HAL Id: hal-01654042
https://hal.archives-ouvertes.fr/hal-01654042
Submitted on 2 Dec 2017
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Distributed under a Creative Commons Attribution| 4.0 International License
Drag and lift reduction of a 3D blu-body using active
vortex generators
Jean-Luc Aider, Jean-François Beaudoin, José Eduardo Wesfreid
To cite this version:
Jean-Luc Aider, Jean-François Beaudoin, José Eduardo Wesfreid. Drag and lift reduction of a 3D
blu-body using active vortex generators. Experiments in Fluids, Springer Verlag (Germany), 2010,
48 (5), pp.771-789. �10.1007/s00348-009-0770-y�. �hal-01654042�
Drag and lift reduction of a 3D bluff-body using active vortex
generators
Jean-Luc Aider
•
Jean-Franc¸ois Beaudoin
•
Jose
´
Eduardo Wesfreid
Abstract In this study, a passive flow control experiment
on a 3D bluff-body using vortex generators (VGs) is pre-
sented. The bluff-body is a modified Ahmed body (Ahmed
in J Fluids Eng 105:429–434 1983) with a curved rear part,
instead of a slanted one, so that the location of the flow
separation is no longer forced by the geometry. The influ-
ence of a line of non-conventional trapezoı
¨
dal VGs on the
aerodynamic forces (drag and lift) induced on the bluff-
body is investigated. The high sensitivity to many geo-
metric (angle between the trapezoı
¨
dal element and the wall,
spanwise spacing between the VGs, longitudinal location
on the curved surface) and physical (freestream velocity)
parameters is clearly demonstrated. The maximum drag
reduction is -12%, while the maximum global lift reduc-
tion can reach more than -60%, with a strong dependency
on the freestream velocity. For some configurations, the lift
on the rear axle of the model can be inverted (-104%). It is
also shown that the VGs are still efficient even downstream
of the natural separation line. Finally, a dynamic parameter
is chosen and a new set-up with motorized vortex generators
is proposed. Thanks to this active device. The optimal
configurations depending on two parameters are found more
easily, and a significant drag and lift reduc tion (up to -14%
drag reduction) can be reached for different freestream
velocities. These results are then analyzed through wall
pressure and velocity measurements in the near-wake of the
bluff-body with and without control. It appears that the
largest drag and lift reduction is clearly associated to a
strong increase of the size of the recirculation bubble over
the rear slant. Investigation of the velocity field in a cross-
section downstream the model reveals that, in the same
time, the intensity of the longitudinal trailing vortices is
strongly reduced, suggesting that the drag reduction is due
to the breakdown of the balance between the separation
bubble and the longitudinal vortices. It demonstrates that
for low aspect ratio 3D bluff-bodies, like road vehicles, the
flow control strategy is much different from the one used on
airfoils: an early separation of the boundary layer can lead
to a significant drag reduction if the circulation of the
trailing vortices is reduced.
1 Introduction
Flow control of separated and complex flows is a challenge
in both academic and industrial research. From the indus-
trial point of view, flow control is a way to increase the
performance of a given vehicle (aeronautics, car manu-
facturer, naval industry) or of the production apparatus
(chemical industry, energy production). From the academic
point of view, it is an exciting theoretical and experimental
problem implying a good knowledge of the target flow in
order to choose the optimal perturbation to control the
flow. The first step is then to define a control strategy to
modify the flow in order to reach the chosen objective. If
the objective is to reduce the drag or lift forces it is then
important to identify the flo w structures that contribute the
most to the aerodynamics forces to be able to choose and to
place properly the actuator.
J.-L. Aider ! J.-F. Beaudoin
PSA Peugeot-Citroe
¨
n, Research and Innovation Department,
route de Gisy, 78943 Ve
´
lizy-Villacoublay, France
J.-L. Aider (&) ! J.-F. Beaudoin ! J. E. Wesfreid
Laboratoire PMMH, UMR 7636, CNRS, ESPCI,
University Paris 6, University Paris 7-10, rue Vauquelin,
75231 Paris cedex 05, France
e-mail: aider@pmmh.espci.fr
1
The strategy to control the flow over a road vehicle is
very different from the one used to control the flow over an
airfoil or the body of an airpla ne (Joslin 1998): the ground
effect, the rotation of the wheels and the complex geom-
etries lead to a fully unsteady and complex 3D flow (Hucho
1998). Moreover skin friction is negligible and the aero-
dynamic forces (especially drag and lift forces) are mainly
governed by pressure losses. The consequence is that the
control of wall turbulence (Bewley et al. 2001; Kim 2003)
has rarely been tested in automotive aerodynamics com-
pared to the control of separation and large coherent
structures (Gad-El-Hak and Bushnell 1991; Greenblatt and
Wygnanski 2000).
As it is difficult to deal with all the complexity of the
flow over a real vehicle, it is important to define a simple 3D
geometry to study the relation betwee n the structures of the
near-wake and the aerodynamic forces. The most famous
bluff-body used in automotive aerodynamics is the so-
called ‘‘Ahmed body’’ (Ahmed 1983), which has a blunt
forepart and a rear part defined with different slant angles,
flat panels and sharp edges (Fig. 1a). The forepart is
designed to avoid separation so that the aerodynamic forces
are mainly governed by the large vortical structures created
on the rear part of the bluff-body (Fig. 1b): a closed or open
separation bubble over the rear slant (for slant angle
15" B h B 30"), a torus on the base of the rear part and two
longitudinal vortices created on the side edges of the rear
slant (the so-called ‘‘C-pillar vortex’’ in automotive aero-
dynamics). The example of the 30" rear slant has been
chosen to illustrate the discussion because, even if it is
highly unstable, there is a competition between the main
flow structures one can expect over the rear of a 3D bluff-
body, unlike the h \ 15" or the h [ 30" where either the
rear slant bubble or the longitudinal vortices are missing.
Here, one should emphasize that our description is a
simplification of the real topology of the flow as demon-
strated by Spohn and Gillie
´
ron (2002) for the h = 25"
configuration and by Vino et al. (2005) for the h = 30"
configuration, but it is sufficient for the following discus-
sions. In particular, the unsteady characteristics of the flow
would not be considered in order to focus on the time-
averaged structure of the near-wake. One of the main dif-
ferences in the description of the flow given by Vino et al.
(2005), compared to the one of Ahmed et al. (1984) is that
the separated flow does not reattach over the rear slant
leading to strong interaction with the recirculation torus as
shown also by Gillie
´
ron and Chometon (1999). One can
also notice that this sketch shows vortex shedding in the
recirculation bubble, which can be witnessed only in
instantaneous visualizations or measurements. They are
cancelled out by time-averaging, which may be an expla-
nation for the fact that these spanwise vortices were not
observed in other studies. Whatever the exact topology of
the flow, one of the main interest of this geometry is that it
reproduces the main features of the near-wake of a hatch-
back vehicle. It was also especially useful to demonstrate
the influence of the rear slant angle on the near-wake
structure and on the drag force. In the perspective of flow
control, and more precisely boundary layer manipulation,
the rear part of the Ahmed body has been modified. The new
geometry is detailed in Sect. 2.
From a general point of view, one can distinguish many
different strategies to control a separated flow. Depending
on the configurations and objectives, one can:
• control the shear layer at the separation (Chun et al.
1999; Aider and Beaudoin 2008; Verzicco et al. 2002).
It is easier when the location of the separation is well
defined like in the case of the backward-facing step or
the Ahmed body (Leclerc et al. 2006).
Fig. 1 a Side view and upper view of the original Ahmed body with a
30" rear slant. b Schematic view of the rear of the model together with
a sketch of the main flow structures expected over the rear of the
Ahmed body with a rear slant angle 15" B h B 30": recirculation
bubble, longitudinal vortices and recirculation torus. This sketch is
taken from an experimental study of Vino et al. (2005) on an Ahmed
body with a 30" rear slant and an a Reynolds number Re
L
= 2.8 9 10
6
2
• control the boundary layer upstream of the separation
(Song and Eaton 2002), which is a less common
strategy in automotive aerodynamics. It is interesting
when the location of the separation is not geometrically
imposed like in the case of a smoothly contoured ramp
(Duriez et al. 2006). The interest, and complexity, of
such a strategy is that controlling the upstream
boundary layer will modify both the location of the
separation and the properties of the shear layer.
• control the flow using actuation along the wall down-
stream the separation, like blowing, suction or both
blowing and suction (synthetic jets). It can be efficient
to control the separ ation (Roumeas et al. 2009), but the
energy balance may be less favorable than its upstream
counter-part.
• control the flow using actuation in the volume down-
stream the separation. For instance, it has been
demonstrated both experimentally (Strykowski and
Sreenivasan 1990; Dalton et al. 2001) and theoretically
(Giannetti and Luchini 2007) that it is possible to
modify the structure of the near wake of cylinder of
diameter d using a smaller cylinder (typically d/10).
The objective of the present study is to modify the
boundary layer properties using vortex generators to
control the separated flow over the rear part of a 3D bluff-
body. There are many ways to produce longitudinal
vortices leading to a large set of mechanical or fluidic
vortex generators that could be appropriate (Betterton et al.
2000; Smith 1994). In this study, an original vortex gen-
erator geometry is proposed and will be discussed later in
the paper.
The parameters defining the VGs have a strong influence
on their efficiency. This is the reason why a parametric
study has been carried out to show the sensitivity of the
drag and lift to the different parameters. Another objective
is also to find one geometric parameter that could be used
as a dynamic parameter in a closed-loop experiment
(Beaudoin et al. 2006).
The paper is organized as follows. In the first section,
the experimental set-up and the 3D bluff-body are descri-
bed. In the following sections, the vortex generators
geometry and the corresponding parameters are presented
before turning to the results of a detailed parametric study.
The first results obtained with motorized VGs is also pre-
sented. It allows an easier two-parameters study leading to
a global representation of the aerodynamic forces as a
function of these parameters. An experimental investiga-
tion of the near-wake of controlled and uncontrolled bluff-
body is then presented to try to understand the mechanisms
associated to the drag and lift reduction. The last section is
the conclusion.
2 Experimental set-up and reference flow
2.1 Description of the bluff-body
In order to deal with a 3D separated flow with a free
separation line, the rear of the original Ahmed bluff-body
(Ahmed et al. 1984), as shown on Fig. 1a, has been mod-
ified. The front part is unchanged (Fig. 2a), but the sharp
edges and flat walls on the rear part are replaced by a
rounded wall (Fig. 2b): the longitudinal cros s-section of
the rear part is now a constant radius circle arc. Thanks to
this rounded slant, the separation line is no longer forced
by the geometry. One can expect that the overall structure
of the flow over the rear of the model should be a little
different from the one of the Ahmed body shown on
Fig. 1b: indeed, one can only expect a competition between
the recirculation bubble induced by the separation over the
rounded slant and the longitudinal vortices created along
the side edges of the rear slant. At that point, it is conve-
nient to introduce a curvilinear coordinate s to define
properly the location of the VGs over the rounded wall. Its
origin s = 0 is located at the beginning of the rounded wall
and it is positive toward the rear of the model (Fig. 2a).
Fig. 2 On the left (a) side view of the bluff-body used in this study.
The front part (on the left of the picture) is similar to the original
Ahmed body while the rear part has a constant radius cross-section
(0.45 m) in order to create an unsteady separation line . The origin of
the curvilinear coordinate s is at the beginning of the rounded wall
and is positive toward the downstream direction. On the right (b)
Upper view showing the 3D geometry
3
The model is 0.29 m high, 0.34 m wide and 0.90 m
long. One shoul d notice here that the dimensions are
different from the original Ahmed body. The objective
was to modify the original dimensions to be closer to the
dimensions of a modern quarter-scale small vehicle
(typically a 206 Peugeot at the time of the study). The
curvature radius of its rear slant is 0.45 m. There are two
other differences with the classic Ahmed body: the height
of the underbody is smaller than the one of the Ahmed
body (0.04 m instead of 0.05 m) to be closer to realistic
configurations, and the struts are profiled to minimize the
perturbations induced by the the original rounded struts.
One should men tion that the same model has recently
been used in the framework of a flow control experiments
with flaps (Aider and Beaudoin 2008). The major differ-
ence is that the rear of the model was a flat slanted rear
panel similar to the Ahmed body. The overall dimensions
were the same as the ones of the model used in this study.
Some information about the near-wake structure and
aerodynamic coefficients of this more classic configura-
tion can be found in this reference.
2.2 Wind tunnel
All the measurements are carried out in the PSA Peugeot-
Citroe
¨
n in-house open wind tunnel (located in La Ferte
´
Vidame, in Franc e) which has a 6 m long closed test sec-
tion, with a rectangular cross-section 2.1 m high and 5.2 m
wide. The main characteristics of the flow in the wind
tunnel are the following:
• free-stream velocity ranging from U
0
= 20 m s
-1
to
U
0
= 40 m s
-1
• zero yaw angle
• Reynolds number Re ¼
U
0
L
m
¼ 1:2 # 10
6
to 2.4 9 10
6
,
L being the length of the model
• turbulence intensity = 1.3%
The coordinate axis are the following: x is the stream-
wise direction and is positive downstream, y is the span-
wise direction and is positive left, while z is the vertical
direction and is positive upward, as it is the convention in
automotive aerodynamics. The axis system origin is loca-
ted on the ground at mid-wheelbase and mid-track. The
velocity components (u, v, w) are then defined respectively
along the (x, y, z) axis.
As can be seen in Fig. 3, a fixed raised floor was used
for the measurements. The interest of such a configuration
is to control the boundary layer thickness. It can be seen as
an alternative to a boundary layer suction device. The
raised floor is 3 m wide and 0.052 m thick. The overall
blockage coefficient, including the bluff-body, the raised
floor and its profiled struts, is about 5%. The leading edge
of the raised floor has been covered with sandpaper to
avoid separation and generate a turbulent boundary layer
over the raised floor. It has been checked through visuali-
zation and hot-wire velocimetry. The boundary layer
thickness d (reached when u = 0.99 9 U
o
) upstream of
the model quickly decreases with tunnel speed: d = 6 9
10
-2
m for U
o
= 20 m s
-1
, d = 3.5 9 10
-2
m for
U
o
= 30 m s
-1
and d = 1.6 9 10
-2
m for U
o
= 40 m s
-1
.
A detailed analysis of the boundary layer over the raised
floor can be found in Golhke et al. (2008) where it is
clearly shown that the boundary layer is turbulent.
2.3 Experimental measurements
2.3.1 Aerodynamic balance
To evaluate the efficiency of the VGs on the aerodynamic
forces, a six-components aerodynamic balance is used.
Only the results on the drag and lift forces will be dis-
cussed. C
d
and C
l
are respectively the global drag and lift
coefficients, while C
lRear
is the lift coefficient applied on
the rear-axle. They are defined as:
Fig. 3 Description of the
experimental facility. The cross-
section of the wind tunnel is
5 m wide. The model is
mounted over a raised floor so
that the incoming turbulent
boundary-layer is smaller than
over the floor of the wind
tunnel. The scales are not
respected
4
