historical overview of flutter

8/2/2019 Historical Overview of Flutter

1/22

NASA Technical Memorandum 4720tV - "J -_

r -.,..3/_ _

A Historical Overview ofFlight Flutter Testing

Michael W. Kehoe

October 1995

(NASA-TN-4?20) A HISTORICALOVEnVIEW OF FLIGHT FLUTTER TESTING(NASA. Oryden Flight ResearchCenter) ZO

N96-14084

Unclas

H1/05 0075823


2/22


3/22

NASA Technical Memorandum 4720

A Historical Overview ofFlight Flutter Testing

Michael W. KehoeDryden Flight Research CenterEdwards, California

National Aeronautics andSpace AdministrationOffice of ManagementScientific and TechnicalInformation Program1995


4/22


5/22

SUMMARY

This paper reviews the test techniques developed overthe last several decades for flight flutter testing of aircraft.Structural excitation systems, instrumentation systems,digital data preprocessing, and parameter identificationalgorithms (for frequency and damping estimates from theresponse data) are described. Practical experiences andexample test programs illustrate the combined, integratedeffectiveness of the various approaches used. Finally, com-ments regarding the direction of future developments andneeds are presented.

INTRODUCTIONAeroelastic flutter involves the unfavorable interaction

of aerodynamic, elastic, and inertia forces on structures toproduce an unstable oscillation that often results in struc-tural failure. High-speed aircraft are most susceptible toflutter although flutter has occurred at speeds of 55 mph onhome-built aircraft. In fact, no speed regime is trulyimmune from flutter.

Aeroelasticity plays a significant role in the design ofaircraft. The introduction of thinner wings, all-movablehorizontal and vertical stabilizers, and T-tail configura-tions increases the likelihood of flutter occurring withinthe desired flight envelope. Today's aircraft designsundergo sophisticated aeroelastic analyses to ensure thatthe design is free of flutter within the flight envelope.These analytical results are often verified by wind-tunnelflutter models and ground vibration tests. Flight fluttertesting provides the final verification of the analytical pre-dictions throughout the flight envelope.

In the early years of aviation, no formal flutter testing offull-scale aircraft was carried out. The aircraft was simplyflown to its maximum speed to demonstrate the aeroelasticstability of the vehicle. The first formal flutter test was car-ried out by Von Schlippe in 1935 in Germany (ref. 1). Hisapproach was to vibrate the aircraft at resonant frequenciesat progressively higher speeds and plot amplitude as afunction of airspeed. A rise in amplitude would suggestreduced damping with flutter occurring at the asymptote oftheoretically infinite amplitude as shown in figure 1. Thisidea was applied successfully to several German aircraftuntil a Junkers JU90 fluttered and crashed during flighttests in 1938.

Early test engineers were faced with inadequate instru-mentation, excitation methods, and stability determinationtechniques. Since then, considerable improvements havebeen made in flight flutter test technique, instrumentation,

Maximumresponseamplltude

m i

iIi0 Vflutte r

Airspeed _c_7_Figure 1. Von Schlippe's flight flutter test method.

and response data analysis. Flutter testing, however, is stilla hazardous test for several reasons. First, one still must flyclose to actual flutter speeds before imminent instabilitiescan be detected. Second, subcritical damping trends can-not be accurately extrapolated to predict stability at higherairspeeds. Third, the aeroelastic stability may changeabruptly from a stable condition to one that is unstablewith only a few knots' change in airspeed.

This paper presents a historical overview of the develop-ment of flight flutter testing, including a history of aircraftflutter incidents. The development of excitation systems,instrumentation systems, and stability determination meth-ods is reviewed as it pertains to flight flutter testing.

FLUTTER HISTORY

The first recorded flutter incident was on a HandleyPage 0/400 twin engine biplane bomber in 1916. The flut-ter mechanism consisted of a coupling of the fuselage tor-sion mode with an antisymmetric elevator rotation mode.The elevators on this airplane were independently actu-ated. The solution to the problem was to interconnect theelevators with a torque tube (ref. 2).Control surface flutter began to appear during World

War I. Wing-aileron flutter was widely encountered duringthis time (ref. 3). Von Baumhauer and Koning suggestedthe use of a mass balance about the control surface hingeline as a means of avoiding this type of flutter. Althoughsome mild instances of control surface flutter wereencountered afterward, these were usually eliminated byincreasing the mass balance of the control surface.

After World War I, higher airspeeds and a shift fromexternal wire-braced biplanes to aircraft with cantilevered


6/22

wingsesulted in more wing flutter incidents. Primary sur-face flutter began to appear around 1925 (ref. 4). Air rac-ers experienced many incidents of flutter from themid- 1920's until the mid- 1930's as attempts were made tobreak speed records. References 3, 4, and 5 give manyexamples of these incidents.

Another form of flutter dealt with in the 1930's wasservo tab flutter. Collar (ref. 3) predicted that this type offlutter would be around for many years. This predictionwas correct, for between 1947 and 1956, 11 cases of tabflutter incidences were reported for military aircraft alone(ref. 6). Even today servo tab flutter is still a problem. In1986, the T-46A trainer experienced aileron flutter duringa test flight that was being flown to find the proper amountof mass balance. These ailerons were free floating anddriven by tabs at the trailing edge of the aileron (ref. 7).

New aeroelastic problems emerged as aircraft could flyat transonic speeds. In 1944, while flight testing the newP-80 airplane, NACA pilots reported an incident of aileronbuzz (ref. 5). From 1947 to 1956, there were 21 incidencesof flutter involving transonic control surface buzz. Proto-types of both the F-100 and F-14 fighters had incidencesof rudder buzz. Today, the transonic flight regime is stillconsidered the most critical from a flutter standpoint.

Chuck Yeager first achieved supersonic speeds in levelflight in 1947. Supersonic flutter then began to be studiedmore seriously as these speeds became routinely flown.Supersonic speeds also produced a new type of flutterknown as panel flutter. Panel flutter involves constantamplitude standing or traveling waves in aircraft skin cov-erings. This type of instability could lead to abrupt fatiguefailure, so the avoidance of panel flutter is important. Inthe 1950's a fighter airplane was lost because of a failedhydraulic line that was attached to a panel that had experi-enced such panel flutter (ref. 5).

The carriage of external stores affects the aeroelasticstability of an aircraft. Seven incidents of flutter from1947 to 1956 involved the carriage of external stores, aswell as pylon-mounted engines (ref. 5). The stores car-riage problem is still significant today, particularly withthe many store configurations that an airplane can carry.Certain combinations of external stores carried by theF-16, F-18, and F-Ill aircraft produce an aeroelasticinstability known as a limit cycle oscillation (LCO)(refs. 8 and 9). Although these oscillations are mostlycharacterized by sinusoidal oscillations of l imited ampli-tude, flight testing has shown that the amplitudes mayeither decrease or increase as a function of load factor(angle of attack) and airspeed.

Much has been learned about the prevention of flutterthrough proper aircraft design. Flight flutter incidences

still occur, however, on primary lifting surfaces as for theF- 117 stealth fighter (ref. 10) and the E-6 Tacamo (ref. 11 )aircraft, both of which experienced vertical fin flutter.

DEVELOPMENT OF FLIGHT FLUTTERTEST TECHNIQUES

Von Schlippe conducted the first formal flight flutter testin Germany in 1935. The objective of his test method wasto lessen the risk associated with flutter testing. The usualpractice at this time was to fly the airplane to themaximum speed and then to observe the stability of thestructure.

Von Schlippe's technique consisted of exciting the struc-ture using a rotating unbalance weight, measuring theresponse amplitude, and then recording the responseamplitude as a function of airspeed. The forced responseamplitude would rapidly increase as the aircraftapproached its flutter speed. Therefore, the flutter speedcould be estimated from data obtained at subcriticalairspeeds.

The Germans successfully used this technique until1938 when a Junkers JU90 aircraft fluttered in flight andcrashed. Inadequate structural excitation equipment andunsatisfactory response measurement and recording equip-ment were identified as probable causes for this accident(ref. 4).

The United States attempted this technique in the 1940'swith flutter tests of a Martin XPBM-I flying boat and aCessna AT-8 airplane (ref. 4). Figure 2, taken from refer-ence 4, shows the response amplitude data as a function ofairspeed. The graph shows that destructive flutter for thisairplane was averted by the narrowest of margins duringthis flight test.

504O30Amplitudeat v

Amplitude at 100 mph 20

AT-8 flight testWrightField, Nov. 1941 Leftwing datao Rightwing data

10J0 320

l

+ i J80 160 240Velocity, v, mphFigure 2.airspeed (ref. 4).

9,0476

Response amplitude ratio as a function of


7/22

In thelate1950's, excitation systems consisted of iner-tia shakers, manual control surface pulses, and thrusters(bonkers). Instrumentation had improved and responsesignals were then being telemetered to the ground for dis-play and analysis. Some programs still displayed responsesignals on oscillographs in the airplane. Many experiment-ers realized the importance of adequate structural excita-tion for obtaining a high signal-to-noise ratio (ref. 12).The use of oscillating vanes to excite the structure wasbeing considered during this time.

From the 1950's until the 1970's, many aircraft wereequipped with excitation systems. Frequency sweeps weremade to identify resonances. These sweeps were often fol-lowed by a frequency dwell-quick stop at each resonantfrequency. In-flight analysis was usually limited to logdecrement analysis of accelerometer decay traces on stripcharts to determine damping.

The F-111 program is an example of this procedure.Figure 3 shows a schematic of the process, taken from

reference 13. Filtered and unfil tered accelerometerresponse data were displayed on strip charts and on X-Yfrequency sweep plotters. Damping was manually deter-mined from the frequency dwell-quick stop decay traces.Computers were not used for analysis of the data.

The P6M aircraft program took a departure from thismethodology (ref. 14). This flight flutter program usedrandom atmospheric turbulence to excite the structure andspectral analysis to analyze the response data for stability.The objective of this technique was to use every minute offlight time for dynamics data and to eliminate special testpoints for flutter.

Since the 1970's, digital computers have significantlyaffected flight flutter testing techniques. The computer hasallowed for the rapid calculation of the fast Fourier trans-form (FTT). Computers have fostered the development ofmore sophisticated data processing algorithms that areuseful for analysis of response data from either steady-state or transient excitation. Frequency and damping are

, "" __Aerotab_,.I "",. []_1"-_ excitation

"'-.WI recor(ler-a-nd I

Ground control and data processing

TM receiving I I Band passfiltersand

tape recording I

Test director rx.Y frequencysweep plotters

Band pass filteredto eccel choiceRecord two st one time

Brush recorders16 channelsaccel and aero tabs

8 channelsflight parametersand control system(unfiltered)

Airplanetape

Postflight

Brush record___ of alltransducers I Manual datareduction(g versus Mach)940477

Figure 3. F-I 11 flight flutter test procedure (ref. 13).


8/22

nowestimatedithparameterdentificationechniques.Suchanalysissdoneonlinen anear-real-timeanner.Frequencynddampingrendsareestablishedsafunc-tionofairspeedrMachnumber.heserendsreextrap-olatedodeterminehestability tthenexthigherairspeedtestpoint.Ascomputerpeedsavencreased,hetimerequiredoconductlightflutterestingpertestpointhasdecreased.heabilitytoanalyzemoredataateachestpointandheincreasedumberfflighttestpointsesult-ing frommoresophisticatedircraftdesigns,owever,havencreasedhetotaltimeto cleartheflightflutterenvelope.Thetesttechniqueypicallyusedstomonitorhetele-meteredesponseignalswithreal-timerequencynalyz-ersandstripcharts.heseataarealsoacquiredydigitalcomputershatprocesshedatausingparameterdentifi-cationechniquesorestimatingrequencynddamping(ref.15).Figure4 showshisprocess.nthisfigure,dataaretelemeteredromtheairplanendsimultaneouslyis-

playedonstripcharts,ndreal-timerequencynalyzers.

Acomputercquireshedataoranalysisodeterminere-quencyanddampingstimates.hetestdirector,whocommunicatesiththetestaircraft,hasaccessoall ofthisinformationomakedecisionsncontinuingheflut-terenvelopexpansion.Althoughlightfluttertesttechniquesaveadvanced,today'sechniquesrestill baseduponthesamehreecomponentssVonSchlippe'smethod:tructuralxcita-tion,responseeasurement,nddataanalysisorstability.Technologyevelopmentssociateditheachcomponentwill bereviewedndadiscussionf theimpactonthesafetyfflightflutterestingwill follow.

EXCITATION SYSTEMS

Structural excitation is a necessary part of the flight flut-ter testing methodology. Detection of impending aeroelas-tic instabilities cannot be made without adequateexcitation. Adequate excitation provides energy to excite

Real-timerequency analyzers

V ,ght.wingtipI I L Right plus

l,z_ llft wingtip

I j Right minus

._ left wingtip0 25 50

Frequency, Hz

Spectral analysis facilityground station

Strip chartsllltStability trends

.12 LDamping,.08

"o I I I I I

Frequency'l_ I

Hz 4_-_0 .8

L I I n 11.0 1.2 1.4 1.6 1.8Mach number940478

Figure 4. Typical modem fl ight flutter test process.


9/22

all of theselectedibrationmodeswithsufficientmagni-tudesoaccuratelyssesstabilityromtheresponseata.TheTransaviaL12/T-400irplaneref. 16)clearlydemonstratedheimportancefadequatexcitationevelsin1986.Thisairplanewasexcitednthenitialflighttestsbycontrolsurfaceulsesndrandomtmosphericurbu-lence.Flutterdidnotoccurduringheflighttest.nasub-sequentflight, the airplaneexperiencediolentoscillationsftherudderandailboomwhentwaslownin roughweatheronditions.heseweatheronditionsprovidedhigherlevelsof excitationhanthe levelsinduceduringheflightflutterest.In the1930's,heGermansecidedhatimproperexciterocationesultednpoorresponseshatpreventedthedeterminationf theonsetof flutterandsometimesresultedn flutteroccurringnexpectedlyref.4).In the1950's,heUnitedStatesearnedhatowexcitationevelstendtogivea largescattern thedampingaluesesti-

matedromtheresponseata.naddition,heestimatedvaluessuggestedoweraerodynamicampinghanactu-allyexisted.Duringfluttertestingof theB-58airplane(ref.17),t wasfoundhatastructuralxcitationevelatleasthreeofourtimeshigherhanwasobtainedyran-domatmosphericurbulenceasnecessaryoprovideanacceptableevelofexcitation.Theexcitationystemmustnotonlyprovideadequateforceevelsbutmustalso(1)providedequatexcitationoverthedesiredrequencyangeof interest,2)belight-weightsoasnottoaffecthemodalcharacteristicsfthe

airplane,nd(3)havepowerrequirementselectricorhydraulic)hatheairplaneanmeet.t isdifficultoranyonesystemomeettheseequirementsimultaneously.Overheyears,manyypesof excitationavebeenriedwithvaryingdegreesf success.omemorecommonmeansncludecontrolsurfaceulses,scillatingontrolsurfaces,hrusters,nertialexciters,erodynamicanes,andrandomtmosphericurbulence.Control Surface Pulses

Manual control surface pulses were the first means ofexcitation. This provided sudden control surface move-ments. Depending on the type of control system, modes upto about I0 Hz can be excited this way. Such pulsesapproximate a delta function that theoretically has a highfrequency content. Two benefits of this type of excitationare that no special excitation equipment is required andthat the transient response signature of the structure iseasy to analyze for stability. Test duration for each pulse isshort, so many can be applied at each test point.

There are several drawbacks, however. First, it is diffi-cult to get repeatable pulses, and thus the degree of excita-tion is inconsistent. Second, either the pilot cannot providea sharp enough input or the control system is unable toprovide a sharp enough disturbance to excite any criticalflutter modes above 10 Hz. Third, such pulses often do notprovide an adequate level of excitation to determine theonset of flutter. The fact that flight was possible beyond theflutter speed without exciting flutter with pulses was dem-onstrated during a flight flutter test in the 1950's (ref. 18).The purpose of this particular flight was to investigate thestability of a vertical stabilizer. The structure was excitedusing rudder pulses and by thrusters (impulse generator).The thrusters excited flutter at a speed 5 knots below thatwhere the structure was previously excited by rudderpulses without incident.In spite of their limitations, control surface pulses have

continued in use to excite the structures of many airplanessince 1950. The F-101 (mid-1950's), the early testing ofthe F-4 aircraft (late 1950's), the A-7A (1965), some of theearly Boeing 747 flutter testing (1969), and low-speedtesting of the DC-10 airplane (1970) all used control sur-face pulses for structural excitation.

Flight control surface pulses are still used today as exci-tation for flight testing. Most modern fly-by-wire flightcontrol systems (analog and digital), however, have low-pass filters in the stick input path that filter out high-frequency signal content. For example, the F-16XL air-plane flight control system has a 1.6-Hz low-pass filter(single pole) in the stick input path that would washoutany sharp stick motion commands to the control surfaceactuators. Manual control surface pulses are still usedtoday on most small aircraft and sailplanes because thisis usually the only affordable type of excitation for theseaircraft.

Oscillating Control Surfaces

Commanded oscillations of the control surfaces werealso used in the 1950's. The XF3H-1 airplane used anoscillating rudder for excitation to investigate a rudderbuzz instability (ref. 19). The rudder was oscillated bysupplying a variable frequency command signal into therudder servo of the autopilot system. This system couldexcite over a frequency range of 5 to 35 Hz and with thefrequency stepped every 3 sec by an automatic rotaryswitch located in the cockpit.

In the mid-1960's, electronic function generators weredeveloped to provide signals to control surface servos inthe autopiiot system. These function generators providedsignals to oscillate the horizontal stabilator and the aile-rons of the F-4 airplane (ref. 19). The stabilator could


10/22

sweep from 8 to 30 Hz and the ailerons from 2 to 16 Hz.The actuators for each surface were modified to providethe required gain and frequency response. A cockpit con-troller provided the capability to adjust the excitationamplitude, mode selection (sweep or dwell), start and stopsweep frequency, and dwell frequency. Further advancesin electronics during the 1970's and 1980's resulted in theability to send excitation signals to the control surfaceactuators that were not sinusoidal. The F-18 aircraftexcitation system can generate sinusoidal andbandpass-filtered pseudo-random commands to the flightcontrol surfaces (ref. 19).

This method of excitation has been successfully usedfor the X-31 and YF-22 airplanes and for the stores clear-ance work on the F-16, F-15, and F-18 airplanes. TheX-31 airplane could sweep frequencies from 0.1 to100 Hz. Although significant actuator roll-off occurredabove 20 Hz, the combination of aerodynamic force at lowfrequencies and control surface inertia forces at higher fre-quencies provided adequate excitation for this airplane(ref. 20).

The primary advantage of this type of system is that noadditional hardware is required except for an excitationcontrol box located in the cockpit. As a result, the flutterspeed of the airplane is not affected as it might be withother types of excitation systems.

A disadvantage of this type of system is the frequencyresponse limitations of the control surface actuators.Often, special actuators are required to excite critical high-frequency modes, as was the case for the F-4 flutter testing(ref. 19).

Thrusters

Thrusters, sometimes known as bonkers, ballistic excit-ers or impulse generators, are an early device circa (1940)used for structural excitation. These small, single-shot,solid-propellant rockets have burn times of 18-26 msecand maximum thrust levels of 400--4,000 lbs (ref. 21).

Thrusters are simple, lightweight devices that generallydo not affect the modal characteristics of the airplane.These devices produce transient responses of short dura-tion, which is important when the airplane has to dive toattain a test condit ion.

The disadvantages for these devices include single-shotoperation, difficulty in firing two or more either in phaseor out of phase with respect to each other, and their inabil-ity to provide a wide frequency band of excitation. Usu-ally required are thrusters with three different burn timesto excite modes in a frequency range that covers 5 to50 Hz.

Thrusters were used for part of the flutter clearance ofthe F-101 airplane in the mid 1950's. Six thrusters weremounted on each wingtip, three on top and three on thebottom. Use of these devices was partially successful forthis program (ref. 18).

Thrusters were also used by Douglas Aircraft Companyin the 1950's (ref. 19) on several airplanes to investigateflutter characteristics. Thrusters were also used for por-tions of the F-4 flutter testing in the early 1960's. Sincethen, thrusters have not been used in the United States forany major flight flutter test program.Inertial Exciters

A large variety of rotating eccentric weight and oscillat-ing weight inertia exciters have also been tried. The rotat-ing unbalance exciter was widely used for flight fluttertesting in the 1940's and 1950's. These systems derivetheir forces from mass reactions. The inertia force is pro-portional to the exciter weight multiplied by the square ofthe rotating speed. As a result, the excitation capabilitymay be limited at lower frequencies and excessive athigher frequencies.

The Martin XPBM-1 flying boat flight flutter test pro-gram used such a system. The precise control of frequencywas difficult with the equipment available; thus the exciterwould not stay tuned on the resonant frequency. Onemethod of tuning the exciter frequency, as done by aConvair F-92 pilot in 1950, was by observing a meter inthe cockpit that measured the response of selected vibra-tion pickups during flutter tests (ref. 4).

The magnitude of the forces required to adequatelyexcite an airplane is usually very large. As a result, thehardware required to produce the unbalance forces oftencould not be contained within the wing contour. In addi-tion, these systems often are very heavy and raise concernsabout the effect on the modal characteristics of the air-plane that they are installed on. For example, the rotatingunbalance equipment designed for (but never installed on)the XB-36 had a maximum force output of 1000 lb, andinstallation in each wing weighed between 400 and 500 lb(ref. 12).

Inertia shakers were used for the B-58 flutter testing inthe early 1950's. These shakers were hydraulically pow-ered and electrically controlled and were used to excite afrequency range of 5 to 40 Hz. The overall dimensionswere 4.5 by 4.5 by 8.5 in., and each unit weighed 25 lb.The force output was 40 lb at 7.5 Hz and 150 lb at 40 Hz;the force level increased linearly between these two fre-quency values (ref. 17). This type of shaker worked well,particularly in exciting the higher frequency modes of theairplane.

6


11/22

TheConvair-102A,whichwaslutterestednthelate1950's,lsousednertialshakers.n thefrequencyangebetweenand50Hz,theforcevariedromapproximately20to300lb.Thisshakerwassufficientlyompactobeinstallednsidehewing-ip,whichhasadepthof4.5in.Theweightof theshakeras8.5lb(ref.22).Afterthe1950's,nertiaexciterswerenotusedexten-sively.However,herehasbeenimiteduseof suchsys-temsoprovidepartialstructuralxcitationortheF-14horizontaltabilizer,heF-111horizontalndverticalsta-bilizer,andtheX-29flaperon.heB-1Aairplanealsousedaninertiashakerystemfig.5)forflightflutterest-ing.Thesystemconsistedf fivehydraulicallyriven,electronicallyontrolled,scillatingmassexciters.Oneexciterwasplacedteachwingtip,oneateachhorizontalstabilizerip,andoneatthetipoftheverticalstabilizer.Eachwingtipandhorizontaltabilizeripcouldbeoper-atedn andoutof phasewithrespecto eachother.Thissystemwascapablef producingmaximumorceof

about550lbof forcewithanexciterweighingpproxi-mately40lb (ref.23).Thissystemdequatelyxcitedhemodesf interestndwasessentialorsafelyexpandingtheflightenvelope.Aerodynamic Vanes

An aerodynamic vane consists of a small airfoil that isusually mounted to the tip of a wing or stabilizer. The vaneis generally mounted on a shaft, driven either electricallyor hydraulically, and oscillates about some mean angle.Oscillation of the vane will result in a varying aerody-namic force acting on the airplane. The amount of forcedepends on the size of the vane, dynamic pressure, andangle of rotation.

movable mass (wand)940479

Figure 5. Inertial excitation system (ref. 23).

Aerodynamic vanes were first used in the 1950's.The YB-52 airplane used a wingtip oscillating airfoilshaker for flight flutter testing (ref. 24). This wingtip unitweighed 150 lb and was mounted on the right wingtiponly. A similar amount of weight was installed on theopposite wingtip. Typical sweep times were approximately7 min. The excitation frequency could be varied from 1.4to 10 Hz.

Since then, many flight flutter test programs haveused the aerodynamic vane as a means of excitation. Theseprograms include the DC-10, L-1011, Boeing 747,Boeing 757, S-3A, F-14, F-Ill, A-10, C-17, and T-46A.Tables 1 and 2 which were taken from reference 13, showthe characteristics of some of these excitation systems.

The advantage of this type of system is that it can excitelow frequencies well; the amplitude at high frequencies islimited only by the response characteristics of the vanedrive mechanism. The excitation frequency and amplitudeat a given airspeed can be controlled, and the force timehistory produced is repeatable.

The main disadvantage is that the maximum force pro-duced varies with the square of the equivalent airspeed.Other disadvantages include the addition of mass, the dis-turbance of the normal airflow around the wingtip or stabi-lizer tip with the vane present, and the large powerrequirements usually needed to operate this system.

There are two notable variations of the oscillating aero-dynamic vane concept: a rotating vane and a fixed vanewith a rotating slotted cylinder attached to the vane trailingedge.

The C-5A airplane excitation system consisted of arotating vane mounted on top of the wing at each tip andon top of each horizontal stabilizer at each tip (ref. 13).The vanes were continuously rotated through 360 degrees,and both sets of vanes were synchronized to provide eithersymmetric or antisymmetric excitation. This system pro-duced periodic excitation to the structure and was success-fully used for flight flutter testing.

The fixed vane with a slotted rotating cylinder attachedto the trailing edge (fig. 6) was developed by W. Reed(ref. 25). The vane/cylinder assembly weighs approxi-mately 10 lb. The device generates periodic lifting forcesby alternately deflecting the airflow upward and downwardthrough the slot. This system was used for the flutter clear-ance of a F-16XL airplane with a modified laminar flowglove (ref. 26). Frequency sweeps were conducted from 5to 35 Hz for this program. The system uses exceptionallylittle electric power and is easy to install.


12/22

Table1.SummaryfaerodynamicxcitationystemssednU.S.Frequency Timeosweep,Airplane Surface Location range sec Sweepaw

747 Wings Externalanest 1.5-7.0Hz 90 ExponentialwingtipsDC-10 Wingshorizontal Externalanesttipsof 1-20Hz 90 ExponentialVerticalail mainsurfaces and1-10Hz 90L-1011 Wingstabilizer Externalanes 1-18Hz 90 Linearperiod3-25Hz 30S-3A Sideoffuselagender Externalanes 1.5-18Hz 90 Linearperiodstabilizer 3-25HzC-5A Wingstabilizer Externalanesntopof .5-25Hz 60normal Exponentialsurfacesearips 30diveonlyF-14 Wingin Aero-tab 5-50Hz 15 ExponentialExternal vaneF- 15 Normal control 2-16 Hz 100-200 Linear

Ailerons 5-10 Hz 45 frequencyStabilator

F-I 11 Wing Aero-tab 35-2 Hz 45 Exponential

Table 2. Summary of inertial excitation systems used in U.S.Frequency Time to sweep,

Airplane Surface Location Range sec Sweep lawF-14 Horizontal tail Right side stabilizer 5-50 Hz 15 Exponential

onlyF- 111 Horizontal, vertical Inboard on stabilizer 35-2 Hz 45 Exponential

tail surfaces near side of fuselage,top of fin

Random Atmospheric TurbulenceAtmospheric turbulence has been used for structural

excitation in many flight flutter test programs (ref. 15).The greatest attraction to this type of excitation is that nospecial onboard exciter hardware is required. Turbulenceexcites all of the surfaces simultaneously, which causesboth symmetric and antisymmetric modes to be excited atthe same time. This method eliminates the need to per-form symmetric or antisymmetric sweeps.

Natural turbulence is the random variation in windspeed and direction. Turbulence is generally produced by

weather-front winds and thermal activities; the extentof turbulence within an air mass can vary widely. Refer-ence 27 provides an excellent description of turbulenceand its use for excitation in flight flutter testing.

This approach was tried in the late 1950's. The P6MSeamaster flight flutter program (ref. 14) used randomatmospheric turbulence to excite the structure and spectralanalysis to analyze the response data for stability. Objec-tives of this technique were to use every minute of flighttime for dynamics data and to eliminate special test flightsfor flutter. The YF-16 also used random atmospheric


13/22

Figure6.F-16XLwithvaneandotatinglottedylinderxcitationystemref.26).

turbulencelongwiththerandomecrementechniqueordataanalysisoclearheflutterenvelopef thebasicair-planeref.28).Althoughhismethodasbeenusedwithsomeuccessoverheyears,hereareseveralisadvantages.heurbu-lencehats foundsoftennotintensenoughoproducesufficientexcitationomparedith thatobtainedwith

onboardexciters.Turbulencesuallyexcitesonlythelowerfrequencymodesor mostairplanes.ongdatarecordsrerequiredoobtainesultswitha sufficientlyhighstatisticalonfidenceevel.Thesignal-to-noiseatioof theresponseatasoftenow,whichmakesataanaly-sisverydifficult.Flighttimeis lostlookingorsufficientturbulence,ndturbulencefteninterfereswithotherengineeringisciplinesata.Figure7 compareshepowerpectrum plots obtained

from the F-16XL airplane excited by random atmosphericturbulence and by a vane with a slotted rotating cylinderattached to the trailing edge. The turbulence was reportedto be light-to-moderate for this flight condition. All of thestructural modes were excited by the vane, while only the8-Hz mode was well excited by turbulence. This datacomparison clearly indicates the poor data quality that istypically obtained with random atmospheric turbulence.

This flight test program also illustrates the lessonslearned during the B-58 flutter test program. Inadequate

excitation levels usually give a large scatter of the esti-mated damping values, and the estimated damping valuesoften indicate lower damping than actually exists. A com-parison of response data damping values from randomatmospheric turbulence and vane excitation (fig. 8) showsthat the turbulence response data, which have a lower exci-tation level, consistently have lower estimated dampingvalues.

Although the flutter envelopes of many modern aircraft(i.e., X-29A, advanced fighter technology integration(AFYI) F-Ill mission adaptive wing, AFTI F-16, Sch-weizer SA 2-37A motor glider, and F- 15 short takeoff andlanding/maneuver technology demonstrator) have beencleared using atmospheric turbulence, caution should beused when using this form of excitation to ensure that thecritical flutter mode has been excited throughout the flightenvelope.

INSTRUMENTATION

The instrumentation used to record the structuralresponses of an airplane to excitation is another criticalcomponent of flight flutter testing methodology. Theresponse data must be measured at enough locations andbe of high enough quality that the flight can be conductedsafely. Included in the instrumentation system is the mea-surement, telemetry, recording, and displaying of the flightdata.


14/22

'03Amplitude, .02 -

g2

.01 --

.04 --

30 40 50 60

.03 -

Amplitude, .02g2

.01

,,.,J0 10 20 0

Frequency, Hz 920768

-L10 20 30 40

Frequency, HzI50

(a) Exciter sweep. (b) Random turbulence.Figure 7. Comparison of excitation of the left-wing aft accelerometer at Mach 0.9 and 30,000 ft (ref. 26).

,1160

920769

.08

.07

.06

.05Structu raldamping [_

.02

.010.6

Atmospheric turbulenceexcitation

Forced excitation using linearsine sweeps

I I I I I I.7 .8 .9 1.0 1.1 1.2Mach number 920770

(a) Symmetric.

.--Ib---

---o---

.08

.07

.06

.05Structuraldamping .04.03

.02.010.6

Atmospheric turbulenceexcitationForced excitation using linearsine sweeps

I I I I I 1

.7 .8 .9 1.0 1.1 1.2Mach number 92o_1(b) Antisymmetric.

Figure 8. Structural damping values for wing bending modes (ref. 26).

MeasurementThe most commonly used transducers to measure the

excited response of a structure have been the accelerome-ter and strain-gage bridge. The selection of the device touse often depends on the ease with which installation canbe accomplished, although today the more commonlyused device is the accelerometer.

Accelerometers used in the early 1940's were large andheavy. As an example, the accelerometers were about3 in. high, 2 in. wide, and weighed about 1 lb. Subminia-ture accelerometers were developed later and these werelighter but still were large (l/2-in. high, and l-in. wide).

The calibration of these devices drifted during operationmainly because of the electronics in use then.

The accelerometers used for the B-58 flutter testing inthe 1950's were of the strain-gage type. These were fluiddamped devices. To minimize the effects of the outside airtemperature on the damping, these units had built-in elec-tric heaters to maintain a constant temperature of 165 F(ref. 17).

Piezoelectric accelerometers have been developed suchthat miniature units today weigh less than one-tenth of anounce, operate in a temperature range of -65 to 200 F,have high sensitivity, have a linear frequency range of 1 to10,000 Hz, and have amplitude linearity from 1 to 500 g.

l0


15/22

Typicallyhedimensionsreassmallas0.25n.wideand0.15n.high.Today'sccelerometersccuratelyeasurethestructuralesponsenalmostnykindofenvironment.Subminiaturenstrumentationasalsobeendevelopedasself-containedeelandstickdevicesref.29).Eachunitcontains battery,ensor,ntenna,rocessor,ndtransmitter.heseevicesonotrequireanywiringandthesignalsmaybetransmitteditheroareceiverocatedonboardheairplanerdirectlyoareceiverocatednthegroundwithinasmalllightradiusof thetransmitter.Advancesuchasthiswill significantlyeducehecostassociatedithflutteresting.

Telemetry and Recording

Recording equipment was not adequate during the1930's and was cited as a possible reason that the Ger-mans lost several airplanes during flutter testing (ref. 4) inthat decade. In the 1930's, a Junker JU86 airplane under-went flight flutter testing to identify the effects of balanceweights on rudder stability. The recording system usedwas a thin wire attached to the rudder, which mechanicallyactuated a recorder installed in the observer's seat of theairplane (ref. 4).

During the 1940's, the accelerometer responses wererecorded on photographic oscillographs mounted in thecockpit or airplane cabin. These devices required a devel-oping time for the paper; the time history responses couldnot be viewed immediately as a result.

In the 1950's, data were commonly FM/FM telemeteredto the ground, recorded on magnetic tape, and then dis-played on strip chart recorders. The telemetry systemsduring this period were small and typically only 8 to12 channels of data could be telemetered to the ground(refs. 30, 31, and 32). As a result, on-board tape recorderscaptured all of the flight data while the recorders on theground captured only the data that could be sent downfrom the airplane. The ground tapes were typically noisierthan the onboard tapes mainly because of data-transmission problems.

Pulse code modulation (PCM) or digital telemetry wasinitiated in the 1960's, although FM/FM telemetry wasstill widely used for flutter testing because of the fre-quency bandwidth required. The PCM telemetry signifi-cantly increases the number of parameters that can betransmitted to the ground but requires a filter to preventfrequency aliasing of the analog response signal duringdigital sampling on board the airplane.

The frequency bandwidth of PCM systems hadincreased significantly by the 1980's. A frequency

bandwidth of 200 Hz is easily attainable and sufficient formost flutter applications. As a result, PCM telemetry isnow usually preferred for most flight flutter testing.

Today the available portable digital recorders are com-pact and can acquire data from flight instrumentation forstorage within the unit's computer memory. Direct storageof the data eliminates the need for expensive equipmentassociated with PCM systems. Although these units canacquire a limited amount of data, they have been used forlow-cost flutter testing (for example, the Pond Racer air-plane). The data from the unit's computer memory aredownloaded after each flight into a digital computer foranalysis.

Displays

Computers were used to manipulate and display data tosome extent during the 1950's. Analog computers weresometimes used to add, subtract, multiply, integrate, andfilter the data signals telemetered (ref. 30) and then to dis-play these signals on strip charts for the flutter engineer.

Sometimes the pilot had a small cathode ray tube oscil-loscope in the cockpit to display the decay trace of a sin-gle, selected accelerometer (ref. 33). The pilot was briefedbefore each flight by the flutter engineer on the anticipatedresponse amplitudes and safe operating limits.

In the 1950's, data were primarily displayed on stripcharts in the control room for analysis by the flutter engi-neer. Strip chart capabilities have greatly increased, andtoday strip charts continue to be the primary device to dis-play real-time accelerometer and strain-gage responsetime histories.

In the 1970's, computer technology had advanced to thepoint that it became feasible to use a computer to performonline stability analysis of the flight flutter test data. Com-puters were also used to provide discretes and alphanu-merics in the control room. Many basic airplaneparameters, such as Mach number, airspeed, angles ofattack and sideslip, and fuel quantities, could now be dis-played on cathode-ray tubes. This display provided theflutter engineer with the ability to more closely monitorthe flight test conditions of the airplane.

In the 1980's, it became possible to provide the pilot areal-time guidance system for maintaining flight condi-tions. With this system the pilot flies the airplane to mini-mize the computed differences between the desired andactual flight condition (Mach number and altitude). Thecomputed differences are telemetered to the airplane froma ground-based computer. The pilot then uses a cockpitdisplay as an aid to reach and hold desired test conditions;

11


16/22

this has resulted in exceptionally accurate stabilized flighttest conditions (ref. 34).

Figure 9 shows how aircraft instrumentation, datarecording, and real-time support in the ground station havechanged since the 1940's. In summary, removing the flut-ter engineer from the airplane during the test, havingincreasingly reliable and accurate instrumentation, andhaving ground stations with highly specialized test capa-bilities has significantly reduced the hazard of flight fluttertesting.

DATA REDUCTION METHODS

The next component of flight flutter testing methodol-ogy is the analysis of the response signal. The responsesignal can consist of random response caused by atmo-spheric turbulence or exciter input, transient responsescaused by either impulse input or exciter frequency dwell-quick stops, or steady-state responses caused by exciterfrequency sweeps. The accurate and timely evaluation of

this data to determine stability is critical to the overallsafety of the flight flutter test program.

In the 1930's and 1940's, the methods used consisted ofmeasuring the response amplitude caused by a frequencysweep or determining the damping from a response causedby a control surface pulse. These data were recorded on anoscillograph recorder. All of this analysis was usually doneby hand between flights, because no computers withsophisticated identification algorithms were available. Thedamping estimation consisted of using the log decrementmethod on the decay portion of a time history response.Occasionally, in lieu of telemetry, the flutter engineer wason board the airplane to do these analyses (ref. 4).

As telemetry became more available in the 1950's, thedata analysis methods just described were still used, butthe analysis was generally conducted on the ground. Itbecame more common to excite the airplane at stabilizedtest points, analyze the response for stability, and thenclear the airplane to the next higher airspeed (refs. 17, 22,and 24).

Year

1990

1980

1970

1960

1950

1940

Instrumentation Data recordingPCM telemetry100s of channels)

Accelerometers FM telemetryand strain gagesOn-board aircraft tape

)utar disk

Accelerometarsand strain gages

PCM telemetry100s of channels)FM telemetry

On-board aircraft tape

Accelerometarsand strain gages

Accelerometersstrain gages

PCM telemetry(100s of channels)FM telemetry

On-board aircrafttape

FM telemetry(10-20 channels)

On-board

Real-time supportDiG stripcharte

graphicsDigital computer for data

manipulation and analysisdiscretas, video

yzersatripcharta

Digital computer for datamanipulation and analysisAlphanumerics, dlscretas,

aircraft tape

analyzersstripcharta

computer to)ulate dataplotters

FiltersI stripchartscomputer

manipulateplottersFilters

On-boardAcceleromaters photographic

and oscilloscopestrain gages


17/22

Data transmission problems usually added noise to thetelemetered data. Filtering of the response data was com-mon to reduce the scatter in the damping estimates. Datawere often passed through a filter to reduce the data to asingle-degree-of-freedom response. Closely spacedmodes, however, proved to be difficult to separate fordamping estimation using these analysis techniques.

Another method for enhancing the data analysis was toadd and subtract time history signals. The symmetric andantisymmetric modes could then be separated, reducingthe modal density of the response signal (ref. 32).

In the 1950's, vector plotting (ref. 31) and spectral anal-ysis (ref. 14) were also used to determine stability. Modaldamping was not estimated from the spectral analysistechnique. This analysis only provided the frequencies andamplitudes of the response signal being analyzed.

These manual or analog analysis techniques continuedto be used until the 1970's. Using tracking filters during anexciter frequency sweep improved the data qualityobtained. Such a technique was used in the 1960's(ref. 19) to filter data provided to an analyzer to produce areal-time plot of frequency and amplitude. Even so, nophase or damping information was available from thisapproach.

By 1970, digital computer systems could be used forinteractive analysis of flight data. During the early 1970's,the fast Fourier transform was implemented on the com-puter, providing the capability to obtain frequency contentof acquired signals in less that a second. The speed ofcomputers then allowed parameter identification algo-rithms to be programmed for estimation of damping fromthe response signals.

The F-14 and F-15 aircraft programs were among thefirst to take advantage of this advance in technology. Forthe F-14, an equation error identification technique wasused to estimate frequency and damping information(ref. 37). The F-15 program used an analysis technique topredict the flutter boundary based on frequency and damp-ing data acquired at subcritical speeds (ref. 38). Other pro-grams, that used the random decrement technique, such asthe YF-16 (ref. 28), also took advantage of the increasedcapabilities of the digital computer to increase the effi-ciency and safety of flight flutter testing.

Since the 1970's, many identification algorithms havebeen developed to estimate frequency and damping fromflight flutter data. References 39 and 40 are excellentreports on modal parameter estimation and providenumerous references for the many different approachestaken.

CONCLUDING REMARKSCurrent State-of-the-Art

Today, the typical approach to flight flutter testing is tofly the aircraft at several stabilized test points arranged inincreasing order of dynamic pressure and Mach number.Data are analyzed at these points only. The number of sta-bilized test points required to clear the flutter envelope ofan airplane is typically high and consequently requiresmany flights to accomplish them. For example, the F-14required 489 shaker sweeps to clear the basic airplaneflight envelope; 177 shaker sweeps were required for theGulfstream III; and 264 shaker sweeps were required forthe Gulfstream II ER airplane (ref. 37). The F-15 required132 shaker sweeps and 156 frequency dwells to clear thebasic airplane flight envelope (ref. 41).

The data obtained at each stabilized test point establish adamping trend as a function of airspeed. Information isthen extrapolated to predict the stability of the nextplanned test point. This practice is questionable becauseactual damping trends can be nonlinear. The most criticalpart of expanding the flutter envelope is the accelerationfrom one test point to the next. During this phase, responsedata are not being quantitatively analyzed. Instead, engi-neers, relying on intuition and experience, are limited toreal-time monitoring of sensor responses on strip charts.

An examination of several flight flutter test programsshows the effectiveness of the techniques used today towarn of the onset of flutter. Figure 10 shows the frequencyand damping trend information obtained, in near-real-time, during the flutter testing of the KC-135 airplane

On Autopilot offe- Autopilot on

Damping, .08g .04

0 I

3.1Frequency, n oHz 2.9 l-2.r I-___I2.5/ I I I I I280 300 320 340 360 380Airspeed, KEAS

940484

Figure 10. Frequency and damping trends establishedfrom flight data.

13


18/22

configuredithwingletsref.42).Theaircraftstructurewasexcitedbypilot-inducedontrolsurfaceulsesnddampingasestimatedsinganFFTalgorithmref.15).Thesubcriticalampingrendfor a2.6-Hzand3.0-Hzmodendicatedhatasairspeedasncreased,hesewomodescoupledandcaused decreasen thedampinglevel.Thedampingeveldecreasedith increasingir-speeduntilit wasnolongersafeocontinuehetest.nthisinstance,hetechniquessedrovidedsufficientndadequatearningof theonsetof fluttermainlybecausethedecreasendampingasgradual.Figure11showshedampingrendobtainednnear-realtimeforanF-16airplaneonfiguredithAIM-9Jmissiles,GBU-8stores,nd370-galxternalueltanks(ref.15).Flightof theairplanewiththisstoreconfigura-tion is characterizedy anLCO.Decayraceswereobtainedyusingheflaperonsoexcitehestructureithfrequencywells-quicktops.Thedampingwasesti-matedusingFFTalgorithms.hissetof dataprovided

uniqueopportunityovalidateheaccuracyf thisalgo-rithmbecausehisconfigurationouldbesafelylownoaconditionf zerodamping. linearextrapolationfthedatarendprovidedninstabilityirspeedredictionhatagreedloselywiththeactualnstabilitynsetairspeedencountered.Although today's techniques appear adequate to warn of

the on-set of flutter for gradual decreases in damping, it isdoubtful that sudden changes in damping, which mayoccur between flutter test points, can be predicted with theaccuracy and timeliness required to avoid flutter.

Future of Flight Flutter Testing

The future of flight flutter testing has been defined atseveral times. The flutter testing symposia held in 1958and 1975 identified future directions and needs (refs. 43

.16 -

.12Damping,

g .08.04

Figure 11.instability.

I0 .2

Damping trend

_ Extrapolated_-Actualinstability

.4 .6 .8 1.0Mach number 940485

for a limited amplitude

and 44). These symposia proceedings will be reviewed toconfirm the progress made toward those needs. Our futureneeds will then be presented.1958 Flight Flutter Test SymposiumThe final sentence in reference 4, which was presented

at the 1958 flight flutter test symposium, was, "It is hopedthat improvements in test techniques will eventually resultin flight flutter tests that will give all the informationwanted and will be considerably less hazardous than theyare today." Reference 17 stated that improvements neededwere to shorten the time required to obtain data and to pro-vide complete and higher quality data. The way to fulfillthis need was to automate the data-reduction equipment.Reference 22 stated the need for completely automaticexcitation, data-recording and data-reduction systems.

1975 Flutter Testing Techniques Symposium

The 1975 symposium contained several papers describ-ing the application of techniques that used computers toestimate frequency and damping from flight flutter testdata. Two future needs identified from the papers pre-sented at the symposium were (1) to further developparameter estimation algorithms that would provide betterestimates from noisy data and closely spaced modes(ref. 28), and (2) to develop effective noise reduction andtransfer function enhancement (ref. 45). The high-speedcomputer was identified as the tool for developingadvanced data analysis methods that would more fully sat-isfy the desired objectives of flight flutter testing (ref. 46).Most agreed that the current (i.e., 1975) techniques werefaster and, more accurate, increased safety, and reducedflight test time when compared with previous methods offlight flutter testing.

The future needs expressed in the 1958 symposium werepartially met at the time of the 1975 symposium. Flightflutter testing was, more automated, and the data werecomplete and of higher quality. The time required toacquire the data was not significantly less because most ofthe flight test organizations were conducting sine sweepsat stabilized test points. Testing was less hazardous in1975 than in the 1950's, although reference 47 warned thatthe current techniques may still not predict flutter forexplosive flutter cases. Reference 48 recommended thatfrequency and damping estimates for clearance to the nextflutter test point be made between flights.Recommendations for Future Research andDevelopmentOnline, real-time monitoring of aeroelastic stability dur-

ing flight testing needs to be developed and implemented.

14


19/22

Techniquesuchasmodal iltering(refs.49and50)canuncoupleesponsemeasurementso producesimplified,single-degree-of-freedomesponses.heseesponsesouldthenbeaccuratelynalyzedwithlesssophisticatedlgo-rithmshataremoreableoruninrealtime.Theidealdis-playwouldshowpredictedrequencynddampingaluesbeingomparedithflighttestvaluesnrealtime.Real-timeonitoringf stabilityliminateshemosthaz-ardousartofflightflutteresting,whichstheaccelerationfromonetestpointtothenext.Suchmonitoringlsoelimi-natesheneedorstabilizedestpoints,whichsextremelytimeconsuming.Broad-bandxcitationechniqueslsoneedobedevel-opedsothata responseignalof sufficientmplitudevertheentirerequencyangeof interestscontinuallyrovidedforreal-timenalysis.Newmethodshouldberesearchedopermita reliabledeterminationf flutterspeedataspeedhatswellbelowtheactualone.ThetechniqueroposedyNissimref.51),whichisbasedn identifyinghecoefficientsf theequa-tionsof motionollowedbysolvingof theseequationsodetermineheflutterspeed,maybeoneapproach.hewholeprocessfflightflutterestingeedsobefullyautomatedoflightflutterestinganbedonemuchasterutmoresafely.

REFERENCES

1Von Schlippe, B., "The Question of Spontaneous WingOscillations (Determination of Critical Velocity ThroughFlight-Oscillation Tests)," NACA TM-806, October 1936(translation).

2Lancaster, EW., "Torsional Vibrations of the Tail of anAeroplane," Reports and Memoranda, no. 276, July 1916, in"AIAA Selected Reprint Series, Volume V, AerodynamicFlutter," I. E. Garrick, ed., March 1969, pp. 12-15.

3Collar, A.R., "The First Fifty Years of Aeroelasticity;'Aerospace, voi. 5, no. 2, (Royal Aeronautical Society),February 1978, pp. 12-20.

4Tolve, L.A., "History of Flight Flutter Testing," in"Proceedings of the 1958 Flight Flutter TestingSymposium" NASA SP-385, 1958, pp. 159-166.

5Garrick, I.E., and Reed, W.H., III, "HistoricalDevelopment of Aircraft Flutter," AIAA 81-0491, J. Aircraft,vol. 18, no. 11, November 1981, pp. 897-912.

6NACA Subcommittee on Vibration and Flutter, "ASurvey and Evaluation of Flutter Research and Engineering,"NACA RM-56I 12, October 1956.

7French, M., NoU, T., Cooley, D., Moore, R., and Zapata,F., "Flutter Investigations Involving a Free Floating Aileron,"AIAA Paper no. 87-0909, April 1987.

8Champion, L.S., and Cabrera, E.A., "F-16C/D Block 40With Advanced Medium-Range Air-to-Air Missile(AMRAAM) Flutter Flight Test Evaluation," AFFTCTR-92-19, December 1992.

9Norton, W.J., "Limit Cycle Oscillation and Flight FlutterTesting," in "Proceedings, Society of Flight Test Engineers,21 st Annual Symposium," August 1990, pp. 3.4-1-3.4-12.

10Farley, H.C., Jr., and Abrams, R., "F-117A Flight TestProgram," in "Proceedings of the 34th Society ofExperimental Test Pilots Symposium," September 1990,pp. 141-167.

llBorst, R.G., and Strome, R.W., "E-6 FlutterInvestigation and Experience," AIAA-92-4601-CE AIAAGuidance, Navigation, and Control Conference, HiltonHead, South Carolina, August 1992.

12Schwartz, M.D., and Wrisley, D.L., "Final Report on theInvestigation of Flight Flutter Testing Techniques for theBureau of Aeronautics, U. S. Navy," Aeroelastic andStructures Research Laboratory, Massachusetts Institute ofTechnology, December 1950.

13Rosenbaum, R., "Survey of Aircraft Subcritical FlightFlutter Testing Methods," NASA CR-132479, August 1974.

14Kachadourian, G., Goldman, R.L., and Roha, D.M.,"Flight Flutter Testing of the P6M," in "Proceedings of the1958 Flight Flutter Testing Symposium;' NASA SP-385,1958, pp. 91-96.

15Kehoe, M.W., "Aircraft Flight Flutter Testing at theNASA Ames-Dryden Flight Research Facility," NASATM-100417, May 1988.

16Goldman, A., Rider, C.D., and Piperias, E, "FlutterInvestigations on a Transavia PL12/T-400 Aircraft,"Aeronautical Research Laboratories, Melbourne, Australia,Aircraft Structures Technical Memorandum 515, July 1989.

17Mahaffey, ET., "Flight Flutter Testing the B-58Airplane," in "Proceedings of the 1958 Flight Flutter TestingSymposium," NASA SP-385, 1958, pp. 121-125.

15


20/22

18Stringham,.H.,Jr.,andLenk,E.J,,"hightFlutterTestingUsingPulseTechniques,"n "Proceedings of the1958 Flight Flutter Testing Symposium," NASA SP-385,1958, pp. 69-72.

19Meany, ].J., "The Evolution of Flutter Excitation atMcDonnell Aircraft," in "14th Annual SymposiumProceedings," Society of Flight Test Engineers, August1983, pp. 4.6-1-4.6-I 1.

20Hodson, C.H., Dobbs, S.K., Brosnan, M.J., and Chen,J.B., "X-31A Flight Flutter Test Excitation by ControlSurfaces," AIAA-93-1538-CP, 34th AIAA Structures,Structural Dynamics and Materials Conference, La Jolla,California, April 1993.

21Laure, P., Millet, M., and Piazzoli, G., "PyrotechnicBonkers for Structural Tests in Flight," ONERA TPno. 1389 E, 1974.

22Dublin, M., and Peller, R., "Flight Flutter Testingof Supersonic Interceptors," in "Proceedings of the 1958Flight Flutter Testing Symposium," NASA SP-385, 1958,pp. 111-120.

23Dobbs, S.K., and Hodson, C.H., "Determination ofSubcritical Frequency and Damping From B-1 hightFlutter Test Data," NASA CR-3152, June 1979.

24Bartley, J., "Flight Flutter Testing of Multi-JetAircraft," in "Proceedings of the 1958 Flight FlutterTesting Symposium," NASA SP-385, 1958, pp. 103-110.

25Reed, W.H., III, "A New hight Flutter ExcitationSystem," in "19th Annual Symposium Proceedings,"Society of Flight Test Engineers, Arlington, Texas, August14-18, 1988, pp. V-I.1-V-1.7.

26Vernon, L., "In-Flight Investigation of a RotatingCylinder-Based Structural Excitation System for FlutterTesting," NASA TM-4512, 1993.

27Norton, W.J., "Random Air Turbulence as a FlutterTest Excitation Source," in "Proceedings, Society ofFlight Test Engineers, 20th Annual Symposium," Reno,Nevada, September 18-21, 1989.

28Brignac, W.J., Ness, H.B., Johnson, M.K., and Smith,L.M., "YF-16 Flight Flutter Test Procedures," in "FlutterTesting Techniques: A conference held at Dryden FlightResearch Center, October 9-10, 1975," NASA SP-415,1976, pp. 433--456.

29Nordwall, B.D., "Mobile Communications to CaptureConsumer Market," Aviation Week & Space Technology,May 31, 1993, pp. 41-42.

30Baird, E.E, Sinder, R.B., and Wittman, R.B.,"Stabilizer Flutter Investigated by hight Test," in"Proceedings of the 1958 Flight Flutter TestingSymposium," NASA SP-385, 1958, pp. 73-81.

31Reed, W.H., III, "A Flight Investigation of OscillatingAir Forces: Equipment and Technique," in "Proceedings ofthe 1958 Flight Flutter Testing Symposium," NASA SP-385, 1958, pp. 41-50.

32philbrick, J., "Douglas Experience in Flight FlutterTesting," in "Proceedings of the 1958 Flight FlutterTesting Symposium," NASA SP-385, 1958, pp. 127-132.

33Laidlaw, W.R., and Beals, V.L., "The Application ofPulse Excitation to Ground and Flight Vibration Tests," in"Proceedings of the 1958 Flight Flutter TestingSymposium," NASA SP-385, 1958, pp. 133-142.

34Meyer, R.R., Jr., and Schneider, E.T., "Real-TimePilot Guidance System for Improved Flight TestManeuvers," AIAA-83-2747, November 1983.

35Rhea, D.C., and Moore, A.L., "Development of aMobile Research Flight Test Support Capability," in"proceedings of the 4th AIAA Flight Test Conference,"San Diego, California 1988.

36Sefic, W.J., and Maxwell, C.M., "X-29A TechnologyDemonstrator Flight Test Program Overview," NASATM-86809, May 1986.

37perangelo, H.J., and Waisanen, ER., "Flight FlutterTest Methodology at Grumman," in "Flight TestingTechnology: A State-of-the-Art Review," Society of FlightTest Engineers, 13th Annual Symposium, New York,September 19-22, 1982, pp. 91-ft.

38Katz, H., Foppe, F.G., and Grossman, D.T., "F-15hight Flutter Test Program," in "Flutter TestingTechniques: A conference held at Dryden hight ResearchCenter," NASA SP-415, 1976, pp. 413-431.

39Allemang, R.J., and Brown, D.L., "ExperimentalModal Analysis and Dynamic Component Synthesis,"AFWAL-TR-87-3069, vol. I, Flight Dynamics Laboratory,Wright-Patterson Air Force Base, Ohio, December 1987.

16


21/22

40Cooper,J.E., "Modal ParameterdentificationTechniquesor Flight FlutterTesting,"n AeronauticalEngineering Group, Aeronautical Report no. 9105,University of Manchester, Manchester U.K., 1991.

41Nash, D.E., Katz, H., and Moody, W.C., "F-15 FlightFlutter Testing: Aircraft Systems and Test Operations,"AIAA Paper no. 75-1031, AIAA 1975 Aircraft Systems andTechnology Meeting, Los Angeles, California, 1975.

42Kehoe, M.W., "KC-135 Winglet Flight Flutter TestProgram," AFFTC TR-81-4, Air Force Flight Test Center,Edwards Air Force Base, California, June 1981.

43NASA Langley Research Center, Proceedings of the1958 Flight Flutter Testing Symposium, NASA SP-385,May 1958.

44National Aeronautics and Space Administration, FlutterTesting Techniques, NASA SP-415, October 1975.

45Lenz, R.W., and McKeever, B., "Time Series Analysis inFlight Flutter Testing at the Air Force Flight Test Center:Concepts and Results," in "Flutter Testing Techniques:A conference held at Dryden Flight Research Center,"NASA SP-415, 1976, pp. 287-317.

46Hurley, S.R., "The Application of Digital Computers toNear-Real-Time Processing of Flutter Test Data," in "FlutterTesting Techniques: A conference held at Dryden FlightResearch Center," NASA SP-415, 1976, pp. 377-394.

47Foughner, J.T., Jr., "Some Experience Using SubcriticalResponse Methods in Wind-Tunnel Flutter Model Studies,"in "Flutter Testing Techniques: A conference held at DrydenFlight Research Center," NASA SP-415, 1976, pp. 181-191.

48Wright, J.R., "Introduction to Flutter of WingedAircraft," von Karman Institute for Fluid Dynamics, LectureSeries 1992-01, December 1991.

49Shelley, S.J., Freudinger, L.C., and Allemang, R.J.,"Development of an On-line Parameter Estimation SystemUsing the Discrete Modal Filter," in "Proceedings of the 10thInternational Modal Analysis Conference," San Diego,California, February 3-7, 1992, pp. 173-183.

50Shelley, S.J., Freudinger, L.C., and Allemang, R.J.,"Development of an On-line Modal State Monitor," in"Proceedings of the l lth International Modal AnalysisConference," Kissimmee, Florida, February 1--4, 1993.

51Nissim, E., and Gilyard, G.B., "Method forExperimental Determination of Flutter Speed by ParameterIdentification," NASA TP-2923, 1989.

17


22/22

REPORT DOCUMENTATION PAGE Form ApprovedOMB No. 0704-0188Puiolic reportng burden for this collection of information is estimated 1o average 1 hour per response, including the time for reviewing instructions, searching exlshng data sources_gathering and maintaining the data needed, and completing and reviewing the collection of informat on. Send comments regarding this burden estimate or any other aspect of thiscollection of information, including suggestions f or r ed uc in g t hi s b ur de n, to Wash in gto n H ead

historical overview of flutter

Documents