/ ..../.12, AIAA-2002-2804 Computational Simulations of a Three- Dimensional High-Lift Wing M. R. Khorrami NASA Langley Research Center Hampton, VA M. E. Berkman, F. Li High Technology Corporation Hampton, VA B. A. Singer NASA Langley Research Center Hampton, VA 20th AIAA Applied Aerodynamics Conference 24-26 June 2002 St. Louis, Missouri For permission to copy or to republish, contact the copyright owner named on the first page. For AIAA-held copyright, write to AIAA Permissions Department, 1801 Alexander Bell Drive, Suite 500, Reston, VA, 20191-4344. Computational SimulationsofaThree-DimensionalHigh-Lift Wing Mehdi R. Khorrami 1 NASA Langley Research Center MS 128, Hampton VA Mert E. Berkman 2 and Fei Li3 High Technology Corporation 28 Research Drive, Hampton VA Bart A. Singer 4 NASA Langley Research Center MS 128, Hampton VA producing fluid dynamical processes at a flap side edge. Abstract The deeper insight gained through this work will guide the development of simplified physics-based models Highly resolved computational simulations of a three- that mimic flow unsteadiness (and thus noise generation dimensional high-lift wing are presented. The steady mechanisms) at the edge. Ultimately, such physics- Reynolds Averaged Navier-Stokes computations are based models will allow efficient design of quiet geared towards understanding the flow intricacies airplanes. associated with inboard and outboard flap side edges. Both moderate and high flap deflections are simulated. In a parallel effort to several companion experiments, Computed surface pressure fields accurately capture the our previous research focused on steady Reynolds footprint of vortices at flap side edges and are in Averaged Navier-Stokes (RANS) simulations of excellent agreement with pressure sensitive paint unswept and untapered high-lift configurations. 3-6 The measurements. The computations reveal that the configurations were generic two-or three-element high- outboard vortex possesses higher rotational velocities lift models that consisted of a two-dimensional (2D) and lower core pressure than the inboard vortex and main element with or without a 2D slat and a part span therefore is susceptible to severe vortex breakdown. flap. These early studies captured the complex nature of the flap side-edge flow field and revealed the Introduction intricacies of shear layer roll-up, multiple vortex formation, vortex merging, and vortex breakdown Projected future growth in air travel and significant processes. Two important aspects of a high-lift quieting of modern jet engines has brought renewed configuration left for future exploration were the effects attention to the non-propulsive (airframe) component of of sweep and taper on a flap side-edge flow field. aircraft noise. Past studies that focused on airframe noise have identified high-lift devices along with the These aspects were addressed in a series of tests landing gears as dominant noise producing conducted at NASA Langley Research Center (LaRC) components. 1'2 Those studies have established flap- in the 14x22 foot wind tunnel during 1998-1999. The side-edges as a potent noise source that deserves tested model is a trapezoidal (trap) wing design that focused attention. The present paper continues our provides a 3D high-lift flow environment. Extensive effort towards uncovering prominent flow structures at acoustic and limited aerodynamic measurements were a flap side edge in high-lift settings. These efforts are obtained. Using a microphone array technique, motivated by our lack of understanding of noise acoustic measurements were obtained by a team from the Boeing Company. Sample ground-ward flap acoustic spectra at moderate and high flap deflections 1Research Scientist, Computational Modeling and are shown in Fig. 1. At a flap deflection angle of 20 Simulation Branch, Associate Fellow AIAA. degrees, the emitted noise from the inboard and 2Research Scientist: currently with ArvinMeritor outboard side-edges are nearly similar both in Exhaust Systems. 3Senior Scientist, Member AIAA. 4Assistant Branch Head, Computational Modeling and Simulation Branch, Senior Member AIAA. amplitudaendfrequenccyontent.At30degreefslap in similar studies of high-lift configurations by deflectionw, hilethe inboardedgeshowsonlya Khorrami et al.4and Berkman et al.5. moderatreiseinnoiselevelsrelativetothelowerflap angle,thereis a markedincrease(8-10dB)in the Table 1. Geometrical settings amplitudoeftheradiatendoisefromtheoutboareddge. Clearly,the measuremenptsoint to significant Parameters Setting differencebsetweetnheestablishefldowfieldsatthe Slat angle, deg 25 twosideedgebsecausoefsweeapndtaper. Flap angle, deg 20, 30 Slat gap, % 1.5 Thecurrentcomputationsatludysimulatetshetrap Flap gap, % 1.5 wingflowfield.Emphasiissplacedonhighresolution Slat overlap, % 1.5 oftheflap'sside-edgeOs.urgoalsherearetocapture Flap overlap, % 0.5 thedifferencetshatmayexistbetweenthetwoflow fieldsaswellastohighlighttheprominencht angetos Two different configurations, corresponding to the thenoiseproducinfglowfeaturetshatmayoccurwith acoustic measurements displayed in Fig. 1, are increasinflgapdeflection. computed. In both cases, the slat deflection angle is 25 degrees and the main element angle of attack is 10 Model Geometry and Computational Grid degrees. The flap deflection angle is 20 degrees in one configuration and 30 degrees in the second The trap wing is a high-lift model consisting of a configuration, representing aircraft approach and constant chord leading edge slat, a main element, and a landing conditions, respectively. part span flap. Figure 2a displays the tested trap wing model which had a simple center body. To ease the A multi-block structured grid is used to simulate the gridding task, for our computational model the center flow past the trap wing. Limited use of patching body was removed and the wing was extended all the strategy is made to reduce the total number of points. way to the tunnel floor. As will be shown later, this High concentrations of grid points occur adjacent to the model alteration has no significant effect on local or model solid surfaces, slat cove, main element leading global aerodynamic characteristics of the trap wing. edge, and flap edges. The surface grid distribution is Figure 2b shows the trap wing geometry as simulated. shown in Fig. 3. Near the flap side-edges, the grids are Both main element and flap are swept and tapered. In clustered in the spanwise direction to resolve edge the stowed position, the leading edge and trailing edge vortices. Chordwise resolution is also high over the sweep are 33.89 and 16.24 degrees respectively. The flap. Grid concentration in the vicinity of the outboard simulated model has a root chord of 53.473 inches flap edge is displayed in Fig. 4 with a similar mesh (1.358m), a tip chord of 21.12 inches (0.536m), a mean clustering occurring at the inboard edge. The total grid aerodynamic chord of 37.3 inches (0.947m), and a span is comprised of 51 blocks with nearly 17.5 million of 85 inches (2.16m). The flap has an inboard chord of nodes. Typically, more than 20 grid points are 13.56 inches (0.344m), an outboard chord of 8.97 clustered in the boundary layers adjacent to the model inches (0.228m), and a span of 42.19 inches (1.07m). surface, with the first point less than 5x10 -6 wing-root- chords off the model surfaces. The relative positions of the slat and the flap with respect to the wing (i.e., the gaps and the overlaps) Flow Solver were fixed to match an experimental setting of interest. The settings for the simulated configurations are presented in Table 1. The tabulated gaps and overlaps As in our previous studies, the CFL3D code 7is used to are normalized with respect to the local stowed chord. perform the computations. CFL3D offers a wide All of the trailing edges of the tested model are blunt. selection of turbulence models, ranging from zero- to To ease the computational task and reduce the number two-equation models. For the present effort, the one- of grid points required, the surfaces of respective equation Spalart-Allmaras (SA) turbulence model s is elements were shaved smoothly (while maintaining the used with a solid-body rotation modification. 9 Because cambers) to produce sharp trailing edges. For a realistic of the strong centrifugal force field, the cores of comparison with the measured quantities, the entire test streamwise vortices displaying solid body rotation section of the 14x22 foot tunnel is modeled to simulate usually exhibit rather weak turbulent fluctuations and the experiment. The wind tunnel walls are treated as behave in a laminar-like manner. The present alteration inviscid surfaces to avoid the additional computational dampens the turbulent viscosity in regions cost of resolving the wall boundary layers. This demonstrating solid-body-rotation and has no effects on assumption had previously been shown to be adequate other regions of the flow field. In previous studies (Dacles-Mariani et al.9 and Khorrami et al._°), the modifiedSAmodewl asshowtnoproducbeetterresults Table 2 Grid-resolution studies forvortexdominatefldowss,uchasthepresenctase. Lift Coefficient Results and Discussion Flap Mid-level Fine-level Difference, deflection mesh mesh % Computations and postprocessing of the results are 20 degree 1.10 1.144 3.8 performed in a non-dimensional fashion. The scales used in our normalizations are the free stream speed of 30 degree 1.24 1.312 5.5 sound, density, kinematic viscosity, and the wing root chord. For the present case, reference flow variables are set to match conditions at the 14x22 test section Table 3Computed and measured lift coefficients entrance. All simulations are obtained for a free stream Lift Coefficient Mach number of 0.2 and the root chord Reynolds number of 5.65 million. Computations are done in a Flap Computed Measured Difference, deflection % fully turbulent mode using the one-equation SA model with the solid-body-rotation modification. 20 degree 1.144 1.125 1.69 30 degree 1.312 1.31 0.15 Because of the large grid, the computations were performed on the SGI clusters of National Aerodynamic Simulation facility at NASA Ames. Each We begin the discussion with a brief presentation of the individual run utilized 52 CPUs and slightly less than global flow before directing our attention to the details 12 GB of memory. A typical run, producing 100 of the flap side-edge flow fields. Figure 5 shows the iterations on the finest level, took nearly 6 hrs of pressure distribution over the entire model for the 30 physical time or approximately 300 hrs of CPU time. degree case. Note that regions of significant pressure To achieve a faster and more efficient convergence rate, suction occur near the wing and flap leading edges and mesh sequencing was employed with two coarser level both flap side edges. Figure 6 illustrates a comparison meshes (where successive removal of every other node of the computed chordwise pressure variation with that provides the lower level meshes). For both 20 and 30 measured over the main wing at the 50 percent span degree flap deflections, in a steady mode, convergence location. Except for the two bad ports near x=0.5, the was assumed when the calculations had no changes in measured Cp are in remarkable agreement with the the lift coefficients to 3-4 significant digits with computed pressure. The small discrepancies near the subsequent iterations. At this stage, the overall global trailing edge may be attributed to the sharpening of the residual displayed 5 orders of magnitude drop before edge in the simulation. The Mach number contours at leveling off. Although the residual remained flat for the the mid-span location for 30 degree flap deflection are 20 degree case, the residual for the higher flap angle presented in Fig. 7. The contour plot shows significant showed a somewhat low amplitude oscillatory acceleration of the flow through the slat and flap gaps behavior, suggesting the onset of an unsteady event. causing the Mach numbers in the vicinity of the main The probable cause of residual oscillation will be element and flap leading edges to approach a value discussed in the following sections. Nevertheless, the nearly twice the freestream Mach number of 0.2. The lift coefficient for the 30 degree case never displayed an recirculating zones and the corresponding low Mach oscillatory pattern and it converged to a finite value numbers in the slat and main element coves are also very smoothly. An estimate (in a global sense) on the vividly visible in Fig. 7. The simulation for 20 degree grid dependency of the computed solutions is obtained flap shows similar global features. via comparison of the lift coefficients. Tabulated in Table 2 are the computed lift coefficients from the mid- As mentioned in the introduction, surface pressure and fine-level grids. More importantly, in Table 3 we distributions near flap edges were obtained using the present a comparison between the fine-level computed Pressure Sensitive Paint (PSP) technique. The and experimentally measured lift coefficients. Good measured pressure distributions for the 20 degree flap agreement is achieved between the two sets of lift deflection are shown in Fig. 8a. The middle section of coefficient. This close agreement suggests that the the flap was not treated with PSP and therefore is removal of the center body from the simulated assigned a value of Cp=0.0. The flap leading edge configuration has had minimal impact on the experiences a higher suction peak towards the outboard aerodynamic characteristics of the wing. segment. In addition, both edges display a pear-shape low-pressure region that is the footprint of the main vortex after the vortex has moved on the top surface. These region of intense low pressures are quite similar tothefootprinotfvorticesforunswepatnduntapered First,vortexlift ismoreprominenatttheinboardedge flapsr,eporteidnreferenc4e-s6.Thecomputepdressure (z=0.39)and its contributionis increasedwith fieldontheflapisplottedinFigure8b.Thesimulated increasinfglapdeflection.Second,vortexlift is fieldisingoodagreemewnitthPSPresultandshows severeldyiminisheadttheoutboareddgeatthehigh thattheprominenflot wfeatureasssociatwediththeflap flapdeflectionG. iventhefactthattheoutboarvdortex arewellresolvedT.hecorrespondiCngpdistributions experiencleoswerpressureinsitscorethantheinboard forthe30degreeflapdeflectioanreplottedinFig.9. vortex,thelift distributiopnlotreinforceosurearlier Onceagainexcellenatgreemebnettweetnhemeasuredassertionthsat,theoutboarvdortexpathremaincsloser andcomputepdressuriessshownM. uchhighesruction to the edgethanthe inboardpath,and vortex peaksattheflapleadingedgeandinthefootprintosf breakdowfnirstoccursattheoutboareddgeandat the edgevorticesareobserved.Theshapeand lowerflapdeflections. orientatioonfthevortex-inducseudctionpeakisndicate that,onceontop,theinboardvortexmovesinwardof Thepresencoef a vortexbreakdowbnecomemsore theedgewhiletheoutboarvdortexremaincslosetothe evidenwthenthestreamwisveelocitydistributioinnthe edge.Fromthepointofviewofnoisegeneratioannd coreof vorticesis viewed.A sampleplotof the scatterinagtsharpedgesth,esesubtledifferencehsave streamwisveelocityforthe20degreceaseisshownin strongramifications. Fig.15.Thetwoplanesshownareslightlyinwardof theflapedges.Recaltlhatthefreestreamvelocityis Baseodntheacoustiacrraymeasuremetnhteso,utboard u=0.2. Duetoextremelowpressurebso,thvortices flapedgewasfoundtobeamorepotenntoisesource attainajet-likeaxialvelocityintheircores.Forthe thantheinboardedgethussuggestinthgattheremay inboardvortex,themaximumvelocitieasretypically existsomefundamentdailfferencebsetweetnhetwo 60-70percenhtighetrhanthefreestreavmelocity.The flowfields.Fortunateltyh,evolumetriCcFDdatabase outboarvdortexcoreshowanadditiona1l0-20percent can be analyzedthoroughlyto highlightthese higheraxialvelocitierselativetotheinboardvortex. differencesT.hepressurceontouraslongtheinboard- Noticethattheinboardvortexmaintaintshejet-like edgechordfor20degrefelapareplottedinFig.10and behavioinrtoregionwsellbeyondtheflaptrailingedge. thecorrespondincogntourfsortheoutboareddgein Ontheotherhandt,heexcesvselocityinthecoreofthe Fig.11. Thecontouprlotsshowtheformationofa outboarvdortexdisappeasrslightlydownstreaomfthe strongvortexnearthebottomsharpedgeandamuch trailingedge.Thecorrespondainxgialvelocityplotfor weakevrortexatthetopsharpcorner.Themerginogf the30degreecaseis displayeidnFig.16. At the thetwovorticesoccursatastationpasttheflapmid- inboardedge,startingatthemid-chordregion,the chordwhenthestrongevrortexhasmovedontotheflap vortexexperiencaesjet-likeaxialvelocityinitscore. top surface.Theoutboardvortexattainsa lower Atthisflapsettingth, emaximumcorevelocityis70-80 pressurien itscore,asit is evidenftromthemore percenhtigherthanthefreestream.Asthetrailing severesaturatioonfthepressurceontour(shollowed edgeoftheflapis approachetdh,eadversperessure regionsinsidethecores).Becausoefthecentrifugal gradienctauseasrapiddeceleratioonfthevortexcore forcebalanceth,elowerthecorepressurderopst,he velocity.Beyondthetrailingedget,hevortexbreaks highetrhevortexrotationavlelocitiebsecomeS.imilar downandthereisaregionofaxialflowreversal.In pressurceontouprlotsforthe30degreeflapdeflection contrastth,eoutboardvortexexperienctehseadverse are displayedin Fig. 12 (inboard)andFig. 13 pressurgeradienmt uchearlierandthereforebreaks (outboardA).ttheinboardedget,hevortexcoreattains downimmediatealyssoonasit movesonthetop alowersuctionpeakrelativetothe20degrecease(Fig. surfacein the mid-chordregion. Themaximum 10)andmaintaintshelowpressurbeeyondtheflap magnitudoef thereverseflow experiencebdy the trailingedge.Attheoutboareddgein, itially,thevortex vorticesis ontheorderof 5-10percenotfthefree showsextremellyowpressureinsitscore.However, streamvelocity.MonitoringoftheCFDdatabasaet oncethevortexmovesonflaptopsurfacew,ithina differingiteratiocnyclesrevealetdheonseotflowlevel veryshordtistanceit,experiencaesseverperessurriese flowunsteadineinssthebreakdowrengion.Giventhe in itscore. Typically,suchrapidpressurreiseis finespatiarlesolutioantthesideedgesit,wouldnotbe associatewdith thecoreexpansiona,ppearancoef surprisingif theunsteadiynternalstructureof the stagnatiopnointinsidethecorea,ndtheonseotfvortex vortexbreakdowins partiallybeingresolved.It is breakdown. believedthattheobservedoscillationin theglobal residuaislrelatetdothistime-dependaecntitvity. Tohelpunderstanthdedifferenceasssociatwediththe twoflapsettingsth,espanwislieft distributioonnthe Thepresencoefvortexbreakdowantaflapsideedge flapforbothdeflectionanglesisplottedinFig.14. wasfirstreporteidnreference[4s,5]andsubsequently Two featuresimmediatelydistinguishthemselves.byBerkmaental6. forunswepatnduntaperefdlaps. Thepresensttudyindicatetshatvortexbreakdowins References notuniquetothosesimplegenericconfiguratioannsdin realityoccurastanyflapsideedgeinahigh-lifstetting. Crighton, D., "Airframe Noise," Aeroacoustics of Thedetectioonfvortexbreakdowanttheoutboareddge Flight Vehicles: Theory and Practice, NASA raisesseveraiml portanqtuestionsIs.vortexbreakdown Reference Publications 1258, Vol.l, edited by H. solelyresponsibfloertheincreasendoiselevelsatthe H. Hubbard, WRDC TR 90-3052, Aug. 1991. flapoutboardedge?Isthecausalitbyetweevnortex breakdowanndincreasendoisedirecotrindirectS?ince . Macaraeg, M. G., "Fundamental Investigations of theinternasltructuroefvortexbreakdowonscillateast Airframe Noise," AIAA Paper 98-2224, June 1998. lowfrequencieosn,emayruleoutthedirecet ffecat nd assumtehatthenoisesourceasremodifiedindirectly Khorrami, M.R., Singer, B.A., and Takallu, M.A., viaincreaseRdeynoldsstresasctivitieasndotherlocal "Analysis of Flap Side-edge Flow Field for flowalterations.Of coursef,urtherexperimenitns Identification and Modeling of Possible Noise conjunctiownithaccuratuensteadfylowsimulations Sources," SAE paper No. 971917. Presented at the areneededto addresssomeof theseimportant 1997 SAE Noise and Vibration Conference and questions. Exposition, Ground Traverse, Michigan, May 1997. Conclusions . Khorrami, M.R., Singer, B.A., and Radeztsky, Steady Reynolds Averaged Navier-Stokes (RANS) R.H.J., "Reynolds Averaged Navier-Stokes simulations of a complex three-dimensional high-lift Computations of Flap-Side-Edge Flowfield," AIAA wing were conducted in an effort to understand the J., Vol. 37, No. 1,January, pp. 14-22, 1999. effect of sweep and taper on the flap-side-edge flow field. Flap deflection of 20 and 30 degrees, . Radeztsky, R. H., Singer, B. A., and Khorrami, M. representing aircraft approach and landing R., "Detailed Measurements of a Flap Side-Edge configurations, were computed. Emphasis was placed Flow Field," AIAA Paper 98-0700, Jan. 1998. on fine resolution of the inboard and outboard flap side edges to highlight the prominent differences between . Berkman, M., Khorrami, M.R., Choudhari, M., and the established flow fields at the two edges. Sadowski, S., "Investigation of High-Lift Flow Comparison between measured and computed lift Field of an Energy Efficient Transport Wing," J. coefficients showed very good agreement. Excellent Aircraft, Vol. 37, No.l, Jan.-Feb., pp. 45-52, 2000. comparison between PSP measurements of flap surface pressure and simulations were obtained both . Thomas, J., Krist, S., and Anderson, W., "Navier- qualitatively and quantitatively. Careful analysis of Stokes Computations of Vortical Flows over Low RANS database revealed the outboard edge vortex to be Aspect-Ratio Wings," AIAA J., Vol. 28, No.2, pp. the stronger vortex, attaining lower pressures and thus 205-212, 1990. higher rotational velocities in its core. Moreover, the outboard vortex path remains closer to the side edge Spalart, P.R. and Allmaras, S.R., "A One-Equation while that of the inboard vortex is more inward of the Turbulence Model for Aerodynamics Flows," edge. The higher suction peaks in the core of the AIAA Paper 92-0439, 1992. outboard vortex make it susceptible to undergo vortex breakdown at lower flap deflections relative to the . Dacles-Mariani, J., Rogers, S., Kwak, D., Zilliac, inboard vortex. G., and Chow, J., "A Computational Study of Wingtip Vortex Flowfield," AIAA Paper 93-3010, Acknowledgements July 1993. The work of the second and third authors was 10. Khorrami, M. R., Li, F., and Choudhari, M., "A sponsored by NASA contract NAS1-00088 (Lockheed Novel Approach for Reducing Rotor Tip-Clearance Martin Corporation subcontract RT46324). The Induced Noise in Turbofan Engines, " AIAA Paper computations were performed on the SGI computer 2001-2148 (also to appear in July 2002 issue of clusters provided by the National Aerodynamic AIAA J.), May 2001. Simulation (NAS) Facility. The authors would like to thank Mr. Michael Wiese formerly of Computer Sciences Corporation for constructing the computational grids. 85- 8O ,-,75 Ill v ..i 7O Ix 65 _ 6o 0 _ 55 -_ 50 Ix 45 4O • d 35 I I I I { I I I I 10 20 30 4050 Frequency (kHz) Figure 1. Effect of flap angle on radiated sound from side edges. Figure 3. Surface grid distribution. l Figure 4. Grid distribution in vicinity of flap inboard edge. a) Tested b) Simulated Figure 5. Computed surface pressure distribution for Figure 2. Trapezoidal wing geometry. 30 de_ee flap. -3 -2 0 [] [] 1 a) Experiment 0 0.2 0.4 0.6 0.8 X Leading Eclgo Figure 6. Chordwise pressure distribution on main element at 50 percent span. b) Computation Figure 7. Mach contours at mid-span location for 30 Figure 8. Comparison between PSP and computed de_ee flap. surface pressure on flap suction surface for 20 de_ee deflection. Inboard edge is on right and outboard edge on left and flow direction is from top to bottom. Pressure Sensitive Paint ....... ._¢_{_iiiii._ii_,.a.:;,_iiiiiiiiiii_i_iiii.,._i_i_i_i{_i_i_i_ _N_i_iIgi_i#iiiiiii_iIiiiiiiiiliiiiiiliii_ i.l....._.{.i_.,._..'.;.._.._.._..._;_._.;.._:_ _;_i{!i_iliiiiil_i_iililiiii!!iii!illiii!i!iililiiiiilii_ii'_ii_{i_ii_i_! a) Experiment Figure 11. Pressure contours along outboard edge for 20 degee flap deflection. Computational Simulation ........_.;.._,#:._m" •"..e_%._ .......... : :: :.......... : ::+ ::. ......:.._.. .......... J_ -0-.5i!{ b) Computation Figure 9. Comparison between PSP and computed surface pressure on flap suction surface for 30 degee Figure 12. Pressure contours along inboard edge for 30 deflection. Inboard edge is on right and outboard edge degee flap deflection. on left and flow direction is from top to bottom. 0 66 Figure 10. Pressure contours along inboard edge for 20 Figure 13. Pressure contours along outboard edge for degree flap deflection. 30 degree flap deflection. 0.3 0.25 0.2 "6o.15 =_ 0 0.1 (,,) ,1=0.05 =,__1 0 ........ flap20 degree ....................f.la.p 30 degree -0.1 I , , L , I , i , , I i , L : I , : , , 0.4 0.6 0.8 1 Flap Span Figure 14. Spanwise lift distribution on flap. Inboard edge is at z=0.39 and outboard edge at z= 1.116. iiiiiiiiiiiiiiiii y i . Figure 15. Streamwise velocity field in planes adjacent Figure 16. Streamwise velocity field in planes adjacent to side edges for 20 degree flap deflection. to side edges for 30 degree flap deflection.