Content uploaded by Gopal Shinde

Author content

All content in this area was uploaded by Gopal Shinde on Mar 27, 2015

Content may be subject to copyright.

1

Numerical Investigations of the DrivAer Car Model using

Opensource CFD Solver OpenFOAM

Gopal Shinde1, shinde.gopal@tcs.com, Aniruddha Joshi, joshi.aniruddha@tcs.com and Kishor Nikam,

kishor.nikam@tcs.com

Abstract:

A lot of the investigations in automotive aerodynamics are still based on strongly

simplified generic bodies such as the Ahmed Body or the SAE body. To close the

gap between these strongly simplified models and highly complex production cars

the new generic DrivAer body model is introduced by the Institute of Aerodynamics

and Fluid Mechanics, Technische Universitat Munchen (TUM). This current study is

focused on three different DrivAer body models namely Fastback, Estateback, and

Notchback and two different underbody types for each model, smooth and detailed.

Hence, total 6 different models are simulated using open source CFD solver

OpenFOAM at two different ground conditions, with ground and without ground effect

(WGS and WoGS). All the models used in simulations are 2.5 scaled down models

as compared with the actual car dimensions. The vehicle velocity considered for this

numerical study is 40 m/s, Reynolds number is 4.87M and turbulence model used is

k-w-SST. The mesh is generated using SnappyHexMesh (SHM) tool of OpenFOAM

and it is around 11 million volume cells for the smooth underbody and 14 million

volume cells for the detailed underbody. The coefficients of drag (Cd) values are

within 0.5% to 12% error band as compared against the experimental values

published by the TUM. The coefficients of pressure (Cp) plots are comparable with

experimental results and also the contribution of individual body part in overall Cd

values is obtained in this study. All the simulations are carried out using OpenFOAM

2.1.1 on Tata Consultancy Services (TCS) High Performance Computing facility.

Keywords : DrivAer body, External Aerodynamics, OpenFOAM, SnappyHexMesh,

CFD, HPC, TCS.

1Author for correspondence, Gopal Shinde, shinde.gopal@tcs.com, Tata

Consultancy Services, Pune, India

2

Introduction

Nowadays with the help of Computational Fluid Dynamics (CFD) and High

Performance Computing (HPC) Technology, vehicle aerodynamics engineers are

reducing the wind tunnel experiments and accelerating the vehicle design cycle. A lot

of the investigations in automotive aerodynamics are still based on strongly

simplified generic bodies such as the Ahmed Body. In an experimental work, Ahmed

et al. [1] set a bluff body, called the Ahmed Model, in the Gottingen open section

wind tunnel and analysed the time averaged flow behaviour for different slant angle

variations. Gilli´eron et al. [2] performed computational simulation and compared

against the experimental ones for the reference Ahmed model. Lienhart et al. [3]

performed experiments for two slant angles, 250 and 350, with slightly lower velocity

40 m/s but of the same order, with Re = 2.8E06. Later, Kapadia et al. [4],

Hinterberger et al. [5], Sinisa Krajnovic and Lars Davidson [6] and Ehab Fares [7]

also contributed to understand the Ahmed body flow physics better. Recently, Ronak

Pandya et al. [8] and Angelina Heft et al. [9] also contributed numerical observations

on Ahmed body.

These simple car models like Ahmed Body or SAE body make it easy to relate the

observed phenomena to specific areas and thus help to understand basic flow

structures. At the same time, more complex flow phenomena, e.g. at the underbody,

wheels/wheelhouses and around the rear view mirrors etc., cannot be reproduced

due to the over simplification of these geometries. On the other hand, it is usually not

feasible to investigate these phenomena on a specific production vehicle, as, due to

its short life span and restricted access, typically little validation data is available.

Recognizing the need for a model combining the strengths of both approaches,

various more or less generic models, such as the VW reference car and the MIRA

reference car, have been proposed [10]. However, while these reference cars mark a

step in the right direction, these models are still too generic to completely understand

the complex phenomena occurring at realistic vehicles.

To close this gap, the Institute of Aerodynamics and Fluid Mechanics of the

Technische Universitat Munchen (TUM), in cooperation with two major car

companies, the Audi AG and the BMW Group, has proposed a new realistic generic

car model called “DrivAer Model” [11]. The body is based on two typical medium

class vehicles (Audi A4 and BMW3 series) and includes three interchangeable tops

and two different underbody geometries to allow for a high universality. To

encourage the use of the DrivAer model in independent research projects, TUM

research group is open to share the geometry and a comprehensive experimental

database is published in different papers [12, 13].

3

The aim of the current work is to use different DrivAer body models and their

experimental results to validate the opensource CFD solver OpenFOAM.

Problem Description

This current study is focused on three different DrivAer body models namely

Fastback, Estateback, and Notchback and two different underbody types for each

model, smooth and detailed as shown in Figure 1. Hence, total 6 different models

are simulated using open source CFD solver OpenFOAM at two different ground

conditions, with ground and without ground effect (WGS and WoGS). In case of

WGS, car wheels are rotating anticlockwise and road is moving in the direction of air

and in case of WoGS both car wheels and road are stationary. All the models used

in simulations are 2.5 scaled down models as compared with the actual car

dimensions. Figure 2 show a sketch and main dimensions of the fastback

configuration of the 1:2.5 DrivAer model. Different parts of the Fastback model

considered for the simulation study are as shown in Figure 3. The individual drag

contribution of these components is also calculated along with the total car drag.

The vehicle velocity considered for this numerical study is 40 m/s, Reynolds number

is 4.87E106 and turbulence model used is k-w-SST.

Fastback

Estateback

Notchback

Detailed underbody

Smooth underbody

Figure 1: DrivAer body models and different underbody types

4

Computational Domain and Mesh Generation

To capture the flow around the car model the computational domain is created

around the car which often called as numerical wind tunnel as shown in Figure 4.

This computational domain is like physical wind tunnel test section. The length of the

computational domain is 28 m, width is 10 m and height is 7 m. While the length of

Figure 2: Typical dimensions of the Fastback model

Figure 3: Different parts of the Fastback model

5

physical wind tunnel test section is 4.8 m, width is 2.4 m and height is 1.8 m. The

blockage ratio used in computational domain is 0.57% while in physical wind tunnel it

was 6.17%.

To capture flow physics around the car more accurately the mesh size kept near the

car is fine and becomes coarse when go away from the car as shown in Figure 5. To

achieve this variable mesh size distribution various refinement boxes are created

with different dimensions as shown in Figure 4. To capture the boundary layer

around the car, multiple layers of very fine and fine element sizes are kept around

full car and along the road which is shown in Figure 6. The y+ obtained is 30 which is

supposed to be good for incompressible simulations using OpenFoam.

SnappyHexMesh (SHM) tool was used in parallel mode for the grid generation. The

meshing was carried out on 4 cores of 32 GB workstation. The mesh generated on

smooth underbody models is about 11 million volume cells and 14 million volume

cells for the detailed underbody.

Figure 4: Different boxes of refinement around the Fastback model

Figure 5: Z-Cut plane of the grid around the Fastback model

6

Boundary Conditions and Solver Settings

Boundary conditions are implemented on the computational domain, car body, car

wheels and road. Velocity was specified at the inlet and pressure was specified at

outlet. Road was given translation speed 40 m/s in the direction of flow in simulations

with ground effect and considered stationary for without ground simulations. Car

wheels are rotating at 320 rad/s anticlockwise in simulations with ground effect and

considered stationery for without ground simulations. All other car parts are defined

as stationary walls.

“simpleFoam” solver from OpenFoam has been used for simulations.

“potentialFoam” solver is used to set pressure field in the domain. Post-processing

has been carried out using the in-built utilities sampleDict, streamlines, and

cuttingPlane. The models considered here are with wheels and mirrors and were

simulated for 10,000 iterations each. The convergence achieved was 5 decades fall

for flow variables P and U and 6 decades fall for turbulence quantities omega and k.

Convergence of drag force is also monitored and achieved upto 15 counts in last 100

iterations.

Some of the solver’s important settings are given below.

Time stepping = Steady state

Gradient scheme = Second order

Divergence scheme = Upwind (First order)

Laplacian scheme = Linear (Second order)

Interpolation scheme = Linear (Second order)

Turbulence model = K-omega-SST[15]

Pressure solver = GAMG

Figure 6: Zoomed view of the grid cut plane around the Fastback model

7

Velocity solver = GaussSiedel

Non-orthogonal correctors = 2

Result Analysis

The total drag values obtained for the three models with smooth underbody

considering with and without ground effect are given in the Table 1. Table 2 shows

the drag values obtained for the three models with detailed underbody considering

with and without ground effect. Percentage error found out in the simulated drag

values is in the range of 0.5% to 12% as compared to the experimental drag values.

Simulated drag value is taken as the average of drag values of last 100 iterations.

The component wise drag plot for all configurations is as shown in Figures 7 and 8.

This plot shows the drag produced by each component, which gives us the important

insight where we should look for design changes in order to reduce the drag.

Reduction of drag directly contributes towards increase of fuel efficiency. These

component wise drag values are not readily available from experiments; however, it

is easily possible in simulations. Cp plots for the top surface of the models at y=0 are

plotted against the experimental Cp distribution as shown in Figure 9 and 10. We

have observed deviation in the simulated Cp distribution at the center of the model.

In experimental setup, at center of the model a vertical strut is mounted.

Model Type

Ground Effect

Expt. Cd

CFD Cd

% Error

Fastback

WoGS

0.254

0.2597

-2.24

WGS

0.243

0.2599

-6.95

Estateback

WoGS

0.296

0.2684

9.32

WGS

0.292

0.2840

2.73

Notchback

WoGS

0.258

0.2567

0.5

WGS

0.246

0.2565

-4.26

Model Type

Ground Effect

Expt. Cd

CFD Cd

% Error

Fastback

WoGS

0.284

0.2925

-2.99

WGS

0.275

0.3029

-10.14

Estateback

WoGS

0.318

0.3199

-0.6

WGS

0.319

0.3546

-11.16

Notchback

WoGS

0.286

0.3042

-6.36

WGS

0.277

0.3097

-11.8

Table 1: Total drag values for Smooth Underbody models

Table 2: Total drag values for Detailed Underbody models

8

Figure 7: Component wise drag contribution plot for Smooth Underbody models

WoGS

WGS

WoGS

9

WGS

Figure 8: Component wise drag contribution plot for Detailed Underbody models

Figure 9: Coefficient of pressure (Cp) plot for Fastback model, WoGS

Figure 10: Coefficient of pressure (Cp) plot for Fastback model, WGS

10

Conclusion

This current study is focused on 3 different models of DrivAer body namely

FastBack, EstateBack, and NotchBack and 2 different underbodies, smooth and

detailed underbody. Hence, total 6 different configurations are simulated using open

source software OpenFOAM at two different conditions with ground and without

ground effect. The coefficients of drag (Cd) values are within 0.5% to 12% error band

as compared against the experimental values published by the TUM. The

coefficients of pressure (Cp) plots are comparable with experimental results and also

the contribution of individual body part in overall Cd values is obtained in this study.

There are huge flow separation / recirculation zones on the back-side of the models.

In order to capture the flow features more accurately, we need to refine the grid in

these zones. Since the flow features in these zones keep on changing with respect

to time, current steady state simulations may not mimic the exact flow situation.

Unsteady state flow simulations with region-wise refined grid and with more stringent

convergence criterions may give more insights.

Future Work

In this current work, steady state simulations using k-w-SST turbulence model are

carried out. We would like to investigate this further for few more turbulence models

and also with transient calculations. Grid refinement may be necessary at the back

side of the models to capture the flow separation / recirculation zones more

accurately, which needs to be investigated.

References

1. S. R. Ahmed, G. Ramm, and G. Faltin, 1984. Some salient features of the

time averaged ground vehicle wake. SAE paper no. 840300.

2. P. Gilli´eron and F. Chometon, 1999. Modelling of stationary three-

dimensional

separated air flows around an Ahmed reference model. In ESAIMproc.,

volume 7, pages 173-182.

3. H. Lienhart, C. Stoots, and S. Becker, 2000. Flow and turbulence structures in

11

the wake of a simplified car model (Ahmed Model). In DGLR Fach Symp. der

AGSTAB.

4. S. Kapadia, S. Roy, and K. Wurtzler, 2003. Detached eddy simulation over a

reference Ahmed car model. AIAA paper no. 2003-0857

5. M. Hinterberger, M. Garcia-Villalba, and W. Rodi, 2004. Large Eddy

Simulation of flow around the Ahmed body. Lecture notes in Applied and

Computational Mechanics, The Aerodynamics of heavy vehicles: Trucks,

Buses, and Trains, R. McCallen, F. Browand, J. Ross (Springer, New York,

2004).

6. S. Krajnovic and L. Davidson, 2005. Flow Around a Simplified Car, Part 1:

Large Eddy Simulation. J. Fluids Eng. 2005, Volume 127, Issue 5.

7. Ehab Fares, 2006. Unsteady flow simulation of the Ahmed reference body

using a lattice Boltzmann approach. Computers & Fluids, Volume 35, Issues

8-9, 2006.

8. Pandya, R., Pavitran, S., Nikam, K., Singh, A., A Turbulence based

Computational Study on the Drag Breakdown of Ahmed Body, IUTAM

Symposium on Bluff Body Flows, December 12-16, 2011, IIT Kanpur, India.

Pg. 337-340.

9. A. I. Heft, T. Indingery, and N. A. Adams, Investigation of Unsteady Flow

Structures in the Wake of a Realistic Generic Car Model, 29th AIAA Applied

Aerodynamics Conference, 27 - 30 June 2011, Honolulu, Hawaii

10. Le Good, G. M., and Garry, K. P., On the Use of Reference Models in

Automotive Aerodynamics, SAE Technical Paper 2004-01-1308, 2004, doi:

10.4271/2004-01-1308.

11. Angelina I. Heft, Thomas Indinger and Nikolaus A. Adams, Introduction of a

New Realistic Generic Car Model for Aerodynamic Investigations, SAE

International, 2012-01-0168

12. Mack, S., Indinger, T., Adams, N. A., and Unterlechner, P., “The Ground-

Simulation Upgrade of the TUM Wind Tunnel,” SAE Technical Paper 2012-01-

0299, 2012, doi:10.4271/2012-01-0299.

13. J. Wojciak, P. Theissen, K. Heuler, T. Indinger, N. Adams, R. Demuth:

Experimental Investigation of Unsteady Vehicle Aerodynamics under Time

Dependent Flow Conditions - Part 2, SAE 2011 World Congress, April 12-14,

2011, Detroit, Michigan, USA, Paper 2011-01-0164

14. OpenFoam User Guide, http://www.openfoam.org/docs/user/

15. Sebastian MÖLLER, Daniele SUZZI, Walter MEILE, Investigation of the flow

around the Ahmed body using RANS and URANS with various turbulence

models, 3rd OpenFoam Workshop, Politecnico Di Milano, Milan, Italy, 10-11

July 2008.

12

Acknowledgements

The authors are thankful to Angelina I. Heft et al. (Institute of Aerodynamics and

Fluid Mechanics, Technische Universitat Munchen, Germany) for sharing the DrivAer

body CAD files and for valuable suggestions during this course of work. Authors

would like to thank the TCS-HPC team for their continuous support to carry out this

exercise. Authors also acknowledge the encouragement and support extended by

Mr. Rajaravisankar Shanmugam and EIS leadership team in pursuing this initiative.