[american institute of aeronautics and astronautics 18th applied aerodynamics conference -...

(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)' Sponsoring Organization.

AOO-39852 AIAA 2000-4010

Integrated Aero-Propulsive CFD Methodology for the Hyper-X Flight Experiment

Charles E. Cockrell, Jr.*, Walter C. Engelundt

NASA Langley Research Center, Hampton, VA

Robert D. Bittner, Arthur D. Dilley, Tom N. Jentink§

FDC-Nyma, Inc., Hampton, VA

Abdelkader Frendi^University of Alabama-Huntsville, Huntsville, AL

ABSTRACTComputational fluid dynamics (CFD) tools

have been used extensively in the analysis anddevelopment of the X-43A Hyper-X Research Vehicle(HXRV). A significant element of this analysis is theprediction of integrated vehicle aero-propulsiveperformance, which includes an integration ofaerodynamic and propulsion flow fields. This paperdescribes analysis tools used and the methodology forobtaining pre-flight predictions of longitudinalperformance increments. The use of higher-fidelitymethods to examine flow-field characteristics andscramjet flowpath component performance is alsodiscussed. Limited comparisons with available groundtest data are shown to illustrate the approach used tocalibrate methods and assess solution accuracy. Inviscidcalculations to evaluate lateral-directional stabilitycharacteristics are discussed. The methodology behind3D tip-to-tail calculations is described and the impact of3D exhaust plume expansion in the aftbody region isillustrated. Finally, future technology development needsin the area of hypersonic propulsion-airframe integrationanalysis are discussed.

INTRODUCTIONHypersonic airbreathing vehicle configurations

are characterized by highly-integrated propulsionflowpath and airframe systems. A significant challengein the development of this class of vehicle is anassessment of propulsion-airframe flow field interactionsand the integrated aero-propulsive performance ofcandidate systems. Advanced experimental, analyticaland computational tools are being developed to aid in the

* Hypersonic Airbreathing Propulsion Branch, AIAA Senior Member.^ Vehicle Analysis Branch, AIAA Senior Member.5 Hypersonic Numerical Applications Group, Hyper-X Program Office' Associate Professor, Department of Mechanical and AerospaceEngineering.

Copyright © 2000 by the American Institute of Aeronautics andAstronautics, Inc. No copyright is asserted in the United States under Title17, U.S. Code. The U.S. Government has a royalty-free license to exerciseall rights under the copyright claimed herein for Governmental Purposes.All other rights are reserved by the copyright owner.

design of configurations that exploit propulsion-airframeinteractions to maximize performance and enhancestability and control characteristics. Presently,capabilities for testing complete engine flowpath/airframe configurations that model all of the pertinentinteractions affecting integrated vehicle performance arelimited. Predictive methodologies, includingcomputational fluid dynamics (CFD) and other analysistools, must encompass a wide range of modelingcapabilities to capture all of the relevant flow physics ofthe complete scramjet flowpath as well as the externalairframe. This analysis is normally accomplished using amulti-level approach, increasing in complexity andfidelity as the design is matured. The preliminaryanalysis phase may employ different tools for the variousflowpath components, which necessitates thedevelopment of force accounting systems appropriate forspecific configurations. CFD is also a valuable tool usedto interpret aerodynamic and propulsion ground testdata.

One objective of the Hyper-X program is todevelop and mature the technologies required forhypersonic airbreathing flight.1 Three flight tests of theHyper-X Research Vehicle (HXRV), or X-43A, arecurrently scheduled to obtain in-flight performance dataon a scramjet-powered hypersonic configuration. Thefirst two of these flight tests will be at Mach 7 testconditions with a third flight at Mach 10. Thedevelopment of the Mach 7 X-43A required a pre-flightassessment of longitudinal and lateral-directional aero-propulsive characteristics near the target flight testcondition.2 The development of this pre-flight data basewas accomplished through extensive aerodynamic wind-tunnel testing and a combination of 3D inviscidairframe calculations and cowl-to-tail scramjet cycleanalyses to generate longitudinal performanceincrements between mission sequences. Theseincrements were measured directly and validatedthrough tests of the Hyper-X flight engine (HXFE) andvehicle flowpath simulator (VFS) in the NASA Langley8-Foot High Temperature Tunnel (8-Ft. HTT).4Predictions were refined with tip-to-tail Navier-Stokescalculations, which also provided information on

161


scramjet exhaust plume expansion in the aftbody region.A qualitative assessment of lateral-directional stabilitycharacteristics was made through a series of tip-to-tailinviscid calculations, including a simulation of thepowered scramjet flight test condition. CFD predictionswere also used to address other aspects of vehicleperformance and flight test development, includingboundary layer trip design and assessment5, thermal andstructural loads and scramjet flowpath componentperformance.

The Hyper-X program represents the firstopportunity to correlate analytical and CFD predictionswith ground-test and flight-test data on an airframe-integrated scramjet configuration. Comparisons withavailable ground test and flight data will be used tocalibrate tools and physical models. Since the CFD andexperimental test techniques used in the Hyper-Xprogram represent the state of the art in hypersonicpropulsion-airframe integration (PAI) research, anexamination of these methods also provides insight intofuture technology development needs for the next phaseof hypersonic vehicle development.

This paper presents an overview of the methodsused in the analysis and pre-flight development of theMach 7 X-43A vehicle. A discussion of CFD codes andother analysis tools is included with their respectivecapabilities and limitations. The appropriateness ofvarious physical modeling approximations and theireffect on performance predictions is discussed. Themethodology for integration of tools for various flowpathcomponents is discussed with limited results and

comparisons to available data. Finally, a discussion offuture challenges for hypersonic propulsion/airframeintegration and predictive methodologies for integratedvehicle performance is presented.

MISSION DESCRIPTION AND ANALYSISREQUIREMENTS

The nominal Hyper-X flight trajectory isillustrated in figure 1. The flight profile begins with thecaptive carry flight of the Hyper-X Launch Vehicle(HXLV) under the wing of a B-52 aircraft. The HXLVconsists of the X-43A mounted to the first stage of aPegasus® booster rocket, manufactured by OrbitalSciences Corporation, with a vehicle-to-booster adapter.Following air-launch of the HXLV from the B-52, thevehicle is boosted to the appropriate flight test conditionand, at burnout, the X-43A separates from the booster.Upon stabilization, the cowl door, which remains closedthroughout the boost phase to block the inlet entranceand protect the internal engine components from highheat loads during boost, opens to establish flow throughthe engine. Following a few seconds of unpoweredoperation, hydrogen fuel is introduced and the poweredportion of the scramjet test is conducted, lastingapproximately seven seconds. A series of parameteridentification (PID) maneuvers are then conducted andthe cowl door closes as the vehicle begins a controlleddescent prior to mission termination, while conductingadditional PID maneuvers to measure lower Machnumber aerodynamic stability and controlcharacteristics. This flight profile necessitates the

Power-On Test 7 sees

100

Altitude(kft)

50

0.47 Mach19,000ft

170psf

Drop Weight:-41,400lbs

-95,000 ft

45,000 ftPush-Over

-1.5g's

/- Kower-UTT larebsecs"̂"°" r- PID 10 sees

/ i- Cowl Closes

• Initial 1g to 10AOADecel - Descent

• Fuel On\ \— Power-Off Tare 5 sees*— Cowl Open

- Free Flyer Staging- Booster Bum-Out

Ascent-AccelTurning Descent

(<2g)to Desired Impact Area

Pull-Up, 2.5g's

Booster Ignition0.49 Mach, 18,700ft

Terminal Glide MissionComplete

OsecONM'

5 sec"0.6 NM*

-80 sec"-70 NM"

-90 sec-80 NM

^ -500 sec^^-700 NM

Figure 1. Hyper-X Flight Trajectory.

162


analysis of three distinct mission phases: cowl-closedunpowered, cowl-open unpowered and cowl-openpowered. Much of the analysis for aero-propulsiveperformance is built upon the prediction of force andmoment increments between the various mission points.The term "inlet-open increment" is used to refer to thedifference in force and moment quantities between thecowl-open unpowered and cowl-closed points and theterm "power-on increment" is used to refer to thedifference between the inlet-open powered andunpowered phases of the flight.

A complete nose-to-tail analysis of the X-43Aat the conditions of interest requires a wide range of flowmodeling capabilities. A summary of the relevant flowphysics and prediction requirements is shown in figure 2.At hypersonic Mach numbers, high temperature gaseffects become important. For the Mach 7 Hyper-X flightconditions, it is sufficient to model the flow field as amixture of thermally-perfect gases, wherethermodynamic quantities vary as a function oftemperature using curve fits for the appropriate species.Composite species models may be used to approximatethe composition as a single species to reducecomputational overhead. Surface pressure and skinfriction predictions are generally required on all externalsurfaces to resolve vehicle forces and moments and toprovide structural loadings on vehicle components. Heattransfer predictions may also be required to assessthermal loads. This implies the need for appropriateturbulence models and knowledge of the boundary layerstate. Accurate computations of forebody flow fields,characterized by shocks, shock-boundary layerinteractions and potentially separated flow regions, arerequired to compute mass capture at the cowl lip station.The inlet flow field is characterized by shock-boundary

External flow field/Airframe• Prediction of surface

pressure, skin friction, heattransfer

• Transition and TurbulenceModeling

Forebody/External Cowl• Shock Shapes/Locations• Viscous Interaction Effects• Transition to Turbulence• Inlet Flow Profiles: Mass, Momentum Flux,

Energy• Shock/Shock Interaction on Cowl LE

layer interactions, flow separation in unfavorablepressure gradients, high leading edge thermal loads andcorner flow regions. Accurate computation of the inlet/isolator region is necessary to provide equivalent IDproperties to evaluate component performance.Computation of the combustor flow field requiresmodeling fuel injection and complex mixing phenomenaas well as finite rate chemical reactions. Downstream ofthe combustor, the high-temperature scramjet exhaustflow field must be modeled by approximating the speciesconstituents of the combustion process. This poweredexhaust plume expands in the aftbody region and mayinteract with vehicle aerodynamic or control surfaces,especially at deflected wing settings or when the vehicleis at non-zero angles of attack or sideslip. Thedetermination of integrated vehicle performance requiresanalysis of both internal and external flow fields and anappropriate accounting of the interactions between thetwo. Another objective of the Hyper-X analysis is todetermine the extent to which these flow features must bemodeled in order to generate quantities of interest withengineering-level accuracy.

ANALYSIS TOOLS

The primary CFD tool used for the pre-flightperformance analysis of the X-43A is the GeneralAerodynamic Simulation Program (GASP), a product ofAeroSoft, Inc.6 GASP is a multi-block, structured-grid,upwind-based, Navier-Stokes flow solver. Mixtures ofthermally-perfect gases are modeled using polynomialcurve fits for thermodynamic properties. GASP canmodel frozen, equilibrium or finite-rate chemistry withmodels for hydrogen-air combustion. The Baldwin-Lomax algebraic turbulence model with the Goldbergbackflow correction has been widely applied forturbulent flows. Various two-equation eddy-viscosity

External Nozzle/Aftbody Region• Exhaust Plume Characteristics: Effects of

Exit Plane Non-Uniformity, Angle-of-Attack, Sideslip Effects.

• Plume Interactions with Airframe/ControlSurfaces

Inlet/Isolator• Shock/BL Interactions• 3D Geometry flow fields• Separated Flow Regions• Kinetic Energy Efficiencies• Starting Characteristics

Figure 2. X-43A Flow Physics and Flow Modeling Requirements.

Combustor• Fuel/Air Mixing• Chemical/Thermal Non-Equilibrium flow fields• Modeling of Injector Geometry

163


formulations are also available in GASP and have beenused for various applications. Convergenceacceleration options include a V-cycle multi-gridalgorithm, mesh sequencing and local time stepping. Alarge calibration data base is available for GASP forhypersonic configurations and scramjet flow fields.Predictions for surface pressure, flow-field quantitiesand integrated forces and moments have comparedwell to available experimental data at unpowered andsimulated powered conditions in previous studies.

Another CFD tool used in the Hyper-Xscramjet flowpath design and analysis is the ViscousUpwind Algorithm for Complex Flow Analysis(VULCAN).14 VULCAN is a Navier-Stokes solvercapable of solving turbulent reacting and non-reactingflows. Physical modeling capabilities include a varietyof one-equation and two-equation turbulence models,compressibility models, finite-rate chemistry andturbulence-chemistry interaction effects. A variety ofnumerical schemes to reduce computational cost arealso available, including wall functions for two-equation turbulence models, multi-grid methods forelliptic and space-marching schemes, meshsequencing and conditioning of governing equations toreduce numerical stiffness. Recent enhancements tothe code include parallel capabilities through the useof message-passing interface (MPI) routines.

Two additional tools are used for analysis ofthe internal propulsion flowpath. The first is thesupersonic hydrogen injection program (SHIP).15'16

SHIP uses the SIMPLE (semi-implicit method forpressure-linked equations) method to solve theparabolized, mass-averaged equations forconservation of mass, momentum, total energy, totalfuel and turbulence fields in a variable area domain ofrectangular cross section. The second tool used forflowpath analysis is the SRGULL code.17 SRGULL iscomprised of a two-dimensional/axisymmetric Eulerflow solver (SEAGULL), which is used to solve theforebody, inlet and external nozzle regions of the lowersurface flowpath, and a one-dimensional chemicalequilibrium cycle analysis code (SCRAM), which isused to approximate the combustor flow field.SRGULL also includes an integral boundary layermethod (HUD) to provide a viscous component to theforces and moments and has a one-dimensionalisolator model used to predict the onset location ofpressure rise ahead of the fuel injectors associated withheat addition due to combustion. Several scalingfactors, based on previous studies and ground test data,are included to account for such factors as massspillage, inlet kinetic energy efficiency, base pressure,combustion efficiency and nozzle thrust multiplier to

account for three-dimensional effects.

PRE-FLIGHT ANALYSIS METHODOLOGYLongitudinal Performance Increments

The development of the X-43A pre-flight aero-propulsive performance data base includes an analysis ofthe post-separation point through the powered flightexperiment. Although not the focus of this paper, the database also supports the ascent, stage separation and post-experiment descent phases of the mission as well. Threemission points are analyzed: cowl-closed, cowl-openunpowered and cowl-open powered. A large body of wind-tunnel data exists for the cowl-closed configuration.Because of model scale and facility limitations, it is notpossible to simulate the flow-through engine or to modelpowered effects in available aerothermodynamic facilities.Therefore, CFD predictions were used to determine theinlet-open and power-on performance increments. Theseincrements were then applied to the experimental data baseto develop predictions for longitudinal performance in eachof the three mission phases. The increments werecomputed using GASP to obtain 3D inviscid flow solutionsfor the X-43 A airframe and SRGULL computations for thepropulsive flowpath surfaces of cowl-open configurationsfrom the cowl leading edge station to the vehicle trailingedge (inlet/isolator, combustor, internal and externalnozzle). Figure 3 illustrates the force accounting systemused in this methodology.

The inviscid calculations were obtained using aspace marching technique with the exception of the bluntnose of the vehicle. Sidewall, cowl and wing leading edgesare treated as aerodynamically sharp. The use of theinviscid approximation reduces computational time andallows multiple points to be analyzed. Solutions wereobtained for the flight-scale X-43 A over a matrix of pointsthat included variations of Mach number and angle of

Top and Chins Surfaces; Airframe

Wing and Ta8 Surfaces:Atrfram*

External Nozzle: Propuision

Figure 3. X-43A Force Accounting Methodology.

164


attack around the Mach 7 test point. A qualitativedepiction of the computed performance increments isshown in figure 4. The predictions indicate a decrease innormal and axial force as well as a nose-down pitchingmoment increment when the cowl door is opened due tohigher surface pressure on the external nozzle and a relieffrom the high pressure on the forward-facing cowl door.A decrease in axial force as well as a nose-down pitchingmoment increment is predicted for the power-onincrement due to the pressurization of the external nozzlesurface from the powered scramjet exhaust plume.

Viscous Predictions

Viscous predictions of the cowl-closedconfiguration were also obtained at various conditions toexamine trends due to Reynolds number effects.Computations with GASP were obtained for a model-scale configuration at ground test conditions and for afull-scale configuration at representative flightconditions. The GASP calculations are performed byspace marching most of the body, with the exception ofthe blunt nose region and the wake region aft of the cowltrailing edge in the aftbody and the vehicle base. TheBaldwin-Lomax algebraic model is used as theturbulence model in these calculations.

Viscous forebody and flowpath computationswere also used to examine various aspects of vehicle andcomponent performance. PNS forebody calculationswere used to predict inlet mass capture for the vehicleand subscale flowpath models. The Baldwin-Lomax

turbulence model was generally used with the transitionlocation fixed based on estimations of the effectivenessof boundary layer transition strips on the forebody.Limited experimental pressure data from ground testsshow good agreement with surface pressure predictions.A series of viscous forebody computations at variousangles of attack and sideslip angles at Mach 6 windtunnel conditions was used to provide a correlation withsurface pressure data used to calibrate an experimentalflush-air data sensors (FADS) system for measurementof Mach number and angle of attack. Figure 5 shows aqualitative comparison of predicted and measuredpressures at one cross-sectional station. Reasonableagreement was obtained considering that thecomputations do not model the boundary layer tripgeometry. Viscous calculations of the inlet/isolatorregion have been used to provide correlations withsurface pressure and schlieren data from various engineflowpath tests. These computations have also been usedto compute equivalent ID properties to evaluate scramjetcomponent performance.

Computations have also been performed toobtain thermal loads on both the cowl-closed and cowl-open configurations. Navier-Strokes solutions have beenused to evaluate turbulent heating amplification, cornerflow effects and shock-shock interaction effects on heatflux predictions in the cowl leading edge and sidewallregions. Figure 6 shows the surface grid topology in thesidewall/cowl leading region used for a Navier-Stokescalculation to obtain heat loads in this region for the

Normal Force Coefficient

0 2 4Angle of Attack, a

Axial Force Coefficient


—— Cowl-Closed (Release 5 Data)—— Cowl-Open Unpowered (From CFD As)—— Cowl-Open Powered (From CFD As)

I/O A = CFD———-CFD^,^= CFDpow,red-CFDunpowerK1

Pitching Moment


Figure 4. CFD Prediction of X-43A longitudinal performance increments at Mach 7.

165


0.140

0.120

O°"0.100„

£ 0.080_G)"Q£ 0.060oO 0.0404)fe 0.0205)S o.oooh.a.

-0.020

-0.040

r

i-'• _,,-— --^— — .

i*''""".̂.- '̂

i. /'

: .̂--""l!.•'•''

- /'- ^.^.-v-.-

- ;h~— -.-- "•. , (,"--—.—..._..—....—...-..... .........-........-.....—jj.-.-.-.r(bx- -_._._._.v._- < '

r !,.._..'- "~" "- •>" •- K

- , , , , i , , , , i , t . , i i ,

• E_ _ _ . c• c..__....... JA C

_._._.. <

"\ T t"\ — — - c

\ * c

if"~-^"v.

\

-^^^i— -J...«,_.-m....-̂ - |

...............................™.._...— .-- — "̂

_._. . - • - ̂

. ..—• —— . ,'\._ 1

i < I i i i i I i , ,

:FD oa-2°, PXO°>ata os-2e, ffcsO0

:FD 0=0', p=o°ata o=0°, M°FD 0=2°, 9=0°ata cce2D, 9=0°FD c»4°, fr=0'ata os4°, ̂ =0°FD 0.8°, &=0°FD o=«°, M°

i

1

•

, |

Cowl Leading Edge Region

-0.20 -0.10 0.00 0.10Y-Station (meters)

0.20 0.30

Figure 5. Comparisons of wind-tunnel surface pressuremeasurements and CFD predictions at forebody station.

cowl-closed configuration.

Code Calibration and Accuracy AssessmentCode calibration and accuracy assessment is

accomplished through appropriate comparisons of CFDpredictions with available experimental data. Figure 7shows a qualitative comparison of force and momentpredictions from 3D GASP inviscid and viscouscomputations of the X-43A cowl-closed configurationwith subscale wind tunnel data at Mach 6. The resultsfrom a viscous computation shown in the figure wereobtained at the same Reynolds number and model scaleas the data. There is an obvious discrepancy with theinviscid CFD axial force prediction. The agreement ismuch better for the viscous computation. Normal forcecoefficient is also slightly overpredicted and smallernose-up pitching moment values are predicted than areindicated by the data base comparisons. There is littlesignificant difference between the inviscid and viscouscomputations in normal force or pitch at 0° angle ofattack. The discrepancies between the data and CFDpredictions are most likely the result of a combination ofuncertainties in corrections made to the data and physicalmodeling approximations in the calculations. The datashown in figure 7 have been corrected for stinginterference effects, base pressurization and other facilityand testing procedure effects. Assumptions regarding theboundary layer and the lack of modeling of the forebodyboundary layer trips may also contribute to theuncertainty of CFD predictions. Despite theseapproximations in physical modeling and consideringthe uncertainties in the data, the agreement is consideredgood. Comparisons of forebody and aftbody surfacepressure predictions with HXFE/VFS full-scale dataobtained in the 8-Ft. HTT also show good agreement.Additionally, inviscid surface pressure predictions do notdiffer substantially from viscous predictions for external

Figure 6. Mach 7 cowl-closed viscous grid used for heattransfer prediction.

airframe surfaces, suggesting that the inviscidapproximation is sufficient to obtain pressure loads onthese surfaces.

Verification of predicted force and momentcoefficients from SRGULL is accomplished primarilythrough comparisons of SRGULL surface pressure andforce and moment predictions with higher-fidelity CFDsolutions for comparable component efficiencies as wellas appropriate comparisons- between predictions anddata. Experimental measurements from various scramjetflowpath tests in Langley scramjet test facilities haveshown good agreement in terms of axial force andsurface pressure predictions. Reasonable agreement withsurface pressures have also been obtained in the inlet/isolator and nozzle regions. Reasonable correlations ofpitching moment have also been obtained.

The only experimental verification of the CFD-computed longitudinal force and moment data basepredictions for the cowl-open configurations wereobtained from tests of the HXFE/VFS model in the 8-Ft.HTT.3 The 8-Ft. HTT is a propulsion test facility thatuses methane-air combustion and oxygen replenishmentto generate a test gas with total enthalpy and Machnumber equivalent to flight conditions.19 The VFS is afull-scale model with that duplicates the flowpath andchine surfaces of the X-43A, but does not model othercomponents. A sketch of the VFS model mounted in thetest section is shown in figure 8. The primary objective ofthese tests was to verify the propulsion thrustperformance, fuel sequencing, and operability of allengine-related subsystems. Force and moment data foreach of the three post-separation mission points (cowl-closed, cowl-open unpowered and cowl-open powered)near the scramjet test point were also obtained. The VFS

166


Normal Force

o

o GASP Mach 6 InviscidMach 6 TunnelGASP Mach 6 Viscous

- 2 - 1 0 1 2 3 4 5 6Angle of Attack, a

Axial Force Pitching Moment

() O

oo

- 1 0 1 2 3 4Angle of Attack, a

O

O

-2 -1 0 1 2 3 4 5Angle of Attack, a

Figure 7. X-43A cowl-closed predictions vs. subscale experimental data base at Mach 6.

configuration models all of the salient features of theflowpath surface affected by these transitions, including3D expansion of the scramjet exhaust plume over theaftbody surface. Because of the differences in geometryand test gas composition, only the cowl-opening andpower-on increments obtained in the test can be utilized.In general, these data show good agreement with thepredicted increments. The magnitude of axial andnormal force as well as pitching moment increments arecomparable and the measurements confirm a small nose-down pitching moment resulting from the cowl-openingsequence. Furthermore, detailed comparisons of surfacepressure distributions from these tests and viscous andinviscid CFD predictions as well as cowl-to-tail scramjetflowpath cycle predictions show that the 8-Ft. HIT testconditions are a good simulation of flight parameters.The comparisons show that the pre-flight database

uFront View Side View

Figure 8. HXFE/VFS model installed in the 8-Ft. HTT.

prediction methodology is sufficient to provide anaccurate prediction of total force and momentincrements. However, higher fidelity Navier-Stokessolutions are required to resolve detailed physics of theflowpath.

Lateral-Directional Stability CharacteristicsComputations were obtained on the X-43A at

the same three mission points analyzed previously inorder to evaluate the effect of the cowl-opening and fuel-on sequences on lateral-directional stability.20 Noexperimental force and moment data is available for thecowl-open configurations at non-zero sideslip angles.Computational cost and gridding requirements areprohibitively large for 3D viscous computations at non-zero sideslip. 3D inviscid computations were obtainedfor the cowl-closed, cowl-open unpowered and cowl-open powered conditions at Mach 7, 2° angle-of-attack,and 0° and 3° sideslip. The unpowered computationswere obtained using GASP, including the internalflowpath without the geometry of the fuel injectors in thecombustor. The powered computations were obtainedusing a ID cycle analysis method for the combustor.Several lateral stations were computed for the combustoranalysis in order to approximate some lateral variation inflow field properties at the 3° sideslip condition. Figure 9shows predicted values for side force, yawing momentand rolling moment derivatives computed from the 0°and 3° predictions. The CFD analysis predicts that thecowl-opening and power-on sequences of the flight have

167

(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)1 Sponsoring Organization.

Side Force Derivative, Cy

-©—— Cowl-Closed Experimental Data* Cowl Closed CFD* Unpowered CFD* Powered CFD

- 2 0 2 4 6Angle of Attack, a

Yawing Moment Derivative, Cn|! Rolling Moment Derivative, CK

0 2 4Ang te of Attack, a

-2 0 2 4Angle of Attack, a

Figure 9. Inviscid CFD Predictions of X-43A Lateral-Directional Stability Derivatives at Mach 7.little significant direct effect on the lateral-directionalstability. There is, however, a significant indirect effect ofthe powered flight condition on airframe stability andcontrol as a result of the horizontal wing deflectionrequired to trim the resultant propulsive-inducedpitching moment. An analysis of the aftbody flow fieldsolution also indicates some impingement of thepowered scramjet exhaust plume on the horizontal wingsurface of the vehicle at the 3° sideslip, 2° angle of attackcondition. No analysis was done to evaluate controlsurface effectiveness under powered conditions or toevaluate exhaust plume interaction effects at non-zerosideslip with deflected wing surfaces.

Tip-to-Tail SimulationsA viscous tip-to-tail calculation, including a

simulation of powered effects, was used to provide themost detailed prediction of performance possible at thetarget flight test point of Mach 7, 2° angle-of-attack.21

This calculation was accomplished using GASP tosimulate both external and internal flow fields, includingmodeling the powered scramjet exhaust effects. A IDcycle analysis was still used to approximate thecombustor flow field due to the complexity of modelingthe geometry of this region as well as the physicalmodeling requirements and computational cost tocompute turbulent reacting flow fields in the combustor.A summary of the methodology is shown in figure 10.External flow fields are typically computed by solvingthe PNS equations except in the regions of the nose andcowl leading edge where bluntness effects are important.

The scramjet exhaust plume is modeled as a single-species thermally-perfect gas. An analysis of the aftbodyflowfield from this calculation shows that the exhaustplume expands beyond the boundaries of the externalnozzle "propulsion" surface as defined in figure 3,creating uncertainty in the force accounting assumptionsused previously to couple propulsion flowpath andairframe CFD predictions. This effect is illustrated infigure 11, which shows density contours of the exhaustplume at several cross-sections along the aftbody. Acomparison of total integrated forces and moments fromthis calculation with those developed from applying theCFD-computed performance increments to experimentalcowl-closed data base values shows only small

Midsection:Forebody: PNS•Elliptic leading edge Laminar external e•PNS forebody•Transition location specifed

Cowl Leading Edge:•Interpolation from forebody•Elliptic region•Laminar leading edge

Inlet:• PNS/TLNSCaculation• Turbulent sidewall and body-side•Transition specified cowl-side

Aftbody:.PNS. Exhaust gas/Air modeled

Internal Nozzle:.PNS• Inflow from Combustor Cycle Analysis• Single Composite Species

Figure 10. 3D Tip-to-Tail Solution Methodology.

168

(c)2000 American Institute of Aeronautics & Astronautics or Published with Permission of Author(s) and/or Author(s)1 Sponsoring Organization.

First Aftbody Plane (1) Horizontal Tail Leading Edge (2)

Trailing Edge of Lower Vertical Fin (3) Trailing Edge of Vertical Tail (4)

AeroSurface

Nozzle(PropulsionSurface)

Base of Vehicle (5)

Wing Surface

Figure 11. Density Contours Showing3D Exhaust Plume Expansion.

differences in predicted axial force and pitching moment,indicating that this "spillage" effect is small at thenominal Hyper-X scramjet flight test condition.

FUTURE HYPERSONIC PAI ANALYSIS

The X-43A pre-flight database methodology,utilizing inviscid airframe CFD computations andscramjet flowpath cycle analyses, was successful inpredicting pre-flight basic longitudinal performanceincrements based on a comparison with availableexperimental data from HXFE/VFS testing. However, anassessment of the methodology illustrates areas ofuncertainty and highlights opportunities for technologydevelopment. The inviscid approximation, which wasnecessary to reduce computational time and memoryrequirements and enable multiple 3D airframecalculations at parametric conditions, necessitates theuse of approximate methods to determine viscous dragforces. The uncertainty in this prediction contributes tothe overall uncertainty in trim performance and net thrustpredictions at the powered scramjet test point.Interactions of the powered scramjet exhaust plume withairframe and control surfaces are not fully captured in thedata base approach. The 3D tip-to-tail analysis illustratedthe three-dimensional exhaust plume expansion in theaftbody region, which makes it difficult to define a forceaccounting system that separates propulsive andaerodynamic forces. Furthermore, the methodology usedfor the X-43A is highly configuration dependent andapplicable only to a point design. Future hypersonicairbreathing systems will require analysis across a broadtrajectory. Predictions, as well as data, for angle of attackor sideslip effects on plume expansion characteristics arelacking. No direct measurements were available for thecowl-open configurations to evaluate lateral-directionalstability, control surface effectiveness or associated trimdrag penalties under powered conditions. Pre-flight

predictions were based on unpowered cowl-closed wind-tunnel data and the effect of any plume interaction wasneglected in these models. Additional data on theseeffects is needed in the development of future systems.

A continued maturation of hypersonicairbreathing vehicle technologies is dependent on thedevelopment of advanced experimental andcomputational techniques to fully examine PAIcharacteristics of candidate designs and to exploitairframe-propulsive interactions to maximizeperformance. Higher fidelity analysis methods arerequired earlier jn the design and development process tofully evaluate trim performance. This implies a need toimplement algorithmic and hardware improvements thatimprove the efficiency of Navier-Stokes codes to reducecomputation times and enable studies over a wide rangeof conditions within reasonable time frames.Implementation of parallel methods offer improvementsin run-time efficiency. Other algorithmic advances, suchas multi-grid methods, dynamic grid adaptation andother convergence acceleration techniques, may provideimprovements. Advanced grid strategies, such as oversetor unstructured grids, may be appropriate for 3D vehiclecalculations and other complex geometries. Continuingadvances in physical modeling capabilities will alsoimprove the fidelity of predictions, including advancedturbulence models, transition prediction, turbulence-chemistry interactions, finite-rate chemistry models(particularly for hydrocarbon fuels) and modeling ofmulti-phase flows. Validation of models will beenhanced by correlations of CFD predictions, groundtest data and flight data for the X-43A. Continuedmaturation of these tools will progress towards the goalof solving full 3D tip-to-tail flowpaths with reacting flowchemistry in reasonable computation times as well as 3Dvehicle analyses over a wider range of conditions.

SUMMARYComputational Fluid Dynamics (CFD) and

other analytical tools have been used in the developmentand pre-flight analysis of the Mach 7 X-43A vehicle.Integrated aero-propulsion performance was predictedusing 3D inviscid airframe computations andengineering cycle analysis tools. Longitudinalperformance increments compare well with measuredincrements from integrated flowpath tests of the Hyper-X flight engine (HXFE). Surface pressure predictionscompare well with limited ground test data. CFD wasalso used to qualitatively assess lateral-directionalstability characteristics. Viscous computations were usedto evaluate scramjet flowpath component performanceand other phenomena. A 3D viscous tip-to-tailsimulation was performed which shows detailed flowfield characteristics of the vehicle at the target scramjetflight test condition. An assessment of analysis methods

169


may be used to highlight future technology developmentneeds for hypersonic airbreathing vehicles.

REFERENCES1. Freeman, Delma C. Jr.; Reubush, David E.;McClinton, Charles R.; Rausch, Vincent J. andCrawford, J. Larry: The NASA Hyper-X Program,Presented at the 48th International AstronauticalConference, October 1997.

2. Engelund, Walter C.; Holland, Scott D.; Cockrell,Charles E. Jr. and Bittner, Robert D.: AerodynamicDatabase Development for the Hyper-X AirframeIntegrated Scramjet Propulsion Experiment. AIAA2000-4006, August 2000.

3. Huebner, Lawrence D.; Rock, Kenneth E.; Witte,David W.; Ruf, Edward G. and Andrews, Earl H. Jr.:Hyper-X Engine Testing in the NASA Langley 8-FootHigh Temperature Tunnel. AIAA 2000-3605, July 2000.

4. Holland, Scott; Woods, William and Engelund, WalterC. Hyper-X Research Vehicle Experimental TestProgram Overview. AIAA 2000-4011, August 2000.

5. Berry, Scott; Auslender, Aaron H. and Dilley, ArthurD.: Hypersonic Boundary Layer Trip Development forHyper-X. AIAA 2000-4012, August 2000.

6. Godfrey, Andrew G.: General AerodynamicSimulation Program Version 3.0 User's Manual.Aerosoft, Inc., May 1996

7. Svehle, R.A. and McBride, B.J.: Fortran IV ComputerProgram for Calculation of Thermodynamic andTransport Properties of Complex Chemical Systems.NASA TND-7056, 1973.

8. Goldberg, U.C.: Separated Flow Treatment with aNew Turbulence Model. AIAA Journal, vol. 24, no. 10,pp. 1711-1713, 1986.

9. Huebner, Lawrence D. and Tatum, Kenneth E.:Computational and Experimental Aftbody flow fields forHypersonic Airbreathing Configurations with ScramjetExhaust Flows. AIAA 91-1709, June 1991.

10. Huebner, Lawrence D. and Tatum, Kenneth E.: CFDCode Calibration and Inlet-Fairing Effects on a 3DHypersonic Powered-Simulation Model. AIAA 93-3041,July 1993.

11. Tatum, Kenneth E.; Witte, David W; Monta, WilliamJ. and Walters, Robert W.: Analysis of Generic ScramjetExternal Nozzle Flowfields Employing Simulant Gases.AIAA 90-5242, December 1990.

12. Srinivasan, Shiva; Bittner, Robert D. and Bobskill,Glenn J.: Summary of the GASP Code Application andEvaluation Effort for Scramjet Combustor Flowfields.AIAA 93-1973, June 1993.

13. Loomis, Mark P.; Venkatapathy, Ethiraj;170

Papadopoulos, Periklis; Davies, Carol B.; Berry, Scott;Horvath, Tom and Campbell, Charles.: Aeroheating andAerodynamic CFD Validation and Prediction for the X-38 Program. AIAA 97-2478, June 1997.

14. White, Jeffrey A. and Morrison, Joe H.: A Pseudo-Temporal Multi-Grid Relaxation Scheme for Solving theParabolized Navier-Stokes Equations. AIAA 99-3360,July 1999.

15. Markotos, N.C.; Spalding, D.B. and Tatchell, D.G.:Combustion of Hydrogen Injected into a SupersonicAirstream (The SHIP Computer Program). NASA CR-2802, April 1977.

16. Ferlemann, Paul G.: Improvements to the SHIPComputer Code and Predictions of Vorticity EnhancedTurbulent Supersonic Mixing. Master of Science Thesis,George Washington University, Washington, D.C.,August 1993.

17. Pinckney, S. Zane.: Program SRGULL: AnAdvanced Engineering Model For the Prediction ofAirframe-Integrated Subsonic/Supersonic HydrogenCombustion Ramjet Cycle Performance. NASA TM1120, January 1991

18. White, J. T: X-43A Real Time Angle-of-AttackEstimation Scheme Utilizing Flush Airdata Sensing(FADS) System Inputs. AIAA 2000-4013, August 2000.

19. Huebner, Lawrence D.; Rock, Kenneth E.; Voland,Randall T. and Wieting, Allan R.: Calibration of theLangley 8-Foot High Temperature Tunnel forHypersonic Propulsion Airbreathing Testing, AIAAPaper 96-2197, June 1996.

20. Frendi, Abdelkader: On the CFD Support for theHyper-X Aerodynamic Database, AIAA 99-0885,January 1999.

23. McClinton, Charles R.; Voland, Randy; Holland,Scott D.; Engelund, Walter C.; White, J. T. and Pahle, JoeW.: Wind Tunnel Testing, Flight Scaling and FlightValidation with Hyper-X, AIAA 98-2866, June 1998.

[american institute of aeronautics and astronautics 18th applied aerodynamics conference -...

Documents