Document
AIAA 9313536
Three Dimensional Aerodynamic
Analysis of a High-Lift Transport
Configuration
Simha S. Dodbele
ViGYAN, Inc.
Hampton, Va
AIAA Applied Aeroaynamics
Conference
August 9-1 1 , 1993 / Monterey, CA
For permlsslon to copy or republish, contact the Amerlcan lnstltute of Aeronautlcs and Astronautics 370 L'Enfant Promenade, S.W., Washington, D . C . 20024 THREE DIMENSIONAL AERODYNAMIC ANALYSIS OF A HIGH LIFT TRANSPORT CONFIGURATION Simha S. Dodbelel ViGYAN Inc.
30 Research Dr.
Hampton, VA 23666 M Freesueam Mach number A bstrnc t P Local static pressure, Ibs/ft2 Two computational methods, a surface panel Free sueam static pressure, Ibs/ft2 method and an Euler method employing unstructured poo 9 Dynamic pressure, Ibs/ft?
grid methodology, were used to analyze a subsonic transport aircraft in cruise and high-lift conditions. The b/2 Semispan, 45 ft.
computational results were compared with two separate SEf Reference area, 980 fL2 sets of flight data obtained for the cruise and high-lift X, Y, Z Dimensional Cartesian coordinate system with the nose of the a q l a n e at X=13O".Y=O, Z=O configurations. For the cruise configuration, the surface a Aircraft angle of attack, deg.
pressures obtained by the panel method and the Euler method agreed fairly well with results from flight test.
0 Circumferential angle, (@=O on the upper However, for the high-lift configuration considcrable centerline)
differences were observed when the computational tan -' [Y/(Z-227.05)], deg
surface pressures were compared with the results from Abbreviations high-lift flight test. TSRV Transport Systems Research Vehicle On the lower surface of all the elements with the 2-D Two-dimensional exception of the slat, both the panel and Eulcr methods 3-D three-dimensional predicted pressures which were in good agreement with Introduction flight data. On the upper surface of all the elements the panel method predicted slightly higher suction compared to the Euler method. On the upper surface of The subject of high-lift systems has always been an the slat, pressure coefficients obtained by both the Euler arca of special interest to aircraft designers. Accurate and panel methods did not agree with the rcsults of the prediction of surface-pressure distributions, confluent flight tests. A sensitivity study of the upward deflection boundary-layers, viscous wakes, separated flow in the of the slat from the 40" flap setting suggested that the cove region, and separated flow regions over multi- differences in the slat dcflection bctween the element high-lift wings play an essential role in the computational model and the flight configuration could design of advanced high-lift systems'. The flow field be one of the sources of this discrepancy.
around a multi-element high-lift wing is very complex the The computation time for the implicit version of and highly interactive. As part of a subsonic-transport Euler code was about 1/3 the time taken by the cxplicit high-lift research program, a multi-phased flight version though the implicit code required 3 times the program is underway using the NASA-Langley memory taken by the explicit version.
Transport Systems Research Vehicle (TSRV) aircraft to obtain detailed full-scale flow measurements of the 5- Nomenclature element high-lift system at various flight conditions2. 3.
The availability of detailed measurements of pressure C Reference wing chord at 2Y/b=0.6, ft.
distributions and boundary layer flow parameters is C, Pressure coefficient, (p-pJqS,, critical to the validation and development of CD Drag coefficient, Drag/qSrcf computational methods. The complexities in the multi- CL Lift coefficient, LifUqS,, element flow field have so far restricted most of the C , Pitching moment coefficient, computational research investigations to two- Pitching moment/cqSref dimensional flow (2-D) and quasi-three-dimensional Cf Skin friction coefficient flow investigation^^‘^. This research paper mainly the fuselage, 89.54 ft.
L Length of focuses on modelling the complex three-dimensional ~~ *. Research Scientist, Associate Fellow AIAA 0 This paper is declared a work of the U.S.Govement and i s not subject to copyright protcction in the United States configuration. Source strengths are solved directly from flow field of the multi-elcment high-lift TSRV the external Neumann boundary condition using the configuration (Boeing 737-100). Using three- dimensional (3-D) methods, it is beneficial to identify normal component of the external flow. A set of linear regions of predominantly 2-D flow, highly interactive equations are obtained with doublet strengths as the regions involving three dimensional flow and vortex unknowns by imposing the internal Dirichlet boundary condition of zero perturbation potential inside the dominated regions. These studies can then be used to configuration. The set of linear equations are solved direct the appropriate computational mcthods for analyzing the flow physics around the complicated either by a direct method or by a blocked Gauss-Siedel iterative procedure depending upon the number of multi-element high lift system. Though the multi- element high-lift systems involve several regions of unknowns in the equations. The gradient of the doublet viscous flow, some of the dominant inviscid flow potential disuibution is used to obtain the surface features can be studied through a 3-D inviscid analysis. perturbation velocities. The wakes downstream of the trailing edges of the multi-element system are modelled These features include aspects such as strong wake as thin wake panels. The wake shapes change at the end vortex roll-up from the spanwise tips of the high-lift of each wake iteration in order to satisfy the force-free systems, which are frequently more powerful than the conditions on the wake panels. A converged wake shape tip vortices associated with cruise wings, significant is obtained when the wake shapes cease to change with spanwise compressibility effects, etc.
the wake iterations after which surface pressure In this paper, two methods have been uscd for computational analyses: a surface panel method based distributions are obtained. The overall aerodynamic on potential flow and an Euler method based on forces can be obtained by integrating the surface unstructured gnd methodology. Before attempting an pressure distributions.
analysis of the complex high-lift configuration, the An Euler method using unsmctured grid simpler cruise configuration was analyzed. This was methodology was used as the other computational done in order to assess the performance of the two method. An efficient Euler equation solver, USM3DI2, analysis methods without introducing the complexities was used to obtain the flow solutions on the associated with the high-lift system. The two rnelhods unstructured tetrahedral grid system around the cruise were then applied to study the high-lift configuration. and the high-lift configurations. This solver is based on The results obtained from the computational analyses of an upwind, cell centered, finite volume method. Explicit the cruise configuration are cornparcd with data as well as the implicit of USM3D code obtained on the fuselage in a flight test'' conducted were used to obtain computational results. The explicit a viscous drag reduction program at NASA under version uses the 3-stage Runge-Kutta time stepping Langley Research Center. "he computational results scheme with local time stepping and implicit residual from the analysis of the high-lift configuration are smoothing. The implicit scheme uses the linearized, compared with data obtained in a different flight test' backward Euler time differencing approach to update conducted under a high-lift program also at NASA the solution at each time step. Details of the implicit Langley Research Center. In h e computational study algorithm are described in Ref. 14. The unstructured involving the cruise configuration, thc fuselage, grids were generated b using a modified version of the horizontal and vertical tail assembly and cruise wing program VGRID3D" which is an interactive were modelled. In the high-lift study, the fuselage, tetrahedral gnd generator based on the advancing-front horizontal and vertical mil assembly, inboard-wing, con~ept'~. Utilizing a new 'structured background grid' mid-wing, outboard-wing, leading edge slats and triple concept16, a smooth grid size variation is achieved by slotted flap system were modelled. The triple slotted flap solving an elliptic partial differential equation on the system for the inboard-wing was not modelled in the uniform Cartesian background grid. The desired high-lift study. Additionally, the engine and pylons were distribution of grid spacing parameters in the field is in either study.
not modelled obtained by specifying a number of 'point' and 'line' sources, and solving a Poisson equation on the Cartesian Comm tational Ana I vses grid. Over the past few years, VGRID3D and USM3D have been successfully applied to several complex 3-D The three-dimensional surface panel method configuration^'^^^^^^^. This is the first time VGRID3D VSAERO" was uscd as one of the computational a 3-D has been applied to develop unstructured grids on methods for analyzing the cruise and high-lift high-lift aircraft configuration and also, this is the first configurations. The program VSAERO uses piecewise time Euler calculations have been made on a 5-element constant source and doublet singularities on high-lift aircraft configuration in subsonic flow using quadrilateral panels representing the surface of the unstructured grids.
chordwise pressure distributions computed by VSAERO ComDiitational Results and U S W D at a spanwise wing section of 2Y/b = 0.6.
Since the main aim of the flight testsl0 was to conduct flow investigation on the fuselage, pressures were
l a ) The Cruise Configuration
measured only on the fuselage and not on the cruise wing. As seen in the figure, on the upper surface near The TSRV cruise configuration is a relatively easy the leading edge, the panel method again predicts higher configuration to model computationally compared to the suction pressures compared to the Euler method.
more complex high-lift configuration. The cruise However, the results from the panel method and the configuration was analyzed using the surface panel Euler method are in good agreement for most of the method, VSAERO and the explicit Euler method, wing section. Figure 5 presents pressure distributions USM3D. For the VSAERO calculations, a total of 2215 quadrilateral panels were generated to represent the obtained using the two computational methods in the cruise configuration. Wakes from the trailing edges of circumferential direction on the top of the fuselage at the cruise wing, horizontal tail, vertical tail and the back the longitudinal station Xb0.402. The experimental end of the fuselage were modelled. The first wake line results are also presented for comparison. As seen in Fig.5, results from both the computational methods are from the cruise wing was fixed to the side of the fuselage and held rigid during the wake iterations. The close to the pressures from the flight experiment, rest of the wake lines were allowed to move freely. A jet although VSAERO tends to overpredict the suction type wake was prescribed for the wake trailing from the levels slightly.
From the above discussion we can infer that for this back end of the fuselage. Surface panels generated on the right side of the cruise configuration are presented simple cruise configuration, surface pressures can be in Fig. 1. The wakes trailing from differcnt components predicted reasonably well by any of the two computational methods, the panel method or the Euler of the configuration are also scen in thc figure. For thc method. The performance of these two computational Euler calculations, the unstructured grid was generated using VGRID3D. Figure 2 illustrates the unstructured methods when applied to a complex high-lift surface grid on the cruise configuration. The entire configuration is discussed in the next section.
computational grid consisted of 685000 tetrahedrons and a total of 124373 nodes representing one half of the f i ) The Hich-Lift Confieurntion complete configuration. A total of 30090 boundary surfaces and 15047 boundary nodes were used to model The more complex, high-lift configuration was the configuration surface, outer boundaries and plane of analyzed using the two computational methods with a symmetry. fixed flap setting of 40'. At this flap setting, the aircraft The results obtained by the two computational is in a landing mode with the inboard slat in a partially methods on the cruise configuration were compared extended position and the two outboard slats in the fully with the flight test res~lts'~. In the flight test, static extended position. The two outboard slats were pressures were measured at several longitudinal stations modelled as a single slat since both the slats are set at along the fuselage centerline and also in the fully extended position and at the same orientation to circumferential direction on the fuselage at several the wing. The inboard high-lift system was not longitudinal sections. Figure 3 presents comparisons of modelled in the study.
pressure coefficients obtained using VSAERO and The surface panel distribution generated on the USM3D, on the top symmetry line of the fusclage.
starboard side of the high-lift configuration by These results correspond to M = 0.5 and a = 6.87'.
VSAERO is presented in Fig. 6. A total of 3787 surface Pressure coefficients obtained from surface pressure panels were generated and distributed in 36 patches on measurements from the flight experiment between the the configuration. Panels were densely distributed on the longitudinal station X/L=0.402 and X/L=0.763 (the high-lift components and sparsely distributed on the beginning of the vertical tail surface) are also shown for fuselage and tail (see Fig. 6a). Wakes were modelled comparison. The panel method predicts higher suction from the trailing edges of the three leading edge slats, pressures compared to the Euler method at locations wing and triple slotted flap system. Again, the first wake downstream of the windshield up to about 25% of the line from the inboard wing was fixed to the side of the fuselage length from the nose. In the mid-section where fuselage and held rigid during the wake iterations. The the fuselage is flatter in the longitudinal direction the rest of the wake lines were allowed to move freely. The panel method and the Eulcr method predict pressures panel method solutions presented here are after 6 wake which are in very good agreement with rhc pressures iterations. A close-up view of the surface panels on the measured in the flight test. Figure 3 presents the triple-slotted flap system is shown in Fig. 6b.
For the Euler calculations, a total of 743,304 better.
The surface pressures on the foreflap, midflap and tetrahedrons were generated in the flow field with aftnap from the two computational methods are 28,073 triangular faces representing the surface of the cornpared with the results of the flight experiment i n configuration. The unstructured surface grid on the Figs. 9c-9e. For up to 50% of the foreflap, the Euler configuration is presentcd in Fig. 7. The surface grid method predicts slightly lower flow acceleration on the was deliberately made sparse on the fuselage and dense upper surface. The panel method on the other hand on the high-lift wing to get a better resoluuon grid on predicts flow acceleration very well on the upper surface the high-lift elements (see Fig. 7a). The grid is stretched of the foreflap. Towards the trailing edge of the foreflap in the spanwise direction to keep the total number of upper surface, the experimental pressure coefficients tetrahedrons reasonable and at the same time getting a level off departing from the predictions from both the good distribution of grid points in the chordwise direction. Detailed view of the unstructured surface grid computational methods. It is reported in Ref. 2 that for this flap setting and all a 2 -0.5 , flow separated near on the mple-slotted flap system is shown in Fig. 7b.
Additional care was taken to concentrate cells near the the upper surface trailing edge of the foreflap as evidenced by near-zero Cf and this separation could lading and trailing edges of the high-lift elements. This have been the result of the complex boundary layer flow was accomplished by carefully choosing proper magnitude and directions for the point and line sources development over the slat and the main wing and its in VGRID3D14. effect on the foreflap. On the lower surface, results from the Euier calculations are in much closer agreement During the flight test under the high-lift program, with the flight test results than the results from the panel static pressures were measured on the upper and lower surfaces of the slat, wing and triple slotted flap system at method.
For the midflap, as seen in Fig. 9d, the surface two different spanwise stations, one on the inboard the other in the midsection, at the spanwise pressures obtained from the panel method and the Euler section and station approximately 2 Y b 0 . 6 (Y=323"). The method are slightly higher than the flight test results on pressures measured at 2Y/b=0.6 are used in this paper to the m i r e lower surface. On the upper surface, the assess the computational results. A sectional view of the suction predicted from the panel method is higher than multi-element wing at 2Y/b=0.6 is shown in Fig. 8. Two the flight results on the entire upper surface. The Euler angles of attack, a= 6.250' and 7.617'. were chosen for andd paneel method calculations predict higher suction this computational study. These angles of attack are near the leading edge of the midnap.
close to aircraft approach angle during landing. In the Figure 9e presents the pressure distributions on the next section computational analysis for these angles of aftflap. On the upper surface, the suction obtained by the attack are discussed. panel method are much higher than the experimental results. The pressures from Euler calculations are in [l> Hirrh-lift Configuration at ci =6.25Q0 excellent agreement with the flight test results in spite of the strong boundary layer flow coming from the forward Computational results obtained for the high-lift lifting surfaces. On the lower surface of the aftflap, both configuration at a = 6.250', M = 0.2420 and at the flap the panel and the Euler calculations predict slightly setting of 40' are presented in Figs. 9a-9e. In the figures higher pressures than the flight pressures.
the pressure dismbutions obtained by both the computational methods are compared with the flight fil High-lift configuration at a = 7.617' data at the spanwise section 2Y/b=0.6.
The implicit version of the code USM3D was used to obtain Euler Computational results were obtained for the results. On the upper surface of the slat, both the panel conditions a=7.617O and Ad= 0.1722 and are presented and the Euler methods predict lower suction compared in Fig. 10. Two sets of flight data are used here for to the flight results (Fig.9a).
comparisons. Flight data used for the slat and wing For the wing, as shown in Fig. 9b, both the methods correspond to the conditions a=7.7Oo0 and M= 0.1956 predict lower suction compared to the flight test results and flight data uscd for the triple slotted system with the panel method predicting slightly larger suction correspand to the conditions a=7.617O and M= 0.1722.
than the Euler results. Both the computational methods It is believed that this small change in the flight fail to predict the suction peak near the leading edge. On conditions on the slat, wing and the triple slotted flap the lower surface, the surface pressures from the Euler system does not change the general conclusions drawn calculations as well as the panel method arc in excellent in this paper. In Fig. 10 results obtained from the panel agreement with flight data except near the trailing edge method and the results from both the explicit as well as where prcdictions from the Eulcr method arc much implicit Euler calculations are presented. As it is seen in and panel methods do not agree with the results of the Fig. 10 there is not much differcncc bctwcen the flight tests. In order to understand this discrepancy, pressure coefficients obtained from the two types of Euler calculations. sensitivity of the slat deflection on the pressures were As seen in Fig. loa, on the upper and lower surfaces studied using the panel method. Figure 11 presents changes in the pressure distribution on the slat due to 5 O of the slat, pressure coefficients obtained by both the and 10' upward slat deflections from the 40' Rap seuing Euler and panel methods do not agree with the results of position for a=6.25O0. From the figure it is seen that 100 the flight tests (Fig. loa). On the wing, the upward deflection of the slat brings the computational computational methods underpredict suction on the pressures closer to the pressures measured in the flight entire upper surface and both the methods fail to capture the suction peak near the leading edge observed in the experiments. This suggested that the discrepancy in the flight test (Fig. lob). On the lower surface of the wing, slat pressures between the computations and the flight except near the trailing edge, there is an excellent experiment shown in Figs. 9a and 10a are possibly agreement of pressures predicted by the computational caused by differences in slat deflections in the methods with the flight data. The results from the Euler computational model and in the actual aircraft. Another calculations are much better than the panel method possible reason for this desrepancy could be the effect of results near the trailing edge. aeroelastic deflections in flight.
Comparison of the surface pressures on h e foreflap Flow Field CharacteristiQ from the computational methods along with the results from the flight experiment are presentcd in Fig. 1Oc. On It is of considerable interest to visualize the wakes the upper surface of the foreflap, the results from the trailing from the slats and the flaps to understand the panel method are closer to the flight test results and the Euler calculations predict lower Row accclcration. But flow physics of the high-lift configurations. Fig. 12 on the lower surface the resuits from Eulcr calculations shows Mach number contours around the 5-element are in good agreement with the flight data. On the configuration at cr=7.617' and at the spanwise station of midflap, as seen in Fig. lOd, the surface pressures 2Y/b=0.6. In the same figure the wakes from surface obtained from the Euler methods are in slightly better panel method are also presented. In Fig. 13 the total agreement with the results of the flight expcriment than pressure contours calculated by the implicit Euler the panel method results. The panel mcthod ovcrpredicts mcthod at ct=7.617O and at X/L=0.828 are presented.
the inboard and outboard spanwise tips the suction on the upper surface. The Eulcr methods also The wake from of the triple slotted [lap system can be clearly seen from predicts higher suction on the first 50% of thc upper surface. the Euler computations. It is also clear from the figure As seen in Fig. IOe, on the upper surface of the tip that the vortex structure from the outboard spanwise aftflap, the results from the Euler calculations agrce very of the triple slotted flap system is stronger than the the uppcr surface, well with the flight test results. On vortex structure from the inboard spanwise tip of the the panel method predicts higher suction than the flap system. The vortex from the spanwise tip of the experimental results and on the lower surfuce, the panel wing is also seen in the figure and is just beginning t o method predict slightly higher pressures than the flight form.
pressures.
From these two angle of attack studies it is Panel Merhod-comnmtional 1 - apparent that the Eulcr method predictcd prcssures on the aftflap which are in excellent agreement with the In the high lift-configuration, one has to deal with flight test results in spite of strong boundary layer flow several highly complex wake shapes uailing from the coming from the forward elements. On part of the upper various elcments of the high-lift components. As seen surfaces of the foreflap Euler calculations predictcd in the previous discussions on high-lift configuration, lower suction than the flight test results. On the lower the panel method, VSAERO predicted higher suction surface of the elements with the exception of the slat, on the upper surfaces than the Euler methods. Some of both the panel and Euler methods predict pressures these differences could be traced to the wake treatments which are in good agreement with flight data. On the in the panel method. When the wake from the slat was upper surface of the wing both the computational initially fixed at a small distance above the wing, they methods underpredict suction.
had a tendency to bend inboard and pass very close to Generally, on the uppcr surface of all the clcmcnts the control points on the panels of the downstream the panel method predicted slightly hiyhcr suction lifting surfaces. This would cause the solution to diverge compared to the Euler method. On the upper surface of after 3 to 4 iterntions. In ordcr to avoid divergence in the the slat, pressure coefficients obtained by both the Euler solutions, the initial wakes from the slat was allowed to and moment coefficients have settled down after about trail downstream at a steep angle up to a distance of 25000 CPU secs for the implicit calculations with about 1/5 wing chord lengths in the longitudinal residuals reduced by about 3 orders of magnitude (see direction and about 1/5 wing chord lengths in the Fig. 14). The explicit code took about 75000 CPU secs direction normal to the wing and then allowed to trail at for the lift, drag and pitching moment coefficients to a constant height of 1/5 chord lengths above the wing.
settle down, about 3 times more than the CPU time This procedure would keep the solutions stable during taken by the implicit version.
the wake iterations and also leads to a converged wake.
However, it was found that slightly different wake shapes as starting solutions would converge to totally different "non-unique" wake.
Two computational methods, a surface panel In the calculations, it was also found desirable to the same method and an Euler method employing unstructured put panels with large mutual inIlucnce in grid methodology were used to analyze a subsonic block during mamx inversion. However, the limitation txansport aircraft in cruise and high-lift conditions. The on the block size in the Gauss-Siedel solution procedure computational results were compared with two separate put a restriction on the number of panels allowed in the sets of flight data obtained for the cruise and high-lift streamwise direction. This limitation resmcted the configurations. For the cruise configuration, the surface number of panels allowed in the streamwise direction.
pressures obtained by the panel method and the Euler This limitation in the block size was found to be very method agreed fairly well with results from flight test on critical for multi-element part-span high-lift the fuselage. However, for the high-lift configuration, configurations.
considerable differences were observed when the It was also found that the "Global" wake-grid- computational surface pressures were compared with planes option present in the panel method restricted the total number of wake panels and hence the size of the the results from high-lift Hight test.
On the upper surface of all the elements the panel panels immediately aft of the trailing edges of thc lifting method predicted slightly higher suction compared to elements. This restriction put a limitation on the size of the Euler method. On the lower surface of all the the wake panels.
elements with the exception of the slat, both h e panel ComDutational Details and Euler methods predict pressures which are in good agreement with flight data. On the upper surface of the The surface panel calculations were done on an SGI slat, pressure coefficients obtained by both the Euler and panel methods do not agree with the results of the flight 4D-320 VGX machine. The computational time for one angle of attack and one wake iteration for the high lift tests. A sensitivity study of the upward deflection of the configuration was typically about 1.75 hours. Each case slat from the 40' flap setting suggested that the was run for 6 wake iterations. differences in the slat deflection between the The Euler calculations were done on a CRAY-YMP computational model and the flight configuration could computer. The convergence history for both the explicit be one of the sources of this discrepancy.
and implicit Euler calculations for the case of one angle In the computational investigations, the panel of attack are presented in Fig. 14. The results presented method, VSAERO predicted higher suction on the in the figure correspond to the case of a = 7.617'. The upper surfaces than the Euler methods. Some of these implicit code required about 4 times the memory dirferences could be rtttributcd to the wake treatments in compared to the explicit version of the program. The the panel method. It was found that the "Global" wake computations were performed with a CFL number of grid planes resmcted the total number of wake panels 3.0. The solutions were started from free stream initial and hence the size of the wake panels immediately aft conditions in both the implicit and explicit calculations.
of the trailing edges of the lifting elements. The block The explicit calculations were stopped after the RMS size in Gauss-Siedel solution put a restriction on the average value of all the residuals &;?-norm) decreased number of panels allowed in the streamwise direction.
by about 1.7 orders of magnitude and since the This limitation was found to be very critical for part- calculations did not show any signs of further reduction span high-lift configurations. Convergence of wakes to in residuals. On the other hand the implicit calculations non-unique wake shapes have raised questions about the were run until the RMS average value of a11 the uniqueness and the accuracy of the solutions.
&-norm) decreased by about 5.2 orders of residuals The computation time for the implicit version of the magnitude. Fig. 15 shows the convergence history of Euler code was about 1/3 the time taken by the explicit the lift, drag and moment ccerficicnts for the explicit version though the implicit code required 4 times the and implicit calculations. As it is seen in Fig. 15 the lift mcmory taken by the explicit version.
Numerical and Physical Aspects of Aerodynamic Flows, California State University, Long Beach, Acknowledgments CA, January 1992.
10. Bertelrud, A.; Watson, R.D.; and McGinley, C.B.: The research was supportcd by NASA Langley Flow Measurements on the Fuselage of a Boeing Research Center under NASA Contract NAS 1-19672 to 737 Airplane. AIAA Paper 89-0209, January, ViGYAN, Inc., Hampton, Virginia. The author wishes to 1989.
acknowledge Mr. Arild Bertelrud and Mr. Long Yip of 11. Maskew, B: Program VSAERO, A Computer NASA Langley Research Center for providing flight Program for Calculating the Nonlinear data. The author would like to thank Mr. Dan Suash of Aerodynamic Characteristics of Arbitrary AMI, Inc., for several useful suggestions during the Configurations. NASA CR-166476, November, course of this work. The author would like thank Dr.
1982.
Shahyar Pirzadeh of ViGYAN, Inc., for assistance in the unstructured grid and Dr. Neal Frink for 12. Frink, N.T.; Pankh, P.: and Pirzadeh, S.: A F a s t generating providing the implicit USM3D code. Upwind Solver for the Euler Equations on Three- Dimensional Unstructured Meshes. AIAA Paper References 91-0102, January, 1991.
13. Frink, N.T.; Personal Communications.
1. Dilner, B.; May, F.W.; and McMasters, J.H.: 14. Anderson, W. K.: Grid Generation and Flow Solution Method for Euler Equations on Aerodynamic Issues in the Design of High -Lift Systems for Transport Aircraft. AGARD CP 365, Unstructured Grids, NASA TM-4295, April, 1992.
May 1984. 13. Parikh, P.; Pirzadeh, S.; and Lohner, R.: A Package 2. Vijgen,P.M.H.W.; Hardin, J.D.; and Yip, L.P.: Flow for 3-D Unstructured Grid Generation, Finite- Prediction over a Transport Multi-Element High- Element Flow Solutions, and Flow Field Lift System and Comparison with Flight Visualization. NASA CR-182090, September 1990.
Measurements. Fifth Symposium on Numcrical and 14. Pirzadeh, S: Recent Progress in Unstructured Grid Physical Aspects of Aerodynamic Flows, Long Generation. AIAA Paper 92-0445, January, 1992.
Beach, CA, 1992 15. Parikh, P.; Pirzadeh, S.; and Frink, N.T.: 3. Yip, L.P.; Vijgen, P.M.H.W.; Hardin, J.D.; and van Unstructured Grid Solutions to a Wing/Pylon/Store Dam, C. P: Subsonic High-Lift Flight Research on Configuration Using VGRID3DmSM3D. AIAA the NASA Transport Systems Research Vchicle Paper 92-4572, August, 1992.
(TSRV). AIAA Paper 924103, August, 1992 4. Morgan, H.L., Jr.: A Computer Program for the Analysis of Multi-Element Airfoils in Two- Dimensional Subsonic, Viscous Flows. NASA SP 347, March 1975.
5. Brune,.W.; and McMasters, J.H.: Computational Aerodynamics Applied to High-Lift systems in Applied Computational Aerodynamics. Progress in Astronautics and Aeronautics, Vol. 125, pp. 389- 433, AIAA 1990.
6. Mavriplis, D.J.; and Martinelli, L.: Multigrid Solution of Compressible Turbulent Flows on Unstructured Meshes Using a Two-Equation Model. AIAA Paper 9 1-0237, January, 1991.
7. Drela, M.: Newton Solution of Couplcdflnviscid Multi-Element Airfoil Flows. AIAA Paper 90- 1470, June 1990.
8. Cebeci, T.; Chang. K.C.; Clark, R.W.; and Ha1seyS.D.: Calculation of Flow over Multielement Airfoils at High Lift. J. Aircraft, Vol.
24,No. 8, pp. 546-551, August, 1987.
9.
Rogers, S. E.; Wiltberg, N. L.; Kwak. D.: Efficient Simulation of Incompressible Viscous Flow over Multi-Element Airfoils. Fifth Symposium on a FLIGHT TEST (25,000 ft) I "
... VSAERO
... VSAERO
- VGRJD3DNSM3D - VGRID3DNSM3D
OSO f
0.25 -
.o.oo - -c, -0.25 - -0.50 - -0.75 - 0.00 0.25 1.00 OSo X/L 0.75 O FLIGHT TEST (25,000 ft) I" - 1 0 ... VSAERO - VGRID3DNSM3D
O 5 t
" $ Fig.5 Pressure distribution on thc frlsclage at longitudinal scction X/L=0.402, a=6.87", M41.5 Surface pnnclc on thc high lift configuration v Unstnictrired surface grid on die high-lift coril'igiiraiiorl 175 I io0 750 ROO X Fig. 8 Scciinnal vicw of the multi-elerncnt L I irig (2Y/hA).6\
' FLIGHT TEST ( a=6.250, M=0.2420, upper surface) '.'
FLIGHT TEST ( a=6.250, M=0.2420, lower surface) *.'
- VGRID3D/USM30 (Implicit) --.VSAERO 4 .O . .
" O ? . . 0 0; -, 3.0 2 . 0 1 .o 0 .o , J I I I I -1 .o 0.0 0.2 0.00 0.10 0.0 0.2 0.00 0.20 0.40 0.60 0.80 0.00 0.20
XIC WC wc
X/C XIC (b) Wing (c) Foreflap (d) Midflap (e) Aftflap (a) Slat Fig. 9 Comparisons of the pressure distributions on the high-lift configuration 40" !=lap sctting, 2 Y h 4 . 6 , a=6.250'
' FLIGHT TEST ( a=7.700. M=0.1956, upper surface) '.'
FLIGHT TEST ( a=7.700, M=O. 1956. lower surface) *.'
0 FLIGHT E S T ( a-7.61 7 , M=O. 1722, upper surface) '.'
FLIGHT TEST ( ce7.617, M=0.1722. lower surface) '.'
- VGRID3D/USM3D (Implicit)
-.- VGRID3DIUSM3D (Explicit)
- - - VSAERO
I I 5 . 0 1
I , I , I I -c,
4.0 t 0
9.
0 1 I 3 .O 2 .o 1 .o 0 . 0 I 1 I I -1.0 0.00 0.10 0.00 0.0 0.2 0.0 0.2 0.00 0.20 0.20 0.40 0.60 0.80 X/C XIC XIC XIC X/C (a) Slat (b) Wing (c) Foreflap (d) Midflap (e) Aftflap Fig. 10 Comparisons of the prcssiire distributions on the high-lift configuration 40" Flap scuing. 2Y/b=0.6, a=7.617" - 40" Flap setting
-. . - 40" Flap setting- 5" deflection
- - - 40" Flap setting- IO" deflection
Upper surface Lowersurface I
x/c 0.1 0
I
Fig. 1 1 Sensitivity of the slat pressures due to slat deflections a=6.250", M=0.2420,40" Flap setting Fig. 13 Contours of total pressure at X L 4 . 8 2 8 for the high-lift configuration a=7.617", Ma. 1722,40" Flap setting Fig. 12 Contours of Mach number around the multi- element wing from Euler calculations a=7.617". M=O.1722,40" Flap setting
0.0 E - Implicit
... Explicit
- .
I
-7.5 0.OEO 2.5E4 5.OE4 7.5E4 1.OE5 CPU (Secs) Fig. 14 Comparisons of residuals from the explicit and implicit Euler solutions (a=7.617, M=O. 1722.40" flap setting) 2.5E4 5.OE4 7.5E4 CPU (Secs) Fig. 15 Convergence history of lift, drag and moment coefficients from the implicit Euler calculations (a-7.617, M=0.1722,40° flap setting)