See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/235946204

Flow and turbulent structures around simplified

cars models
Conference Paper January 2012

CITATIONS

READS

160

5 authors, including:
David Aljure

Ivette Rodriguez

Polytechnic University of Catalonia

Polytechnic University of Catalonia

8 PUBLICATIONS 9 CITATIONS

72 PUBLICATIONS 456 CITATIONS

SEE PROFILE

SEE PROFILE

Oriol Lehmkuhl

R. Borrell

Polytechnic University of Catalonia

Termo Fluids S.L.

120 PUBLICATIONS 456 CITATIONS

49 PUBLICATIONS 225 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

SEE PROFILE

Available from: David Aljure

Retrieved on: 17 July 2016

Conference on Modelling Fluid Flow (CMFF12)

The 15th International Conference on Fluid Flow Technologies
Budapest, Hungary, September 4-7, 2012

F LOW AND T URBULENT S TRUCTURES A ROUND S IMPLIFIED C AR M ODELS

David ALJURE,1 Ivette RODRIGUEZ, 2 Oriol LEHMKHUL, 2 , 3 Ricard BORRELL, 3
Assensi OLIVA 3
1 Corresponding Author. Centre Tecnol`
ogic de Transfer`encia de Calor (CTTC) , Universitat Polit`ecnica de Catalunya (UPC). ETSEIAT, Colom
11, 08222, Terrassa, Barcelona, Spain. E-mail: davidal@cttc.upc.edu
2 Centre Tecnol`
ogic de Transfer`encia de Calor (CTTC) , Universitat Polit`ecnica de Catalunya (UPC). ETSEIAT, Colom 11, 08222, Terrassa,
Barcelona, Spain. Fax: +34 93 739 89 20. E-mail: cttc@cttc.upc.edu
3 Termo Fluids, S.L. Av. Jaquard, 97 1-E, 08222 Terrassa (Barcelona), Spain. E-mail: termofluids@termofluids.com

ABSTRACT

1.

External car aerodynamic study has great

importance in overall car efficiency and ride stability,
making it a key element in successful automotive
design. Flow over car geometries fully show three
dimensional, unsteady turbulent characteristics.
Additionally, vortex shedding, flow reattachment
and recirculation bubbles can be found around the
buff body.
These phenomena greatly influence
the key factors in aerodynamic study, the lift and
drag coefficients, vital for ride stability and energy
efficiency, respectively. The purpose of the present
study is to assess and compare the different large scale
turbulent flow structures present in car aerodynamics
using two basic car geometries.
Flow around two model car geometries is resolved,
the Ahmed and the Asmo car. These generic buff
bodies reproduce the basic fluid dynamics features of
real cars using simplified geometries. Firstly, the flow
over both geometries is studied and compared against
experimental results to validate the code, afterwards,
various LES models are used to study the flow in detail
and compare the structures found in both geometries.

Computational fluid dynamics (CFD) have

evolved greatly in the past two decades becoming
a vital tool in industrial research, development and
investigation. Due to the nature of fluid flows, most
practical applications deal with turbulent motion; the
modelling of this phenomenon is of vital importance
within the CFD industry, where vast resources are
invested into researching it. Within the different
possible applications of the CFD technology come
aerodynamics and automotive design. The automotive
industry makes great advances every year; engine
performance has increased greatly in the past decade,
along with weight reduction and safety measures.
External aerodynamics in automotive developments
bear increased importance as such advances are
made. Aerodynamic resistance increased its relative
importance by 124% when an audi r8 automobile was
such advances [1]. Additionally, as weight reduction
becomes more popular, aerodynamics become more
important to ensure vehicle drive stability.
The bodies to be studied using Challenging Large
Eddy Simulations (CLES) in the present paper are
the widely studied (Lehmkuhl [2], Krajnovic [3][4],
Minguez [5]) Ahmed car model (with a 25 angle of the
rear slanted surface) and the Asmo car model ([6],[7]).
The Ahmed body car is a semi-rectangular vehicle with
a rounded front and a slanted back. The simplified
topology of this model allows easy modelling, meshing
and comparisons between experimental and numerical
results. The second model, the Asmo car is a model
created by Daimler-Benz in the 90s to investigate low
drag bodies in automotive aerodynamics.

Keywords: Automotive aerodynamics, CFD, CLES,

Turbulence models.

NOMENCLATURE
cD
cL
C
D
G
M
cP
Q
u
urms

[]
[]
[]
[]
[]
[]
[]
[1/s]
[m/s]
[m/s]

INTRODUCTION

Total drag coefficient

Total lift coefficient
Convective operator matrix
Diffusive operator matrix
Gradient operator matrix
2. NUMERICAL METHOD
Divergence operator matrix
P
Adimentional pressure U 2
2.1. Domain and Boundary Conditions
Second invariant of U
The geometries to be studied are the Ahmed car
Velocity vector
[8] and the Asmo car [7]. The computational domain
Root mean square stream-wise velocity is a 9.1944 1.87 1.4m rectangular channel with
the front of the car located 2.1014 m downwind from

the inlet boundary. The outlet boundary is at a distance

of 6.048 m for the Ahmed car and of 6.282 m for the
Asmo car, measured from the rear of the body. To
simulate the same case as Lienhart [8] a 34 open wind
tunnel was simulated using the following boundary
conditions:
Inlet: The inlet condition is a uniform flow such
as the reference Reynold using the model height
is Re = 7.68 105 .

3. RESULTS
3.1. Mesh Study
3.1.1.

Ahmed Car

First, to validate the code, results of the Ahmed

car simulations using the VMS turbulence model were
compared to the experimental results obtained by
Lienhart [8].

Outlet: A convective outlet boundary condition

was used.
Floor: A no-slip - 0 velocity condition was used
for the floor.
Top, front and back: To complete a open wind
tunnel these three boundaries where modelled
using a symmetry condition.

2.2.

Governing
Discretization

Equations

The
incompressible-isotherm
equations are solved:

and

Figure 1: Average stream-wise velocity profiles over

the slant wall in the mid plane.

Navier-Stokes

u
+ C(u)u + Du + Gp = 0
t
Mu = 0

(1)

As we can see in figure 1, the obtained results are in

good agreement with the experimental results; a mesh
of approximately 1 106 control volumes shows good
results.

(2)

3.1.2.

where u is the three-dimensional velocity vector

field and p is the pressure scalar field. The matrices
C, D, G and M are the convective, diffusive, gradient
and divergence operators respectively. It is important
to note the non-linear relation in the convective term of
the equation C(u)u. The velocity-pressure coupling
is solved by means of a fractional-step algorithm;
the temporal discretization for the convective, difusive
and derivative parts of equation (1) was made using
a second order self-adaptative scheme [9]; whereas
a back-ward euler scheme was used for the pressure
gradient. The spatial discretization was made on a
collocated mesh by means of a conservative scheme
similar to that presented by Felten [10] where
skew-symmetry is ensured for the convective operator
as well as a positive-definite symmetry diffusive
operator.
Turbulence modelling is carried out by using
different eddie-viscocity models; the wall-adapting
local-eddy viscocity (WALE) [11], a WALE model
with a variational multiscale approach (VMS) [12], the
singular values subgrid model (SIGMA) [13], and the
q-r model that uses invariants q and r of the filtered
strain tensor to model eddy viscocity (QR) [14].
The simulations have been performed using the
TermoFluids high performance parallel code [15] on a
hybrid tetra-prism unstructured mesh generated using
the Ansys IcemCFD software and ran in the JFF
computing cluster in the UPC. METIS software was
used for domain partitioning.

Asmo Car

Figure 2: Average pressure coefficient in the

underbody
Figure 2 shows the mesh independence of the
simulations performed on the Asmo car, it can be seen
that a 2.5 106 mesh shows good results in agreement
with the experiments performed by Volvo [6].

3.2.

LES Analysis

Four different LES models were used to analyse

the flow over the simplified cars, the QR, WALE,
SIGMA and VMS models.
3.2.1.

Ahmed Car

The first geometry studied, the Ahmed car, allowed

the comparison of the results with the experimental
data obtained by Lienhart [8]. Stream-wise velocity

Figure 3: Average stream-wise velocity profiles over

the slant wall in the mid plane.

Figure 5: Average pressure coefficient in the

underbody

profiles shown in figure 3 evidence that the VMS

model and SIGMA model results are closer to
the experimental data than the other two models.
Moreover, it can be seen that the WALE, SIGMA and
VMS models yield good results in the boundary layer
of the slant wall, where the QR model fails to predict
the flow behaviour correctly. The results in the near
wake are a little different, having the QR show the best
results.

These eddies extract energy from the mean flow

through the vortex stretching mechanism. However,
when the flow is bounded by a wall, the eddies
are restricted and form double-cone eddies. These
two structures are ideal, however they bear great
resemblance to the structures observed in the wake of
the Ahmed car flow. Several authors have discused the
different methods to identify vortical structures within
the flow [16],among those is the second invariant of
the velocity gradient tensor Q:
1
(||W 2 || ||S 2 ||)
2
1
W = (u (u)T
2
1
S = (u + (u)T
2

Q=

(3)
(4)
(5)

Figure 6a and 6b show, by means of the second

a

Figure 4: Average stream-wise RMS-velocity

profiles over the slant wall in the mid plane.
Figure 4 shows the stream-wise root mean square
velicity profile urms along the slant back of the
Ahmed model. In this case, the WALE model yielded
the best results of all four models. Again the less
accurate data is the one resulting from the QR model.
3.2.2.

Asmo Car

The same 4 models where used to solve the flow

over the Asmo car. As figure 5 shows, the results
obtained by the WALE and SIGMA models show the
highes error whereas the QR and VMS models results
are the better, however fail to fully predict the pressure
spikes in the front and back.

3.3.

Flow Structures

3.3.1.

Lifted streamwise vortices:

According to Townsends attached eddy

hypothesis [16], the average large eddy motion
consists of pairs of inviscid parallel roller vortices.

Figure 6: Streamwise vortices (a) Ahmed car Q=0.1

(b) Asmo car Q=0.1 (c) Ahmed car P=0 (d) Asmo car
P=(-0.01,0.01)
invariant Q, two longitudinal vortices generating in
the back of the cars. The presence of this vortices
can also be verified by means of a local presure
minimum, shown in figure 6c and 6d. This turbulent
structure is found both on the Ahmed and the Asmo
geometry, however, when comparing the vortices in
both geometries the magnitude and strength of these

structures in the Asmo car is lower, they dont display

a strong a profile as the ones present in the Ahmed
car. This fact is specially evident in in figures 6c
and 6d, where, compared to the pressure isosurface
of the Ahmed car, the vortex structure in the Asmo
car is not unified, making this eddy much weaker in
comparison to the one displayed by the Ahmed body
flow. It can also be seen in figure 6 that the lower
part of the vortices are bounded by the floor of the
channel, thus, restricting the overall size of the eddy.
Figure 7 shows the mean streamlines where the main
a

Figure 8: Recirculation bubble behind the Ahmed

Body.

vortex takes the shape of a horseshoe thoroidal vortex

bounded by the flow coming from the sides and the top
of the body. The flow from the underbody, after passing
through the diffuser, reduces its speed thus reducing its
influence in the recirculation bubble.

Figure 7: Mean streamlines in the back of the car

models
longitudinal vortices are generated. It is important to
note that the geometry in both cases is very different,
the slant back in the Ahmed body and the curved
profile in the Asmo car, however, the vortex generation
mechanism seems to be the same. In both cases there
is a high speed stream moving along the side wall of
the car, a slow speed stream in the back of the car
and the main flow travelling along the top of the body.
In the Ahmed body the slow speed stream consists of
fluid recirculating along the slant wall, whereas in the
Asmo car the slow speed stream consist in the main
recirculation bubble in the back of the car. The vortex
is then generated by the interaccion between these three
streams, as the high speed stream passes the end of the
side wall it flows towards the mid plane (pushed by the
low pressure zone found in the back), crashing into the
slow speed stream reducing its speed before colliding
against the main flow present close to the midplane
(where the pressure has a higher value). This second
collision changes the mixed stream direction giving it
a twirling motion, and thus, creating the longitudinal
vortex1 .
3.3.2.

Recirculation Bubble

Figures 8 and 9 show the recirculation bubbles

present behind the bodies.
The recirculation
adimentional length for the Ahmed body is around
0.69H (0.65H in [3]) whereas for the Asmo body is
around 0.48H using the bodys height as the reference
length. Besides from the size, the recirculation bubbles
in both bodies differ in shape; in the Ahmed body it is
composed of two straight spanwise vortices generated
by the flow coming down the slant back and the flow
coming from the underbody. In the Asmo body the
1 The results in this section where obtained using a different mesh
containing more control volumes in the wake area

Figure 9: Recirculation bubble behind the Asmo

Body.

3.3.3.

Pressure Field and Pressure Gradients

Flow detachment is generally identified by a low

pressure zone. The places where fluid separation
occurs are accompanied by steep pressure gradients
that generate vortices and eddies, which in turn
augment drag on the aerodynamical body. Other
factors afecting the aerodynamic drag, and which are
partially responsable for the pressure drag a body
experiences, are the high pressure bubbles in the front
of the geometry and low pressure bubbles in the back.
Figures 10 and 11 show the adimensional pressure cP
contours for the two bodies; this figure shows that
the pressure gradients present in the Ahmed body are
much greater than those present in the Asmo body.
Additionally, it can be seen that the Ahmed body has
several zones of detachment. Furthermore, the pressure
difference around the Ahmed car is greater than that
present in the Asmo car.
The mild pressure gradients present in the Asmo
body allow the flow to remain attached for most of the
cars length, thus producing a smaller wake as seen
before. The steeper pressure gradients present in the
Ahmed body are strong enough to produce separated
flows increasing the size and pressure losses in the
wake, thus, increasing the pressure drag.
Figure 12 shows the high pressure and low pressure
bubbles around the bodies. The low profile nose of
the Asmo car allows the formation of a smoother flow
pattern thus diminishing the stagnation in the flow
and reducing the size of the high pressure bubble;

Figure 13: Q = 20 instantaneous isosurfaces in the

front of the Bodies.
Figure 10: Mean pressure contours in the Ahmed
body.

Figure 11: Mean pressure contours in the Asmo

Body.

whereas in the Ahmed car the wider nose puts up a

bigger resistance to the flow increasing the size of the
high pressure bubble. It is also important to notice
that at the slant back the edges are in a low pressure
zone with high flow separation. This accounts for
the formation of the big lifted longitudinal vortices
shown before. There is also a mid-body low pressure
zone present in both geometries, in the Ahmed car it
is found near the front of the car, immediately after
the nose, whereas this zone is spread wider in the
Asmo geometry, accounting for the smoother pressure
gradients present in this geometry.
a

Figure 12: Mean pressure surfaces around the

Ahmed and Asmo bodies.
The effects of pressure gradients can be classified
in two mayor groups, positive and negative relative to
the direction of the flow. These zones can readily be
seen on figure 10 and 11.
Figure 13 show the Q isosurfaces in the front of

the model cars. The first structure in the front of

the Ahmed car is smooth and resembles a laminarized
flow. This part of the geometry has a negative pressure
gradient (pressure decreases in the direction of the
flow); this gradient pushes the fluid back, increasing
the velocity gradient near the wall, thus, increasing the
local wall shear stress. The flow structure in front of
the Asmo car is different. After passing the stagnation
point (zone with no structure visible), the mild negative
pressure gradient along the nose of the car permits the
flow to be more stable than that found in the Ahmed
car.
As the fluid moves back, it reaches the local
pressure minimum and, thus, a zone of positive
pressure gradient. Figure 13a shows this point as
the transition of the smooth coherent structure to the
unstable flow along the top of the car. This zone is
characterized by a separation zone and a lower wall
shear; figure 13a shows this zone as a collection of
small vortical structures away from the wall. Several
authors, such as Minguez [5] and Krajnovic [3] state
there is a recirculation zone here; eventhough the
meshes used in the present study are not fine enough for
a complete resolution of the flow in this area, evidence
(pressure field and coherent structures) suggests the
presence of recirculation zone in this place. Further
back, the pressure gradient becomes small and the
coherent structures in the Ahmed car exhibit the
turbulent behaviour to be expected from the streak
inestability cycle.
The flow over the Asmo car show very little
instabilities, however, evidence of the behaviour
explained earlier can be found. As the fluid travels
throught the top of the nose it reaches a local pressure
minimum. As the fluid closes on this point the pressure
gradient becomes larger and a smooth structure, such
as the one mentioned earlier, forms. Upon reaching the
low pressure point the structure breaks down due to the
negative pressure gradient. Furthermore, the vortical
structures in the bottom of the car, shown by figure 13b,
are vortices shedding around the tyres.
Studying closely figure 13a a small structure,
generating in the lower corner of the car, can be seen.
As was exposed by Krajnovic [4], this vortical structure
is generated by the interaction of the boundary layers
of the floor and the underbody of the Ahmed car. The
presence of the wheels in the Asmo car prevent this

structure from forming.

The flow structures in the back of the car are
also very different. The slant back is a geometry
feature unique to the Ahmed car and it displays very
turbulent flow. Figure 14a shows the Q isosurfaces in
this region, it shows small vortical structures forming
on the the surface, right at the edge where the
slanted surface starts as well as several big vortical
structures. The big vortical structures present are the
vortices that form the recirculation bubble, the big
lifted streamwise vortices (shown earlier) and large
horseshoe like vortices shedding from the slant wall.
a

Table 1: Ahmed car drag coefficients

Model
VMS
QR
WALE
SIGMA
Exp[17]

Pressure
0.302
0.301
0.310
0.281

Skin friction
0.008
0.008
0.011
0.010

Total
0.310
0.309
0.321
0.292
0.298

performs well in predicting the drag, specially the

pressure, whereas the WALE shows the worst results;
furthermore, the VMS and the QR model perform very
similarly. All four models perform, when resolving the
viscous drag, in a similar fashion. As can be seen in
table 2, the Ahmed geometry has a high lift coefficient,
it is of the same magnitude as the drag coefficient.
Table 2: Ahmed car lift coefficients

Figure 14: Q = 200 instantaneous isosurfaces in the

slant back of the (a) Ahmed body (b) Asmo Body
These small vortical structures are the result of the
change in surface direction and the subsequent negative
pressure gradient this change generates. As was
explained earlier, this pressure gradient creates flow
instabilities. The combination of the flow instabilities
and the geometry present add up to vortex shedding in
this section.
The turbulent structures present in the back of
the Asmo car are different in shape and size as
those seen in the Ahmed car. Figure 14b shows
less vortical structures being generated in the back
of the car, the vortex shedding phenomenon can be
witnessed readily in this geometry; large structures
detach from the bottom of the car, as well as small
vortices detaching from the wheels, the diffuser and the
trailing edge of the car. As it is to expect in flow over
buff bodies, both geometries show vortex shedding,
however, in the Ahmed car, specially in the slant back,
this phenomenon is very strong.

3.4.

Drag & Lift Coefficients

As stated earlier, two of the most important

aerodynamic parameters are the lift and drag forces.
The drag coefficient quantifies the advancing resistance
posed by the air (surrounding fluid) onto the vehicle;
whereas the lift coefficient cuantifies the force exerted
onto the vehicle in the direccion perpendicular to the
flow. Reducing the drag coefficient of a vehicle will
reduce its overall fuel consumption, reducing the lift
coefficient will improve ride stability and cornering
performance.
3.4.1.

Ahmed Car

Table 1 shows the drag coefficients obtained

for the different LES models. The SIGMA model

Model
VMS
QR
WALE
SIGMA

Pressure
0.313
0.328
0.256
0.313

Skin friction (105 )

3.8
25.6
6.9
6.8

Total
0.313
0.328
0.256
0.313

3.4.2. Asmo Car

Table 3: Asmo car drag coefficients

Model
VMS
QR
WALE
SIGMA
Exp. Volvo [6]

Pressure
0.154
0.154
0.168
0.161

Skin friction
0.0027
0.0027
0.0036
0.033

Total
0.157
0.156
0.171
0.164
0.158

Tables 3 and 4 show the force coefficients on the

Asmo geometry. As it can be easily seen, the drag
coefficient of the Ahmed car is nearly double the drag
coefficient of the Asmo car. The drag coefficient
calculated with the VMS and QR models are quite
close to the experimental values reported by Volvo.
The WALE and SIGMA models slightly overpredict
the drag. The QR and VMS models yield very similar
lift coefficients, whereas the WALE and the SIGMA
report quite different results. Comparing tables 2 and 4
it can be seen that the lift coefficient for the Asmo car is
one order of magnitude smaller than the lift coefficient
found for the Ahmed car.

4.

CONCLUTIONS

CLES simulations where carried out in two

different geometries, the Asmo car an the Ahmed car;
using 4 different turbulence models, the QR, the VMS,
the SIGMA and the WALE (to the authors knowledge,

Table 4: Asmo car lift coefficients

Model
VMS
QR
WALE
SIGMA

Pressure
0.0279
0.0283
0.0143
0.0728

Skin friction (105 )

31.7
31.7
22.4
34.3

Total
0.0282
0.0286
0.0.0145
0.0.0731

this is the first time the SIGMA and VMS models are
used in resolving flow around simplified car models).
When comparing the results obtained in the Ahmed
geometry with the experimental results it can be seen
that the agreement in the slanted back is accepable,
however, due to the mesh sizes used in the present
simulations the results arent in good agreement with
the experimental results in the near wake as other
authors with finer meshes have achieved. The results
in the underbody of the Asmo car are also in good
agreement with experimental results , however, the
pressure distribution in the front part has some error.
The coherent structures in the flow around the
Asmo car are much weaker and exhibit less turbulent
behaviour than those found around the Ahmed car. The
biggest structures in the flow, the lifted streamwise
longitudinal vortices in the wake, are present in both
geometries using the same generation mechanism,
however, the geometry features of the Ahmed body
allow for the formation of a much stronger vortex.
The recirculation bubble present in both
geometries differ both in shape and size; the one
present in the Asmo car is smaller, shows a horseshoe
shape and is formed by flow coming from the top and
the sides of the car; the bubble found in the Ahmed
car is formed by two spanwise vortices and is created
by the flow coming from the top and the bottom of the
geometry. The differences in this structure allow for
different magnitudes in the pressure fields, being the
pressure behind the Ahmed car lower than that found
in the Asmo car.
Another difference in pressure, this time in the
nose of the car, is also found, the low profile nose
of the Asmo car results in less stagnation and lower
pressures in the front. The Asmo car also exibits
lower pressure gradients around the entire body, this
produces a more laminarized flow around the entire
geometry, less vortical structures dettaching and less
vortex shedding in its back.
All the differences in the flow features explained in
the previous paragraphs account for the results shown
in tables 1, 2, 3 and 4; the drag coefficient for the
Ahmed car is greater than that calculated around the
Asmo car. Additionally, the Ahmed car shows a large
positive lift coefficient, opposed to the smaller value
found for the Asmo car.
Furthermore, the VSM model was the slowest
model of the four tested, restricting the timestep by
predicting a higher turbulent viscocity, whereas the QR

model was the fastest, permiting a higher timestep due

to lower turbulent viscocity. The SIGMA and VSM
models performed very similarly and better than the
WALE and QR models in the Ahmed car, where the
geometry changes and corners heavily influence the
flow. VMS and QR models performed well in the
Asmo car; however the SIGMA and WALE models
didnt, they show problems when dealing with flows
heavily influenced by the boundary layer.

ACKNOWLEDGEMENTS
This work has been financially supported by the
Ministerio de Educacion y Ciencia, Spain (Project:
Development of high performance parallel codes for
the optimal design of thermal equipments, reference
ENE2010-17801) and Termo Fluids S.L.

APPENDIX A: Fine Meshes

All the previous data, with exception to those
that show the big lifted streamwise vortices, where
done with coarse meshes. The results are satisfactory,
however it is important to contrast the results found
with finer meshes to ensure their validity. A 6M control
volume mesh was used for the Ahmed car and a 4.5M
control volume mesh was used for the Asmo car, both
using the VMS model.
The turbulent structures around the Ahmed car are
quite more complex than those found in the Asmo
car. Figure 15 shows the mean Q isosurfaces present
around the front of the cars. The structures found
in the figure 15a, when compared to 13a are much
more complex. The finer mesh allows the resolution
of smaller scales thus showing much more structures,
however, the structures found in the coarser mesh are
consistent with the structures in this section. Figure
13b show the structures present in the Asmo car; also
consistent with those found in the finer mesh shown in
figure 15b.
a

Figure 15: Q = 20 instantaneous isosurfaces in the

front of the bodies.

Figure 16: (a) Q = 200 instantaneous isosurfaces in

the back of the bodies

Figures 16 show the structures found in the back of

the geometries. Once again, when comparing to figure
14 the main difference is the smaller scales shown in
the results from this section; however, the large scale
structures found are found in both coarse and fine
meshes. It is important to note that vortex shedding
is largely influenced by both large and small length
scales. As can be seen when comparing figures 16b
and 14b, there is a vortex structure shedding from the
bottom of the car, the coarse mesh represents it as a
single large structure, when in reality it is a collection
of different sized structures.
Table 5: Lift and drag coefficients
Geometry
Ahmed Car
Asmo Car

cD (exp)
0.2925(0.298)
0.1576(0.158)

cL
0.2950
0.01646

Finally, the lift and drag coefficients for the fine

meshes are shown in table 5. When compared to the
results shown in tables 1, 2, 3 and 4 the parameters
dont vary much. The force coefficients calculated with
the coarser meshes are close to the results obtained in
this appendix. When comparing to the experimental
results for the Ahmed car [17] and by Volvo [6] for the
Asmo car we see that the results obtained for both cars
are lower, 1.85% and 0.25% for the Ahmed and Asmo
car respectively.

References
[1] Islam, M., 2009, Application of OpenFOAM
Technology
for
Vehicle
Aerodynamics
Development OpenFoam Proceedings.
[2] Lehmkuhl, O., Borrell, R., Perez-Segarra, C.D.,
Oliva, A., and Verstappen, R., 2010, LES and
Regularization Modeling of the Turbulent Flow
Over an Ahmed Car. ERCOFTAC Workshop.
[3] Krajnovic, S., and Davidson, L.,2005, Flow
Around a Simplified Car: Part1: Large
Eddy Simulation ASME: Journal of Fluids
Engineering, 127, pgs 907-918.
[4] Krajnovic S., and Davidson, L., 2005, Flow
Around a Simplified Car: Part2: Understanding
the Flow ASME: Journal of Fluids Engineering,
127, pgs 919-928.
[5] Minguez, M., Pasquetti, R., and Serre, E., 2008,
High Order Large Eddy simulation of flow over
the Ahmed body car model. Physics of Fluids
20.
[6] Perzon, S., and Davidson, L., 2000, On
Transient Modeling of the Flow Around Vehicles
Using the Reynolds Equation ACFD 2000
Beijing, China, pp 720-727.

[7] Nakashima, T., Tsubokura, M., Nouzawa, T.,

Nakamura, T., Zhang, H., and Oshima, N., 2008,
Large-Eddy Simulation of Unsteady Vehicle
Aerodynamics and Flow Structures BBAA VI
International Colloquium on: Bluff Bodies
Aerodynamics & Applications Milano, Italy.
[8] Lienhart, H., Stoots, C., and Becker, S., 2000,
Flow and Turbulence Structures in the Wake of
a Simplified Car Model (Ahmed Model) DGLR
Fach. Symp. der AG STAB, Stuttgart University.
[9] Trias, F., and Lehmkuhl, O., 2011, A
self-adaptive strategy for the time integration of
navier-stokes equations Numerical heat transfer,
Part B, fundamentals, 60 - 2, pgs. 116-134.
[10] Felten, F.N., and Lund,T. S., 2006, Kinetic
Energy Conservation Issues Associated with the
Collocated Mesh Scheme for Incompressible
Flow. Journal of Computational Physics, vol.
215, pgs 465-484.
[11] Nicoud, F., and Ducros, F., 1999, Subgrid-Scale
Stress Modelling Based on the Square of the
Velocity Gradient Tensor. Flow, Turbulence and
Combustion, vol. 62, pgs 183-200.
[12] Lehmkuhl, O., Baez, A., Rodriguez, I., and
Perez-Segarra, C.D., 2011, Direct Numerical
Simulation and Large-Eddy Simulations of the
Turbulent Flow Around a NACA-0012 Airfoil
7th International Conference on Computational
Heat and Mass Transfer.
[13] Nicoud, F., Baya, H., Cabrit, O., Bose, S.,
and Lee, J., 2011, Using Singular Values to
Build a Subgrid-Scale Model for Large Eddy
Simulations Physics of fluids, 23.
[14] Verstappen, R., 2009, When does eddy viscosity
damp subfilter scales sufficiently? Quality and
Reliability of Large Eddy Simulations II.
[15] Lehmkuhl, O., Perez-Segarra, C.D., Borrell,
R., Soria, M., and Oliva, A., 2007,
TERMOFLUIDS: A new Parallel unstructured
CFD code for the simulation of turbulent
industrial problems on low cost PC Cluster
Proceedings of the Parallel CFD 2007
Conference. Ismail H. Tuncer, p. 1-8.
[16] de Villiers, E., 2006, The Potential of Large
Eddy Simulation for the Modeling of Wall
Bounded Flows Department of Mechanical
Engineering Imperial College of Science,
Technology and Medicine.
[17] Kapadia, S., Roy, S., Vallero, M., Wurtzler,
K., and Forsythe, J., 2003, Detached-Eddy
Simulation Over a Reference Ahmed Car Model
41st Aerospace Sciences Meeting and Exhibit.