PmwEr*LkfM AstaLL ET AL Rcr23 WOOD EF AL Rd25

0.000 .I--l 0.0

0.2

0.4

0.6

0.8

YtYT

1.0

AND FUTURE

DEVELOPMENTS

These are divided into processes, aerodynamic developments and flow physics and will be covered very briefly. PROCESSES CFD. A current design challenge for the military ahcraft wing designer is to obtain satisfactory aerodynamic performance from novel configurations which achieve low observability. This also poses a challenge to the CFD codes, where existing evaluations against conventional experimental data bases may not be suitable. In addition body faceting, chines, sharp leading edges will provoke separated regions at low incidences thus calling for an improvement in separated flow predictions. To meet this challenge both Euler and Navier Stokes codes have been evaluated on a number of such configurations. An example for a 50 lambda wing is shown in fig. 484, B, C. The first figure shows good agreement with the force data to a high incidence and the theoretical surface skin friction lines (fig 48B) are close to the experimental data. The experimental data lies in between the results from the two turbulence models. The wings in this example were of symmetric section and the fuselage was of simple circular cross section and did not strongly influence the wing flow. Predictions at high subsonic Mach number are shown in fig 48C, where potential problem regions are indicated over the rear of the outer wing. It remains to be seen whether the N - S code will be as successful on full configurations and this topic is the subject of ongoing investigation. However on the simple configuration presented here the use of Navier Stokes for design with separated flows is a possibility.

Fig 48b - Skin

Friction

Lines

M = 0.3 THEORY
SECONDARY SEPARATION LINE

BL + DS
a, -----= A

THEORY

JK

SECONDARY

19-18

Fig 48~ - Transonic Navier Stokes

Prediction - Mr0.95

Fig 49 - Current

and

Future

Ways

of Working
-I

OUTER LOOP
CONFIGURATION LAYOT EMPIRICAL METHODS/ SIMPLE cm DATABASB3 MoO(LBvELI)

TRATLMG

WGE

,
INCIDENCE = 6.5

INNER LOOP
CONFIGETRATION/COMPONMT OPTIMISATION ADVANCED cm hfDo(LEvELZ)

,! I-+ I
AERODYNAMIC

lTNFORMXTION ELOW

DEVELOPMENTS

INCIDENCE

= 8.5

NUMERICAL OPTIMISATION AND MULTIDISCIPLINARY OITIMISATION As will have been noticed from this paper the process of aerodynamic design is significantly iterative and it is natural to expect that numerical optimisation could play a major role in future developments. It is only recently that such methods have been developed for routine(?) use for the non linear flow codes e.g. Euler. The system which is most familiar to the author is that of Ref. 27 and has produced, again with user defined empirical limits and user nursing, a good standard of design on a number of configurations. It is likely that such methods will be needed at an earlier stage in the design process, as long as proper evaluation of the underlying CFD has taken place, due to the less certain data base available on novel configurations. Eventually it might be expected that the outer (coarse) and inner (detailed) design loops shown in fig 49 will merge. In the same time frame it is expected that MD0 (Multi Discipline Optimisation) procedures will be developed, these merging across the disciplines. Pilot schemes are already in place linking aerodynamics, structures and low observability.

To alleviate the strength of shock waves and to possibly increase buffet penetration external ramps / bumps have been proposed (Ref 28). These are still under evaluation but little data is available on their effectiveness or otherwise on 3D configurations. Wing body blending holds promise of reduced supersonic drag but is configuration dependent and the subject of detail design. From the current military perspective the emphasis is on SIOVL with supersonic capability. As such these are non ideal shapes for supersonic flight and it is a matter of doing the best one can at supersonic speeds from an inherited poor datum shape. FLOW PHYSICS

A significant development over the last 5 years has been the growth in collaborative, and technology exchange, activities between the FSU and western countries. BAe have identified a number of topics which may hold promise in aerodynamic, structural and chemical areas. One topic which could influence supersonic and hypersonic flight is the use of plasmas to reduce drag - wave drag and possibly skin friction drag. This has aroused interest in both the U.K. and the U.S.A and is the subject of ongoing collaborative programs in both countries. A principal short term aim is to repeat and validate the FSU claims for supersonic drag reduction and to apply the effects to flight conditions of interest to BAe. A longer term aim is to develop an understanding of the observed phenomenon and to increase plasma efficiency. CONCLUSIONS Essentially for combat aircraft the numerous design requirements and constraints demand that a broad look is initially taken on wing design aspects before starting the detail design. It is often beneficial to undertake global calculations on aminimum/ maximum range of wing parameters so that the initial wing base line definition can be tailored to suit the evolving configuration. When the configuration, especially

19-19 fuselage, intake, afterbody are fiing up then detail design can begin. This takes a different form depending on the type of aircraft, and hence the type of wing flow, under consideration. In the future it is expected that increased use of Navier Stokes codes will figure in the control of separated flow, by design, while the role of numerical optimisation for detail design will spread into the early phases of configuration design. REFERENCES 1. AGARD Flight Mechanics Panel WG-19 AR-314. Operational Agility 2. Carmichael, R.L. and Woodward. F.A., An integrated approach to the analysis and design of wings and wing - body combinations in supersonic flow, NASA TN D - 3685, Ott 1966 3. Woodward, F. A., Tinoco, E. N. and Larsen, .I. W., Analysis and design of supersonic wing - body combinations including flow properties in the near field. part 1 - Theory and applications, NASA CR - 73106,1967 4. Middleton, W. D. and Lundry, J. L. A computational system for aerodynamic design and analysis of supersonic aircraft, NASA CR - 2715, March 1976. 5. Hall. M.G. and Firmin. M.C.P. Recent developments in methods for calculating transonic flow over wings. I, I.C.A.S. Paper 74-18, 1974 6. Albone. C.M. Numerical solutions for flow past wing body combinations. IUTAN symposium Transsonicum II Ed. K. Oswatitsch and D. [Link]-Verlag, 1976. 7. Forsey. C.R. The calculation of transonic flow over wing body combinations using the full potential equations - ARA Report 59 8. EJ65 user guide - Euler Viscous coupled single block code 9. EJ83 user guide - Euler multi block code 10. Lucking. M. The Warton contribution to the BAe M-BNavier Stokes assessment exercise BAe-WAe-RP-GEN00828 March 1993. 11. Benton J.J. 3D multiblock N-S code RANS-MB Current status and possible future developments. ADE-ETRR-RES-CM-3769 Jan 1993. [Link]. G.D. Further assessment of the unstructured FLITE3D suite of codes using T40,W4,SMARM and generic missile test cases. 13. Holt, D. R. Some particular configuration effects on a thin supercritical variable camber wing, AGARD CP - 285 Subsonic/transonic configuration aerodynamics(May 1980) 14. Probert, B. and Hoit, D. R. Advanced combat aircraft wing design, ICAS Congress( 1980) 15. Bradley. R.G. Practical aerodynamic problems - Military aircraft transonic perspective - NASA/Ames Research CentreMoffet Field ,CA 94035(Feb 18-20,1981) 16. Rogers, E. W. E. and Hall, I. M., An introduction to the flow about plane swept -back wings at transonic speeds, Journal of the Aeronautical Society, Vol. 64, Aug 196O,pp.449-464. 17. Kulfan, R. M. and Sigalla, A. Real flow limitations in supersonic airplane design, AIAA Paper 78 - 147 Jan 1978 18. Kulfan, R. M. Wing geometry effects on leading - edge vortices, AIAA - Aircraft Systems and Technology Meeting (Aug 20-22,1979,NY) 19. Carlson , Mann. Survey and analysis of research on drag supersonic due to lift minimisation with recommendations for wing design, NASA TP 3202 1992 20. Mason, W. H. aand Dafomo. G. Opportunities for supersonic performance through non linear aerodynamics, AIAA paper 79-1527,July 1979 21. Mason W. H. SC - A wing concept for supersonic manouvre, AIAA 83 - 1858 July 1983 22. Stanniland, D. R. Supersonic manoeuvre design guidelines for a highly swept military aircraft wing, ARA Model Test Note Q85/3, June 1990. 23. Ashill, P. , Fulker. J. , Simmons. P, A study of flows over highly swept wings designed for manoeuvre at supersonic Aerodynamic design for speeds, RAeS Conference supersonic flight, April 1988. 24. Pittman, J.L.,Miller. D. S.,Mason. W. H. Supersonic non linear attached flow wing design for high lift with experimental validation, NASA TP 2336 25. Wood. R. M.,Bauer. S.S.,Evaluation of a three dimensional empirically derived wing at supersonic speeds. AIAA-88-0481 1988 26. Fulker. J.L. ,Ashill. P.R.,Shires. A. A theoretical and experimental investigation of the flow over the supersonic leading edge of wing body configurations. DRA/AP/TM9341/1.O(Aug 1993) 27. Lovell. D. A. and Doherty. J.J. Aerodynamic design of aerofoils and wings using a constrained optimisation method, DERA,EAC 1994 paper 3.21,Oct 1994. 28. Ashill. P.R. ,Fulker. J. L.,Shires. A novel technique for controlling shock strength of laminar flow aerofoils, DRA Tech Memo.