Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

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

net/publication/257253171

Study of aerodynamics for a simplified car model with the underbody shaped
as a Venturi nozzle

Article  in  International Journal of Vehicle Design · March 2012


DOI: 10.1504/IJVD.2012.045927

CITATIONS READS

20 3,864

3 authors, including:

Gabriela Huminic Adrian Soica


Universitatea Transilvania Brasov Universitatea Transilvania Brasov
50 PUBLICATIONS   1,298 CITATIONS    48 PUBLICATIONS   83 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Adrian Soica on 19 June 2014.

The user has requested enhancement of the downloaded file.


Int. J. Vehicle Design, Vol. 58, No. 1, 2012 15

Study of aerodynamics for a simplified car model


with the underbody shaped as a Venturi nozzle

Angel Huminic* and Gabriela Huminic


Department of Thermodynamics and Fluid Mechanics,
Aerodynamics Laboratory,
Transilvania University of Brasov,
Brasov 500036, Romania
E-mail: angel.h@unitbv.ro
E-mail: gabi.p@unitbv.ro
*Corresponding author

Adrian Soica
Department of Automotive Engineering,
Passive Safety and Testing Laboratory,
Transilvania University of Brasov,
Brasov 500036, Romania
E-mail: a.soica@unitbv.ro

Abstract: This paper presents new results concerning the flow around the
Ahmed body fitted with a rear underbody diffuser without endplates, to reveal
the influence of the underbody geometry, shaped as a Venturi nozzle, on the
main aerodynamic characteristics. The study is performed for different
geometrical configurations of the underbody, radius of the front section,
length and the angle of the diffuser being the parameters, which are varied.
Later, based on a theoretical approach, the coefficients of the equivalent
aerodynamic resistances of the front section of underbody and diffuser are
computed, which help to evaluate the drag due to underbody geometry.

Keywords: vehicle aerodynamics; Ahmed body; underbody geometry; Venturi


effect; CFD.

Reference to this paper should be made as follows: Huminic, A., Huminic, G.


and Soica, A. (2012) ‘Study of aerodynamics for a simplified car model
with the underbody shaped as a Venturi nozzle’, Int. J. Vehicle Design, Vol. 58,
No. 1, pp.15–32.

Biographical notes: Angel Huminic received his MS and PhD in Mechanical


Engineering from Transilvania University of Brasov, Romania, in 1996 and
2005, respectively. During 1996–2000, he was a Project Engineer at the Design
and Research Department of IAR Brasov, Romanian Company of Helicopters
and Aircrafts. Currently, he is an Associate Professor with Transilvania
University, Thermodynamics and Fluid Mechanics Department, where he is
charged with the Aerodynamics Laboratory. His academic and research
interests include fluid mechanics and vehicle aerodynamics. He has been a
member of SAE International USA since 2004.

Gabriela Huminic received her MS and PhD in Mechanical Engineering from


Transilvania University of Brasov, Romania, in 2001 and 2005, respectively.

Copyright © 2012 Inderscience Enterprises Ltd.


16 A. Huminic et al.

Currently, she is a Lecturer with Transilvania University, Thermodynamics


and Fluid Mechanics Department, Aerodynamics Laboratory. Her academic
and research interests include applied thermodynamics and industrial
aerodynamics.

Adrian Soica received his MS and PhD in Mechanical Engineering from


Transilvania University of Brasov, Romania, in 1997 and 2002, respectively.
Currently, he is an Associate Professor with Transilvania University,
Automotive Engineering Department, where is charged with the Passive Safety
and Testing Laboratory. His academic and research interests include accident
analysis and reconstruction for passenger vehicles, human injury tolerance to
impact loading and design of vehicle bodies.

1 Introduction

The study of flow around road vehicles became a subject of significant importance in the
automotive industry. One obvious way of improving fuel economy of vehicles is to
reduce aerodynamic drag by optimising the body shape. In this way, execution of good
aerodynamic design under safety and, not lastly, stylistic constraints requires an extensive
understanding of the flow phenomena and, especially, how the aerodynamics is
influenced by changes in body shape.
Till recently, the exterior shape of the vehicles represented the main preoccupation of
the engineers, from aerodynamics point of view. The underbody geometry was playing
a secondary role, being neglected for some vehicles, as for off-road ones. In the last
decade, to optimise the flow around cars, the flow management of the underbody also
become a major problem of the designers, as stated by Sumantran and Sovran (1996,
p.49).
To reduce the drag and to increase the downforce, many modern cars have the
underbody geometry shaped in such a way that the Venturi effect is generated in
proximity of the ground. Katz (2006, p.200) explained the reduction of the pressure of
fluid when the fluid flows through a constricted section of tubes (or stream tubes,
generally speaking). A rear-slanted underbody surface, which works as a diffuser, is the
most common method used to generate the Venturi effect for road vehicles.
Katz (2006, p.222) also showed that there is a similarity between the flow through a
Venturi tube and the underbody flow, even the latter does not exactly reproduce a
Venturi flow, which is internal. Thus, the diffuser of cars works to bring the low pressure
of air below the cars back to the ambient atmospheric pressure without inducing
turbulence. Another advantage of the Venturi underbody is that it offers to designers the
ability to control the centre of pressure of vehicles by using an adequate geometry of the
diffuser: angle and length.
Concerning the flow through the underbody diffusers with endplates, previous works
have been performed in this field using different types of bluff bodies. In this way, using
a simple Rover body fitted with a rear diffuser, Howell (1994) examines the effect of the
moving ground on aerodynamic characteristics. He concludes that the effectiveness of the
diffuser at reducing drag is dependent on the upper body shape. For a square back car,
a shallow diffuser has a small benefit, whereas a rear body generating lift benefits from
a much steeper diffuser. Cooper et al. (1998) in an extensive study investigate the
Study of aerodynamics for a simplified car model 17

contribution of each parameter of the diffusers for several values of ground clearance and
propose some physical arguments to explain the performances of the diffusers. Later,
Cooper et al. (2000) propose an analytical model that facilitates the selection of optimum
length and area ratio for diffusers of flat-bottomed racing cars. Desai et al. (2008) and
Breslouer and George (2008) have continued the first work of Cooper et al. for a similar
body with the addition of nearby stationary wheels.
Similar works in this field were also reported by George (1981), Senior and Zhang
(2001) and Zhang et al. (2004). All these studies emphasised the influence of underbody
shape in the mechanism of generation and increasing of downforce. Using a more
concrete body, Buchheim et al. (1981) examine the effect of the underbody rear slant
on the drag and rear lift of a VW concept car. There is shown that the lift at the rear axle
is reduced with increasing of the slant angle. Drag is reduced up to a slant angle of 4°.
Above this value drag increases.
Being a continuous subject of research, as shown previously, the authors present
in this paper new aspect concerning the flow around the Ahmed body fitted with a simple
rear underbody diffuser, without endplates, and having a variable radius of the lower
front section. In this way, it revealed the influence of the underbody geometry, shaped as
a Venturi nozzle, on the main aerodynamic characteristics, lift and drag. The study is
performed for different geometrical configurations of the Venturi underbody, radius of
the front section, length and the angle of the diffuser being the parameters, which are
varied in ranges that are relevant for hatchback passenger cars. Later, based on a
theoretical approach, the coefficients of the equivalent aerodynamic resistances of the
front section of underbody and diffuser are computed, which help to evaluate the drag
due to underbody geometry. Thus, dividing the underbody into components that permit
computation of coefficients of equivalent aerodynamic resistance, an optimisation of flow
beneath bodies in ground effect can be performed. This approach can help to evaluate the
contribution of the underbody on total drag, even in an early stage of the design process.
The present contribution is a companion paper of Huminic and Huminic (2010).
The option for the Ahmed body was made taking into consideration that this generic
shape of car, used for a very first time in 1984 (Ahmed et al., 1984), is much studied
from aerodynamic point of view. It is also used to validate investigations using CFD,
which has become a valuable tool for fine-tuning of the external shape of the road
vehicles (Guilmineau, 2007). The hatchback profile was considered based on conclusion
of Howell (1994), mentioned above.

2 Theoretical approach

The aerodynamic performances of the vehicles are characterised using specific


dimensionless indicators, as drag and lift coefficients. Using these as measure of the state
of the art in the vehicle aerodynamics, continuous progress is possible in this field. In this
context, because the decomposition of the aerodynamic forces into measurable
components would facilitate the optimisation of the design process, in a previous study
(Huminic and Huminic, 2010), a theoretical method for computation of the drag of an
underbody shaped as a Venturi tube/nozzle was presented. Thus, there was proposed the
decomposition of the total drag into two following components:
D = Dext + Dub, (1)
18 A. Huminic et al.

where
Dext drag due to the airflow on external upper surfaces of the vehicle, having the flow
rate Qext, as shown in Figure 1;
Dub drag due to the flow under the body of vehicle, in the space between the lower
surface of the vehicle and road, having the flow rate Qub.

Figure 1 Main geometrical characteristics of the nozzle at the level of underbody

The sum of Qext and Qub represent the volume of air what is enveloping the vehicle,
dislocated in unit time:
Qext + Qub = Q = v∞A, (2)
where
v∝: reference velocity of the free stream
A: reference area of the vehicle.
The main components of Qub are given by:
• the air in atmospheric conditions flowing through the Venturi nozzle; this term
contains also lateral inflows into the underbody region by means of free
ejection
• the lower branch of the stream generated through impact at the leading edge, which
flows under the vehicle; a significant fraction of this airstream is used, generally, for
cooling in the engine compartment.
Considering that the resultant fluid is homogeneous in the entire cross section of the
nozzle b × h, for the second component of the drag of vehicles, Dub, was determined
analytically the following equation:
ρ v3
Dub = ζ ub bh (3)
2 v∞

where
ζub: coefficient of the equivalent aerodynamic resistance of the nozzle
(underbody geometry)
v: average velocity of the air through the constant section of the nozzle.
Also, the following dimensionless indicators were proposed to characterise the
underbody airflow:
Study of aerodynamics for a simplified car model 19

KD: coefficient that represent the ratio between underbody drag and total drag defined
as product of three dimensionless factors, equation (4)
KQ: coefficient that shows the contribution of the underbody flow rate on total
flow rate, equation (5)
3
D ζ bh v 
K D = ub = ub   , (4)
D cD A  v∞ 

where
ζ ub
represents the relative drag
cD

bh
relative area
A
v
relative velocity.
v∞

Qub
KQ = . (5)
Q

In this way, the underbody drag coefficient cDu can be expressed with the following
equation:
3
bh  v 
cDu = K D ⋅ cD = ζ ub   . (6)
A  v∞ 

There are two variables in the equation (6): the coefficient of the equivalent aerodynamic
resistance of the nozzle and the average velocity of the air through the section of the
nozzle. Dividing the underbody geometry into the following three sections, as in
Figure 2, the coefficients of the equivalent aerodynamic resistance of the nozzle can be
computed as:
ζ ub = ζ i + ζ m + ζ d (7)

where

ζi coefficient of the aerodynamic resistance of the inlet section

ζm coefficient of the aerodynamic resistance of middle (constant) section

ζd coefficient of the aerodynamic resistance of diffuser.

By knowing these coefficients, the underbody drag can be analytically evaluated using
equation (6), and some reductions of the underbody drag can be operated in the
design stage by manipulation of ζub. In this paper, using the results of several CFD
analyses, the coefficients of the aerodynamic resistance of the inlet section and diffuser,
ζi and ζd respectively, are computed. Being the result of the pressure losses in a stream
20 A. Huminic et al.

tube, such of coefficients can be computed using similar procedures provided by Idelcik
(1984, p.30):
∆ptot ∆p ( p − pout ) tot
ζ = = 2 tot2 = 2 in , (8)
pdyn ∞ ρ∞ v∞ ρ∞ v∞2

where
∆ptot variation of the total pressure between inlet and outlet sections
of stream tube
ρ ∞ v∞2
pdyn ∞ = dynamic reference pressure of the stream.
2

Figure 2 The main geometrical characteristics of the Venturi underbody nozzle

3 CFD methodology

The flow field around a vehicle is physically very complex. In consequence, the
efficiency of an aerodynamic CFD simulation depends on many factors. Creation of the
model geometry and its integration in a physical domain, grid generation and choice of a
suitable numerical computing scheme are significant factors that can determine the level
of success of the simulation process. The CFD methodology for these simulation
processes of this study were detailed in the previous work by Huminic and Huminic
(2010). In the following sections, the main steps of the performed simulations are briefly
presented.

3.1 CAD model


The Ahmed reference model was originally developed in 1984 for a time-averaged
vehicle wake investigation. It is a bluff body, with a curved front section, a straight
middle section and an angled rear end, representing a simplified 1 : 4 scale model of
lower-medium size hatchback vehicles, as illustrated in Figure 3. The geometry of this
bluff body was designed to be such that an experiment could be conducted with reference
to only one significant aerodynamic feature, namely the flow over the slanted rear end,
as flow was expected to remain attached over the other sections.
In this sense, the model was studied for various specific angles of the back end,
between 0° and 40°, in 5° increments. This variation in geometry provides a range of
flow characteristics over the back end of the model. Detailed test cases descriptions can
Study of aerodynamics for a simplified car model 21

be found in many studies, as those of Ahmed et al. (1984), Guilmineau (2007) and
Strachan et al. (2007).

Figure 3 Ahmed body, simplified scale model of a hatchback vehicle

Thus, it has been shown that the flow over the angled upper back section is dependent on
the specific slant angle. There have been found to be two critical angles, 12.5° and 30°,
at which the flow structure changes significantly, revealed also by changing of curvature
of the drag variation. Below 12.5°, airflow over the angled surface remains fully attached
before separating from the model when it reaches the vertical surface of the back end.
The flow from the angled section and the side walls produces a pair of counter rotating
vortices, which continue downstream. For angles between 12.5° and 30°, the flow over
the angled section becomes highly three-dimensional. Two increased counter-rotating
lateral vortices are again shed from the sides of the angled section. The increased vortex
size affects the flow over the whole back end, causing a three-dimensional wake.
These vortices are also responsible for maintaining attached flow over angled surface up
to an angle of 30°, and it has been shown by Strachan et al. (2007) that they are
extended up to half of the length of the model (0.48 l) beyond the model rear end.
Close to the second critical angle, a separation bubble is formed over the back end.
The flow separates from the body, but reattaches before reaching the vertical back
section, near the middle of the slant. At this point, the flow again separates from the
model. Above 30°, flow over the angled section is fully separated. There remains though
a weak tendency of the flow to turn around the side edge of the model, depending on the
separation positions of the flow over model top and over the back end side edges.
When the flow is in this state, a near constant pressure was found across the backlight,
as shown by Strachan et al. (2007). As consequence, changes in drag are not significant
for rear slant upper surface above 30°.
On the basis of the previously described and Howell’s (1994) conclusion mentioned
above, current work is performed on the Ahmed body having a rear slanted upper surface
of 35°. This angle corresponds to a low-drag configuration, quasi-two-dimensional wake,
as shown by Guilmineau (2007). The side vortices are also weaker for this case.
The body was located 50 mm above the ground, as in original work of Ahmed, but
not mounted on stilts, to benefit from the facilities of used code to simulate relative
motion between bodies and ground, as in recent experiments of Strachan et al. (2007).
An underbody diffuser and a variable rounded front section were also considered, as
shown in Figure 4.
22 A. Huminic et al.

Figure 4 Ahmed body with underbody diffuser and variable rounded lower inlet section;
dimensions are in mm

The CFD code used for this study was ANSYS CFX-12.0, which is fully integrated fluid
analysis software of ANSYS Workbench platform (ANSYS Inc., Canonsburg,
PA, USA). It combines CAD (modelling and input), complex meshing solutions, fast
solution algorithm and post-processing facilities. The surfaces of the Ahmed body were
drawn as parameterised CAD data with the aid of the design module and integrated into
computational domain, a rectangular enclosure. This was of such dimensions that the
adverse pressure effects between the body and the walls are negligible (Huminic and
Huminic, 2010). To avoid the possible blockage effect, the cross-sectional size was set
to be relatively larger than usually in the wind tunnel facility. The assigned blockage ratio
was smaller than 2%.

3.2 Grid resolution and boundary conditions


The grid was generated using a multi-block scheme, with tetrahedral elements, and
hexahedral elements nearest to surfaces of the body to solve accurately the flow in the
proximity of the latter. In this sense, the side length of the elements on the surface of the
vehicle was in range of (0.001 ÷ 0.004) m. Also, the maximum distance from surface
of the vehicle to the first layer of grid points was set to fulfil the condition of y + ≤ 100,
and for computations the ‘wall function’ approach was used. A resolution of the boundary
layer of 30 points was assigned. For a half of model, the dimensions of the computational
grids were more than:
• 1,275,000 grid points for entire computational domain
• 32,500 grid points for surfaces of body.
Boundary conditions for the computation were applied based on wind tunnel tests
described by Strachan et al. (2007), used for validations, too:
• a uniform and constant velocity vx = v∝ (velocity of air free stream) and vy = vz = 0
were imposed at the inlet boundary
• same conditions were considered for the surface, which represent the ground, and
in addition, the option ‘solid moving wall’ was activated
Study of aerodynamics for a simplified car model 23

• at the outlet boundary, a zero pressure condition was imposed: p = 0


• no slip conditions on surfaces of the Ahmed body: vx = vy = vz = 0
• fluid walls and free slip conditions were assigned for the rest of the surfaces.

3.3 Model governing equations and conditions of simulations


The hardware on which the analyses were conducted is the cluster of the Aerodynamics
Laboratory from Thermodynamics and Fluid Mechanics department of Transilvania
University of Brasov. It is a parallel computer system of distributed memory, and it is
composed of 20 processor cores with 100 GB memory RAM, and it represents a practical
tool for engineering problems involving fluid dynamics.
For computing the variables of flow, the used code solves the full Reynolds-Averaged
Navier–Stokes (RANS) equations in their conservation form: mass (continuity) equation,
momentum equation and energy equation, if necessary. For this study, the steady-state
three-dimensional RANS equations of mass and momentum conservation for an
incompressible, viscous, isothermal flow of a Newtonian fluid, in Cartesian coordinates,
in partial differential equation and conservation form with the Boussinesq’s hypothesis
are:
∂vi
=0 (9)
∂xi

∂vi v j ∂p ∂  ∂v ∂v  ∂
ρ∞ =− + µ∞  i + j  − ρ ∞ (v 'i v ' j ) (10)
∂x j ∂xi ∂x j  ∂x ∂x  ∂xj
 j i 

where
xi, xj: Cartesian coordinates
vi: ith component of the averaged air velocity
vi′, v′j Fluctuating velocity parts

p Average pressure
ρ∝ Density of air free stream, which is assumed constant
µ∝ Dynamic viscosity of air free stream, which is assumed constant.
The indexes i and j are used to distinguish Cartesian components {1, 2, 3}. If an index
appears twice in a term, summation over that index is required. The turbulent Reynolds
stress tensor, vi′v′j , in equation (5) is modelled by adopting the extended Boussinesq’s
hypothesis, which relates the turbulent stresses to the mean rate of deformation
(Versteeg and Malalasekera, 1995):

µt  ∂vi ∂v j  2
v 'i v ' j = −  +  + kδ ij (11)
ρ  ∂x j ∂xi  3
24 A. Huminic et al.

where
µt Turbulent viscosity
δij Kronecker delta: δij = 1 if i = i and δij = 0 if i≠j
k Turbulence kinetic energy per unit mass: k = (vi′v′j ) / 2.

The Shear-Stress-Transport (SST) closure model of Menter (1994) was used to solve the
simulation processes. This is a two-equation eddy-viscosity model, based on k–ω model
of Wilcox (1986). The SST model was developed to effectively blend the robust and
accurate formulation of the k–ω model in the near-wall region, with the free stream
independence of the k–ε model in the far field. This model uses a blending function to
switch from k–ω to k–ε in the wake region to prevent the model from being sensitive
to free stream conditions. The definition of the turbulent viscosity is modified also to
account for the transport of the turbulent shear stress, and the modelling constants are
different. These features result in a major improvement in terms of flow separation
predictions, and the performances of this model have been demonstrated in a
large number of validation studies, including previous experiences of the authors,
Huminic and Huminic (2009, 2010).
In this study, the analyses were performed for the following conditions of the
International Standard Atmosphere: air pressure p∝ = 101,325 Pa, air temperature
t∝ = 15°C. These were used for computation of the rest of the air-free stream parameters,
as density and viscosity. The reference velocity of air-free stream was v∝ = 40 m/s
(Re = 2.63 ⋅ 106, Reynolds number computed with the Ahmed body length), and the
turbulence intensity was set to 0.2%. The modelling constants used in model equations
are the standard values that can be found in the literature.

3.4 Validation of computational procedure


In a first step, the accuracy of the code to solve the flow around vehicle for the used
computational grid and procedure was checked vs. experimental results. Thus, the
flow around the Ahmed reference model for 35° angle of the back end was simulated
in similar conditions as described by Strachan et al. (2007). In the experimental
configuration, the model was supported from above by an aerodynamic strut, but
due to the lack of data about its profile, this could not be reproduced in the current
work.
The solution was considered finished when the variations of normalised rate of
change for the variables of processes were insignificant for the final steps of iterations.
These variables include the components of velocity, pressure and turbulence quantities.
The main convergence criteria checked were:
• decreasing of the residuals below 10–4, and variations of the aerodynamic loads
(acting on the body) smaller than 0.5% for the final iterations
• a value of y+ < 100 to the first grid points above the surfaces of the vehicles
• a continuous and physically realistic distribution of the variables of the process.
The used CFD code allows calculating the aerodynamic loads, D and L (drag and lift),
that are acting over surfaces of vehicle, and the aerodynamic coefficients, cD and cL are
Study of aerodynamics for a simplified car model 25

computed. Figure 5 shows results concerning the aerodynamic coefficients of this work
vs. experimental values.
D L
cD = 2 ; cL = 2 . (12)
ρ∞ v∞2 A ρ∞ v∞2 A

Figure 5 Current and previous experimental results for the Ahmed reference model for 35° angle
of the back end

As can be shown, the accuracy of computational procedure with respect to prediction


of the aerodynamic loads is very good for the test case and, especially concerning the
drag. A deviation occurs for lift coefficient, which can be explained by the presence of
the sting in the experimental works. According with Hetherington and Sims-Williams
(2006), a sting induces errors, more significant on (rear) lift. In conclusion, the adopted
CFD procedure proves to be correct.

4 Results and discussions

In the first stage, the influence of the diffuser on drag and lift was studied for five values
of diffuser length, ld, each of them for the following angles αd = 1°, 3°, 5°, 7°, 9°.
The values of ld are corresponding to the following normalised lengths: (ld/l) = 0.1, 0.2,
0.3, 0.4, 0.5. Additional simulations were performed to establish the influence of the
radius of the front section of the body, ri = (0, 20, 40, 60, 80, 100) mm. The choice for
ranges of the studied parameters was made based on results of Buchheim et al. (1981),
mentioned above, and taking into consideration that for common cars, hd and ri are
limited by the presence of structural components, as front and rear bumpers. Force and
pressure data computed for the model, are presented in this paper, in terms of
dimensionless coefficients.
Concerning the aerodynamic loads on body with underbody diffuser, the results are
presented graphically in Figure 6 as lift and drag increment, ∆cL = cL – cLref and
∆cD = cD – cDref. As references were considered the values of the aerodynamic
coefficients of the Ahmed body obtained in the analysis of the Ahmed reference model
for 35° angle of the back, used for validation.
As stated in previous works, and also shown in above figures, for the bodies having
an underbody diffuser, due to the effect of the ground which constrains the flow beneath
the model, a Venturi effect occurs for small and moderate values of the angle of the
diffuser. This result in a negative lift (downforce), and the decreasing of the lift is
accompanied also by a decreasing of drag. While downforce is continuously increasing
with αd and ld, drag has a predictable behaviour only for angle of the diffuser up to
αd = 7°. Above this value, separations of airstream may occur, which can affect
26 A. Huminic et al.

significantly the aspect of flow behind body, mainly due to the increasing of the vortex
structures induced on the sides of the diffuser surface by lateral inflows into the
underbody region, as show in Figure 7 for three angles of diffuser.

Figure 6 Increment of lift and drag coefficients vs. normalised length of diffuser

Figure 7 Vortex structures induced on the sides of the diffuser surface (ld/l) = 0.3

No significant changes of the aerodynamic loads are recorded due to variation of ri,
but a curved front section influences the pressure variation on underbody, as shown
below.
The pressure coefficient variations due to the changes in shape of the underbody are
shown in Figures 8 and 9, in the symmetry plane of the body. These variations are useful
for understanding the pressure field, downforce generation and the role of the diffuser.
Dashed-line curves represent the results obtained in the analysis of the Ahmed reference
model for 35° angle of the back, used also for validation. Thus, starting from the lower
edge of the front face, with a pressure coefficient close to unity, characteristic of
stagnation points, the flow accelerates around the lower rounded surface producing
negative pressures, which peak in the beginning part of the flat underbody. This peak
decreases significantly with the radius of the lower front section, ri, and increases slightly
with the length of the diffuser, ld (see Figure 8). Next, a pressure recovery follows over
the flat part of the underbody. While this recovery is described by a uniform and
continuous increase of the pressure for a flat underbody, pressure recovery occurs in two
stages for the underbody with diffuser. The flow accelerates also near the diffuser inlet,
pressure decreases and another peak of the pressure coefficient is recorded at the level of
the first edge of diffuser. This second peak of the pressure increases with both length
and angle of the diffuser, more significantly with the latter parameter. Decreasing the
pressure coefficient results in increasing of the downforce generated by underbody,
accomplished by a reduction of the resultant drag of the body, which is the main benefit
for passenger cars.
Study of aerodynamics for a simplified car model 27

Figure 8 Pressure coefficient variations with the radius of the lower front section

Figure 9 Pressure coefficient variations with the angle and length of the underbody diffuser

Changes on the flow aspect over the upper side of the body appear if the angle of the
diffuser is increased, as shown in Figure 9. Drag reductions are not significant for such
situations, rising of the downforce being the only one advantage.
Benefiting by a continuous and predictable variation of drag for αd ≤ 7°, as shown in
Figure 6, the coefficients of the aerodynamic resistance of the inlet section and diffuser
may by computed analytically with the aid of theory presented above. Thus, the equation
for pressure loss induced by changes of the flow section was used for computation of
ζi (Idelcik, 1984, p.105), adjusted by a coefficient to capture the influence of curvature.
For underbody diffusers, Cooper et al. (1998) show that the streamwise pressure variation
is also a function of the ratio of the inlet area and outlet area of the diffuser. A similar
equation was considered for the computation of ζd:
m m n
 h  r  1  α  h 
ζ i = Cir 1 −  = Ci 1 −  ; ζ d = Cd  1 −  (13)
 h + r   1 + r / h   hd 

where
Cir , Cdα : adjustment coefficients; Cir = 0.5 if r = 0

m, n: exponents which depend on the inlet conditions; m, n = (0.75 ÷ 1); m , n ≅ 1


for uniform and constant velocity profiles; for this study m = n = 0.75, as for
non-uniform velocity profiles.
28 A. Huminic et al.

In a first stage, ζi and ζd were computed with the equation (8) using the information
provided by CFD analyses about weighted average total pressure in the inlet and outlet
cross sections of the lower front part, respectively, of the diffuser, Sdin and Sdout as in
Figure 7. In this way, the coefficients Cir and Cdα can be computed as function of ri,
respectively αd, as shown in Figure 10, and later, they can be expressed analytically.

Figure 10 Variations of Cir and Cdα vs. ri/h, respectively, αd

For this study, power fit equations were considered:

Cir = e[ a1 ⋅lg( r / h ) + b1 ] ; Cdα = e[ a2 ⋅lg(α d ) + b2 ] , (14)

where
a1 = –0.34333, b1 = –3.16865 and a2 = –0.92255, b2 = –1.59543.
In this way, using equations (13), the coefficients of the aerodynamic resistance, ζi and ζd,
can be analytically computed as functions of the geometrical characteristics of the
underbody. Figures 11 and 12 show a three-dimensional map of ζd, and the distribution
of deviations for analytical results from CFD: XCFD – Xanalytic.

Figure 11 3D map of Cdα


Study of aerodynamics for a simplified car model 29

Figure 12 Distribution of deviations of analytical results from CFD

5 Summary and conclusions

In this paper, the authors present new aspects concerning the flow around the Ahmed
reference model for 35° angle of the back, fitted with a simple rear underbody diffuser,
without endplates and heving a variable radius of the lower front section. The influence
of the underbody geometry shaped as a Venturi nozzle on the main aerodynamic
characteristics, lift and drag, is revealed. The study is performed for different geometrical
configurations of the underbody, radius of the front section, length and the angle of the
diffuser being the parameters, which are varied in ranges that are relevant for hatchback
passenger cars.
Concerning the lower front section, no significant changes of the aerodynamic loads
are recorded due to variation of its radius, but it is shown that the curved front section
influences the pressure variation on underbody. A coefficient of the aerodynamic
resistance of the inlet section is analytically computed. Study of the influence of the
diffuser on the main aerodynamic characteristics reveals that there is continuous and
predictable variation of lift and drag for angle of the diffuser αd ≤ 7°. Above this value,
separations of airstream may occur due to the increasing of the vortex structures induced
on the sides of the diffuser surface by lateral inflows into the underbody region, which
can affect significantly the aspect of flow behind the body. Benefiting by a continuous
and predictable variation of drag with diffuser angle, a coefficient of the aerodynamic
resistance of the diffuser is also analytically expressed as function of parameters of
diffuser, with adequate accuracy, as shown in Figure 12.
Thus, an optimisation of the flow beneath bodies in ground effect can be performed
by dividing the underbody, shaped as a Venturi nozzle, into components that permit
computation of coefficients of aerodynamic resistance. This approach can help to
evaluate the contribution of the underbody to the total drag, even in an early stage of the
design process.
Not least, thanks to increases in computational power, CFD has become a valuable
tool for fine-tuning the external shape of the road vehicles by providing much more
complex information than from a routine experiment, before a testable model even exists.
Also, for ground effect situations, Berber (2006) shows that the numerical simulation is a
feasible option. Simulation of the relative motion between vehicle and road is also easy to
accommodate, comparatively as in wind tunnels.
30 A. Huminic et al.

Acknowledgement

This work was supported by the Romanian Research Agency (CNCSIS), PN II ID 758
and ID 130 programmes, using the resources of Aerodynamics Laboratory of
Thermodynamics and Fluid Mechanics Department, Transilvania University of Brasov.

References
Ahmed, S., Ramm, G. and Faltin, G. (1984) Some Salient Features of the Time-Averaged Ground
Vehicle Wake, SAE Technical Paper 840300.
Berber, T. (2006) ‘Aerodynamic ground effect: a case study of the integration of CFD
and experiments’, International Journal of Vehicle Design, Vol. 40, pp.299–316,
ISSN 0143-3369.
Breslouer, J.O. and George, A.R. (2008) Exploratory Experimental Studies of Forces and Flow
Structure on a Bluff Body with Variable Diffuser and Wheel Configurations, SAE Technical
Paper 2008-01-0326, Vehicle Aerodynamics, 2008, SAE SP-2151.
Buchheim, R., Deutenbach, K.R. and Laxckoff, H.J. (1981) Necessity and Premises for Reducing
the Aerodynamic Drag of Future Passenger Cars, SAE Technical Paper 810185.
Cooper, K.R., Bertenyi, T., Dutil, G., Syms, J. and Sovran, G. (1998) The Aerodynamic
Performance of Automotive Underbody Diffusers, SAE Technical Paper 98-0030.
Cooper, K.R., Sovran, G. and Syms, J. (2000) Selecting Automotive Diffusers to Maximise
Underbody Downforce, SAE Technical Paper 2000-01-0354.
Desai, S., Lo, C-M. and George, A.R. (2008) A Computational Study of Idealized Bluff Bodies,
Wheels, and Vortex Structures in Ground Effect, SAE Technical paper 2008-01-0327, Vehicle
Aerodynamics, SAE SP-2151.
George, A.R. (1981) ‘Aerodynamic effects on shape chamber, pitch, and ground proximity on
idealized ground vehicle body’, ASME Journal of Fluid Engineering, Vol. 103, pp.631–638,
ISSN 0098-2202.
Guilmineau, E. (2007) ‘Computational study of flow around a simplified car body’, Journal of
Wind and Industrial Aerodynamics, Vol. 96, pp.1207–1217, doi: 10.1016/j.jweia.2007.06.041.
Hetherington, B. and Sims-Williams, D. (2006) Support Strut Interference Effects on Passenger
and Racing Car Wind Tunnel Models, SAE Technical Paper 2006-01-0565, Vehicle
Aerodynamics, 2006, SAE SP-1991.
Howell, J.P. (1994) ‘The influence of a vehicle underbody on aerodynamics of a simple car shapes
with an underfloor diffuser’, Vehicle Aerodynamics, R.Ae.S. Conference, Loughborough, UK,
pp.36.1–36.11.
Huminic, A. and Huminic, G. (2009) CFD Study Concerning the Influence of the Underbody
Components on Total Drag for a SUV, SAE Technical Paper 2009-01-1157, Vehicle
Aerodynamics, 2009, SAE SP-2226.
Huminic, A. and Huminic, G. (2010) Computational Study of Flow in the Underbody Diffuser for a
Simplified Car Model, SAE Technical Paper 2010-01-0199, Vehicle Aerodynamics, 2010,
SAE SP-2269.
Idelcik, I.E. (1984) Guide of Hydraulic Resistances Computation, Editura Tehnica, Bucuresti.
Katz, J. (2006) Race Car Aerodynamics – Designed for Speed, 2nd ed., Bentley Publisher,
ISBN 978-0-8376-0142-7.
Menter, F.R. (1994) ‘Two-equation eddy-viscosity turbulence models for engineering applications’,
AIAA Journal, Vol. 32, pp.1598–1605, ISSN 0001-1452.
Senior, A.E. and Zhang, X. (2001) ‘The force and pressure of a diffuser-equipped bluff body in
ground effect’, ASME Journal of Fluid Engineering, Vol. 123, pp.105–111, ISSN 0098-2202.
Study of aerodynamics for a simplified car model 31

Strachan, R., Knowles, K. and Lawson, N. (2007) ‘The vortex structure behind an Ahmed reference
model in the presence of a moving ground plane’, Journal of Experiments on Fluids,
Vol. 42, No. 5, pp.659–669, ISSN 0723-4864.
Sumantran, V. and Sovran, G. (1996) Vehicle Aerodynamics, PT-49, SAE International, USA,
ISBN 1-56091-594-3.
Versteeg, H.K. and Malalasekera, W. (1995) An Introduction to Computational Fluid Dynamics.
The Finite Volume Method, Prentice-Hall, Englewood Cliffs, NJ, p.257.
Wilcox, D.C. (1986) ‘Multiscale model for turbulent flows’, AIAA 24th Aerospace Sciences
Meeting, American Institute of Aeronautics and Astronautics, Reno, Nevada, USA,
Paper 86-0029.
Zhang, X., Senior, A. and Ruhrmann, A. (2004) ‘Vortices behind a bluff body with an upswept aft
section in ground effect’, International Journal of Heat and Fluid Flow, Vol. 25, pp.1–9,
ISSN 0142-727X.

Nomenclature
A Reference area of vehicle (body) (m2)
b Width of vehicle (body) (m)
cD Drag coefficient of vehicle (body)
cDu Underbody drag coefficient
cL Lift coefficient of vehicle (body)
D Total drag of vehicle (body) (N)
Dext Drag due to airflow, which envelopes the external upper surfaces of vehicle (body) (N)
Dub Underbody drag, drag due to flow beneath vehicle (body) (N)
KD Coefficient that represent ratio between underbody drag and resultant drag
KQ Coefficient which shows contribution of underbody flow rate on total flow rate
h Ground clearance of vehicle (body) (m)
hd Height of underbody diffuser (m)
k Turbulence kinetic energy per unit mass (m2s–2)
L Lift (downforce) of vehicle (body) (N)
l Length of vehicle (body) (m)
ld Length of underbody diffuser (m)
p Pressure (Nm–2)
ptot Total pressure (Nm–2)
–2
pdyn∝ Dynamic pressure of free stream (Nm )
Qext Flow rate of air, which envelopes the external upper surfaces of vehicle (body) (m3 s–1)
Qub Flow rate of air beneath vehicle (body) (m3 s–1)
Re Reynolds number;
r Radius (m)
v Velocity (m s–1)
ν Component of velocity (m s–1)
ν′ Fluctuating velocity parts (m s–1)
v∝ Velocity of free stream (m s–1)
32 A. Huminic et al.

t∝ Temperature of free stream (°C)


x Cartesian coordinate (m)
X Result
y+ Dimensionless wall coordinate
αd Angle of underbody diffuser (°)
ε Turbulence eddy dissipation (m2s–3)
δij Kronecker delta
µt Turbulent viscosity (kg m–1s–1)
µ∝ Dynamic viscosity of free stream (kg m–1s–1)
ω Turbulence eddy frequency (s–1)
ρ∝ Density of air free stream (kg m–3)

ζi Coefficient of aerodynamic resistance of vehicle (body) inlet section

ζm Coefficient of aerodynamic resistance of vehicle (body) middle (constant) section

ζd Coefficient of aerodynamic resistance of vehicle (body) diffuser

View publication stats

You might also like