•o/j - ̂

NASA Technical Memorandum 102859

Comparisons of Elastic and RigidBlade-Element Rotor Models UsingParallel Processing Technology forPiloted SimulationsGary Hill, Ames Research Center, Moffett Reid, CaliforniaRonald W. Du Val, John A. Green, and Loc C. Huynh, Advanced Rotocraft Technology, Inc.

Mountain View, California

(MASA-TM-10P859) COMPARISONS OF ELASTIC AND N91-32071RIGID BLADE-ELEMENT ROTOR MODELS USINGPARALLEL PROCESSING TECHNOLOGY FOR PILOTEDSIMULATIONS (NASA) 19 p CSCL 018 Unclas

G3/01 0045751

June 1991

NASANational Aeronautics andSpace Administration

Ames Research CenterMoffett Reid. California 94035-1000

https://ntrs.nasa.gov/search.jsp?R=19910022757 2018-06-21T10:51:01+00:00Z

NASA Technical Memorandum 102859

Comparisons of Elastic and RigidBlade-Element Rotor Models UsingParallel Processing Technology forPiloted SimulationsGary Hill, Ronald W. Du Val, John A. Green, andLoc C. Huynh

June 1991

NASANational Aeronautics andSpace Administration

Comparisons of Elastic and Rigid Blade-Element Rotor Models Using ParallelProcessing Technology for Piloted Simulations

GARY HILL, RONALD W. DU VAL,* JOHN A. GREEN,* and LOG C HUYNH*

Ames Research Center

Summary

A piloted comparison of rigid and aeroelastic blade-element rotor models was conducted on the Crew StationResearch and Development Facility (CSRDF) at AmesResearch Center. FLIGHTLAB, a new simulation devel-opment and analysis tool, was used to implement thesemodels in real time using parallel processing technology.Pilot comments and quantitative analysis performed bothon-line and off-line confirmed that elastic degrees of free-dom significantly affect perceived handling qualities.Trim comparisons show improved correlation with flighttest data when elastic modes are modeled. The resultsdemonstrate the efficiency with which the mathematicalmodeling sophistication of existing simulation facilitiescan be upgraded using parallel processing, and theimportance of these upgrades to simulation fidelity.

Introduction

Developing the next generation of advanced rotorcraft—capable of higher speeds and greater maneuverability,with precise, superaugmented control systems and hinge-less rotors—will require expanded-bandwidth, real-timesimulations that include aeroelastics and structuraldynamics. These disciplines can no longer be left uncou-pled, and more engineers on the design staffs will needaccess to these higher-fidelity simulations. AdvancedRotorcraft Technology (ART), Inc., the U.S. Army Aero-flightdynamics Directorate, and NASA Ames's MilitaryTechnology Office have pursued parallel processing as acost-effective means of meeting this need. In a series ofresearch activities, rotorcraft models of increasing com-plexity have been parallelized for real-time operation.Under a Small Business Innovative Research Program(SBIR), a piloted workstation was developed, which wasdriven by a MicroVAX n computer convened to a paral-lel processor by adding four processor boards to theMicro VAX's backplane. On that parallel processor, theRotor Systems Research Aircraft/X-Wing (RSRA/X-Wing), with possibly the most complex aerodynamicinteractions ever modeled, was flown through conversions

Advanced Rotorcraft Technology, Inc., Mountain View,CA 94035-1000

between rotary and fixed wing (ref. 1). Later, the samemodified MicroVAX n replaced the aerodynamic rotormap with a blade-element representation on the UH-60Black Hawk training simulator at Fort Ord, California,and piloted comparisons were performed (ref. 2). Thispaper reports on an extension of this technology to ablade-element rotor model that adds aeroelastics andstructural dynamics. Again with the UH-60 as the subject,this "global" simulation model with blade flexibility wasdemonstrated in real time on two commercial parallel pro-cessor computers, and interfaced with the Army's CrewStation Research and Development Facility (CSRDF) atthe NASA Ames Research Center.

The authors would like to thank the Army Aeroflightdy-namics Directorate at Ames Research Center and, in par-ticular, the Crew Station Research and DevelopmentBranch for providing both access to the CSRDF and thenecessary support personnel to carry out this investiga-tion. They would also like to thank Alex Gallico andGuy Beauchesne of CAE Electronics for their help andsupport. -

Objectives

As a follow-on to the effort at Fort Ord (ref. 2), this studyhad four objectives. The first was to develop a free-flightrotorcraft simulation model with aeroelastic and structuraldynamics modeling; the second was to design a parallelprocessing architecture for this global model to permitreal-time operation; the third was to run the real-timesimulation interactively with a full-simulation facility; andthe fourth was to perform piloted simulations comparingthe model with conventional rigid body modeling.

Methods

In this study, the global simulation was developed andadapted for parallel processing using ART, Inc.'sFLIGHTLAB system, a tool for developing optimal paral-lel software architectures for real-time operation from amodular library of elements. FLIGHTLAB includes

1. A library of simulation elements to synthesize adesired vehicle configuration, a dynamic assembler to

set up data structures, and a solver routine forintegrating solutions.

2. An interactive programmer's workstation for devel-oping and optimizing parallel architectures by decom-posing existing codes into modules, mapping datadependencies of these modules, and profiling theirexecution times.

3. An engineer's workstation for on-line, real-time dataaccess and engineering analysis. This workstation isalso interactive, with symbolic data access, high-levelengineering analysis, and graphical display capabilities.

4. A pilot's workstation with a three-axis side arm con-troller and an out-the-window visual scene with a head-up display (HUD).

Blade-element aerodynamic rotor models divide eachrotor blade into segments and use airfoil tables to computeeach segment's aerodynamic load from the angle ofattack, Mach number, and dynamic pressure produced bylocal motion and flow. Airloads from the segments aresummed to derive forces about the blade degrees of free-dom and to calculate reactions at the hub. Both the rigidand the elastic blade modeling include hub/fuselagemotion, flapping and lead-lag hinge articulation, and bladefeathering.

This global model is a free-flight, blade-element, real-time simulation that includes component elasticity. Aero-elastics and structural dynamics are modeled by modeshapes derived from finite-element analysis. When themodal representation was added to the rigid blade-elementmodel, the number of blade elements had to be doubledfrom five segments per blade to ten in order to obtainproper high-frequency-dynamics fidelity. The globalmodel also contains hinge articulations in addition tocoupled elastic degrees of freedom.

FLJGHTLAB's free-flight rigid blade-element model wasvalidated to match the UH-60 blade-element model of theNASA version of Sikorsky Aircraft's Master GeneralHelicopter Program (GENHEL) (ref. 3). This simulationmodel of the UH-60 helicopter, obtained from Sikorsky,has been updated at NASA Ames and validated againstflight test data (ref. 4). The FLIGHTLAB aeroelastic rotormodel was validated against Lockheed's REXOR aeroe-lastic rotor model (ref. 5). and the global (free-flightaeroelastic) model was validated against available flighttest data (ref. 6). FLIGHTLAB's data-driven configuringcapability facilitated the comparative analysis and valida-tion. Modifications such as the interchange of elastic andrigid modules and altering the hinge sequence in the mod-eling would require extensive receding in most simulationmodels.

After creating and validating the global model, parallelarchitectures were designed to provide a real-time capabil-ity. FLIGHTLAB's parallelizing features determined thatan eight-node (each node is an individual CPU) parallelcomputer, each node having a CPU speed approximately17 times faster than the Micro VAX n, would suffice. Theglobal model was then benchmarked in serial mode oncommercial parallel processors of the RISC class. Speedsof up to 21 times that of the Micro VAX n were obtained;thus it appeared that real time was feasible.

The next step, that of actually running the massive globalmodel in real time, was facilitated by the cooperation ofthe computer manufacturers, who generously providedtechnical assistance and free access to their computers.The two parallel-processing computer systems used werea Silicon Graphics IRIS (SGI) 4D/280 GTX with eightprocessors, and a BBN TC2000 with sixteen processors.Identical parallel architectures were installed on both, andreal time was confirmed by timing checks and simulatedflight from the pilot workstation. Interface to theCSRDF's VAX 8650 host computer was through Ether-net. Existing visual and cockpit protocols remained intactwhile the vehicle's simulation math model, on the hostcomputer, was replaced by the math model running underFLIGHTLAB on the parallel platforms.

The complete Black Hawk simulation model was updatedevery 10° of rotor azimuth travel, which resulted in a6-msec integration cycle. Only the first two blade flappingand inplane bending modes were used in this experimentThe parallel"processing architecture as implemented onthe eight-processor SGI computer and eight of the sixteenBBN computer processors is shown in figure 1. It usedfive processors for the math model: four processors dedi-cated to the rotor (one per blade), and one processor forthe fuselage, tail, control system, and engine model. Asixth processor drove an engineer's workstation, and aseventh served the Ethernet interface to the CSRDF hostVAX 8650 at 60 Hz. Integration of the parallel processorwith the CSRDF is shown schematically in figure 2.

Pilot tasks were chosen to assess the effects of aeroelasticdegrees of freedom on perceived fidelity. Hover maneu-vers included bob up/bob down and lateral dash/quickstops. High-speed flight maneuvers included slalommaneuvers that require rapid bank reversals to controllateral position, and dolphin maneuvers with rapid collec-tive changes to control vertical position. The pilots per-formed frequency sweeps to collect engineering data.

Results and Discussion

The following analysis compares trim, stability, anddynamic responses of the rigid and aeroelastic

blade-element models. There are differences between thetwo simulation models in all three areas, especially at highspeed.

Longitudinal trim of the rigid and elastic models are com-pared with flight test data (ref. 6) in figure 3. Pitch atti-tude, longitudinal cyclic position, and collective positionare plotted for flight speeds from -50 to 160 knots. At lowspeeds, the rigid and elastic models show little differencein their trim characteristics. At hover, pitch attitude andlongitudinal cyclic position agree well, but results for bothrigid and elastic models deviate from flight test results attransition speeds. One explanation could be the extremesensitivity of trim to horizontal stabilator incidence set-ting, which varies with airspeed in the transition range.The flight test report gave no indication whether the stabi-lator position had been calibrated or not. Trim compar-isons for lateral/directional axes are not included, asdifferences were inconclusive or less significant.

The rigid model variations from flight test results increasewith higher airspeeds. This airspeed-dependent effect isrelieved in the elastic model by the added high-frequencycontent The widest differences occur at 160 knots, andare seen in pitch attitude (4°) and collective position (2in.), on the right side of figure 3. Significantly morecollective control is required to produce the same thrust,because not all of the force from flexed blades is gener-ated in the direction of rotor thrust This increasedin-plane component due to blade flexure, and the addi-tional airfoil drag from higher collective pitch combine tosignificantly increase the effective rotor drag, requiringmore pitch-down attitude to balance drag with thrust

The accuracy of both the NASA Ames and Sikorsky Air-craft UH-60 GENHEL simulation models has beenimproved by the use of a quasi-static approximation toblade flexibility derived from flight test data. In figure 4,.this "dynamic twist" approximation is added to theFUGHTLAB rigid blade model and significantlyimproves the fit at high speed. The aeroelastic model,however, still provides a closer fit to the flight test data,and does not require tuning since its improved fit resultsfrom higher fidelity physical modeling. We shall see thatthe global model also represents transient elastic effects.

To compare stability characteristics of the rigid and elasticmodels, a six-degree-of-freedom linear model with eightstate variables was constructed by making perturbationsabout the trim points. Roots of the linearized models werecollected a sufficient time after the perturbations to allowhigher-frequency modes to reach steady state, assuring anaccurate quasi-static approximation for these modes.

Damping ratio and natural frequency of the roll/pitch,short period, and dutch roll modes are plotted against air-

speed for the rigid and elastic models in figure 5. High-frequency elastic modes would not be expected to affectthe six-degree-of-freedom response, but these resultsshow that elasticity has an impact on the stability charac-teristics. The greatest differences occur at higher airspeed,as is the case with trim. These linearized models representthe open-loop dynamics, so the effects are not related tocontrol-system coupling. Airspeed dependence suggeststhat a periodic coefficient effect could be responsible forthe coupling of elastic effects into the handling-qualitiesrange.

The roll and pitch gyroscopic modes are aperiodic at air-speeds below 100 knots, but at higher speeds coalesce intoa coupled roll/pitch mode. The damping and natural fre-quency of this second order pair are plotted in figure 5(a).At low speed the overdamped system has a pair of realroots. Their real axis locations are plotted as natural fre-quency. The pitch roots of the rigid and elastic versionsare indistinguishable, but differences in the roll root as aresult of elasticity are noticeable. When the system isunderdamped the natural frequency of the resulting com-plex root pair is plotted instead of the real axis locations,and rigid and elastic each become a single line. The short-period mode (fig. 5(b)) is unstable for most of the speedrange for both models, although the elastic model hasmore stability than the rigid model does at high speed.The dutch-roll mode (fig. 5(c)) is stable for most of thespeed range in both models although a significant reduc-tion in damping for the elastic model is seen at speedsgreater than 100 knots, resulting in instability at140 knots, whereas the rigid model remains reasonablystable. The heave mode and the spiral mode do not exhibitmuch difference between rigid and elastic models, and arenot shown.

Next we compare dynamic responses of the rigid andelastic models. Figures 6 and 7 show the pitch rateresponse to a 3-2-1-1 longitudinal cyclic input for bothopen loop control and with the stability augmentationsystem (SAS) on. Figure 6 shows data for the models athover, where a small, static trim difference exists at timezero. Response differences increase with time in the openloop system because of the unstable short period pitchmode identified in figure 5. But when the SAS provideslow-gain feedback, no difference between elastic and rigidmodels is observed. The SAS stabilizes the models with-out coupling the elastic response into the handling-qualities range.

Open-loop and SAS-on responses at ISO knots forwardspeed are similarly compared in figure 7. The rigid modelexhibits significantly different open-loop time responsethan the elastic model, confirming the stability analysis infigure 5. Rigid-model pitch response is divergent as

predicted by linear analysis of the short-period pitchmode. Response to the 3- and 2-sec steps by the elasticmodel indicates less stick sensitivity for low-frequencyinputs. Response of the elastic model to the higher-frequency 1-sec doublets in the 3-2-1-1 input showssignificant elastic excitation and coupling with the air-frame degrees of freedom. With the SAS on, the elasticmodel's pitch rate drops to almost nothing at the end ofthe first pulse. Response to high-frequency inputs isalmost identical for both models. The SAS suppressesopen-loop excitations by higher-frequency inputs, butlow-frequency differences, such as trim, are stillsignificant. It is obvious that the higher-order dynamicsdo affect handling qualities, and need to be considered inthe formulation of flight control laws for higher speeds.This assessment was later confirmed during the pilotedsimulations.

More evidence of elasticity's impact on closed-loop con-trol is provided by the SAS-on frequency responses athover and ISO knots. Figure 8 shows the pitch-rateresponse to longitudinal cyclic control for the rigid andelastic models. The magnitude of the transfer function ispresented in the upper plots and the phase is given in thelower plots, both as functions of the frequency of theinput This comparison again shows virtually no differ-ence at hover. At ISO knots the two models differ signifi-cantly at low frequency, but they exhibit no differences athigh frequency. This is consistent with the time-domainresults of figure 7 and further indicates that the SASsuppresses the rotor modeling differences in the higher-frequency range.

From the analytic phase, the study proceeded to thepiloted simulation. Installation and interface of the paral-lel processing computers at the CSRDF was accomplishedin only three weeks and was followed by two weeks ofpiloted evaluation. Several NASA and Army test pilotsparticipated in the evaluation, some of whom wereinvolved in the rotor-map versus blade-element compari-son at Fort Ord (ref. 2). Data was collected on-line usingthe engineer's workstation and was correlated with pilotcomments, off-line comparisons, and flight tests.

The CSRDF simulator cockpit is equipped with dual,four-axis side-stick controllers in lieu of conventionalcyclic and collective. The UH-60 SAS and the flight-pathstabilization (FPS) control system were modified to usethe CSRDF side arm controllers in place of the centerstick and collective lever. The opportunity was taken toalso integrate the advanced control system known asADOCS (Advanced Digital Optical Control System) fromthe facility (ref. 7). Since the UH-60 control laws were notdesigned for a side stick inceptor, is it not surprising thatthe ADOCS control system was preferred by the research

pilots. However, one must remember that the objectivewas to compare rigid and elastic simulations and to dis-cern their effect on handling qualities and flight controldesign, and not just to calibrate another Black Hawksimulation model.

The research pilots performed bob up/bob down andlateral dash/quick stops at hover, and slalom and dolphinmaneuvers in high-speed flight. The aeroelastic degrees offreedom were turned on and off, both in-the-blind and onrequest, to assess fidelity perception. Every pilot was ableto detect the differences indicated by mathematical analy-sis. Those differences were more noticeable at higherspeeds, and during performance of exacting, high-gaincontrol tasks. The dominant effect was increased damp-ing, which is the logical consequence of adding structuralflexibility to a physical representation.

A more surprising result was obtained from frequencysweeps performed by the pilots to compare elastic andrigid modeling with the low-gain SAS and advanced high-gain ADOCS. Figure 9 shows pitch-rate amplitude andphase response to longitudinal cyclic control, at hover.The left side of figure 9 (SAS on) duplicates the left sideof figure 8 except for scale. No differences between rigidand elastic models are noted. With the ADOCS controller(right side of figure 9), the static gain is five times greaterthan with the SAS, and a significant difference is exhib-ited between the rigid and elastic models, particularly inthe higher-frequency range. The elastic model is lesssensitive to pilot inputs than the rigid model is, an effectconsistently confirmed by the pilots. Since there are noperiodic coefficient effects at hover, these differences aremost likely a result of coupling between elastic and rigidbody modes induced by the high-gain ADOCS controller.For future applications employing high-gain controllers,the need for elastic modeling, even at hover, isdemonstrated.

More analytic comparisons between rigid and elasticmodels with the ADOCS configuration are shown infigures 10 through 12. The formats are taken from therecently issued ADS-33C handling qualities specification(ref. 8). The equivalent time delay versus bandwidth forthe elastic and rigid models is plotted in figure 10. Thefrequency response data for the pitch attitude response tolongitudinal cyclic control at 80 knots is shown at thebottom of the figure. The bandwidth and equivalent timedelay shown in the top graph were taken from these plots.Elastic and rigid models are both to the right of dielevel-l/level-2 border, but inclusion of blade elasticsreduced the margin by one-half.

Figure 11 shows the vertical response to collective inputwith the ADOCS controller, at hover. Time response ofthe rigid and elastic models to a collective step is shown

at the bottom of the figure with the curve fit of the low-order equivalent altitude response model overlaid. Thecurve fit is good in both cases. The identified time con-stant and time delay parameters of the low-order equiva-lent response models are shown in the tables at the top ofthe figure for the rigid and elastic models and comparedwith the ADS-33C specification for level-1 and level-2response. Responses were similar for both elastic andrigid models, and are well within the level-1 category forboth cases.

Figure 12 shows an "attitude quickness" comparison ofrigid and elastic models at 80 knots using the ADOCS.The time response plots at the bottom of the figure showthe longitudinal cyclic input and the resulting pitch-rateand pitch-attitude responses for specific elastic and rigidtrials. The increased level of control input required toachieve the same pitch response with the elastic model isa result of the reduced sensitivity of the elastic model.This effect was previously noted, in the time-responseplots of figure 7 and the frequency-response plots of fig-ures 8 and 9. This characteristic was repeatedly confirmedby the pilots, and they felt that the elastic model responsewas more realistic. The top plot, of the peak pitch rate topeak attitude change ratio versus peak attitude change, isrepresentative of maneuverability or quickness; it shows asignificant difference between rigid and elastic models forthe 80-knot flight condition, although both models arewell above the level-1 boundary.

Conclusions

The following conclusions were drawn from thisinvestigation.

1. The ability to process the aeroelastic model in realtime on both the SGI and BBN parallel-processing sys-tems was successfully demonstrated.

2. The ability of parallel processors to support a real-time interface with a full-simulation facility while main-taining real-time operation was demonstrated.

3. The aeroelastic free-flight model compared well withflight test data in off-line comparisons, and exhibitedimprovements both statically and dynamically over therigid blade model.

4. Pilots perceived the effect of the aeroelastic degreesof freedom on the handling qualities and rated thiseffect as an enhancement of simulation fidelity.

The comparisons between the rigid and aeroelastic modelsdemonstrated two mechanisms for coupling between thehigh-frequency elastic modes and the low-frequency air-frame response. A speed-dependent effect, probably due

to interharmonic coupling produced by the periodic-coefficient rotor model, resulted in significant differencesbetween rigid and elastic trim conditions at high speed;comparison with flight test data confirmed the elasticmodel to be more accurate. A gain-dependent effect wasdemonstrated by comparison of SAS and ADOCSresponses at hover. The high-gain ADOCS was seen toproduce significant coupling of the elastic degrees offreedom with the airframe response, whereas the low-gainSAS suppressed dynamic differences in the rotor modelswithout producing additional coupling. A comparison ofopen-loop stability characteristics versus airspeed showeda speed-dependent difference that tended to stabilize theshort-period mode and destabilize the dutch-roll mode.Comparison of rigid and elastic models using the ADS-33C handling qualities criteria further demonstrated sig-nificant differences between the two models and quantita-tively supported the pilot's qualitative assessment that theelastic model was significantly less sensitive to controlinputs than the rigid model was.

The major impact of parallel-processing technology onrotorcraft simulation is expected to be greater productiv-ity. The most immediate impact will be elimination of thereceding and verification usually required when the simu-lation model is transferred to a large simulation facility.The fact that smaller, less expensive computers are able todrive larger facilities like CSRDF means there can bemore backup computers for continuous operation of theexpensive visual and motion bases, and additionalmachines for preparing the next entry, at no additionalcost The common on-line engineering-and-analysis toolwill also realize substantial productivity advantages.

A second advantage is that combining aeroelastics andstructural dynamics in the simulation model increases thefidelity of rotorcraft real-time simulations. Inclusion ofthese disciplines is a necessity for the development ofhigh-speed and high-agility rotorcraft. Hingeless rotors,high-gain flight control systems, higher harmonic control(active vibration suppression), all require concurrentengineering.

A third advantage is achieved by lowering the cost ofcomputers capable of running blade-element models.These more physical representations add fidelity to train-ing simulators, tie the trainer simulation model to theengineering simulation model, reduce expensive tuning,and provide greater flexibility to modify the model foraerodynamic changes.

The FLJGHTLAB system provides a unified tool for thedevelopment, parallelization, analysis, and real-timeoperation of flight simulations. Its use in this applicationexpedited the cost-effective utilization of the CSRDFfacility.

References

1. Du Val, R. W.: "A Real-time Blade Element HelicopterSimulation for Handling Qualities Analysis,"Proceedings of the Fifteenth European RotorcraftForum, Amsterdam, Sept 1989.

2. Corliss, L.; and Du Val, R. W.: "A Comparison ofBlade-Element and Rotor-Map HelicopterSimulations Using Parallel Processing,"Proceedings of the 46th Annual Forum of theAmerican Helicopter Society, Washington, D.C.,May 1990.

3. Ballin, Mark G.: "Validation of a Real-Time Engineer-ing Simulation of the UH-60A Helicopter," NASATM-88360, Feb. 1987.

4. Ballin, MarkG.: "Validation of the Dynamic Responseof a Blade Element UH-60 Simulation Model inHovering Flight," Proceedings of the 46th Annual

Forum of the American Helicopter Society,Washington, D.C.. May 1990.

5. Reaser, J. S.; and Kretsinger, P. H.: REXOR n Rotor-craft Simulation Model-Volume I—EngineeringDocumentation, Lockheed-California Co., NASACR 145331, June 1978.

6. Singer Link Trainer Test Procedures and ResultsReport (TTPRR) for the UH-60 Black HawkTrainer, Singer Link, Mar. 1989.

7. Advanced Digital/Optical Control System (ADOCS)Flight Demonstrator Program, Interim TechnicalReport—Detailed Design, Boeing Vertol CompanyReport No. D358-10045-1, May 1983.

8. Aeronautical Design Standard, Handling QualitiesRequirements for Military Aircraft, United StatesArmy Aviation Systems Command, St. Louis, MO,Directorate for Engineering, ADS-33C, Aug. 1989.

One cycle

D 1

Cntl

sERi1A

2 3 4 5 6 7 8 9 1 0 1 1 1 2 1

Blade 2Blade 3

• — Blade 4

One cycle

n i o i 4 C C T a

3 14 15 16 17 18 19 2(

Engine// Fuselage

Eng -—A 1

/ Solver j

pARALLEL

CPU1

CPU 2

CPU 3

CPU 4

CPUS

Cntl Eng | Fus | Tall

Blade 1

Blade 2 |—

BladeS [•-•

Blade 4

Solver |

Tall Include*:horiz & vertstabilizer* andtall rotor

Figure 1. The parallelizadon process.

Crew Station Researchand Development Facility

Compuscene IV

Crew station

VAX 8650(AOOCS)

Parallel processors

Blade 1 Blade 2 Blade 3 Blade 4Control syst,era, fus, &tail rotor

Interface Scope

Shared memory

Figure 2. CSRDF/paraUel processor architecture.

GW. = 17140 IbsAlt = 9720 ftton. CG = 347.6 In.Lat. CG s -0.2 In.

10 r

-2

-10

10

I 8

GW. = 16420 IbsAlt. = 6080 ftton. CG s 347.1 In.Lat. CG s 0 In.

Rigid modelElastic modelFlight test

10 r

oo

-50 -30 -10 10 30 50 .60 80 100 120 140 160

Velocity (knots)

Figure 3. Trim comparison of rigid and elastic models with flight test data.

GW. = 16420lbs; long. CG = 347.1 in.Alt. = 6080 ft; lat. CG = 0 In

10

6

2

-2

-6

-10

10

8

Elastic modelRigid model with dynamic twistFlight test

li

co

i

6

4

2

0

10

8

6

4

2

60 75 90 105 120 135 150Velocity (knots)

165

Figure 4. Trim comparison using dynamic twist

1.5

1.0

o1 .5

-.5

-1.0

Overdamped

Stable

t

Rigid modelElastic model Unstable

(a) Roll/pitch mode

1.5 r

Unstable

(b) Short period mode

1.5

rat

1.0

01

ftS o

Stable

Unstable

0 30 60 90 120 150Velocity (knots)

(c) Dutch roll mode

-1.5

4.5

3.3

2.1

.9

-3

-1.5

4.5

oS 3.3

2.1

.9i|

!' -3

-1.530 60 90 120 150

Velocity (knots)

Figure S. Damping and natural frequency for the open-loop roll/pitch, short period, and dutch roll modes.

10

ffe

a ^i =

22£U

£

7

6

5

4

3

.3

.15

-.15

-.3

-.45

Control Input Rigid modelElastic model

Open loop

.070 r

1

£ -.

0

.035

.070

105

SASon

_TL_r

1.5 3.0 4.5 6.0Time (sec)

7.5 9.0 10.5

Figure 6. Dynamic comparison of rigid and elastic models at hover.

11

IfIt3

I

4

3

2

1

0

.4

0

-.4

-.8

-1.2

-1.6

Control input —— Rigid modelElastic model

10.5

5 6 7 8 9

!I

-.2

SASon

1.5 3.0 4.5 6.0Time, (sec)

7.5 9.0 10.5

Figure 7. Dynamic comparison of rigid and elastic models at ISO knots.

12

Hover

•? .10

•a. -05

3 0

« -100a.

-200

Rigid modalElastic modal

SASON

150 knots

1.0 10.0 100.0 .1

Fraq (rad/s)

1.0 10.0 100.0

Figure 8. Comparison of rigid and elastic pitch-rate transfer function at hover and at ISO knots.

SASon AOOCS

1.25

7 1.002 .75

| .50

< .25

0

100

I °£ -100

* -200

-300

——— Rigid modal— — — Elastic modal

.1 1.0 10.0 .1Fraq (rad/s)

1.0 10.0

Figure 9. Comparison of rigid and elastic pitch-rate transfer function at hover for S AS and ADOCS controllers.

13

Combat

Level 3 Level 1

-90

-180

a. -270

-360.1

Elastic Rigidmodel , modal

1 2 3

Bandwidth (rad/sec)

Rigid model——— Elastic model

6dB Gain marginJ

100.0

Difference betweencross-over frequencyand twice cross-overfrequency

. .I1.0 10.0 100.0

Freq (rad/sec)

Figure 10. Short-term pitch response to control inputs using the ADOCS controller (80 knots) (air combat limits).

14

Maximum values for height Actual values for height responseresponse to collective controller to collective controller

Level

1

2

T*eq(sec)

5.0

oo

T*eq(sec)

0.20

0.30

Rigid model

I

i

-2

Level

Rigid

Elastic

T*eq(sec)

1.7

1.9

<heq(sec)

0.17

0.16

2

1ioo

-.40

-.48

-.56

- /

/

//

1 1 1 1

- Time to equalize climb rate

Tn - Phase delay to equalize climb rateeq

Elastic model

-.135

-.225

-.315

—— Simulation -5— — — Estimation

-104 0

Time (sec)

Figure 11. Height response to collective inputs using the ADOCS controller (hover).

15

Level 1

Level 2

5.0 7.5 10.0 1i5 15.0 17.5 20.0Minimum attitude change (deg)

Elastic model

Peak

22£ 25.0

1.0 1.5 2.0 2J 0Time (sec)

Figure 12. "Attitude quickness" comparisons for rigid and elastic models using the ADOCS controller (80 knots).

16

NASANatkml Atnnufca and Report Documentation Page

1. Report No.NAS ATM-102859

2. Government Accession No. 3. Recipient's Catalog No.

4. Title and Subtitle

Comparisons of Elastic and Rigid Blade-Element Rotor ModelsUsing Parallel Processing Technology for Piloted Simulations

5. Report Date

June 19916. Performing Organization Code

7. Author(s)

Gary Hill, Ronald W. Du Val,* John A. Green,* andLoc C. Huynh**Advanced Rotorcraft Technology, Inc., Mountain View, California

8. Performing Organization Report No.

A-90266

10. Work Unit No.

505-61-919. Performing Organization Name and Address

Ames Research CenterMoffett Field, CA 94035-1000

11. Contract or Grant No.

12. Sponsoring Agency Name and Address

National Aeronautics and Space AdministrationWashington, DC 20546-0001

13. Type of Report and Period Covered

Technical Memorandum

14. Sponsoring Agency Code

15. Supplementary Notes

Point of Contact: Gary Hill, Ames Research Center, MS 237-11, Moffett Field, CA 94035-1000(415) 604-6577 or FTS 464-6577

This paper was presented at the Sixteenth European Rotorcraft Forum, Glasgow, Sept. 18-20,1990.16. Abstract

A piloted comparison of rigid and aeroelastic blade-element rotor models was conducted at theCrew Station Research and Development Facility (CSRDF) at Ames Research Center. FLIGHTLAB,a new simulation development and analysis tool, was used to implement these models in real time usingparallel processing technology. Pilot comments and quantitative analysis performed both on-line andoff-line confirmed that elastic degrees of freedom significantly affect perceived handling qualities.Trim comparisons show improved correlation with flight test data when elastic modes are modeled. Theresults demonstrate the efficiency with which the mathematical modeling sophistication of existingsimulation facilities can be upgraded using parallel processing, and the importance of these upgradesto simulation fidelity.

17. Key Words (Suggested by Author(s))

Parallel processingReal-time simulationAircraft modeling

18. Distribution StatementUnclassified-Unlimited

Subject Category - 01

19. Security Classif. (of this report)

Unclassified

20. Security Classif. (of this page)

Unclassified

21. No. of Pages18

22. Price

A01

NASA FORM 1626 OCT86For sale by the National Technical Information Service, Springfield, Virginia 22161