AEROELASTICITY MATTERS: SOME REFLECTIONS ON TWO DECADES OF TESTING

IN THE NASA LANGLEY TRANSONIC DYNAMICS TUNNEL

by

Wilmer H. Reed IIINASA Langley Research Center

Hampton, Virginia 23665

Presented at the InternationalSymposium on Aeroe1asticity

Nuremberg, GermanyOctober 5-7, 1981

AEROELASTICITY MATTERS: SONE REFLECTIONS ON TWO DECADES OF TESTINGIN THE NASA LANGLEY TRANSONIC DYNAt1ICS TUNNEL

Wilmer H. Reed 111*

ABSTRACT

Testing of wind-tunnel aeroelastic models hasbecome a well established, widely used means ofstudying flutter trends, validating theory andinvestigating flutter margins of safety of newvehicle designs. The Langley Transonic DynamicsTunnel was designed specifically for work ondynamics and aeroelastic problems of aircraft andspace vehicles. This paper presents a cross sec­tion of aeroelastic research and testing in thefacility since it became operational more than twodecades ago. The paper illustrates, by means ofexamples selected from a large store of experience,the nature and purpose of some major areas of workperEormed in the Transonic Dynamics Tunnel. Theseareas include: specialized experimental techniques;development testing of new aircraft and launchvehicle designs; evaluation of proposed "fixes" tosolve aeroelastic problems uncovered during devel­opment testing; study of unexpected aeroelasticphenomena (i.e., "surprises"); control of aero­elastic effects by active and passive means; and,finally, fundamental research involving measurementof unsteady pressures on oscillating wings andcontrol surfaces.

I. INTRODUCTION

Aeroelastic problems encountered by high-speedaircraft and launch vehicles most often arise inthe transonic speed range, the very range where,regrettably, aerodynamic theory is least developed.Designers and researchers, therefore, must placeheavy reliance on wind-tunnel models to aid inclearing new designs for safety from flutter andbuffet, evaluating solutions to aeroelastic prob­lems, and studying aeroelastic phenomena at tran­sonic speeds. The lack of suitable wind-tunnelfacilities prompted A. A. Regier in 1951 to proposethat the NACA construct a large transonic windtunnel to be dedicated specifically to work ondynamics and aeroelasticity problems associatedwith development of high-speed aircraft. (1) Herecommended that such a tunnel should: (1) be aslarge as possible; (2) be capable of operating overa wide density range; (3) use air or Freon as thetest medium; and (4) operate through the criticaltransonic speed range, up to Mach number 1.2.Regier's proposal began to become reality in 1955when work started on the conversion of a large sub­sonic tunnel, the Langley 19-foot pressure tunnel,to a l6-foot (4.87-m) transonic tunnel withFreon-12 or air as the test medium. This newfacility, designated the Transonic Dynamics Tunnel(TDT), became fully operational in 1960 and hasserved ever since as a National facility dedicatedalmost exclusively to work on dynamics and aero­elasticity problems of flight vehicles, particu­larly in the critical transonic speed range.

For more than two decades the unique capabili­ties of the TDT have been applied in a greatvariety of aeroelastic investigations. The scopeof this work ranges from flutter-proof tests ofhigh-performance aircraft to fundamental researchon aeroelastic phenomena. Figure 1 depicts someof the major technical areas currently under studyin the TOT. For example, the facility is used toverify, by means of dynamic models, the fluttersafety and aeroelastic characteristics of most U.S.high-speed military aircraft and commercial trans­port designs; to explore flutter trends and aero­elastic characteristics of new configurations; Eoractive control of aeroelastic response of airplanesand rotorcraft; for ground wind loads, flutter andbuffet testing of space launch vehicles; and forunsteady aerodynamic load measurements on oscil­lating wings and control surfaces.** These andother specific areas of work performed in the TDThave been reviewed in various prior publica-tions. (2-11) A survey by Reed(8) indicates thatthe predictions from a variety of aeroelastic modelstudies in the TOT have, in general, been substan­tiated in full-scale flight tests. The intent ofthis paper is to portray the broad picture of aero­elastic testing and research performed in the TDTsince it became fully operational in 1960.

The paper is organized along the followinglines. First, to set the stage, some salient fea­tures of the facility are described followed by adiscussion of specialized testing techniques devel­oped for the study of aeroelastic problems. Then,drawing from a considerable store of research andtesting experience acquired over the years, exam­ples are selected and used to illustrate the natureand purpose of major activities performed in theTOT. These activities include: development test­ing of new aircraft and launch vehicle designs;evaluation of proposed "fixes" to solve aeroelasticproblems uncovered during development testing;study of unexpected aeroelastic phenomena (i.e.,"surprises"); control of aeroelastic effects byactive and passive means; and, finally, fundamentalresearch involving measurement of unsteady aero­dynamic pressures due to dynamic motions of liftingsurfaces.

It should be understood that the paper, being inthe nature of a general survey, presents only aglimpse of these many-faceted aeroelastic studies.In most cases, however, more detailed informationmay be found in the references cited in the paper.

II. TRANSONIC DYNAMICS TUNNEL FACILITIES

The operating characteristics of the TOT andsome major features which make this facility par­ticularly suited for experimental work on dynamicsand aeroelasticity problems are shown in Figure 2.

*Chief Scientist, Loads and Aeroelasticity Division, NASA Langley Research Center.,~*

The oral version of the paper included a short kaleidoscopic-type motion picture showing a "diary" offlutter-model testing in TOT from 1960 to present.

1

ACTIVE CONTROLS::..- -"

Figure 1. Some aeroelastic technology areas supported by the LangleyTransonic Dynamics Tunnel.

These features include a slotted 4.9 M x 4.9 M(16' x 16') test section, variable Mach numberfrom 0 to 1.2, variable total pressure from 0.01to 1.0 atmosphere using either air or Freon-12 asthe test medium. The large test section is impor­tant because large models allow more accuratesimulation of pertinent structural details, such ascontrol surfaces and greater Reynolds numbers. Awide range of density variation is required inorder to simulate altitude changes which oftenaffect flutter. The use of Freon-12 gas(dichlorodifluoromethane) as a wind-tunnel testmedium for dynamically-scaled aeroelastic modeltesting has several extremely desirable features,the most important of which is its high density(the molecular weight of Freon is about four timesgreater than that of air). A denser test mediummakes it easier to satisfy the density ratio scal­ing parameter which requires that the density ofthe model relative to the density of its surround­ing fluid be the same as for the full-scale air­plane in the atmosphere. Thus, the use of Freonpermits heavier and consequently more rugged, lessexpensive dynamically-scaled models. Also, becauseFreon has a low speed of sound (about one-half thatof air), the scaled vibration frequencies of modelsin Freon are half what they would be in air. Thishas the advantage of easing data acquisition fre­quency requirements and of reducing the scaledrotation'speeds of model propellers and rotors.

2

Further advantages associated with Freon as com­pared with air are a nearly three-fold increase inReynolds number for comparable dynamic pressuresand much reduced tunnel drive power requirements.

TUNNEL CHARACTERISTICS -

• TEST SECTION ------------- 4. 9m x 4. 9m (16' x 16')• MACH RANGE --------------0 TO 1.2• TEST MED IUM------------- AIROR FREON -12• TOTAL PRESSURE ----------0.01 TO 1.0 ATMOS.• REYNOLDS NO. (MAX)- 3x l06/m (l071ft)

SPECIAL TESTING FEATURES -

• COMPUTERIZED DATA ACQUISITION SYSTEM• "Q-STOPPER" FOR FLUmR TEST! NG• TUNNEL-FAN SAFETY SCREEN• SUSPENSION SYSTEMS FOR "FREE-FLYING" MODELS• GUST GENERATOR

Figure 2. Transonic Dynamics Tunnel features.

In addition to the Freon test medium, otherfeatures which make the facility uniquely suitedfor work on dynamics and aeroelastic problems aregiven in Figure 2. These include a computerized

data acquisition system especially designed torapidly process large quantities of dynamic data

for use in guiding the progress of tests. (10)Another special feature is a capability to rapidlydecrease the tunnel Mach number and dynamic pres­sure by means of quick opening valves which bypassa portion of the flow around the test section.This capability reduces the risk of damage or lossof expensive models when flutter is encountered.However, despite all reasonable precautions, modelsdo, from time to time, experience catastrophicflutter, and therefore safety screens are providedto protect the tunnel fan from model debris. Also,to enable simulation of airplane free-flightdynamic motions in the wind "tunnel, special modelmount systems have been developed. And, to studyairplane gust response problems, a system ofoscillating vanes at the entrance of the test sec­tion is used to generate a sinusoidal variation intunnel flow angle.

III. EXPERIMENTAL TECHNIQUES

Wind-tunnel tests of aeroelastic models requirespecialized experimental techniques seldom foundin other types of wind-tunnel studies. Over thepast years, a variety of new or improved experi­mental techniques has been developed by the staffof TDT and others to broaden capabilities forstudy of dynamics and aeroelasticity problems ofaircraft and space launch vehicles. Although mostof the techniques currently in use have beendescribed in earlier publications, it is felt thata brief review of a few of these would provide use­ful background for the present paper.

Model Mount SystemsIn wind-tunnel-model studies of dynamics and

aeroelastic problems careful attention must begiven to the manner in which models are suspendedin the tunnel. A model rigidly mounted in thetunnel does not, in general, properly simulate thedynamic characteristics of its free-flying counter­part. On the other hand, a "soft" suspension sys­tem that provides a model the freedom of motionneeded to simulate free-flying conditions may atthe same time introduce instabilities of its own.Considerable effort has therefore gone into thedevelopment of suitable "free-flight" mount systemsfor use with aeroelastic models of both launchvehicles and airplanes.

Launch Vehicle Models. An effective techniquefor predicting the buffet response of launchvehicles by use of a suitably scaled aeroelasticmodel is described by Hanson and Jones. (3) Thistechnique involves a complete-vehicle model sus­pended in the wind tunnel on a sting mount whichprovides the model freedom to respond essentiallyin its "free-free" bending modes. Two such sting­mounted models are shown in Figure 3. On the leftside of the figure is shown a O.OB-scale Saturn 1­Apollo used in buffet response and aerodynamicdamping studies in the transonic speed range. Themount system is designed to restrain the model inthe longitudinal and lateral directions but softsprings are provided to give the model a rigid­body-pitch degree of freedom which simulates thatof the full-scale vehicle with its control system.The difference in the weight and lift force actingon the model is compensated for by means of acable system through the sting mount and remotelylocated adjustable springs. Another important

3

feature of this mount system is a water-cooledelectromagnetic shaker installed inside the modelwith field coils attached to the sting and movingcoils attached to the model. This shaker is usedto excite model vibration modes for the purpose ofdetermining the aerodynamic damping in each mode.

Figure 3. Launch vehicle buffet models.

Shmm on the right side of the figure is aO.OSS-scale model of the space shuttle used influtter and buffet studies. This model was iso­lated from the sting by means of a pair of pneu­matic springs which allowed freedom in pitch andplunge modes. A locking system was provided torestrain the model in the event of structuralfailure or dynamic instability of the model on itssoft suspension system. During the tests themodel was "flown" at low lift coefficients byadjusting the sting angle and positioning theorbiter elevons to unload the wing.

Airplane Models. Large complete airplanemodels are routinely used in the TDT to study avariety of dynamic aeroelastic problems includingflutter, buffet, gust response, and stabilityderivative measurements. In most instances thefree-flight dynamic characteristics of the air­craft play an important role in these studies mak­ing it necessary that such characteristics besimulated in the model tests. To satisfy thisneed the so-called two-cable mount system wasdeveloped and has been used extensively in the TDTand in other tunnels as well. This basicallysimple model-mount system was first described and

(12)analyzed by Reed and Abbott. Although ini-tially developed primarily for use in flutterwork, the two-cable mount, with variations, hassince proven its versatility and usefulness inother areas as will be discussed later.

A schematic diagram of the basic two cablemount system is shown in Figure 4. Two loops ofsmall-diameter cable extend in mutually perpen­dicular planes from the model to the tunnel walls,one loop upstream and the other downstream.Cables pass through pulleys located within themodel contour and tension is applied by stretchinga soft spring in the rear cable. In addition, asnubber cable system (not shotm in Fig. 4) isprovided for emergency restraint. These smalldiameter cables cause little aerodynamic inter-

Figure 5. Buffet response measurements (Ref. 15).

need to simultaneously satisfy both Mach numberand Froude number (gravity scaling parameter). InFreon this is possible but only when the modellength scale factor is about 1/4. Except forsmall fighters most models tested in the TDT havelength scale factors less than 1/4 and consequentlyfly at smaller angles of attack (lift coefficient)than does the airplane. To permit simulation ofthe proper load factor on cable-mounted models alift-balancing device (see Fig. 4) has been devel­oped to counteract the lift in excess of the modelweight. This device, described by Hanson, (15) con­sists of a soft pneumatic spring which, by meansof a cable attached near the model center ofgravity, applies a relatively constant force withminimum restraint to model motion. This point­force simulation of gravity is only approximate,of course. Although the total lift is correct,the total lift distribution may not be, due toinertia and pitch rate effects. Comparative wind­tunnel/flight buffet studies by Hanson(ll) of theF-lll variable sweep fighter indicate, however,that the buffet response predicted by the modelcorrelate well with that measured in flight asindicated by Figure 5. This figure shows variousbuffet response quantities obtained in high-gflight maneuvers and in the wind tunnel on alI8-scale flutter model of the F-lll "flown" tosimulate these high-g conditions.

L~•0 •

.~

W.2 .4 .6 .8 1.0 1.2

CN

• AIRPLANE

o MODELcY. 0

INDICATED AIRPLANE fi).BUFFET ONSET~

.75

1.251.0.75.5

.25Ol.-_L------_L------_~:JLL--L--l

.75 ~.50

.25o ----'------'--1.......<

1. 00

WI NG ROOTBEND ING MOMENT.. 50PERCENT DESIGN .25

HORIZONTAL TAILBEND ING MOMENT,PERCENT DES IGN

o

C.G. ACCELERATION.PERCENT DES IGN

PNEUMATICLI FT BALANC INGDEVICE

Figure 4. Two-cable mount system.

Although in principle the mount system issimple, in practice a detailed stability analysisof the model/mount system is needed to guide thechoice of design parameters (e.g., cable geometry,cable tensions, and model e.g.) for each new model.In fact, a stability analysis for the cable mountedmodel is usually more complex than that of the air­plane because of the added degrees of freedom ofthe mount which may become unstable. Stabilityanalysis procedure for cable mounted models may befound in references such as 12, 13, and 14.

In addition to the requirement for a mount sys­tem stability analysis, it is also considered goodpractice for the pilot and test crew to gain"flying" experience on a "dummy" model prior torisking the more expensive aeroelastically scaledmodel. The dummy model is built much stiffer thanthe aeroelastically scaled model but has the samegeometry, and total mass and inertia propertjes.

Remotely operable trim controls on the modelare provided to keep the model centered in thetunnel throughout the test range. Pitch and rollare usually sufficient and a single operator or"pilot" can fly the model using a miniature air­plane-type control stick which positions the modelcontrol surfaces.

ference and have negligible mass, compared I'i ththat of the model.

As mentioned, the two-cable mount system hasbeen adapted for studying a variety of aeroelasticproblems in addition to flutter. Some of thesenon-flutter aeroelastic applications relating toaircraft stability, control, and loads have beendiscussed in review papers by Rainey and Abel, (7)Reed, (8) Abbott, (6) and Hanson. (15) The followingsection describes an adaptation of the mount sys­tem to enable study of high angle-of-attack phe­nomena, in particular, aircraft buffet response.

Aircraft BuffetFor some aeroelastic phenomena such as buffet,

the model lift coefficient becomes an importantscaling parameter. Models flown in the two-cablemount are usually trimmed to be in equilibrium atthe center line of the test section where aerody­namic lift on the model exactly balances the modelweight. This "l-g" condition on the model doesnot in general represent a l-g condition for theairplane. The reason for this has to do with the

Gust ResponseThe need for a wind-tunnel technique for study

of the dynamic response of airplanes to atmo­spheric turbulence, particularly in the transonicspeed range, stimulated the development of aunique airstream oscillator system for use in theTDT. This technique involves measuring the fre­quency response function of a "free-flying" aero­elastic model using as input a sinusoidal verticalgust field generated by oscillating vanes locatedupstream of the test section as indicated inFigure 4. The airstream oscillator system con­sists of two sets of biplane vanes on each sideof the test-section entrance. The vanes areoscillated sinusoidally in pitch at frequenciesup to 20 Hertz. Trailing vortices from the vanetips pass downstream near the side walls of thetest section and induce a reasonably uniform dis­tribution of vertical velocity components acrossthe model span. The model, suspended on the two-

4

o

MODEL! 0 TEST--CALC.

FLIGHT TEST.

16

100

CBM£g 8

PER deg

-200 ;--'----;!:-:----'------:-:~_~---L-----!--:-----L---.Ja .04 . 08 .12 . 16

REDUCED FREQUENCY, k

cable mount system, is free to respond to these"gust" inputs in approximately the same manner aswould the full-scale flexible airplane in freeflight. Guidelines for designing a model mountsystem that is not only stable but also producesnegligible distortion of the airplane short periodfree-flight mode have been developed and demon­strated in preliminary experimental gust responsestudies by Gilman and Bennett. (13)

To further verify the validity of wind-tunneldata obtained by the airstream oscillator tech­nique a comparative wind-tunnel/flight-analysisstudy was undertaken jointly by NASA Langley, theAir Force Flight Dynamics Laboratory, and theBoeing Company (Wichita Division). Because con­siderable flight data were available for the B-52Eairplane in the form of gust frequency responsefunctions, this airplane was chosen as the testvehicle. These flight-measured frequency responsefunctions were determined from simultaneous mea­surements of the random atmospheric turbulenceinput and the associated airplane response usingpower spectral analysis techniques such asdescribed in reference 16.

Shown in Figure 6 are the 1/30-size model ofthe B-52E, mounted on cables in the TDT togetherwith the gust vanes upstream of the test sectionand a gust measurement probe in the vicinity ofthe model. Some results from the study reportedby Redd, Hanson, and wynne(17) are presented inFigure 7, which shows frequency response for anondimensional wing bending-moment coefficient atthe midwing span per degree of vertical gust angleas a function of reduced frequency k (based onmean aerodynamic semichord). Shown for comparisonwith the measured flight data are the measured andcalculated frequency response functions for themodel.

The major response in this case comes from theshort period mode at reduced frequency k; 0.08.At very low frequencies the model response isaffected by the mount system vertical plunge modeand the airplane response by spurious pilot-inducedmotions; at higher reduced frequencies (k > 0.14)the low level of gust input in the wind tunnelleads to measurement inaccuracies and scatter inthe model test data. The overall satisfactorycorrelations between wind tunnel, flight, andanalytical predictions indicate this to be a use­ful valid wind-tunnel technique for airplane gustloads research.

Figure 6. B-52 model used in wind-tunnelgust response studies (Ref. 17).

Figure 7. B-52 model and airplane gustfrequency response measurements.

Stability Derivative MeasurementsAeroelasticity has naturally a major influence

on the stability and control characteristics offlexible aircraft at high speeds. The loss ofaileron control effectiveness or the change inlift-curve slope with increasing speed are promi­nent examples. Paralleling the methods developedfor flight testing, wind-tunnel testing techniqueshave been developed in the TDT for extracting air­plane stability derivatives from aeroelasticallyscaled models. These techniques again employ"free-flying" cable-mounted models. In the \~ind

tunnel, as in flight, the model response to knowninputs, such as control surface deflections orexternal forces applied through the suspensioncables, is measured and used in the equation ofmotion of the model and suspension system to solvefor the unknown aerodynamic coefficients. Two suchtechniques for identifying aerodynamic parametersfrom tests of cable mounted models are describedbelOl~ .

The first, and simpler, of these techniquesenables one to determine the aileron effectivenessand roll damping from aeroelastic model tests.This approach developed by Abel(7,18) is based onthe assumption that the dynamic roll response ofthe model to sinusoidal aileron oscillations can berepresented by a single-degree-of-freedom system inroll. Using a flutter model of a large cargotransport aircraft, the C-14l, Abel obtained aile­ron effectiveness and damping in roll data in thewind tunnel which agreed well with flight measureddata. The aileron effectiveness of this model wasalso determined successfully by an alternate method

described by Grosser(19) in which the model wassupported by a sting-mounted pylon with springs toprovide the model freedom of motion.

The second method is patterned after modern sys­tem identification techniques in current use forextracting airplane stability derivatives fromflight test data. This technique has been applied

5

by Bennett, Farmer, Mohr, and Hall(20) to deter­mine both longitudinal and lateral stabilityderivatives using dynamic models of the F-14 andspace shuttle orbiter. In these studies the two­cable mount system was modified by the addition ofservo torque motors and load measuring cells ineach of the cable loops as indicated in Figure 8.

IIII I

.1. I

~:-:-__ :-:::::-i:ULl£YSI VERTICAL ",

CABLE '

Figure 8. Active two-cable mount system.

By means of this "active" mount system, dynamicexcitation forces can be applied to the modelthrough the cables to provide known inputs suitablefor the parameter identification analyses. Thelongitudinal and lateral derivative sets determinedfor both modes were in reasonable agreement withderivatives determined from other sources. Table Ishows, for example, estimates of the stabilityderivatives for the shuttle orbiter determined bythe test technique of Reference 20 together withinitial estimates by the manufacturer based onother wind-tunnel tests and theoretical analysis.

TABLE I. SHUTTLE ORBITER STABILITY DERIVATIVES

Stability Initial Ref. 20derivative estimate estimate

Longitudinal:

CZCl -2.51 -3.04

Cm + Cmci -2.40 -2.68q

Crna 0.164 0.164

Lateral:

CYs

-0.920 -0.935

ClS-0.046 -0.032

Clp-0.288 -0.306

Cl r0.127 0.103

Cns 0.044 -0.022

Cn 0.194 0.011p

Cnr -0.231 -0.167

The active mount can be used also to augment thestability of an otherwise unstable model configura­tion. In this mode electrical signals proportionalto such variables as the torque motors' angularposition and rotational rate and/or model pitch and

6

yaw rates (determined from rate gyros in the model)are suitably mixed and fed back as input commandsto the torque motors. These stability augmenta­tion features of the active mount system have beenanalyzed and modeled by Chin and Barbero(2l) anddemonstrated in the wind-tunnel studies of Bennettet al. (20) using the shuttle orbiter model. Inthis case the model with its c.g. moved aft toproduce static longitudinal instability was effec­tively stabilized by means of feedback propor­tional to the output of a model pitch rate gyro.

Subcritical Testing TechniquesTo reduce risk of damage or destruction of

expensive aeroelastically scaled models due to thesudden encounter of aeroelastic instabilities,various subcritical testing techniques have beendeveloped to aid in predicting instabilities fromresponse measurements obtained under stable safetest conditions. With regard to flutter testing,a comprehensive syrnposium(22) on this subject washeld in 1975 which covered many aspects of wind­tunnel flight flutter testing. Most of the sub­critical flutter testing techniques in current usein the TDT were presented by various authors atthis symposium: notably, by Foughner; Hammond andDoggett; Bennett and Desmarais; and by Houbolt.Many of these procedures have been implemented onTDT data acquisition systems and others are beingdeveloped to aid the test engineer in guiding theconduct of flutter test on a near-real-time basis.

Recently, an upsurge of renewed interest inforward swept-wing concepts, which are usuallymore prone to static divergence than to flutter,sparked the development of subcritical staticdivergence testing techniques. Using a series ofsimple flat aluminum-plate models in the TDT

Ricketts and Doggett(23) evaluated six differentsubcritical divergence testing techniques. Fourof these were based on measurements of static datasuch as strain-gage measured mean bending momentsat the wing root; the other two methods were basedon dynamic measurements of such quantities asmodal frequencies and peak response amplitudes.

Two of these static methods are illustrated inFigure 9. Both use as input data the load/angle­of-attack gradient, Ai' measured at dynamic pres­sures that are well below the divergence pointwhile holding Mach number constant (Fig. 9a). Inone method a so-called "divergence index" param­eter, 6, is calculated using measured values ofthe loadiangle-of-attack gradient and dynamicpressure as indicated by the equation in Figure 9b.When plotted against dynamic pressure, q, thedivergence index is a straight line which passesthrough unity at q = 0 and crosses the q-axis atthe predicted divergence dynamic pressure. Experi­ence with the method has shown that the predicteddivergence condition is accurate even when extrapo­lated from subcritical data acquired at dynamicpressures well below the divergence point.

The other static method is illustrated in Fig­ure 9c. It is based on the observation by Flax(24)that experimental procedures developed in the1930's by R. V. Southwell for the prediction ofcolumn buckling loads from measurements of the rateof change of column lateral deflection with loadwere equally applicable to aeroelastic studies suchas aileron reversal and static divergence. By the

STATIC RESPONSE DATA

WING IKqn

\3 qzBENDING qlMOMENT I. I

lal

BDI ,

7BI ,

76I I

74I I

72I I

70, I

68I I

66, ,64I

1960 621 I I I

LOCKHEED ELECT.----"PYC·141 Yn't' ". 't' '"' 't'f'lll --------'l".~".'"'".I'-".Y_-...".'"'".~WIY_".".'-_.".'----- .........~.LOCKHEED HELICOPTER----..BELL HELICOPTER--------".......'"C-5A-----------'III,"',,,.,,,.,-,,.Y_----....BOEING SST-----------' --".L...C........- ••LOCKHEED SST--------'"747 ". '"L-lOlI ". W

F-14 YWI' '"OC·ID ". '"S·JA .....XV·15 TILT ROTOR ". "..F-15 _

B·I ".,. "..".".LIGHT IVT. fiGHTER IGD I ".". '"B-52CCV ... ....C-5A ACTI VE LOAD CONTROL '"f'16 --------------------_."."l".,."'""'''''-'I".--~'''

SPACE SHUTIl£ =============::::::".~=::::::".~==~".="'"''"GULfSTREAM ill .,YF-I7 STORES fLUDER SUPPR. --------------...__- ••'-"1''''Yf'16 STORES fLUDER SUPPR. --------------'-'I".--~"'..-_'t'B-767-----------------------_'"HELICOPTER ROTORS--------------------__'t''t'

DIVERGENCEINDEX.

qDIV 11

lei L.:I.:::b)--l..._.l.---L_..L-----l---..:~___l

I.iq DYNAMIC PRESSURE. q

ANGlE OF ATTACK

SOUTHWELL METHOD DIVERGENCE INDEX METHOD

SLOPE.I.

Figure 9. Subcritical divergenceprediction methods.

Southwell method the gradient A is plottedagainst A/q (see Fig. 9c). The projected diver­gence dynamic pressure is simply the slope of aleast squares straight-line fit of the measureddata points.

These subcritical divergence prediction methods,developed using inexpensive flat-plate models, werethen employed with a high degree of success intransonic tests of aeroelastically tailoredforward-swept wing models as will be discussedlater in the paper.

IV. VEHICLE DEVELOPMENT TESTING

On the average about one-t~ird of all testsscheduled in the TDT ar~ ~~ s~?,ort aerospacevehicle develop:r.ent .,LO ;ra. ,s. ·.~l;ese stun.iesinvolve dynamic, aeLoe~asticallv scalei ~oiels o~

vehicle pro to types a:u.l aLe COi1'~UCted alo:1~~ ,:.fi t'1analyses to aS8ttLC t~;at n;~p :i~~ifl';:'.;' l·!:i-ll '1::J:J~ a~e"

quate margins of safety from flutter and otheraeroelastic problems, particularly in the impor­tant transonic speed range. As indicated in Fig­ure 10 such studies have supported most of themajor high-performance aircraft· developed in theU.S. since 1960. The TDT is utilized also inlaunch vehicle aeroelastic investigations such asbuffet at transonic Mach numbers and ground windload effects on the vehicle while erected on thepad prior to launch. (3,4)

Figure 10. Aircraft flutter investigations.

C-141 Jet Cargo TransportAeroelastic model investigations in TDT of the

C-141 involved three models in nine separate wind­tunnel entries over a period extending from 1962through 1968. During early phases of the programthe major focus of these studies was on T-tailflutter. To permit better simulation of struc­tural details believed to be important in T-tailflutter a relatively large model of the aft fuse­lage and T-tail was used. A view of this ratherunusual model suspended in the TDT test section isshmm in Figure 11.

Figure 11. Models used in C-141 flutterclearance program.

To illustrate how aeroelastic model studies inthe TDT have been used to support development offlight vehicles, two examples selected from thoseshown in Figure 10 will be discussed briefly inthis section. These vehicles are the C-141, amilitary jet transport. developed in the earlysixties, and the space shuttle, a hybrid airplane/space launch vehicle whose recent highly successfulfirst flight represented the culmination of over10 years of extensive research and development.

Because the lower order vibration modes of thewing and fuselage interact with the T-tail fluttermode, the structural dynamics of the forepart ofthe fuselage and wing were scaled using beamswhich simulate the mass and stiffness but not theaerodynamic surfaces. Also, free rigid-body modeswere simulated by supporting the model at its e.g.by means of flexible rods and cables as shown inthe figure.

The other two C-141 models were a completecable-mounted flutter model shown in Figure 11 anda so-called "dummy" flut ter model ,,,hich was simi­lar in size and weight to the flutter model butwhich was much more rigid to avoid the risk of

7

flu t ter. The use of such dummy models is highlydesirable in flutter studies involving model sus­pension systems designed to simulate free-flightconditions.

Construction of the C-14l flutter model was thepod and spar method typically used in high-aspect­ratio flutter models. The stiffness is providedby the spar and removable pods provide the aero­dynamic surfaces. Removable weights within thewing allowed different fuel loadings to besimulated.

In addition to satisfying the primary objectiveof the test which was to demonstrate the requiredflutter margin (132% of maximum dynamic pressure)at Mach numbers up to the airplane maximum divevalue the flutter model also provided usefulbyproduct information such as, mentioned earlier,the development of new testing techniques and dataon aileron effectiveness and roll rate.

Space ShuttleSpace shuttle, the product of a marriage of

airplane and launch vehicle mated in a configura­tion never before tried, stimulated intensiveinterest and study in many areas of aeroelasticity.Early exploratory wind-tunnel experiments usingsimple models were made to assess the likelihoodof the shuttle's encountering heretofore unimpor­tant aeroelastic phenomena. Included, for example,

were studies by Reed(2s) and Hess(26) to determinewhether the shuttle, having a noncircular crosssection and large-area surfaces normal to windflow while on the launch pad, might encounter cer­tain wind induced aeroelastic instabilities suchas "galloping"--a lateral instability of the wholevehicle akin to the galloping phenomena of icedtransmission lines--or "stop sign" flutter--alongitudinal torsional instability of the wholevehicle associated with separated flow and akin tostall flutter of wings. Other early studies, sum­marized by Runyan and Reed, (27) investigatedtransonic wing buffet characteristics associatedwith reentry at angles of attack approaching90 degrees as well as the transonic flutter ofwings in close proximity.

Once the shuttle configuration had essentiallygelled, a series of aeroelastic model investiga­tions in the TDT were undertaken. Starting in 1972,with ground wind load tests of an early configu­ration and flutter tests of semispan orbiter wingmodel, the series was successfully concluded in1979 with flutter/buffet-model tests of the com­plete space shuttle vehicle which demonstrated therequired flutter margin over the Mach number range0.6 to 1.15. Shown in Figure 12 is the family ofaeroelastic models used in these studies whichcover ground wind loads, flutter, buffet, and as aspinoff mentioned earlier, stability derivativemeasurements on the orbiter.

Figure 12. Space shuttle aeroelastic model studies in TDT.

8

V. AEROELASTIC "FIXES"

DES IGN BEND I NG MOMENT1.0 ,..-L----'------'-----'L--'--~----'--.L.---L-.L..iIo__L__

RESULTANT LOADING.!Figure 14. F-14 aeroelastic model tests

revealed flutter problem on over­."ing fairing.

F-14 AEROELASTIC MODEL TESTS'REVEAL FLUTTER PROBLEMON OVER-WING" FAIRINGS

Another example in which problems were firstuncovered and later solved by means of fluttermodel tests relates to flutter of wings withexternally mounted stores. An extensive series ofwind-tunnel tests of a one-quarter-scale F-16flutter model was performed to define the fluttercharacteristics associated with various externalstore configurations the airplane must carry.From the large number of store configurationstested a few were found to be flutter criticalwithin the design operating envelope. As reportedby Foughner and Bensinger}28) satisfactory solu­tions were developed, evaluated in the wind tunnel,and eventually implemented on the airplane. Oneof these flutter-critical store configurationsincluded air-to-air missiles mounted on launchersat each wing tip and on pylons at the outboardunder-."ing station. Flutter occurred at a Nachnumber of about 1.1 near the required flutter mar­gin of safety boundary. On the basis of analyticalpredic tions three possible "fixes" ."ere evaluatedon the model: moving the under-wing missiles for­ward, stiffening the missile launchers, and addingballast weight to the missile launchers. Of thesethe latter solution was chosen for implementationon the F-16. The addition of 4.7 pounds of bal­last weight in the wing tip launchers completelyeliminated the flutter tendencies of this storeconfiguration.

FlutterThe effects of high angle of attack on flutter

and buffet loads of the F-14 were investigatedusing the cable-mounted flutter model. Duringthese tests, which preceded the first flight ofthe prototype, it was discovered that the flowover the "over-.,ling" 'fairings caused the fairings,which were essentially cantilevered from a pointnear the swing-wing hinge line, to deform andoscillate in a "hand-clapping" manner. Severalpotential fixes were evaluated. Of these a stiff­ening "strake" was proven to be effective on themodel and ."as later incorporated in the airplanedesign. Figure 14 shows the F-14 flutter model(without stiffeners) and the airplane with thestiffeners installed.

w ~ ~ ® ~ ~

SIMULATED WIND VELOCITY. knots

-o-c/cc = 0.0191 DYNAMIC

-0- clc = 0028 LOAD Sc .--- STEADY LOADS

VEH ICLEBASE

BEND INGMOMENT

If aeroelastic-type problems are uncovered dur­ing the course of a vehicle development program,model studies in the TDT often playa key role infinding and evaluating effective "fixes." In thissection some examples are given in which suchfixes suggested by model tests were subsequentlyimplemented on the full-scale counterparts.

Ground Wind-Induced Oscillationsn"o of these examples relate to the hazards of

wind-induced oscillations of erected structures atthe launch pad. In ground-wind-load studies usinga O.03-scale model of the Saturn V-Apollo andumbilical tower, it was found that certain combi­nations of wind velocity (about 57 knots) anddirection caused near sinusoidal oscillations thatwere severe enough to exceed the structural designlimits of the vehicle. Various possible solutionsto the problem were investigated such as the addi­tion of aerodynamic spoilers or damping devices tothe structure (Farmer and Jones(4)). Of thesedamping was selected as being the most feasiblesolution. Figure 13 shows the effectiveness ofdamping on reducing dynamic wind-induced loads toacceptably safe levels. This solution was thenimplemented on the vehicle in the form of aviscous damper strut connecting the Apollo capsuleatop the vehicle to the adjacent umbilical towerstructure. The damper remained attached to thevehicle until just before lift-off.

Figure 13. Wind-tunnel predicted response ofSaturn V to ground-wind loads.

The other example of a wind-induced oscillationproblem concerns the Titan III umbilical mast, aISO-foot "A" frame structure used in transportingthe erected vehicle to the launch pad. Hurricanewinds at Cape Kennedy induced violent oscillationsof the mast to the extent that they rocked thebase, damaged the supporting foundation piers, andthreatened the main structure. The wind-tunnelinvestigation, using a 7-1/2 percent dynamicallyscaled model of the mast and transporter, revealedthat by altering the upper quarter of the modelmast from a closed to an open-grid type structurethe oscillations were greatly reduced. On thebasis of these findings, the Air Force implementedthe solution on the actual mast at Cape Kennedywhich significantly extended the range of wind con­ditions under which Titan III ground-handlingoperations can be conducted safely.

9

Another wing-store flutter problem identifiedin the F-16 model studies could be correctedsimply by revising the fuel usage sequence ofexternally mounted fuel tanks. These tanks arecompartmented into separate forward, mid, and aftsections from which fuel can be dra,vn. The origi­nal fuel sequence, which was found to be fluttercritical, called for first emptying the forwardand aft compartments and then the center compart­ment. As shown in Figure 15, by merely reversingthe order of fuel usage from the tank, i.e., byemptying the center compartment first, the flutterproblem was eliminated .

...,1...........,15""'.~ Jf <:: I ~

!",."'""" c-t", RumA

FLUTIER BOUNDARY~<::::_-'I"-':'-"-T-_~

EMPTY FULL EMPTY

figure 16. Lockheed Electra propellerwhirl flutter model.

of safety from flutter. The engine mount systemswere redesigned to provide "fail-safe" redundan­cies such that the failure of anyone component inthe mount system would not cause flutter. Whirlflutter avoidance has now also become a designconsideration for prop-rotor V/STOL aircraft (seeKva ternik and Kohn (33)) and modern ,~ind turbinegenerators.

7(. "

" "'.,

.8

MACH

.4o

1.0

Figure 15. External tank fuel usage sequencechange increases flutter speed of F-16(Ref. 28).

VI. AEROELASTIC "SURPRISES"

During the development or early service life ofnew vehicle configurations aeroelastic "surprises"are sometimes encountered which, to understand andaccount for in future designs, warrant extensivewind-tunnel testing and analysis. Three such sur­prises encountered on new aircraft designs will bediscussed in this section.

Propeller Whirl FlutterSoon after coming into service two Lockheed

Electra turbopropeller-driven transports were lostin mysterious accidents. The suspected cause ofthese accidents was a new form of aeroelasticinstability that had been discovered analyticallyin the late thirties by Taylor and Brmvne(29) in astudy of vibration isolation of aircraft enginesbut was found to be of no significance for air­planes of that time. Later to become known aspropeller whirl flutter, the instability involvesa coupling of the aerodynamic and gyroscopic forcesof the propeller with the stiffness and inertiaforces of the mount resulting in a precession-typemotion of the propeller. A wind-tunnel modelinvestigation was urgently carried out in the newlycommissioned TDT using a complete flutter model ofthe airplane. Figure 16 shows the model in the TDTtest section mounted on a rod suspension system tosimulate free flight. Results from the model testsby Abbott, Kelly, and Hampton(30) and analyses by

(31) (32)Reed and Bland and by Houbolt and Reedindicated that propeller whirl flutter could occurat high forward speeds but only if the power plantsupport stiffness is severely reduced due to someform of damage to its mount structure. In anundamaged condition the airplane had ample margin

Nonlinear Aerodynamic EffectsAlthough conventional linear aerodynamic theory

predicts flutter to be independent of the steady­state aerodynamic loads, wind-tunnel and flighttests have sometimes shown otherwise. Duringdevelopment testing of two different prototypeaircraft, flutter-type instabilities were observedin severe maneuvers whereas in earlier flight­flutter clearance tests at similar Mach numbersand altitudes but in I-g flight, the flutter modeswere well damped. In both instances the insta­bility phenomena encountered in flight were lateressentially duplicated in the TDT using fluttermodels of the prototype aircraft.

T-Tail Deflected-Elevator Flutter. The firstincident involved T-tail empennage flutter of alarge cargo transport airplane, the C-141. Theinstability occurred during high-altitude testsat a Mach number near 0.8 but only in maneuverswhen the elevator was deflected more than 8 degreesin either direction. The instability was charac­terized by limited amplitude oscillations involv­ing coupling between elevator rotation and stabi­lizer torsion. Subsequent flight investigationsof various proposed solutions, including vortexgenerators, dampers, and elevator mass balance,led to the selection of mass balance.

An experimental study using the existing C-141high-speed flutter model was undertaken in theTDT. Results from this study by Sandford andRuhlin(34) are summarized in Figure 17. It wasfound that the basic instability phenomenonencountered on the airplane in flight was repro­duced in the wind tunnel, although at higherspeeds, and the elevator mass-balance solution forthe airplane also eliminated flutter on the model.

10

WING SWEEP fi7 DEli.\1 (,34 0"'10 ;lUi.

==::==-:~--s:::::~:\..~:~:'~ -3=:1' '- -\ ---'':'!\If \"'DEf:RfASE a

fUll SeAL[. nEO . .:. 2.1 HI

W1NG SWEEP 55 DEG.M= 0.91 o=.a OEG.

--=--->TIME

SCALED FHEQ.:= 2.2 H7

WING'1 P

ACCEL. I

c:::~---=:Z+ °e

500

400 AIRPLANE FLUTTER

IOel >80

VKEAS .300 FLI GHT ENVELOPE

knots200

6 x 103 m

100 \(9 x 103

m14 x 103

m

a .4 .6 .8 1.0MACH NUMBER

Figure 17. Comparison of flight measurements andmodel-predicted flutter of T-tail withdeflected elevator (Ref. 34).

Shock-Induced Flutter. The other incidentinvolving self-excited oscillations due to maneu­vers occurred during high-altitude flight-loaddemonstration tests of the B-1 bomber. In thisinstance the instability appeared as a limitedamplitude bending oscillation of the outer-wingpanels. The oscillations occurred near criticalMach number conditions for the airfoil and only athigh positive angles of attack. A qualitativeexplanation for these so-called "shock-inducedself-excited bending oscillations" is given byStevenson. (35) Some simplified calculations by

Ashley(36) relate the instability to chordwiseshock movement with angle of attack and importantphase lags known to be present in the shock oscil­lation. This instability does not represent alimitation to the B-1 as it occurs outside therange of normal flight operations.

To study the phenomena further the B-1 fluttermodel, which was shown to be free of flutter prob­lems in earlier tests under simulated I-g condi­tions, was tested again in the TDT. As with theC-141 model/flight correlation studies describedpreviously, attempts were made to simulate theload factor and flight conditions for which theB-1 encountered shock-induced oscillations.Results from both the flight and wind-tunnel testsare shown in Figure 18. Again, the ins tabili typhenomenon encountered in flight was demonstratedin the tunnel, although at slightly different con­ditions than in flight.

VII. CONTROLLING AEROELASTIC EFFECTS

Aircraft design options for controlling aero­elastic effects can be broadened significantlythrough application of advanced concepts such asactive control technology and aeroelasticallytailored composite structure. Potential aero­elastic benefits made possible by these advancedtechnologies are being evaluated experimentally inthe TDT using dynamic-elastic models. Three suchresearch investigations have been selected for dis­cussion in this section of the paper. Thesestudies concern flutter suppression of wings withexternal stores, helicopter vibration control, anddivergence of forward swept wings.

Figure 18. B-1 shock-induced instabilitystudied in TDT.

Wing/Store FlutterHigh-speed strike aircraft are required to

carry an ever-increasing number of wing-mountedexternal stores such as fuel tanks and armament.Out of the many combinations of store loadingspossible some inevitably cause reduction in theairplane's flutter speed and as a consequencerestrict its operating envelope. Two promisingconcepts for alleviating wing/store flutter prob­lems have been demonstrated in recent investiga­tions in the TDT using the models shown in Fig­ure 19. The first involves active controls tech­nology wherein the onset of flutter is sensed byaccelerometers on the wing which are used, bymeans of computer-implemented feedback controllaws, to activate control surfaces that produceaerodynamic forces to oppose flutter (seeRefs. 37 and 38, for example). The second fluttersuppression method is a basically passive conceptcalled the decoupler pylon which dynamicallydecouples the store from the wing by means of aself-aligning store suspension system that is softin pitch. (39)

/ :~!liCtANGULAR3!1ijFigure 19. Wing/store flutter suppression

studies in the Langley TDT.

11

Helicopter Vibration Control

LATERALACCELERATION. A

9 .3

.2

.1

.2 .3 AADVANCE RATIO

~.~;:~;~~;~,,;lJ9 .2

.1

D~.7

.6

.5

':.423Il~LONGITUDINALACCELERATION,

9

o .20 .30 AOADVANCE RATIO

Wind-tunnel model investigations of this con­cept have been reported by Hammond(41) and Molusis,Hammond, and Cline. (42) The model, shm"n in Fig­ure 21, is known as the Aeroelastic Rotor Experi­mental System (ARES) and is an outgrowth of amodel used in earlier rotor aeroelastic investiga­tions in the TDT. (9) The rotor control systemconsists of a conventional swash plate that isremotely driven by three electro-hydraulic servoactuators to provide the necessary high-frequencyresponse characteristics. The closed-loop higherharmonic control system makes use of digitaloptimal control theory, the controlled quantitiesbeing either the vibratory forces and moments asmeasured by a strain-gage balance on which therotor was mounted(41) or vibratory accelera-tions. (42)

Figure 21. Test results of closed-loopadaptive controller for helicoptervibration reduction (Ref. 42).

repair of helicopter components. Oscillatoryloads transferred from the rotor to the airframeare the primary contributors to airframe vibra­tions. By means of active feedback control, thevibratory loads are reduced at their source--therotor--in contrast to some conventional passivemeans of vibration control which are designed toisolate the airframe from the vibration source.In this concept, known as higher harmonic control,the rotor blade pitch angle is commanded to oscil­late at the blade passage frequency (e.g., 4 perrevolution for a four-bladed rotor) with appro­priate amplitude and phase so that the transmittedloads are reduced to a minimum. By use of anadaptive automatic control system and optimum con­trol theory the blade pitch control inputs arecontinuously updated to provide "optimum" vibra­tion reduction under changing flight conditions.

Also shown in Figure 21 are data indicating theeffectiveness of the higher harmonic control inreducing the vibration response for variousadvance ratios. It can be seen from these datathat the system is highly effective in reducingthe vibration levels in the vertical and hori­zontal directions over the entire range of advanceratios tested. The lateral vibration results aremixed, the controller being effective at thehigher advance ratios but causing increasedresponse at lower advance ratios. These reduc­tions in vibration level were accompanied by

.96 .7 .8~\ACH NU~\B[R

.5

Figure 20. Damping trends of F-16model (Ref. 40).

JDEC~PLERPYLON

.~.---

~G,-__SNOMI~:L~ESIG~-G .5 1.0 1.5 2.0 2.5

RELATIVE DYNAM IC PRESSURE. Q/Qnom

These active and passive flutter suppressionconcepts have both been investigated throughout thetransonic speed range using aeroelastic models ofthe YF-17 and the F-16. Some typical flutter-modedamping trends measured on the F-16 model withstores are shown in Figure 20. Damping measuredfor an active flutter suppression system, thedecoupler pylon, and, for comparison purposes, thenominal design are plotted as a function of dynamicpressure and Mach number, holding the tunnel pres­sure altitude constant. For the active system theflaperon control surfaces on each wing were actu­ated by signals which had been suitably filtered,mixed and fed back from accelerometers mounted oneach wing panel at the location indicated by thesketch in Figure 20. For the decoupler-pylon caseonly one of the three stores on each wing was somounted, tha t being a GBU-8 guided bomb ,,,hich waslocated between the tip-mounted missile and theinboard-mounted fuel tank. The figure indicatesthat for both systems the flutter mode was welldamped and there were no signs of impending flutterup to dynamic pressure nearly 100% greater thanthat for which the nominal design encounteredflutter. Also, studies by Reed, Foughner, andRunyan(40) have shown that in addition to increas­ing the flutter speed, the decoupler pylon makesflutter relatively insensitive to store center ofgravity and inertia changes.

DM1PING

The feasibility of using the decoupler pylon onthe F-16 airplane has been studied under a NASAcontract by General Dynamics, Fort Worth, in whichfactors other than flutter, such as maneuver loads,store ejection, gust response, etc., were con­sidered. The study indicated the concept to befeasible for flight applications and a program hasbeen initiated to evaluate the decoupler pylon onan F-16.

Another application of active control technologyunder study in the TDT concerns the reduction ofhelicopter vibration through rotor loads control.Most present-day helicopters experience excessivevibrations in some regions of operation. Thesevibrations lead to pilot fatigue, passenger annoy­ance, and the need for frequent maintenance and

12

increased blade and pitch link loads but they weresmall relative to design limits for the model.These tests are believed to be the first time useof an adaptive control system employing optimalcontrol theory for such purpose. Preparations areunderway to demonstrate the system in flight on anOH-6A helicopter.

Divergence of Forward-Swept WingsAs with active controls, the use of aeroelasti­

cally tailored composite structure opens up newdesign options for controlling aeroelastic response.In this approach the orientation of the fibers infibrous composite materials in wing skins becomesa design variable by means of which aeroelasticitymay be put to beneficial uses. A noteworthyexample is the recent rekindling of interest inforward-swept wings made possible by the applica­tion of aeroelastically tailored composites. Bysuch means Krone(4j) showed in analytical studiesthat the low static divergence speed of forward­swept wings, a problem which had originally led torejection of the concept, could now be avoided withrelatively small structural weight penalties.

This rebirth of interest in forward-swept wingsbrought about by aeroelastic tailoring technologyhas led to considerable research effort by the AirForce Flight Dynamics Laboratory, the DefenseAdvanced Research Projects Office, and NASA tobetter understand the aeroelastic characteristicsof forward-swept wings. Aeroelastically tailoredmodels of forward-swept wing designs by GrummanAerospace Corp. and Rockwell International weretested in the TDT to provide correlation withanalysis and to gain confidence that divergence canbe efficiently avoided within the design flightenvelope. Results from these tests together witha discussion of the evolution and background ofaeroelastic tailoring as applied to forward-sweptwings have been Eresented by Hertz, Shirk, Ricketts,and Weisshaar.(4) Photographs of the O.S-scaleGrumman model and the O.6-scale Rockwell modelmounted in the \~ind tunnel are shmm in Figure 22.Also shown in the figure are the experimentaldivergence speed boundaries (projected from testdata at speeds below the divergence speed asdescribed in Reference 23) and calculated diver­gence boundaries. Thus far divergence has beenthe major focus of aeroelastic studies of forward­swept wings. Other aeroelastic characteristicssuch as buffet, flutter, and gust response, willno doubt become the subject of future studies forthis class of aircraft.

VIII. UNSTEADY PRESSURE MEASUREMENTS

Up to this point, the paper has dealt with theuse of scaled flexible models to study variousaeroelastic stability and response problems. Thesemodels play the role of mechanical analogs of thefull-scale structure. When mounted in a wind tun­nel that properly simulates the flow field, suchmodels perform the difficult time and space inte­grations of the aerodynamic loads to produceresponse characteristics that would be expected onthe full-scale article. Whereas tests of this kindshow the net effects of the aerodynamic flow fieldaround the model, they give little insight into thephysics of the flow itself. Because unsteady aero­dynamic forces, particularly in the transonic range,are probably the weakest link in the chain of tech­nologies used in aeroelastic design and analysis,

13

~.l ''''''00

.ANALYSIS / /-......, IoT!< ..,..... 160

'".........., ANALYSIS /suacamCAl- 120

"""'IOU' eo EXl'U:IMlNTSU8QlTICAl

""n(.H~Uf

0- ~- ...---,----, 00 .2 .4 .6 ..8 1.0 1.2 0 ., .. .. .. 1.0 "...01 NUMlISl MA9!,~U~R

Figure 22. Divergence tests in TDT ofaeroelastically tailored forward­swep t wings (Ref. 44 ) .

there is much need for experimental data on surfacepressures and flow field measurements on bodies andlifting surfaces. Such measurements are needed togain better understanding of the physics of aero­dynamiC flow fields about realistic configurationsat transonic speeds, and to obtain a data base foruse in the validation of aerodynamic theories andin the design of active controls.

Until recently most of the unsteady pressuremeasurement work had been performed in Europeanwind tunnels (notably by the NLR in TheNetherlands, ONERA in France, RAE in England, andDFVLR in Germany), whereas in the United States,the emphasis was on large aeroelastic models. Atpresent, there is a sizable program underway andplanned for measuring steady and oscillating pres­sure distributions on a variety of lifting sur­faces at Langley in the TDT. In a recent surveypaper on unsteady pressure measurement programsconducted in European wind tunnels, 01sen(4S)observes that this new interest in unsteady pres­sure measurements in the U.S. appears to be matchedby a growth of interest in flutter model testingin Europe.

The planforms presently under investigation inthe TDT include a clipped delta-wing model ofinterest for supersonic transport and fighterapplications and a high-aspect-ratio (10.7) sweptwing of interest for energy efficient transportswith active controls. Both models are equipped\~ith active leading- and trailing-edge active con­trol surfaces and a large number of static orificesand dynamic pressure transducers on their upperand lower surfaces.

The test program involving the high-aspect-ratiomodel is described, and some selected results frominitial wind-tunnel tests ara presented bySandford, Ricketts, Cazier, and Cunningham. (46)This model has five leading-edge and five trailing­edge control surfaces which can be oscillatedindividually and in various combinations. It isinstrumented with 2S2 static orifices and 164 insitu dynamic pressure transducers. Figure 23 showsa photograph of the model mounted in the TDT andsome unsteady pressure measurements associated withthe oscillation of an outboard leading-edge and atrailing-edge control surface. The data are for

200fPHAS£IOOL~ANGLE

o •

100­

PHASEO~ANGLE IW

-100

operational more than two decades ago. The many­faceted aspects of experimental aeroelasticitywere illustrated by way of examples selected froma broad base of testing experience in the TDT.The coming decade should see continued productionuse of this unique National aerospace facility.

X. ACKNOI.JLEDGEMENTS

The author is indebted to Mr. Andrew D. Careyof the Photographic Branch and Mr. John H. Clineof the Configuration Aeroelasticity Branch fortheir untiring efforts devoted to the preparationof film clips used in the oral version of thispaper.

XI. REFERENCES

IX. CONCLUDING REMARKS

This paper has documented that aeroelasticitymatters by presenting a cross section of aero­elastic research and testing performed in the NASALangley Transonic Dynamics Tunnel since it became

Mach number 0.6 and a reduced frequency, based onroot semichord, of k = 0.21. The calculatedresults,which show good correlation with experi­ment everywhere except in the regions of the con­trol surface hinge, were obtained using a subsonickernel function method (RHO IV) for wings withoscillating controls. (47)

3. Hanson, P. W., and Jones, G. W., Jr.: On theUse of Dynamic Models for Studying LaunchVehicle Buffet and Ground Wind Loads. Pro­ceedings of Symposium on Aeroelastic andDynamic Modeling Technology, U.S. Air ForceSystems Command, RTD-TDR-63-4197, Part 1,1964.

4. Farmer, M. G., and Jones, G. W., Jr.: Summaryof Langley Wind Tunnel Studies of Ground­Wind Loads on Launch Vehicles. Compilationof Papers Presented at the NASA LRC Meetingon Ground Wind Load Problems in Relation toLaunch Vehicles, TMX-57779, June 7-8, 1966.

8. Reed, W. H. III: Comparisons of FlightMeasurements with Predictions from Aero­elastic Models in the NASA Langley TransonicDynamics Tunnel. AGARD Conference Proceed­ing No. 187 on Flight/Ground Testing Facili­ties Correlation, Valloire Savoie, France,1975.

7. Rainey, A. G., and Abel, I.: Wind TunnelTechniques for the Study of AeroelasticEffects on Aircraft Stability, Control, andLoads. AGARD Proceedings No. 46 - Aero­elastic Effects from a Flight MechanicsStandpoint, Marseilles, France, April 21-24,1969.

6. Abbott, F. T.: Some Current Techniques inExperimental Aeroelasticity. Proceedings ofASME 1967 \Yinter Annual Meeting Symposium onSolid-Fluid Interaction Problems inMechanics, Nov. 12-16, 1967.

5. Guyett, P. R.: The Use of Flexible Models inAerospace Engineering. Royal AircraftEstablishment Technical Report No. 66335,Nov. 1966.

1. Garrick, I. E., and Reed, W. H. III: Histori­cal Development of Aircraft Flutter. AIAAJour. of Aircraft, Vol. 18, No. 11,November 1981. Also available as AIAApaper no. 81-0591-CP.

2. Regier, A. A.: The Use of Scaled DynamicModels in Several Aerospace Vehicle Studies.Proceedings of ASHE Winter Annual Meeting,Symposium on Use of Models and Scaling inShock and Vibration. Philadelphia, Pa.Edited by Welfred and Baker, published byASME, N.Y., 1963.

=4

.80c.15c

6CpL~~o 10

fRACTION Of CHORD

.sOc.15t

6Cpl~Lo fRACTION Of CHORD 0 "\0

Figure 23. Unsteady pressure measurements onmodel with oscillating control surfaces.

Finally, mention should be made here of a coop­erative unsteady pressure measurement projectbetween Lockheed, the U.S. Air Force, NLR (TheNetherlands), and NASA referred to as the "LANN"wing. This wing has a supercritical airfoil sec­tion and is highly instrumented for measuring sur­face pressures. The wing is scheduled for testingat NLR in late 1981 and in the Langley NationalTransonic Facility, at flight Reynolds numbers, in1983. The data will be useful for examiningReynolds number effects on steady and unsteadyaerodynamics of supercritical wings. These testswill require significant advances in the state ofthe art of unsteady pressure measurements due tothe high dynamic pressures and the extremely lowtemperatures required to achieve high Reynoldsnumbers.

Other experimental studies underway at Langleyinclude unsteady pressure measurements on a two­dimensional dynamic-stall model that can be oscil­lated at large pitch amplitudes in order to simu­late the dynamic-stall response of helicopterblades. Also, steady and unsteady pressures are tobe measured for a pitching rectangular-wing model.This rectangular wing is especially suited for usein validating 3-D transonic unsteady aerodynamictheories which are currently limited to simple wingplanforms.

14

9. Hammond, C. E., and I.]eller, I.]. H.: Hind-TunnelTesting of Aeroelastically Scaled HelicopterRotor Models. Presented at the 1976 ArmyScience Conference, Hest Point, New York,June 22-25, 1976.

10. Cole, P. H.: Wind Tunnel Real-Time DataAcquisition System. NASA TM-80081, April1979.

11. Hanson, P. W.: Evaluation of an AeroelasticModel Technique for Predicting AirplaneBuffet Loads. NASA TN D-7066, 1973.

12. Reed, H. H. III, and Abbott, F. T., Jr.: ANew Free-Flight Mount System for High-Speed\.]ind-Tunnel Flutter Models. Proceedings ofSymposium on Aeroelastic and Dynamic ModelingTechnology. RTD-TDR-63-4197, Pt. I, U.S. AirForce, Mar. 1964, pp. 169-206.

13. Gilman, J., Jr., and Bennett, R. M.: A IhndTunnel Technique for Measuring Frequency­Response Functions for Gust Load Analyses.J. Aircraft, Vol. 3, No.6, Nov.-Dec. 1966,pp. 535-540.

14. Barbero, P., and Chin, J.: Users Guide for aComputer Program to Analyze the LRC 16'Transonic Dynamics Tunnel Cable-Mount System.NASA CR-132313, 1973.

15. Hanson, P. W.: The Prediction of StructuralResponse to Buffet Flow: A State-of-the-ArtReview. NASA TM X-72627, 1974. Also seeProceedings of the 39th AGARD Structural andMaterials Meeting, Munich, Germany,Oct. 7-12, 1974.

16. Houbolt, J. C., Steiner, R., and Pratt, K. G.:Dynamic Response of Airplanes to AtmosphericTurbulence Including Flight Data on Input andResponse. NASA TR R-199, 1964.

17. Redd, L. T., Hanson, P. H., and Hynne, E. C.:Evaluation of a Wind-Tunnel Gust ResponseTechnique Including Correlations withAnalytical and Flight Tests. NASA TP-1501,Nov. 1979.

18. Abel, I.: Evaluation of a Technique for Deter­mining Airplane Aileron Effectiveness andRoll Rate by Using an Aeroelastically ScaledModel. NASA TN D-5538, Nov. 1969.

22. Many Authors, Flutter Testing Techniques,NASA SP-415, 1976.

23. Ricketts, R. H., and Doggett, R. V., Jr.:Hind Tunnel Experiments on Divergence ofForward-Swept Ihngs. NASA TP-1685, Aug.1980.

24. Flax, A. H.: The Influence of StructuralDeformation on Airplane Characteristics.Jour. of the Aero. Sci., Vol. 12, Jan 1945,pp. 94-102.

25. Reed, W. H. III: Hind Effects on SpaceShuttle Vehicles Erected on Launch Pad.Proceedings of the Third International Con­ference on Wind Effects on Buildings andStructures, Japanese Organizing Comm., eds.,Saiko Co., Ltd. (Tokyo), 1971.,pp. 1127-1140.

26. Hess, R. W.: Hind-Tunnel Roll-Damping Measure­ments of a Winged Space Shuttle Configurationin Launch Attitude. NASA TN D-7394, Dec.1973.

27. Runyan, H. L., and Reed, W. H. III: ShuttleDynamics and Aeroelasticity--An Appraisal.AlAA Astronautics and Aeronautics, Vol. 9,No.2, Feb. 1971, pp. 48-57.

28. Foughner, J. T., Jr., and Bensinger, C. T.:F-16 Flutter Model Studies with External HingStore. NASA TM-74078, Oct. 1977.

29. Taylor, E. S., and Bro\voe, K. A.: VibrationIsolation of Aircraft Power Plants. Jour.Aero. Sci., Vol. 6, Dec. 1938, pp. 43-49.

30. Abbott, F. T., Jr., Kelly, H. N., and Hampton,K. D.: Investigation of Propeller-Power­Plant Auto-Precession Boundaries for aDynamic-Aeroelastic Model of a Four-EngineTurboprop Transport Airplane. NASA TND-1806, June 1963.

31. Reed, W. H. III, and Bland, S. R.: An Analyti­cal Treatment of Aircraft Propeller Preces­sion Instability. NASA TN D-659, Jan. 1961.

32. Houbolt, J. C., and Reed, H. H. III:Propeller-Nacelle Whirl Flutter. Jour.Aerospace Sci., Vol. 29, March 1962,pp. 333-347.

33. Kvaternik, R. G., and Kohn, J. S.: An Experi­mental and Analytical Investigation ofProprotor Whirl Flutter. NASA TP-I047,Dec. 1977.

35. Stevenson, J. R,: Shock-Induced Self-ExcitedAirfoil Bending Oscillations. Los AngelesDiv., Rockwell International, paper pre­sented to Aerospace Flutter and DynamicsCouncil, Las Vegas, Nev., Oct. 19-20, 1978.

19. Grosser, H. F.: A Transonic Speed Wind TunnelInvestigation of the Rolling Effectiveness ofa Large Swept-Wing Transport Aircraft withConventional-Type Ailerons and VariousSpoiler Configurations, AlAA paper No. 65-789,Nov. 1965.

20. Bennett, R. M., Farmer, M. G., Mohr, R. L., andHall, W. E., Jr.: Wind Tunnel Technique forDetermining Stability Derivatives from CableMounted Models. J. Aircraft, Vol. 15, No.5,May 1978, pp. 304-310.

21. Chin, J., and Barbero, P.: User's Guide for aRevised Computer Program to Analyze the LRCTransonic Dynamics Tunnel Active Cable-MountSystem. NASA CR-132692, 1975.

15

34. Sandford, M. C., and Ruhlin, C. L.:Tunnel Study of Deflected-ElevatorEncountered on a T-Tail Airplane.TN D-5024, Feb. 1969.

Wind­FlutterNASA

36. Ashley, H.: Role of Shocks in the "Sub­Transonic" Flutter Phenomenon. Jour. ofAircraft, Vol. 17, No.3, Marc'h 1980,pp. 187-197.

37. Noll, T. E., Huttsell, L. J., and Cooley,D. E.: Investigation of International Con­trol Laws for Wing/Store Flutter Suppression.A Collection of Technical P~pers - AlAA/ASME/ASCE/AHS 21st Structures, StructuralDynamics and Materials Conference, Part 2,May 1980, pp. 585-594. (Available asAlAA-80-0764.)

38. Peloubet, R. P., Jr., Haller, R. L., andBolding, R. M.: F-16 Flutter SuppressionSystem Investigation. A Collection ofTechnical Papers - AlAA/ASME/ASCE/AHS 21stStructures, Structural DynamiCS and MaterialsConference, Part 2, May 1980, pp. 620-634.(Available as AIAA-80-0768.)

39. Reed, 1". H., III, Foughner, Jerome T., Jr.,and Runyan, Harry L., Jr.: Decoupler Pylon:A Simple, Effective Wing/Store Flutter Sup­pressor. J. Aircraft, Vol. 17, No.3,March 1980, pp. 206-211.

40. Reed, W. H. III, Cazier, F. W., Jr., andFoughner, J. T., Jr.: Passive Control ofWing/Store Flutter. NASA TM-81865, Dec.1980.

41. Hammond, C. E.: Wind Tunnel Results ShowingRotor Vibratory Loads Reduction Using HigherHarmonic Blade Pitch. A Collection ofTechnical Papers - AlAA/ASME/ASCE/AHS 21stStructures, Structural Dynamics, andMaterials Conference, Part 2, May 1980.(Available as AIAA 80-0667.)

42. Molusis, J. A., Hammond, C. E., and Cline,J. H.: A Unified Approach to the OptimalDesign of Adaptive and Gain Scheduled Con­trollers to Achieve Minimum Helicopter RotorVibration. Presented at the 37th Annual AHSForum, New Orleans, LA, May 1981. (Availableas AHS preprint 81-20.)

43. Krone, N. J., Jr.: Divergence Elimination withAdvanced Composites. AlAA paper No. 75-1009,Presented at AlAA 1975 Aircraft Systems andTechnology Meeting, Los Angeles, August 1975.

44. Hertz, T. J., Shirk, M. H., Ricketts, R. H.,and Weisshaar, T. A.: Aeroelastic Tailoringwith Composites Applied to Forward SweptWings. Presented at Fifth DOD/NASA Confer­ence on Fibrous Composites in StructuralDesign, New Orleans, LA, Jan. 27-29, 1981.

45. Olsen, J. J.: Unsteady Pressure Measurementson Oscillating Models in European WindTunnels. Tech. Memo. AFDV AL-TM-80-1-FIBR,March 1980.

46. Sandford, M. C., Ricketts, R. H., Cazier,F. W., Jr., and Cunningham, H. J.: Tran­sonic Unsteady Airloads on an Energy Effi­cient Transport \.Jing with Oscil1a'tingControl Surfaces. Journal of Aircraft,Vol. 18, July 1981, pp. 557-561.

16

47. Rowe, W. S., Sebastian, J. D., and Petrarca,J. R.: Reduction of Computer Usage Costs inPredicting Unsteady Aerodynamic LoadingCaused by Control Surface Motions - Analysisand Results. NASA CR-30009, Jan. 1979.