1 st Automotive Workshop, Oxford, UK Paul Batten, Jian Wang & - PowerPoint PPT Presentation

1 st Automotive Workshop, Oxford, UK Paul Batten, Jian Wang & Oshin Peroomian Dec. 11/12, 2019 1

Overview Challenges • Software • Case 1 • Case 2A • Summary • 2

Challenges Attached BLs Massive Small/shallow separation separation N.B. In anticipation of the various separation types and extents, hybrid RANS/LES simulations that used IDDES imposed a box over the forward sections of the simulated vehicles in which IDDES was maintained in RANS 3 mode (this is referred to as an “LES deactivation box”)

Software • CFD Solver Used: CFD++ by Metacomp Technologies, Inc. • Steady-state and transient finite-volume solutions • Linear eddy-viscosity RANS models • Non-linear eddy viscosity (EARSM) RANS models • Hybrid RANS/LES (DDES and IDDES) • Meshing Software: MIME by Metacomp Technologies, Inc. • Multipurpose Intelligent Meshing Environment • General size automation and curvature-based refinement • Hex-dominant meshes • Solve-to-wall meshes created for each MIME mesh 4

Case 1 - Models • RANS: • SA • SARC+QCR • Realizable k-epsilon (RKE) • Cubic k-epsilon (CKE/cubic k- e EARSM) • SST • Hellsten (quartic k- w EARSM) • Hybrid RANS/LES • IDDES + LES deactivation box 5

Case 1 - Meshes • Workshop RANS Mesh: 4 M cells • Half-model mesh (plus symmetry plane) • y + > 1 over much of the body • Large spacings near underside wedge/ramp • • MIME RANS Meshes: Three half-model meshes generated: coarse (5M), medium (10M), fine (15M) • Only fine-mesh results presented in subsequent slides • y + < 1 everywhere, for all three meshes • 6 MIME fine RANS half-mesh

Case 1 - Meshes • Workshop Hybrid RANS/LES Mesh: 30.6 M • Full-model mesh • Growth rate 1.15, 30 cell layers in BL • Cell size at rear refined region = 1.7 mm, base (1.2 mm), underbody (2.4 • mm), ground (4.8 mm), rear window (1.2 mm), top and front (2.4 mm) First layer height = 3.7e-5m • 7

Case 1 - Meshes • MIME Hybrid RANS/LES Mesh: 99 M • Full-model mesh • Growth rate 1.17, 45 cell layers in BL • Cell size at rear refined region = 1.7 mm, base (1.2 mm), underbody (2.4 • mm), ground (4.8 mm), rear window (1.2 mm), top and front (2.4 mm) First layer height = 7.5e-6 m • 8

Case 1 - RANS Convergence on Fine MIME Mesh Realizable k-epsilon 9 Cubic k-epsilon

Case 1 - Convergence on Fine MIME RANS Mesh SA 10 SARC+QCR

Case 1 - Convergence on Fine MIME RANS Mesh SST 11 Hellsten

Case 1 - IDDES Force History for Fine MIME Hybrid RANS/LES Mesh IDDES 12

Case 1 - Selected Mesh- Convergence Plots: C d 0.24 CKE 0.22 SARC+QCR SA 0.2 Cd RKE 0.18 SST 0.16 0.14 0 2 4 6 8 10 12 14 16 Million Cells 13

Case 1 - Selected Mesh- Convergence Plots: C l -0.055 CKE -0.065 SARC+QCR -0.075 SA CL -0.085 RKE -0.095 SST -0.105 -0.115 0 2 4 6 8 10 12 14 16 Million Cells 14

Case 1 - Cp Centerline Distribution Realizable k-epsilon Cubic k-epsilon N.B. Red dots = Experiment 15

Case 1 - Cp Centerline Distribution SA SARC+QCR N.B. Red dots = Experiment 16

Case 1 - Cp Centerline Distribution SST Hellsten N.B. Red dots = Experiment 17

Case 1 - Cp Centerline Distribution DDES IDDES + box N.B. Red dots = Experiment 18

Case 1 - Wake Profiles Exp. Realizable k-epsilon 19

Case 1 - Wake Profiles Exp. Realizable k-epsilon 20

Case 1 - Wake Profiles Exp. Cubic k-epsilon 21

Case 1 - Wake Profiles Exp. Cubic k-epsilon 22

Case 1 - Wake Profiles Exp. SA 23

Case 1 - Wake Profiles Exp. SA 24

Case 1 - Wake Profiles Exp. SARC+QCR 25

Case 1 - Wake Profiles Exp. SARC+QCR 26

Case 1 - Wake Profiles Exp. SST 27

Case 1 - Wake Profiles Exp. SST 28

Case 1 - Wake Profiles Exp. Hellsten 29

Case 1 - Wake Profiles Exp. Hellsten 30

Case 1 - Wake Profiles Exp. DDES 31

Case 1 - Wake Profiles Exp. DDES 32

Case 1 - Wake Profiles Exp. IDDES+box 33

Case 1 - Wake Profiles Exp. IDDES+box 34

Case 1 - IDDES: Normalized Q-Criterion IDDES+box 35

Case 1 - Fine MIME Mesh Results Test Cases Cd Cl Cm Realizable k-epsilon 0.1917 -0.08385 -0.1195 Cubic k-epsilon 0.1844 -0.07547 -0.1121 SA 0.2297 -0.1022 -0.1097 SARC+QCR 0.2339 -0.08895 -0.1189 Hellsten 0.1962 -0.07850 -0.1280 0.2223 ( d =0.006) -0.08457 ( d =0.014) -0.1406 ( d =0.013) SST (Deep convergence not achieved) IDDES coarse (28 M cells) 0.1991 -0.07384 -0.1231 IDDES fine (99 M cells) 0.2002 -0.07624 -0.1218 D. Wood, SAE 2014-01-0590, 2014 0.210 0.055 ß ? D. Wood, PhD Thesis, 2015 0.210 -0.035 (-0.0250 ~ - ß ? 0.0465) 36

Case 2A - Models • RANS: • SA • Realizable k-epsilon (RKE) • SST • Hellsten (quartic k- w EARSM) • Hybrid RANS/LES • IDDES (unmodified - for demonstration purposes only) • IDDES + LES deactivation box 37

Case 2A RANS Forces – Coarse Grid Instantaneous C d - Coarse Grid Instantaneous C l - Coarse Grid 0 . 5 0 . 2 SA SA 0 . 4 RKE RKE SST SST 0 . 1 Hellsten Hellsten 0 . 3 C d C l 0 . 2 0 . 0 0 . 1 200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400 Iteration Iteration 38

Case 2A RANS Forces – Medium Grid Instantaneous C d - Medium Grid Instantaneous C l - Medium Grid 0 . 5 0 . 2 SA SA 0 . 4 RKE RKE SST SST 0 . 1 Hellsten Hellsten 0 . 3 C d C l 0 . 2 0 . 0 0 . 1 200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400 Iteration Iteration 39

Case 2A RANS Forces – Fine Grid Instantaneous C d - Fine Grid Instantaneous C l - Fine Grid 0 . 5 0 . 2 SA SA 0 . 4 RKE RKE SST SST 0 . 1 Hellsten Hellsten 0 . 3 C d C l 0 . 2 0 . 0 0 . 1 500 1000 1500 2000 2500 500 1000 1500 2000 2500 Iteration Iteration 40

Case 2A IDDES Forces – Fine Grid 41

Case 2A Mean Forces – Coarse Grid Cumulative Mean C d - Coarse Grid Cumulative Mean C l - Coarse Grid 0 . 2 SA SA RKE RKE 0 . 1 SST SST Hellsten Hellsten 0 . 25 C d 0 . 0 C l − 0 . 1 0 . 20 200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400 Iteration Iteration 42

RANS Mean Forces – Medium Grid Cumulative Mean C d - Medium Grid Cumulative Mean C l - Medium Grid 0 . 2 SA SA RKE RKE 0 . 1 SST SST Hellsten Hellsten 0 . 25 C d 0 . 0 C l − 0 . 1 0 . 20 200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400 Iteration Iteration 43

RANS Mean Forces – Fine Grid Cumulative Mean C d - Fine Grid Cumulative Mean C l - Fine Grid 0 . 2 SA SA RKE RKE 0 . 1 SST SST Hellsten Hellsten 0 . 25 C d 0 . 0 C l − 0 . 1 0 . 20 500 1000 1500 2000 2500 500 1000 1500 2000 2500 Iteration Iteration 44

Hellsten Model Forces – All Grids Hellsten Model - C d Hellsten Model - C l 0 . 250 0 . 08 Coarse Coarse 0 . 245 Medium Medium Fine 0 . 06 Fine 0 . 240 C d C l 0 . 04 0 . 235 0 . 02 1000 2000 3000 4000 5000 1000 2000 3000 4000 5000 Iterations Iterations N.B. C l still not well converged on fine mesh! 45

Case 2A - Mesh Convergence 46

IDDES Mean Forces – Fine Grid 47

Wake Profiles – Fine Mesh SA RKE SST IDDES (no box) IDDES + box Hellsten 48

Centerline Cp – Upper Surface 49

Centerline Cp – Lower Surface 50

Separation Isosurfaces Realizable k-epsilon SST SA IDDES (no box) IDDES + box Hellsten 51

Spanwise Vorticity Realizable k-epsilon SA SST Hellsten IDDES (no box) IDDES + box 52

IDDES - No Box : NQcrit Normalized Q-Criterion 53

IDDES + Box : NQcrit Normalized Q-Criterion 54

Case 2A - Fine Workshop Mesh Results Test Cases Cd Cl Realizable k-epsilon 0.2097 0.01375 * SA 0.2231 0.01870 * SST 0.2363 0.004877 ** Hellsten 0.2418 0.003583 * IDDES (default, no box) 0.2668 -0.003056 * IDDES + box 0.2512 0.003518 Exp. 0.243 * Results not well converged 55

Summary Case 1: Convergence (other than SST) + mesh convergence demonstrated Hellsten RANS model looks promising IDDES required an LES-deactivation box to avoid partial collapse of the boundary layer prior to the main (rear) separation Case 2: Deep convergence challenging with RANS Mesh convergence not convincingly demonstrated using the three workshop-supplied grids Hellsten RANS model again looks promising IDDES again required use of an LES-deactivation box 56

Thank you J 57