AIAA 2003-0749

DEVELOPMENT OF A FREE-TO-ROLLTRANSONIC TEST CAPABILITY

(Invited)

F. J. Capone, D. B. Owens, and R. M. HallNASA Langley Research CenterHampton, Virginia

41st AIAA Aerospace Sciences

Meeting and Exhibit

6-9 January 2003Reno, Nevada

For permission to copy or republish, contact the American Institute of Aeronautics and Astronautics

1801 Alexander Bell Drive, Suite 500, Reston, VA 20191

Development of a Free-To-Roll TransonicTest Capability

F. J. Capone t, D. B. Owens, and R. M. Hall*NASA Langley Research Center, Hampton, Virginia

AIAA-2003-0749

ABSTRACT

As part of the NASA/Navy Abrupt Wing Stall Program, a relatively low-cost, rapid-access wind-tunnelfree-to-roll rig was developed. This rig combines the use of conventional models and test apparatuses toevaluate both transonic performance and wing-drop/rock tendencies in a single tunnel entry. A descriptionof the test hardware as well as a description of the experimental procedures is given. The free-to-roll testrig has been used successfully to assess the static and dynamic characteristics of three differentconfigurations--two configurations that exhibit uncommanded lateral motions, (pre-production F/A-18E andA V-8B), and one that did not (F/A-18C).

SYMBOLS AND ABBREVIATIONS INTRODUCTION

AWS

CA

C,

CN

Cm

Ca

Cy

FTR

LERX

M

t

(_te

16-ft -I-I-

abrupt wing stall

axial force coefficient

rolling moment coefficient

normal force coefficient

pitching moment coefficient

yawing moment coefficient

side force coefficient

free to roll

leading edge root extension, percent

Mach number

time

trailing edge flap angle, deg

model roll angle, deg

model pitch angle, deg

16-Foot Transonic Tunnel

tSenior Research Engineer,

Configuration Aerodynamics Branch§Research Engineer*Senior Research Engineer,Configuration Aerodynamics Branch, Member AIAA

This material is declared a work of the U. S.Government and is not subject to copyrightprotection in the United States.

A joint NASA/Navy Abrupt Wing StallProgram (AWS) was established after several pre-production F/A-18E/F aircraft experienced severewing-drop motions during the development stage.A Blue Ribbon Panel determined that a poorunderstanding of these phenomena existed andmade the recommendation to: "Initiate a national

research effort to thoroughly and systematicallystudy the wing drop phenomena." The problemarea addressed by the AWS Program 1-17 is theunexpected occurrence of highly undesirablelateral-directional motions at high-subsonic andtransonic maneuvering conditions. One of therecommendations from reference 2 was to

"develop a relatively low-cost, rapid-access wind-tunnel approach that combines the use ofconventional models and test apparatuses toevaluate both transonic performance and wing-drop tendencies (using free-to-roll approach) in asingle tunnel entry."

The overall objective of free-to-roll (FTR)testing is to identify early the potential ofexistence of uncommanded lateral motions.

Together with the force and moment data, free-to-roll testing can: determine the severity of modelmotions; assess the impact of unsteady andnonlinear aerodynamics (rate and amplitude); anddetermine dynamic aerodynamic data (rolldamping). Figure 1 shows a schematic of the free-to-roll test setup. With the model given a degreeof freedom in roll, kinematic variations of angles ofattack and sideslip occur during the rollingmotions.

The free-to-roll test technique has beenused for subsonic studies for over 30 years. Forexample, a research effort by NASA-Langleyutilized free-to-roll testing to evaluate the F-4 wingwing-rock behavior 2,18. For this investigation, themodel used a bearing and a dummy force balance

AIAA-2003-0749

to providea singledegreeof freedomin roll.Northrop and the Ames ResearchCenterconductedstaticandsemi-free-to-rolltestsonaF-5A model19to studywing-rockmotionsseeninflight. Forthese tests,a specialstingwithatorsionalspringandvariabledamperwasused.NASA-Langleyhas used single-degree-of-freedomfree-to-roll wind tunnel tests tosuccessfullypredictrollingmotionsat lowspeedandhighanglesof attackLow-speedtestingoftheX-29forwardsweptwingaircraft2°discoveredwingrockbeforeflight.Thisenabledchangestobemadetotheflightcontrolsystemto handlethewingrock.Otherpublishedfree-to-rollstudiesbyLangleyincludeanassessmentof theeffectsoffuselageforebodygeometryon stabilityandcontrol_ andwingrockcharacteristicsof slenderdeltawings22.

As partof theAWSprogram,pathfindertestswereconductedin the LangleyTransonicDynamicsTunnelon a 9-percentlightweightcompositemodelof the F/A-18E(fig.2). Theobjectivesof this investigationwereto:measurestaticforcesandmomentswithatraditionalforcebalance;conductforcedoscillationtestsusingarolloscillationbalance;anddo a free-to-rolltestwithoutaforcebalance.Thistestusedasmuchexistinghardwareas possible.The lightweight,lowroll inertiamodelwas requiredbecauseoflimitationsof the rolloscillationbalance.Sometypicalresultsfromthatinvestigationaregiveninfigure3. For8= 7°,themodelwasdampedwithonlyminordeviationsinrollingmotioncausedbytunnelturbulence.At8= 7.5°, intermittentlarge-amplitudewing-dropeventsoccurred.At _= 8°,themodelexhibitedlarge-amplitude(_+40°) rollingmotions.Ingeneral,FTRactivitywasobservedinregionswhereseverebreaksin theliftandrollingmomentcurvesoccurred.

Based on these encouraging results, adesign effort was initiated to develop a free-to-rolltest rig that could handle a variety of conventionalwind tunnel models in a transonic tunnel. The

AWS program identified three models for whichfree-to-roll testing was desired. These includedthe AV-8B, F-18C, and the preproductionF/A18E. These models had wing areas thatranged from 1.33 to 5.18 sq ft and had weightsthat ranged from 55 to 490 Ibs. This paper willdiscuss the requirements imposed on the newfree-to-roll rig as well as describing the hardwarearrangement. In addition, the safety analysis,experimental setup and testing performed duringcheckout phases without a model will bediscussed. Results of tests of the three

configurations are presented in references 9, 10and 14.

In order to obtain approval for releasingthis paper to the public, quantitative informationhas been removed from most vertical scales as perguidelines from the Department of Defense.

FREE-TO-ROLL DESIGNREQUIREMENTS

A major objective of the AWS project wasthe desire to conduct free-to-roll testing on threedifferent fighter aircraft configurations of varyingsizes and weight as shown in figure 4. Thesketches of the models are to the same scale in

order to convey the relative model sizes used inthe tests. Two of these configurations wereknown to have "wing drop" problems and theother two did not. The Langley 16-Ft. TransonicTunnel was chosen because its large size permitstesting of a wide range of model sizes.

In order to achieve this objective, thefollowing requirements were imposed on thedesign of the free-to-roll rig:

The rig had to meet all Langley safety andLAPG 1710.15 design requirements _.

Must be able to use existing industry highstrength models for testing at higher Reynoldsnumbers. Thus, no separate model would needto be fabricated.

Must be able to handle wind tunnel models

of varying sizes and weights.The force balance was to be retained and

used during testing.

No set time required removing the forcebalance to install the free-to-roll rig or to test inthe free-to-roll mode.

In addition, the following mechanicaldesign requirements were also specified:

The free-to-roll rig would replace thestandard sting butt in the 16-Ft TransonicTunnel

The maximum normal force would be 4000Ibs.

A high-response device was needed tomeasure model roll angle.

Some means of fixing the rotatingmechanism to the fixed hardware so that

conventional static testing could be done.A mechanical hard stop with damping.An independent braking system.A counterweight to adjust the model c.g. to

the axis of rotation.

Remotely actuated fins were desired fortrimming the model

2American Institute of Aerodynamics and Astronautics

AIAA-2003-0749

FREE-TO-ROLL DESIGN

Description of Free-To-Roll Rig

The free-to-roll rig was designed toreplace the standard sting butt 24 in the 16-Ft.Transonic Tunnel such that a model could freelyrotate about the longitudinal axis. A sketchshowing the rig installed in the wind tunnel isshown in figure 5. A cross-sectional sketch of therig is shown in figure 6 and a solid body andassembly representation of the mechanism ispresented in figure 7. Since the free-to-rollreplaced the standard sting butt, its the overalllength was 33.93 inches and maximum diameterwas 13.47 inches.

As shown in figure 6, the rig consists of arotary head that is supported in a stationary headby a spherical roller bearing, and an aft needlebearing. This arrangement allows thermalexpansion and some shaft flexibility withoutbinding on the bearings and also allows significantloading capability in excess of currentrequirements. The spherical bearing has a loadrating of 102,000 Ibs and the needle bearing has aload rating of 15,600 Ibs. The rotary head is heldonto the stationary head with a large carbon steeldraw nut at the forward face of the stationary head.

The initial design of the free-to-roll rigcalled for dampened hard stops to be located at+40 °, +60 °, and +80 ° rotation from the top centeras shown in figure 7. However, during thefabrication, this was changed to a single stoplocated at the top center location as shown infigure 6. In addition, a locking bar was provided tofix the rotating head to the stationary head whenstatic testing was required.

The independent braking systemconsisted of four 24 VDC electric brakes. As

shown in figures 6 and 7, the brakes are mountedin the space between the stationary and rotaryheads. The brakes are mounted to the draw nut

via pin posts much like brake calipers onautomotive disc brakes. The brakes applypressure to a wear plate mounted on the back ofthe rotary head. Brake pressure is applied byincreasing current to the electromagnets insidethe brakes. This design is not only streamlined,but also allows the capability to provide a soft stopto the rotary head.

Four actuated fins are provided for use intrimming the model in roll. The fins were eachmounted in a pivot block that contained the finbearings, gearing and actuator. Each fin waspowered by a harmonic drive electric motor with a

feedback encoder for fin position. Each fin can beactuated independently or in unison.

An adjustable counterweight assembly asshown in figure 6 can be used to balance themodel to ensure that the model's vertical center of

gravity location is on the centerline of the supportmechanism. The assembly consisted of a verticalrod attached to which varying weights can besecured to reach the desired result.

A roll resolver was used to measure roll

angle. A 5:1 mechanical gear ratio amplifies theroll angle measurement. This roll resolver has atracking rate of 30 deg/sec.

Auto Braking System

Initially, control of the brakes was to beaccomplished with the use of a simple on/offtoggle switch. However, concerns were raisedthat very high torque loads would be imposedfrom applying the brakes during a sudden stop.While there was no indication that the brakes

would grab when applied--that is, result in largerthan expected braking torque for a given currentlevel--this risk could probably be mitigated by"ramping up" the current provided to the brakes.In order to achieve this mode of operation, acomputer-controlled auto braking system wasdesigned.

For a symmetric model, the rolling moment isexpected to be a zero under operationalconditions with no sideslip. However, experiencehas shown that large asymmetries in rollingmoments can occur under abrupt wing stallconditions. For a very large model such as the AV-8B, these asymmetries can be as high as 150 ft-Ibs of torque and are indicative of the type offorcing functions present. However, under free-to-roll conditions, the total rolling moment to bebraked by the system model is determined by thecombined inertia and aerodynamic rolling-moments. For the design of the braking controlsystem, the nominal dynamic stopping torque wasassumed to be 140 ft-lbs (at 5 rad/sec velocity)and the maximum peak torque was 260 ft-lbs.

Brake Control System. The brake control

system was designed to both deploy the brakes ata specific roll angle and then hold the model atsome position in presence of a rolling torque.Hence an appropriate algorithm for braking isnecessary. Initially, the brakes are deployed at anominal braking current of 54-percent when somepredetermined roll angle is reached and then thebrake current is ramped up to a full 100-percent.The brake control law is only dependent onmeasured roll angle e, and does not account for

3American Institute of Aerodynamics and Astronautics

AIAA-2003-0749

roll rate changes that can occur when the model isexposed to unanticipated roll accelerations fromunsteady flow conditions at high angles of attack.

A schematic of the computer control used isshown figure 8. A PC computer is used forimplementing the control law. This system

Provides 28V 400 Hz drive to the rollresolver

Accepts the resolver signal conditioninginstrument on an ISA bus

Analyzes roll angle and roll rate datacontinuously

Estimates the braking current required forthe given roll conditions

Drives the brakes through the poweramplifier

Provides a throughput of above 30-40Hzclosed loop for braking current

The response time of the brake is afunction of the gaps, the coil inductance and itsresistance. From available data, the time constantis expected to be about 10-millisecond. Thisdictates that the bandwidth of the power amplifierdriving the coils must be at least 0-100 Hz. Hencethe choice of an amplifier is a linear power amplifierand not a switching mode amplifier. A fourquadrant linear power amplifier with a 0-100 Hzbandwidth that can provide a current of 2 ampspeak was chosen for the brake coil drive.

Model roll angle is determined with a rollresolver with an accuracy of 7 minutes (gear rationot accounted for). A "resolver card" in the brakecomputer determines the final accuracy. This"resolver card" provides this excitation voltage aswell as reading the roll angle signal. The overallaccuracy when combining the accuracy of theresolver card, the resolver, and the 5:1 gear ratio is4 arc minutes or 0.067 degrees. The resolver cardwas a custom made piece because we had tohandle the resolver shaft rotating at up to 5000deg/sec. The resolver itself has no speedlimitation only the resolver card in the brakecomputer. Roll angle can be resolved to 0.067degrees at FTR rig rates up to 1000 deg/sec.Since the amount of angle estimation on theresolver signal conditioner is high, a digital signalprocessor based PC computer ISA buscompatible instrument will be required. A gearratio of five exists between the resolver roll angleand model roll angle.

Control Interface. The computer

program was written with a graphics mode screen,which shows the system schematic (fig. 9). Thegraphic displayed on the computer screenrepresents an animated picture of the status of the

free-to-roll rig that shows the roll attitude of themodel in real time with an update rate of higherthan 30 frames/sec, which is acceptable from apersistence of vision point of view. The displayalso shows the various other elements of thesystem such as the raw resolver angle, brakecurrent percent, and the status of mode control.The control shows three modes of operation:

Auto mode: In this mode, the model angle e iscompared with the set el, e2 and based on thecontrol logic; a current is imposed on the brakes asper control law.

Manual Mode: Once on manual mode, thereare two choices available, the brake-off mode orthe brake-on mode.

Pulse mode: The pulse mode works while onManual mode and, in the Automode only whenthe brakes are ON. The pulse mode, momentarilyswitches the brakes ON or OFF, to a stateopposite of existing state.

Reliability and Safety Issues. Fromthe onset of the design of the computer-controlled braking system, the central issue for theFree-To-Roll rig was reliable, accurate and quickdetermination of the model angle, and then whennecessary, impose the desired brake currentbased on the prescribed control law. With thefollowing precautions, the reliability of the logicpart of the system was high and was onlydependent on the quality of brake and itsresponse

A dedicated resolver should be used for the

brake system. Sharing resolver signals withother data devices would have created issues of

reliability. However, this did not turn out to be apractical solution and in the end, signal sharingwas used.

The excitation to the resolver was internallygenerated by the computer and hence provideshigh reliability as long as computer was ON

The computer program was a single taskprogram and can run as long as computer haspower.

The brake amplifier was an industrial classlinear amplifier with known reliability.

The brake power supply was an industrialclass switching mode device with high reliability.

The computer power and brake power wasprovided from an uninterrupted supply at the16-Ft Transonic Tunnel.

Protected wiring from resolver, to the brakepower system was provided.

The computer was prevented frominadvertent reset by locking the power ON.Provision for keyboard lock was also desirable.

4American Institute of Aerodynamics and Astronautics

AIAA-2003-0749

SAFETY ANALYSIS

Required Safety Analysis

All models tested in various wind tunnelsat the Langley Research Center must adhere tothe strict requirements set forth in reference 23.This guide contains criteria for the design,analysis, quality assurance and documentation ofwind-tunnel model systems to be tested. A wind-tunnel model system includes model supporthardware including force balances and stings. AModel Systems Engineer serves as the residentexpert for the review of the model systems designand analysis. His findings are reported to theFacility Safety Head who is the final approvalauthority for all models to be tested in the facility.Since these requirements are incorporated in theoverall Langley Safety manual, further reviews by asafety engineer may be required, which was thecase for the free-to-roll rig.

All the parts of the free-to-roll rig met therequired safety factors of 4 on ultimate and 3 onyield. Both the stationary head and rotating headwere analyzed using hand calculations and finiteelement analysis.

All model systems are designed for asystem failure event, that is, the design shall besuch that after an initial failure, the model shall notexperience any further failure that would causefacility damage during the tunnel shutdownprocess. Under this requirement, a failure of theindependent braking system was not considereda safety issue, since any rotation of the modelwould be halted with the mechanical hard stop. Aconcern was raised concerning impact loading onthe force balance and an attempt was made toremove it. However, in the absence of anyanalysis, the safety engineer wanted the hard stopto be retained even though having the modelessentially spinning was not considered a safetyissue.

It should be noted, that additional stressanalyses are required for the wind-tunnel modelitself to account for inertial loads that would be

imparted to key model components due to highaccelerations in roll if the model is experiencingwing drop or rock.

Brake Analysis

The braking analysis performed to assurethat the brakes could arrest the model assumed

(1) a worst-case forcing function, (2) took intoaccount that the braking effectiveness was afunction of time, and (3) integrated the resultingequation of motion about the body axis to predict

a final bank angle, e, at which the model wouldhave come to rest.

The first task was to assume a veryconservative forcing function that would modelthe acceleration the model might experiencewhen it was in a wing-drop region. The basis forthis assumption was data from the proof-of-concept experiment in the TDT. The data areillustrated in figure 10 and relate the rollacceleration of the model, d2e/dF, as a function ofe. For analysis purposes, it is assumed that themodel only experiences an aerodynamicpropelling function for -20 ° < e < 20 °. The forcingfunction in the data of figure 10 is seen to actbetween 8 ° and 30 ° of e and spikes to a value of2500 deg/sec 2 for d2e/dF. To be on theconservative side, the notional forcing function forthe analysis was assumed to occur over a range ofbank angles between 0 ° and approximately 45 °and to be of a magnitude of between 4300 and4900 deg/sec 2, depending on Mach number.This region is approximated by the dotted line infigure 10 and illustrates the extremelyconservative characterization for the forcingfunction.

The second task was to characterize thebrake effectiveness as a function of time. The first

model was assumed to be a ramp function, whereat some preset bank angle, e, the brakes wouldinstantly have 50% of the assumed maximumbraking power. This power would then linearlyramp up until it reached 100% of assumedmaximum braking power. A second model waslater employed when it was realized that thebrakes took a finite time to energize as the coil wascharged. Both models are graphically described infigure 11.

Results for both braking models areillustrated in figure 12. The top two figures are forthe coil brake effectiveness model. The model

being used for this example was the AV-8B. Withthe assumed forcing function of 4300 deg/sec 2,the model can achieve a roll rate of 120 deg/secbefore the forcing function ceases near e = 45 °.With the brakes being actuated at e = 50 °, thecalculation predicts a value of e for model at rest ofjust over 160 °. Using the ramp-braking modelpredicts a lower stopping rotation of less than140 °.

These calculations, which assumed idealbrake effectiveness, were not intended to beprecise predictions of what would happen duringthe experiment. They were intended to beengineering estimates that would show that thebrakes could safely arrest the model before it

5American Institute of Aerodynamics and Astronautics

AIAA-2003-0749

would reach bank angles beyond 180 °, which heldtrue during the experiments.

One result indicated by the brake analysiswas that if a model was halted by the mechanicalhard stop, the impact load was such as to overloadthe roll component on the force balance. Inaddition, there was some question how to analyzethe loads imparted to the model due to a suddenacceleration. With this new information, a newassessment of the hard stop was made and adecision was reached to remove it. The only riskidentified to a model spinning would be twisting ofthe various instrumentation leads with a possibilityof the leads breaking. This also was notconsidered a safety issue.

Risk Analysis

Both NASA and NAVAIR personnel notassociated with the required analysis previouslydiscussed made an independent risk assessment.The initial assessment, shown in figure 13,indicated three high and three medium risk itemspresent. It is very unlikely that a Facility SafetyHead would allow any testing to occur with anyitem identified as a high risk. After carefulconsiderations, all the high and medium risk itemswere downgraded to a low risk status.

The first item was eliminated with the

decision to remove the hard stop. The seconditem was downgraded because of the extra stressanalysis required to account for inertial loads. Inaddition, roll Ioadings are continuously monitoredduring testing by safety systems in the 16FTT. Ifany load or combination of loads exceeds 100% ofbalance/model limits, testing is automaticallyterminated at that condition. The third and sixth

risks were addressed by conducting a groundvibration test rap assessment before FTR testingfor each of the 4 configurations. Frequencieswere monitored during testing to ensure that nocoalescing would take place. Again, there is inplace at the 16F-I-I-, a safety system that had thecapability to not only monitor these variousfrequencies, but also to automatically reducetunnel speed in the event of an occurrence. Avery stringent analysis is already performed on anymodel support system as part of the requiredsafety analysis. There little likelihood ofoverstressing either the FTR rig or sting becauseof the safety monitoring systems used. The fifthrisk was eliminated because of the installation of

the automatic braking system. Figure 14 showsthe resulting risk assessment after the riskmitigations.

SYSTEM CHECKOUT

Extensive pretest testing of the free-to-roll rig was performed in the model build up bay atthe 16-Ft. Transonic Tunnel without a modelmounted to the rig. Functional tests of the finactuation system, brakes and the automaticbraking system were performed. Sting bendingcharacteristics of the rig with the locking barattached were done and these results were

compared to those measured previously with thesting mounted to the standard sting butt. Onevery important part of these tests was to help indeveloping test techniques and to give theoperators training in the use of the rig.

During this time, it was found that thebrakes did not completely disengage when thecurrent to the brakes was turned off. As a result,sets of external springs were installed to pull thebrakes back when they were not actuated.

In addition, system friction characteristicswere measured wind-off, and wind-on fin tareswere determined. Measurements were made to

determine the friction from the bearings and anyresidual brake contact with the rotor. It is essential

to know the magnitude of the friction momentsince this moment can be a source of error when

determining the roll damping for the variousmodels. Friction is determined by using the rollangle time history resulting from a release of thecounter weight from an initial angulardisplacement. Ideally, one would want to load therig to those exhibited by the model in the windtunnel. However, no safe way was beendetermined to measure the friction with the

bearings loaded at the wind-on forces generatedby the models.

Estimates of rolling friction from the FTRrig were made by pendulum tests with no modelprior to testing. Additional estimates were madeduring wind tunnel tests with another model bymeasuring the rolling moments from an internallymounted balance, and comparing those outputswith the calculated rolling moments based on rollacceleration of the model. These estimates

showed some variation of friction between runs,and some differences with loads on the rig. Theestimates across the range of conditionsevaluated resulted in friction factor ranging fromapproximately # = 0.25 to 0.65 ft-lb/rad/sec.

A wind-on fin checkout was conducted todetermine roll control effectiveness from the fins

and to ascertain if any uncommanded fin behaviorsuch as fin movement when no control inputs aremade or fin "buzz" was present.

6American Institute of Aerodynamics and Astronautics

AIAA-2003-0749

Since aerodynamic moments generatedby fins represent a source of error, an assessmentof this error was made. With the rig andcounterweight alone on the strut, the rig wasdisplaced to some initial roll angle and held inposition by the brake. After the brakes arereleased, the roll angle time history of the resultingmotion is analyzed to determine the rollingmoment generated by the fins. This was repeatedat various Mach numbers, sting angles and findeflection angles.

There were two major findings of thesetests: 1.) The damping from the fins was large andincreased with Mach number; and 2.) During someof these damping runs, the fins would sometimesgo in the wrong direction or not damp out in asinusoidal fashion, i.e. a meandering or thedecelerations were not smooth. As a result of

these finding, it was decided to remove the finsfrom the free-to-roll rig and proceed to testing thethree models.

EXPERIMENTAL PROCEDURES

Static Force Testing. A block diagramof the experimental setup is presented in figure15. The unfiltered signals from the force balanceand other model instrumentation are sent to the

wind tunnel data acquisition system (NEFF), ananalog amplifier-conditioning unit (analog-to-digitalconverters). The NEFF filters and amplifies theanalog signal and passes it on to the wind tunnelcomputer (ModComp).

This NEFF unit also sends unfiltered and

buffered signals on to other devices such as theRMS converters, the Balance Dynamic DisplayUnit (BDDU) and the Model Protection SafetySystem (MPSS). The RMS converters generate aDC signal representing a RMS value of theunfiltered signal that is sent back to the NEFF. TheNEFF then filters and amplifies the signal andsends the filtered signal to the Modcomp. Thefiltered signals are then recorded by theModComp depending on the mode operation.

The BDDU normalizes and multiplexes theunfiltered and buffered signals from the NEFF sothat the displayed output on an oscilloscopedenotes percent of full-scale design loads in sixsequential horizontal locations. Two-level visualand audible alarms are incorporated to indicatewhen a signal exceeds 70% and 100% of thestatic plus dynamic design load. This systemrequires some predetermined operator actionwhen 100% design load is reached.

The Model Protection Safety System(MPSS) monitors corrected balance component

inputs to identify excessive loads on the balance,model or sting. This system gives a continuousdisplay in percent of the static plus dynamicdesign loads. The MPSS allows for a number ofconfigurable load limits including combinations ofthe forces and moments. If one of the limits are

exceeded or the balance voltage is out oftolerance the MPSS sounds an audible alarm and

begins driving the tunnel fan speed down toreduce the tunnel dynamic pressure

Free-To-Roll Testing. During free-toroll testing, all of the systems just described arealso used. Since free-to-roll testing is inherentlytime history in nature, the free-to-roll testingrequires three additional forms of "data"acquisition. One form is the videotaping of therolling motions of the model. Overlays on thevideo of tunnel conditions, run numbers, etc. anda flag of when roll angle time histories are beingacquired by the data acquisition make postprocessing and syncing of the video to roll angletime histories possible. The second form is thedigitizing of the roll angle signal. A computer inthe control room where the engineer can quicklyanalyze the signal for frequency, amplitude, rollrates, and roll accelerations content processesthis signal immediately. This information is thenfed to the final form of "data" acquisition. This finalform of data acquisition is written "pilot" commentsin the form of a run log. These comments can beverbally recorded on the videotape as well. Thiswritten run log allows a quick assessment of thenature of the rolling motions during the free-to-rollphase.

Modes of Operation. There are twomodes of operation with the free-to-roll rig. First,conventional static testing is accomplished byhaving the locking bar in place. The signals fromthe NEFF are passed through 1 Hz filters and datais then recorded for 5 seconds at 10

frames/second. Model roll angle in this mode isdetermined from the standard instrumentation

located within the model strut support system.Testing in the static mode is then conductedusing the standard procedures in place at the 16-Ft. Transonic Tunnel.

For free-to-roll testing, the locking bar isremoved. Prior to starting the wind tunnel, thebrakes are deployed with the model at roll angle of0°. In this mode, the signals from the NEFF arepassed through 10 Hz filters and data is thenrecorded continuously at 100 frames/second.Model roll angle is determined from the rollresolver. Three separate test techniques areemployed during FTR testing. First the model ispitched continuously from some low pitch angle tothe maximum angle desired. Next, a series of

7American Institute of Aerodynamics and Astronautics

AIAA-2003-0749

pitch-pause tests are conducted. With the brakesset, the model is pitched to the desired pitchangle and then the brakes are released. For thethird technique, the model is displaced to someinitial roll angle and held in position by the brake.After the brakes are released, the roll angle timehistory of the resulting motion is analyzed todetermine the roll damping characteristics of themodel. Further details of these techniques as wellas results of the tests conducted on the threewind tunnel models can be found in references 10and 14.

One of the major requirements in thedesign of the free-to-roll rig was the force balancewas to be retained and used during testing. Therewere risks involved with this requirement but it wasfelt that the having the force balance would proveto be invaluable in particular from a safetystandpoint in being able to monitor model loadsduring FTR testing. The risk posed by damage tothe balance has been discussed previously.There was always a concern that the forces andmoments measured during free-to-roll testingwould be erroneous. However, this turned outnot to be the case. Comparisons were made ofthe measured forces and moments between FTR

and static testing for different cases when a modeldid or did not exhibit any free-to-roll activity.Although not shown, excellent agreement wasfound to exist for the longitudinal data for the twotest modes. A further discussion of these resultscan be found in reference 14.

ACCOMPLISHMENTS

A new free-to-roll test rig has beendesigned and built for transonic testing. This is anational test asset that can be used to determine

potential uncommanded lateral motions early inthe design stage. The free-to-roll test rig hasbeen used successfully to assess the static anddynamic characteristics of three differentconfigurations (fig. 16).

During these tests, the safety procedures,general test approach and interpretation of resultshave been very successful.

ACKNOWLEDGEMENTS

The authors would like to thank Mr. Joe

Chambers for motivating the use of the free-to-rolltechnique in the transonic speed regime. Mr.Sundareswara Balikrishna of Vigyan, Inc.designed the computer-controlled automaticbraking system.

REFERENCES

1. Hall, R. M.; and Woodson, S: Introduction tothe Abrupt Wing Staff (AWS) Program. AIAA-2003-0589, January, 2003.

2. Chambers, J. R.; and Hall, R. M.: HistoricalReview of Uncommanded Lateral-Directional

Motions at Transonic Speeds. AIAA-2003-0590. January, 2003.

3. McMillin, S. N; Hall, R. M.; and Lamar, J. E.:Understanding Abrupt Wing Stall withExperimental Methods. AIAA-2003-0591,January, 2003.

4. Woodson, S.; Green, B; Chung, J.; Grove, D.;Parikh, P.; and Forsythe, J.: UnderstandingAbrupt Wing Staff (AWS) with CFD. AIAA-2003-0592, January, 2003.

5. Schuster, D.; and Byrd, J.: TransonicUnsteady Aerodynamics of the F/A-18E atConditions Promoting Abrupt Wing Stall.AIAA-2003-0593, January, 2003.

6. Forsythe, J.; and Woodson, S.: UnsteadyCFD Calculations of Abrupt Wing Staff UsingDetached- Eddy Simulation. AIAA-2003-0594, January, 2003.

7. Parikh, P; and Chung, J.: A ComputationalStudy of the A WS Characteristics for VariousFighter Jets: Part I, F/A-18E & F-16C. AIAA-2003-0746, January, 2003.

8. Chung, J.; and Parikh, P.: A ComputationalStudy of the Abrupt Wing Staff (AWS)Characteristics for Various Fighter Jets: Part II,AV-8B and F/A-18C. AIAA-2003-0747,January, 2003.

9. Lamar, J. E.; Capone, F. J.; and Hall, R. M.:AWS Figure of Merit (FOM) DevelopedParameters from Static, Transonic ModelTests. AIAA-2003-0745, January, 2003.

10. Owens, B; Capone, F. J.; Hall, R. M.; Brandon,J; and Cunningham, K: Free-To-Roll Analysisof Abrupt Wing Staff on Military Aircraft atTransonic Speeds. AIAA-2003-0750,January, 2003.

1 1 .Roesch, M, and Randall, B.: Flight TestAssessment Of Lateral Activity. AIAA-2003-0748, January, 2003.

12.Green, B; and Ott, J.: F/A-18C to E WingMorphing Study for the Abrupt Wing StaffProgram. AIAA-2003-0925, January, 2003.

8American Institute of Aerodynamics and Astronautics

AIAA-2003-0749

1 3.Kokolios, A.; and Cook, S.: Use of PilotedSimulation for Evaluation of Abrupt Wing StallCharacteristics. AIAA-2003-0924, January,2003.

14. Capone, F. J.; Hall, R. M.; Owens, B.; Lamar,J. E.; and McMillain, S. N.: RecommendedExperimental Procedures for Evaluation ofAbrupt Wing Staff Characteristics. AIAA-2003-0922, January, 2003.

15. Woodson, S.; Green, B.; Chung, J; Grove, D.;Parikh, P; and Forsythe, J.:Recommendations for CFD Procedures forPredicting Abrupt Wing Stall (A WS). AIAA-2003-0923, January, 2003.

16.Cook, S.; Chambers, J.; Kokolios, A.;Niewoehner, R.; Owens, B.; and Roesch, M.:An Integrated Approach to Assessment ofAbrupt Wing Stall for Advanced Aircraft. AIAA-2003-0926, January, 2003.

17. Hall, R. M.; Woodson, S.; and Chambers, J.R.: Accomplishments of the AWS Programand Future Requirements. AIAA-2003-0927,January, 2003.

18. Chambers, Joseph R.; and Anglin, Ernie L.:Analysis of Lateral-Directional StabilityCharacteristics of a Twin-Jet Fighter Airplaneat High Angles of Attack. NASA TN D-5361,1969.

19.Hwang, C.; and Pi, W.S.: "Investigation ofSteady and Fluctuating Pressures AssociatedWith the Transonic Buffeting and Wing Rockof a One-Seventh Scale Model of the F-5A

Aircraft." NASA Contractor Report 3061,November, 1978. NOTE: Subsequentlysummarized, reduced, and published as:"Some Observations on the Mechanism ofAircraft Wing Rock." AIAA Paper 78-1456,AIAA Aircraft Systems and TechnologyConference. Los Angeles, CA. August 21-23,1978.

20. Murri, D. G.; Nguyen, L. T.; and Grafton, S. B.:Wind-Tunnel Free-Flight Investigation of aModel of a Forward-Swept Wing FighterConfiguration. NASA TP-2230, February1984.

21. Nguyen, L. T.; Yip, L.; and Chambers, J. R.:Self-Induced Wing Rock of Slender DeltaWings. AIAA-81-1883, August 1981.

22. Brandon, J. M.; Murri, D. G.; and Nguyen, L.T.: Experimental Study of Effects of ForebodyGeometry on High Angle of Attack Static andDynamic Stability and Control. ICAS, 15 thCongress Proceedings, vol. 1, pp 560-572.September 1986.

23. Wind-Tunnel Model Systems Criteria. NASALAPG 1710.15, September 27, 2000.

24.Capone, Francis J.; Bangert, Linda S.;Asbury, Scott C.; Mills, Charles T.; and Bare,

E. Ann: The NASA Langley 16-FootTransonic Tunnel- Historical Overview, FacilityDescription, Calibration, Flow Characteristics,Test Capabilities. NASA TP-3521, 1995.

gAmerican Institute of Aerodynamics and Astronautics

.f

:+:7_ m . m

.!

Wings Level¢=0 °a=O

/3=0 °

0 Rolling Right¢>0 °a<O

/3<0 °

Figure 1. Notional sketch of the free-to-roll test technique.

iiiiiiiiiiiiiiiiiiiii :::::: ::

A

Aceelerometers

Figure 2. Photograph of the pre-production F/A-18E model

Tested In the Langley Transonic Dynamics Tunnel.

10

8~ 7°Damped motions

CL

8~7.5 °Intermittent wing drop

8, deg

Figure 3. Typical test results from

iii -:i?

:_:.i_!!!ii_ii_i!iiiiii....!iiiiiii_.......

8~8 oLarge amplitude wing rock

iiiii -_FTR activity observediiili .

Ithe Transonic Dynamics Tunnel.

iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

o

iiiiiiiiiiiiiiiiiiiiiii_iiiiiiiiiiiiiiiiiiiiiiiii

iiiiiiiiiiiiiiiiiiiii!_!i_iiiiiiiiiiiiiiiiiii

iiiiiiiiiiiiii!,iiii i! ;;iiiiiiiiiiiii

iiiiiiiiiiiiiiiiiiiiiiiiii_ili!iiiiiiiiiiiiiiiiiiiiiiiii_

AV-8B F-18E/F F-18C F-16CLength, in. 81.27 54.99 39.18 37.25

Wing Area, ft2 5.18 3.2 1.33 1.67Span, ft 4.55 3.34 2.25 2.07

Weight, Ibs 490 185 55 56

Inertia, slugs-ft2 4.0 1.2 0.2 0.2Scale, % 15 8 6 6.67

Figure 4. Geometric characteristics of the four models tested.

11

Tunnel Top Wall

_'_Free-to-roll rig replaces 16-ft TT

standard sting butt

- _ Tunnel _. ----7

.........................................................._)_Ce ter of rotation _--33193-_

Figure 5. Sketch of free-to-roll rig installed inLangley 16-Foot Transonic Tunnel.

I

< 33.93 >1

Brake Pad- iry HeadHead Brake

Stop

13.47

Sperical Bearing Needle Bearing

ght

U

Figure 6. Cross-sectional sketch of the free-to-roll rig.

12

Sperical

Needle Bearing

"y Head

Rotary

ig Bar

Hard StopLocation

Counterweight

Figure 7. Solid body representative of the free-to-roll rig.

115 V.._.

iPowersu YlI115 VAC.,._ 5 amps max .__f

Power Supply f

2 amps max

PC Computer, Kbd, Monitor

ISA Bus

I

Data Translation

Processor

12 bit command

0-5 VDCto brake

Power amplifierDC to 5 kHZ bandwidth

0 to 2 amps

current monloring

4 65 ohm coils

I

Digital SignalProcessor

)k

ResolverGear Ratio 5

I

Figure 8. Schematic of computer-controlled automatic braking system.

13

Figure 9. Braking system display monitor.

Forcing function 5000restoring

2500

b2_ / bt2, 0deg/sec2

-2500

Forcing function _

propelling -5000-80

_lmb-

(-

-60 -40 -20 0 20e, deg

..,_,, Forcing function.j,,_- propelling

_.. _ V_ _ "--__Assumed forcing_ _ _v,,_ function propelling

V _ " Forcing functionrestoring

40 60 80

Figure 10. Typical forcing function for brake-stopping analysis.

Ramp AssumptionI, max

I, full

c_o-- I, on

.¢L--"mO

Coil Delay 1oo

E° 80 f-o ,/"6 60 / (1 - e-t/0"33 __o_ jr

40 /o r

_° 20 /

0 ,_ 00 _, on _, full 180 rn 0 20 40 60 80

_, deg Time, milliseconds

Figure 11. Brake effectiveness as a function of time.14

100

10000

Ramp Assumption

300 200

5000 I..... 225 _ 160

_,deg 120_/_, o 15odeg/sec 2 • _._ 80

-5000 75

-100000 .1 .2 .3 .4

Time, sec

10000

m j5000 3 /

deg/sec2 I

-5ooo I

-lOOOO I0 .1 .2 .3 .4

Time, sec

40 j

o _0

Coil Assumption

, ,1(p J \• ,it

i.1 .2 .3

Time, sec

I

.50

1000

8OO

600 6_/6t

400

200

0.4 .5

300 200 1000

160 • _ " '800225 ._

_- _, deg 120 /. -,, _# 600 6_16t

150 _ 80 r /,, 400

S J •• _r *" 20075 rn 40 / • *.

O0"_f • 0.1 .2 .3 .4 .5

Time, sec

Figure 12. Typical brake performance.

l. Model/balance damage due to ihitting the hard stop

(brake failure)

Highly likely-or- > 5

no confidence in

predictive capabilities4

oo

J_= 3

._1

Not likely-and- _ 1

high confidence in

predictive capabilities 1 2

co

_ Overstress of model/balance Idue to AWS-induced

accelerations

3. Interactions between model 1

dynamics and FTR rig

4. Overstress of FTR rig/sting

5. Model/balance damage

due to sudden/unplanned

application of brake

6. Model/balance damage

due to coupling between

pitch and roll dynamics

Minimal Severe impact

Figure 13. Initial risk assessment.

15

Highly likely

-or- _ 5no confidence in

predictive capabilities4

oo'=- 3

...i

2

Not likely-and- > 1

high confidence in

predictive capabilities

Minimal impacl

Overstress of model/balance

due to AWS-inducedaccelerations

3. Interactions between model ii

dynamics and FTR rig iii

4. Overstress of FTR rig/stingii i

Model/balance damagei

due to sudden/unplanned iapplication of brake iii

Severe impact

Figure 14. Risk assessment after risk mitigations.

I Top ViewMonitor and VCR

Side View IMonitor and VCR

IIRMS '_

C ,nverterslI IWind Tunnel Data

I Balance Dynamics I j Acquistion I_

DataUnit _" ! System I iI ModelProtection I_

Safety System

l Wind TunnelComputer

System

J End View IMonitor and VCR I

>'!

[ Roll Resolver I

I Force Balance IWing bending

Pressures

Resolver DriverInterface

H FTR DataAcquistion

System

Figure 15. Schematic of experimental setup

16

Figure 16. Photographsof the modelstested.

17