TAU Experiences with Detached-Eddy Simulations

Transcription

1 TAU Experiences with Detached-Eddy Simulations Herbert Rieger & Stefan Leicher EADS Deutschland GmbH Military Aircraft Flight Physics Department Ottobrunn, Germany

2 Outline The Typical Design Problem of Military Aircraft Some Examples From Simple to Complex NACA0012/21 M219 cavity FA-5 Aircraft The Future Needs & Expectations Remarks on TAU code Page 2

3 Outline The Typical Design Problem of Military Aircraft Some Examples From Simple to Complex NACA0012/21 M219 cavity FA-5 Aircraft The Future Needs & Expectations Remarks on TAU code Page 3

4 The Design Problem - Combat Aircraft Flow Phenomena vortex burst flow separation fin buffeting J.F. Campbell, J.R.Chambers, NASA SP-514,1994 Manoeuvre Performance strongly dependent on vortex-induced (nonlinear) lift maximum of vortex-induced lift dependent on vortex stability Flight Control and Stability roll stability depends on asymetric vortex breakdown Aeroelasticity unsteady vortices influence flutter speed & limit-cycle oscillations Structural Dynamics unsteady vortices influence structural fatigue of controls and stabilizers Page 4

5 Aerodynamic Longitudinal Control Characteristics: F/A-18E F/A-18E Abrupt Wing Stall Problem Lift Rolling Moment α AIAA : J.R. Forsythe & S.H. Wodson - Unsteady CFD Calculations of Abrupt Wing Stall Using DES α Page 5

6 Aerodynamic Longitudinal Control Characteristics: F/A-18E F/A-18E Abrupt Wing Stall Problem What is the right time step size to resolve inertial and important energy ranges properly? AIAA : J.R. Forsythe & S.H. Wodson - Unsteady CFD Calculations of Abrupt Wing Stall Using DES Page 6

7 Outline The Typical Design Problem of Military Aircraft Some Examples From Simple to Complex NACA0012/21 M219 cavity FA-5 Aircraft The Future Needs & Expectations Remarks on TAU code Page 7

8 NACA0012/21 NACA0012: Mach 0.2, a = 20, Re/c = TAU-SADES Strelets - Shur (Low Re Mod.) EXP. C D (3%) 0.30 C L (1%) 0.66 Page 8

9 NACA0012/21 O-Grid 161x113x41 (1 chord span 41pts.) instantaneous solution vorticity magnitude NACA0021: Mach 0.1, a = 60, Re/c = TAU-SADES turbulent to laminar eddy viscosity ratio Page 9

10 NACA0012/21 NACA0021: Mach 0.1, a = 60, Re/c = Tau - SADES: EADS-Grid FLOWer- SADES EADS-Grid TAU - SADES NTS-Grid Time-Averaged Surface Presssure Distribution Total Forces Tau EADS-Grid FLOWer EADS-Grid TAU NTS-Grid Experiment Cl 0.89 (-4.3%) 0.98 (+5.3%) 0.87 (-6.5%) 0.93 Cd 1.45 (-6.5%) 1.52 (-1.9%) 1.39 (-10.3%) 1.55 Page 10

11 NACA0012/21 Cl NACA0021 Mach 0.1, a = 60, Re/c = Cd Problems to resolve the 2 nd dominant frequency Hz RMS spectral density of Lift & Drag Page 11

12 Outline The Typical Design Problem of Military Aircraft Some Examples From Simple to Complex NACA0012/21 M219 cavity FA-5 Aircraft The Future Needs & Expectations Remarks on TAU code Page 12

13 M219 cavity Sketch Sketch of of M219 M219 experimental experimental configuration configuration Location Location of of pressure pressure tabs tabs cavity cavity ceiling ceiling and and front/rear front/rear plate plate Page 13

14 M219 cavity SADES XLES M = 0.85 Vorticity Magnitude SADES XLES Turbulent eddy viscosity to laminar viscosity ratio Instantaneous Solutions at longitudinal middle section Unstructured Hybrid Grid (6.2 Mil. Nodes), Dt = sec Page 14

15 M219 cavity SADES XLES M = 0.85 K20 K20 SADES XLES K29 Instantaneous Solutions at cross sections Unstructured Hybrid Grid (6.2 Mil. Nodes), Dt = sec K29 Vorticity Magnitude Page 15

16 M219 cavity SADES XLES M = 0.85 K20 K20 SADES XLES K29 Instantaneous Solutions at cross sections Unstructured Hybrid Grid (6.2 Mil. Nodes), Dt = sec K29 Turbulent eddy viscosity to laminar viscosity ratio Page 16

17 M219 cavity M = 0.85 Normalized RMS- Pressures [kpa] K20 K29 Static Pressure RMS Values at various longitudinal stations Frequency [Hz] Page 17

18 M219 cavity SADES M = 1.35 Normalized RMS- Pressures [kpa] K20 K29 Static Pressure RMS Values at various longitudinal stations Frequency [Hz] Page 18

19 First Steps into DES methods M219 cavity Remarks cavities are models for internal store compartments most important are structural fatigue & store release problems SADES and XLES results are general in good agreement RMS pressure data are in fair agreement with experiments amplitudes of RMS data are generally overestimated positions of tone frequencies of RMS and SPL data are in good agreement with experiments up to approx Hz better results may require higher mesh resolution and shorter time steps as well as higher number of data samples for statistics Page 19

20 Outline The Typical Design Problem of Military Aircraft Some Examples From Simple to Complex NACA0012/21 M219 cavity FA-5 Aircraft The Future Needs & Expectations Remarks on TAU code Page 20

21 Full Aircraft FA-5 Start of a small DES project in 2003 within EADS-M Partner : TUM experiments TUB & DLR numerical simulations General Aim : Validation of DES simulation for full aircraft configuration Codes: ELAN-3D, FLOWer, TAU Technical Aims: What relevant parameters w.r.t. computational mesh and CFD methodology haveto be usedto obtain good time-averaged results for forces and moments to engineering accuracy. What are the time step resolution requirements to involve thetechnically (FCS-) relevant modes Page 21

22 Full Aircraft FA-5 FA-5 5 Aircraft Geometry & Model Characteristics Page 22

23 Full Aircraft FA-5 Flow Velocity Reynolds Number Flow Conditions Angle of Attack Sideslip Angle Canard Position Configuration Parameters Slat Position Flap Position Aquisition Rate (per channel) Low Pass Filter Measurement Period / No. of Samples Data Aquisition Parameters FA-5 5 Aircraft Configurations, Flow Conditions & Data Acquisition System Page 23

24 Full Aircraft FA-5 a=24 zero control Turbulence Intensity of u-velocity u Component canard 20, slats 20 Page 24

25 Full Aircraft FA-5 FA5 Configuration Block-Structured Mesh 3 level multigrid possible 225 mesh blocks 10 7 grid points (half configuration) 10 5 grid surface points Page 25

26 Full Aircraft FA-5 FA-5 5 Trial Computations FA5 Wing FLOWer, TAU, structured mesh, Mach 0.2, Re lµ = 0.97*10 6, α=15, 28 FA5 Full Configuration (E1) FLOWer, TAU, structured mesh, Mach 0.2, Re lµ = 0.97*10 6, α=15, 28 FA5 Full Configuration (E1) TAU, structured mesh, Mach 0.7, Re lµ = 8*10 6, α=28 Page 26

27 Full Aircraft FA-5 TAU-SADES Mach 0.2, α =15o, Relmue = Instantaneous turbulent eddy viscosity structure timetime -averaged lambda2lambda2 -field Page 27

28 First Steps into DES methods Full Aircraft FA-5 FA5 Configuration: TAU/SADES Mach 0.2, a =15 o, Re lmue = stations: x/c=0.8 / 0.9 Page 28

29 Full Aircraft FA Velocity spectral analysis at x/c=0.9 Page 29

30 Full Aircraft FA-5 Power Spectral Density - U Power Spectral Density - V Power Spectral Density - W Pts. 1 Experiments TAU-SADES Experiments TAU-SADES Experiments TAU-SADES Frequency [Hz] Frequency [Hz] Frequency [Hz] Page 30

31 Full Aircraft FA-5 DES-EASMEASM RANS-EASM Experiment DES-EASM Experiment U/U u rms /U DES-EASMEASM RANS-EASM Experiment DES-EASM Experiment V/U v rms /U DES-EASMEASM RANS-EASM Experiment DES-EASM Experiment W/U Time-Averaged Velocity- and Turbulence Intensity Components at x/c=0.9 w rms /U Page 31

32 Full Aircraft FA-5 A Power Spectral Density of Velocity Fluctuations FA5 Configuration: ELAN3D: a =15 o, Re lmue = Page 32

33 Full Aircraft FA-5 Remarks firstresults show qualitatively correctvortex structures for mean flow quantities quantitative comparisons with experiments exhibitdeficiencies w.r.t. - mean velocity components (l.e. vortex region, vortex burst) - turbulence intensities - spectral content (amplitudes) at control stations there are indications that a background turbulence model using turbulence length scales as a switch between RANS and LES region may be beneficial further basic validation efforts are necessaryto get industrial confidence (mesh refinement, time step, background turbulence model, etc.) Page 33

34 Outline The Typical Design Problem of Military Aircraft Some Examples From Simple to Complex NACA0012/21 M219 cavity FA-5 Aircraft The Future Needs & Expectations Remarks on TAU code Page 34

35 Future Needs & Expectations (1) DES methodology as an industrial tool is a promising approach For baseline configurations (cylinder, profiles at high AoA, cavities) DES provides already now solutions with sufficient accuracy w.r.t. to spectral quantities For complex configurations the situations is - for the moment - unclear whether or not DES will definitely better the solution quality w..r.t. to overall forces and moments However basic issues have still to be resolved: - control of RANS / LES switching - transition handling - grid dependent method properties giving rise to damaged phyics (grid induced separation, influence oft turbulence transport from RANS interface, etc.) Page 35

36 Future Needs & Expectations (2) From industrial point of view unsteady methods like DES are needed for the following aircraft design items : - high AoA flows - buffet & fatigue control - aeroelasticity - aeroacoustics - inlet compatibility (dynamic distortions) - manoeuvre simulation - FCS check-out and optimisation - pre- and post flight test analysis For industrial use one may expect the development of the hybrid RANS-LES methodology which allows to resolve the engineering relevant scales in a systematic way. As aircraft design needs analysis support with known uncertainty limits for future development processes hybrid RANS-LES methodology could become a key technology Page 36

37 Outline The Typical Design Problem of Military Aircraft Some Examples From Simple to Complex NACA0012/21 M219 cavity FA-5 Aircraft The Future Needs & Expectations Remarks on TAU code Page 37

38 Remarks on TAU Code TAU code is able to produce acceptable results already for industrial relevant but simple configurations Progress of TAU code s quality for DES computations may evolve as projects like DESIDER provide further modeling insight However DES computations are presently lengthy undertakings. For statistical purposes a large number of time steps (5-30 k) are needed Assuming for each time step sub-cycles computations come into the order of 10 6 cycles in an industrial environment such efforts are too costly! Key point is to raise TAU code solver efficiency to lower considerably the computational work for one time-step Therefore industry demands sustained research in solution techniques of utmost efficiency!! Page 38

39 END Thanks for your attention Page 39

40 Page 40