Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
2nd Automotive CFD Prediction Workshop
Olivier Thiry, Daniel Ocaña Blanco
August 26th, 2021
Auto-CFD 2
© 2021 Cadence Design Systems, Inc. All rights reserved.2
IntroductionAbout Cadence - Numeca
• Numeca is part of Cadence Design Systems (and has been so since February 2021)
• 28 years of experience (Founded by Prof. Charles Hirsch in 1993)
• Worldwide presence with 150+ and now part of a 9000+ employees' group
• Best Exporter Award 2013 & Award for the most innovative company in 2001 and 2013
• Expertise in the simulation of fluid flows, heat transfers and Multiphysics
• Sponsoring the future. In the development of state-of-the-art technologies, Cadence works in partnership on several R&D projects with many universities worldwide.
More info:https://www.numeca.com/home
Follow us here:https://www.linkedin.com/showcase/cadence-computational-fluid-dynamics/
© 2021 Cadence Design Systems, Inc. All rights reserved.3
About Cadence - Numeca
Introduction
INTELLIGENCE
SYSTEMS
IP / SoCs
Numeca OMNIS™
Leader in CFD, multi-physicssimulation and optimization
Fluids and
Thermal
(CFD)
Simulation
and
Analysis
Electro-
magnetic
Thermal
(Electronics)
Electro-
mechanicalenablement
Cadence Design SystemsLeader in Intelligent Systems
© 2021 Cadence Design Systems, Inc. All rights reserved.4
• Integrated, parallel & Multiphysics environment for end-to-end
simulations (1D, CAD, CFD, CAA, FEA, MDO)
• Automatic CAD clean-up and high-quality meshing tool
• Solver agnostic mesher with full layer coverage + high level of
compatibility with solvers and in-house / research codes
• HPC / distributed memory + CPU and GPU friendly
• Wide scope of automotive applications
• Today’s focus on External Aerodynamicso Coupled pressure-based solver
o Steady RANS kw-SST with extended wall functions
o Transient DDES-SST* scale resolving (implicit 2nd order)
OMNISTM
Introduction
*Gritskevich, M.S., Garbaruk, A.V., Schütze, J., Menter, F.R. (2012) “Development of DDES and IDDESformulations for thek-ω shear stress transport model”, Flow, Turbulence and Combustion, vol. 88, pp. 431-449.
TM
© 2021 Cadence Design Systems, Inc. All rights reserved.5
Case 1: Windsor Body
© 2021 Cadence Design Systems, Inc. All rights reserved.6
Case 1: Windsor Body
Variant Mesh size [M]Mesh size
(committee) [M]Max
skewnessMax expansion
ratio
nW (WMRANS) 6.79 6.69 0.818 8.85
nW (WRRANS) 11.03 11.26 0.848 8.84
wW (WMRANS) 7.72 7.74 0.924 10.99
wW (WRRANS) 14.43 14.92 0.942 12.47
Variant Mesh size [M]Mesh size
(committee) [M]Max
skewnessMax expansion
ratio
nW (WRER) 46.98 46.76 0.847 8.83
wW (WRER) 50.20 50.08 0.943 11.92
• Mesh generated in OMNISTM/Hexpress following workshop guidelines.
• Two configurations considered:o No wheels (nW)
o With wheels (wW).
• Three meshes generated for each configurations:
o High Reynolds Wall Modeled RANS mesh (WMRANS)
o Low Reynolds Wall Resolved RANS mesh (WRRANS)
o Low Re Eddy Resolved mesh (WRER).
• Full hexahedral meshes generated with excellent quality
• Commitee mesh sizes given for reference
RANS grids
DES grids
Mesh through a centerline slice
nW wW
Meshes used
© 2021 Cadence Design Systems, Inc. All rights reserved.7
• No clear gain in using low Reynolds meshes with the RANS approach.
• DES better captures drag in absolute value, and the variant inter-ranking for both lift and drag.
Case 1: Windsor BodyForces
Reference: Varney, Max. "Base drag reduction for squareback road vehicles." PhD diss., Loughborough University, 2019.
Drag coefficient
Lift coefficient
Delta drag coeff. (wW vs. nW)
Delta lift coeff. (wW vs. nW)
DataCD
(nW)
CD
(wW)
ΔCD
(wW-nW) [counts]
CL
(nW)
CL
(wW)
ΔCL
(wW-nW) [counts]
Varney 2019 (Exp) 0.280 0.372 92 -0.0271 0.1372 164.3
WMRANS 0.2699 0.3461 76.2 -0.1176 0.1233 240.9
WRRANS 0.2740 0.3400 66.0 -0.1099 0.1033 213.2
WRER (DDES) 0.2819 0.3802 98.3 -0.1210 0.0588 179.8
© 2021 Cadence Design Systems, Inc. All rights reserved.8
Case 1: Windsor BodyPressure coefficient on Y and Z cuts
No
wh
ee
lsW
ith
wh
ee
ls
Z=259.4mm
Y=0.0mm
Experiment(Varney 2019)WMRANSWRRANSWRER (DDES)
© 2021 Cadence Design Systems, Inc. All rights reserved.9
Case 1: Windsor BodyPressure coefficient on the backface on Y cuts
Plane 1
No
wh
ee
lsW
ith
wh
ee
ls
Plane 2 Plane 3 Plane 42 43
1
Experiment(Varney 2019)WMRANSWRRANSWRER (DDES)
© 2021 Cadence Design Systems, Inc. All rights reserved.10
Case 1: Windsor BodyWake pattern comparison: streamwise velocity
WMRANS (6,8M)Experiment (Varney 2019) WRRANS(11M) WRER (47M)
No w
hee
lsW
ith
wh
ee
ls
Isosurface of Q-criterion=1e5 in the
DDES-SST nW simulation (WRER)
Y=0.0mm
Isosurface of Q-criterion=1e5 in the
DDES-SST wW simulation (WRER)
© 2021 Cadence Design Systems, Inc. All rights reserved.11
Case 1: Windsor BodyVelocity profiles in the wake
ux Ux,rms
Uz,rms
ux
Streamwise velocity: Exp., WRRANS, WRER (DDES) Velocity fluctuations: Exp., WRER (DDES)
No
wh
ee
lsW
ith
wh
ee
ls
Y=0.0mmY=0.0mmY=0.0mmZ=194mm
© 2021 Cadence Design Systems, Inc. All rights reserved.12
Case 2: DrivAer Notchback
© 2021 Cadence Design Systems, Inc. All rights reserved.13
Case 2: DrivAer NotchbackMeshes used
• Mesh generated in OMNISTM/Hexpress following workshop guidelines.
• One configurations considered:o Notchback version of the open cooling DrivAer with detailed underbody
• Two meshes generated :o High Reynolds Wall Modeled mesh (WM)
o Low Reynolds Wall Resolved mesh (WR)
• Full hexahedral meshes generated with excellent quality
• Committee mesh sizes given for reference
VariantMesh size
[M]Mesh size
(committee) [M]Max skewness
Max expansion ratio
Notchback (WM)
136 128 0.935 45.29
Notchback (WR)
236 244 0.929 82.34
View of the mesh in the center plane
© 2021 Cadence Design Systems, Inc. All rights reserved.14
Case 2: DrivAer NotchbackAerodynamic Forces
• The experiment from FKFS, with a smallerblockage ratio, is more in line with the simulations.
• No clear gain in using low Reynolds meshes withthe RANS approach.
• DES improves the prediction of drag and rear lift.
Ecara website: https://www.ecara.org/fileadmin/media/dateien/DrivAer/FKFS/FKFS_OCDA_MOD-B_N_EB_wM_wW_woL_oG_KBS.pdf
Hupertz, Burkhard, Chalupa, Krueger, Howard, Glueck, Lewington, Chang, and Shin. “On the Aerodynamics of the Notchback Open Cooling DrivAer: A
Detailed Investigation of Wind Tunnel Data for Improved Correlation and Reference”. SAE Technical Paper, 2021.
CD CLF CLR
FKFS reference(Exp) 0.275 0.015 0.127
RANS-SST (WM) 0.270 -0.068 0.179
RANS-SST (WR) 0.268 -0.060 0.184
DDES-SST (WR) 0.277 -0.098 0.122
Rear Lift coefficient Front Lift coefficient
Drag coefficient
© 2021 Cadence Design Systems, Inc. All rights reserved.15
Case 2: DrivAer NotchbackPressure coefficient at y=0
• On the front part of the car, no significant difference is observed, as RANS and DES exhibit the same turbulence modeling behavior.
• On the rear section and under the car, the DES model acts as an LES and is closer to the experiment.
Isosurface of Q-criterion=5e4 in
the DDES-SST simulation
Y=0.0mm
Exp. (Hupertzet al. 2021) ; RANS-SST (WM) ; RANS-SST (WR) ; DDES-SST (WR)
© 2021 Cadence Design Systems, Inc. All rights reserved.16
Pressure coefficient in horizontal cuts
• DES shows a clear advantage in predicting the wake pattern on the rear section
Z=500.0mmZ=150.0mm
Exp. (Hupertzet al. 2021) ; RANS-SST (WM) ; RANS-SST (WR) ; DDES-SST (WR)
Case 2: DrivAer Notchback
© 2021 Cadence Design Systems, Inc. All rights reserved.17
Total pressure coefficient in the wake (x=4m)
• Main body wake better predicted by the DES, but also the wheels wake close to the ground, as well as the underbody wake.
Exp (Hupertz et al. 2021) RANS - SST (WR) DDES - SST (WR)
Isosurface of Q-criterion=5e4 in
the DDES-SST simulation
Ad
im. v
elo
city
Tota
l pre
ssur
e co
eff.
RANS - SST (WM)
X=4000.0mm
Case 2: DrivAer Notchback
© 2021 Cadence Design Systems, Inc. All rights reserved.18
Flow in the front wheel’s wake (x=0.4m)
Exp (Hupertz et al. 2021) RANS - SST (WM) RANS - SST (WR) DDES - SST (WR)
Isosurface of Q-criterion=5e4 in the
DDES-SST simulation
X=400.0mm
Adi
m. v
elo
city
Tota
l pre
ssur
e co
eff.
Case 2: DrivAer Notchback
© 2021 Cadence Design Systems, Inc. All rights reserved.19
Streamwise velocity and pressure under the car (z=-0.2376)
Exp. (Hupertz et al. 2021)
RANS - SST (WR)
DDES - SST (WR)
Isosurface of Q-criterion=5e4 in the DDES-SST simulation
Adim. velocity Total pressure coeff.
Exp RANS (WR) DES (WR)
CD 0.275 0.268 0.277
CLR 0.127 0.184 0.122
RANS - SST (WM)
Z=-237.6mm
Case 2: DrivAer Notchback
© 2021 Cadence Design Systems, Inc. All rights reserved.20
Conclusions
© 2021 Cadence Design Systems, Inc. All rights reserved.21
• Overall good drag trends prediction using RANS for the Windsor Body and DrivAer Notchback
• No influence of mesh refinement on RANS cases past a certain level
• High level of accuracy reached with DDES (preferred approach) on both lift and drag
• Systematic capturing of flow physics in agreement with measurement around wheels, underbody and wake regions
• Robust end-to-end workflow fully automated with embedded best practices
Conclusions
© 2021 Cadence Design Systems, Inc. All rights reserved worldwide. Cadence, the Cadence logo, and the other Cadence marks fou nd at https://www.cadence.com/go/trademarks are trademarks or registered trademarks of Cadence
Design Systems, Inc. Accellera and SystemC are trademarks of Accellera Systems Initiative Inc. All Arm products are registered trademarks or trademarks of Arm Limited (or its subsidiaries) in the US and/or elsewhere. All MIPI
specifications are registered trademarks or service marks owned by MIPI Alliance. All PCI-SIG specifications are registered trademarks or trademarks of PCI-SIG. All other trademarks are the property of their respective owners.