Art-1. Exp Fluids-2010, Vol.48 (Pages 1–16)

download Art-1. Exp Fluids-2010, Vol.48 (Pages 1–16)

of 16

Transcript of Art-1. Exp Fluids-2010, Vol.48 (Pages 1–16)

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    1/16

    RE S E A RCH A RT I CL E

    Aerodynamic drag reduction by vertical splitter plates

    Patrick Gillieron

    Azeddine Kourta

    Received: 8 June 2008/ Revised: 13 June 2009/ Accepted: 15 June 2009 / Published online: 2 July 2009

    Springer-Verlag 2009

    Abstract The capacity of vertical splitter plates placed at

    the front or the rear of a simplified car geometry to reducedrag, with and without skew angle, is investigated for

    Reynolds numbers between 1.0 9 106 and 1.6 9 10

    6. The

    geometry used is a simplified geometry to represent estate-

    type vehicles, for the rear section, and MPV-type vehicle.

    Drag reductions of nearly 28% were obtained for a zero

    skew angle with splitter plates placed at the front of models

    of MPV or utility vehicles. The results demonstrate the

    advantage of adapting the position and orientation of the

    splitter plates in the presence of a lateral wind. All these

    results confirm the advantage of this type of solution, and

    suggest that this expertise should be used in the automotive

    field to reduce consumption and improve dynamic stability

    of road vehicles.

    List of symbols

    LA Length of the Ahmed body

    lA Rear window length

    wA Width of the Ahmed body

    HA Total height of the Ahmed body

    h Height of the geometry front partw Width of the geometry front part

    Re Reynolds number based on the geometry

    length

    sl Viscous shear stress tensor

    st Turbulent shear stress tensor

    Pio Farfield total pressure

    P Static pressure

    dr Surface element

    n~ Normal vector unit

    x~ Vector unit in the longitudinal plane

    R Surface of the outlet boundaries around

    Ahmed body (R RL Se Ss Sc)RL Lateral surface

    Se Inlet section (engine compartment)

    Ss Outlet section (engine compartment)

    Sc Body surface

    V~0 Upstream velocity vector

    V~ Local velocity vector

    Vx, Vy, Vz Velocity components

    x~; y~; z~ Vector unit system related to the model

    S Transversal section immediately downstream

    of the bluff body

    Cp Static pressure coefficient

    q Density

    a Angle between the rear window and the

    upstream flow direction

    b Skew angle

    k Orientation angle of the splitter plate related to

    the body

    h Angle between the splitter plate and the

    velocity VoCd Aerodynamic drag coefficient

    Cdref Reference aerodynamic drag coefficient

    P. Gillieron

    Research Division, Fluid Mechanics & Aerodynamics,

    Renault Group, 1, avenue du Golf (TCR AVA 058),

    78288 Guyancourt, France

    e-mail: [email protected]

    A. Kourta (&)

    Institut PRISME, ESA, PolytechOrleans, 8 rue Leonard de

    Vinci, 45072 Orleans Cedex 2, France

    e-mail: [email protected]

    123

    Exp Fluids (2010) 48:116

    DOI 10.1007/s00348-009-0705-7

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    2/16

    1 Introduction

    The growing request on fossil energy transforms today the

    world energetic system becoming incompatible with the

    available resources and the necessity to reduce gas emis-

    sion with greenhouse effects. As an example, the number

    of Chinese road vehicles in 20 years from now will be

    approximately 270 millions30 times more than that in2002 (Passenger Cars 2006). East Europe and Africa

    follow the same procedure at more or less equivalent

    proportion and at more or less long term. In this context,

    the gas with greenhouse effects emission will increase of

    about 57% in 2030 with environmental and climate strong

    repercussion (IEA 2007). Solutions have hence to be

    searched in different physical domains, by and for transport

    industry, to reduce significantly the CO2 emission. In this

    situation and independently of the energy used (fossil

    combustible, electric, hydrogen, etc.), the aerodynamic

    flow control for the road vehicles becomes necessary to

    optimise the loaded energy. Scientific researchers work toreduce the vehicle drag of about 30% and to reduce the

    CO2 emission of about 12 (New European Driving Cycle as

    NEDC) to 24 g/km (customer real cycle). All solutions and

    results obtained previously and currently need to be actu-

    alised, adapted and improved in the perspective use for the

    road vehicle.

    Amongst the solutions retained in previous studies

    (Gad-el-Hak 1996), the use of splitter plates can be a good

    and an interesting solution because it constitutes a simple

    device without any electronic (no sophisticated actuator).

    The use of these splitter plates has been performed before

    without testing their effects on specific shape related to

    automotive vehicle (front right side for a bus, inclined for

    monospace (Fastback) type, rear right side for utility

    vehicle (square back), inclined rear window for estate car,

    and so on) for Reynolds number higher than 106. Their use

    for real car depends on the results on simplified car

    geometry in the presence of lateral wind. This study is

    related to this subject and hence concerned by the evalu-

    ation of splitter plate effect on simplified car geometry with

    or without side wind.

    Roshko and Koenig (1978) demonstrated that it is

    possible to reduce by 97% the drag on a cylinder with

    its axis parallel to the incident flow direction V0 using

    circular discs placed perpendicularly upstream to the

    velocity direction of V0. The result is obtained for a

    Reynolds number based on the cylinder diameter equal

    to 5 9 105. Mair (1965) analysed the effect of splitter

    discs set downstream of the bases and perpendicular to

    the incident flow. Experiments performed on a torpedo-

    type obstacle equipped with a splitter disc downstream

    of the base for a Reynolds number equal to 6 9 105

    demonstrated that aerodynamic drag may be reduced by

    35%. A second splitter disc placed downstream of the

    first disc achieved drag reductions of nearly 55%.

    However, these geometries remain very far from the

    road vehicle geometries and, the ground effects have never

    been considered. Significant results have to be obtained on

    representative and simplified geometry representing better

    the automotives to convince the industry for the splitter

    plates use. The work proposed here is conducted on theAhmed body, geometry commonly used as one represen-

    tative of automotive aerodynamics. Some studies have

    been conducted on this geometry with longitudinal splitter

    plates (Baudoin and Aider 2008) and vertical splitter plates

    (Levallois and Gillieron 2005).

    This work completes previous studies and aims to

    characterise the influence of splitter plates on aerodynamic

    drag with simplified geometry hatchback and MPV- or

    utility-type vehicles. Swept angle effect on the efficiency

    of the splitter plates placed front or behind the model (with

    inclined front and square base) is analyzed. The experi-

    ments were performed in a wind tunnel, around Ahmedbody, varying the position, orientation and dimensions of

    the splitter plates. The splitter plate, with different skew

    angle, placed at front or rear of the geometry is also

    examined.

    2 Theoretical bases

    Aerodynamic drag is defined as an integral over the

    surface vehicle of the static pressure, friction and tur-

    bulence stresses. It can be also obtained by a simplified

    analytical model based on the momentum equation

    applied to the air inside a stream tube enclosing the

    vehicle (see Fig. 1).

    From the pressure, viscosity and turbulent stress distri-

    bution over the vehicles boundary surface Sc (Fig. 1), the

    aerodynamic drag is given by

    Fx

    ZSc

    PIn~dr

    ZSc

    sl stn~dr

    264

    375x~ 1

    Fig. 1 Integral momentum balance

    2 Exp Fluids (2010) 48:116

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    3/16

    where sl;, st and PI represent the viscous, turbulent and

    pressure stress tensors, respectively. The vector n~ is a unit

    vector moving towards the outside of the fluid domain, and

    x~ is a unit vector with the same direction, collinear to the

    far field velocity V0!

    .

    Viscous and turbulent stresses are connected to the

    formation and development of boundary layers on the

    vehicle surface. The aerodynamic pressure forces dependon the vehicle geometry and also on a distribution and

    evolution of the pressure related to the wake vortices. The

    aerodynamic drag given by Eq. 1, and averaged over a

    duration Dt, was measured in a wind tunnel using an

    aerodynamic balance. For an estate car, the aerodynamic

    pressure and friction contributions represented, respec-

    tively, 90 and 10% of the total aerodynamic drag (classical

    result).

    The aerodynamic pressure forces deduced from Eq. 1

    and expressed as a function of the static pressure coeffi-

    cient Cp is defined by

    Cp P P0q2

    V202

    where P0 is the static pressure, V0 is the upstream farfield

    velocity and q is the density of air, and the aerodynamic

    drag is given by

    Fx q

    2V20

    ZSc

    Cpn~x~dr 3

    With a rounded front face-like Rankines half-oval without

    flow separation, aerodynamic drag is directly controlled by

    a static pressure distribution on the base. In general, on an

    automotive vehicle, the static pressure coefficient Cp on the

    base is between -0.05 and -0.20. Controlling the flow

    therefore consists of obtaining a zero static pressure coef-

    ficient on the base.

    Aerodynamic drag can also be analysed using pressure,

    viscous and turbulent information distributed over the

    boundary surface R (Fig. 1), and contained essentially in

    the wake (classical momentum equation, see Eq. 1). In this

    case, its expression is given by

    Fx ZR

    PIn~dr ZR

    sl stn~dr24 35x~

    ZR

    qV~x~V~n~dr 4

    where V!

    is the local velocity vector.Onorato et al. (1984)

    proposed a simplified analytical model based on the

    momentum equation applied to the flow inside a stream

    tube enclosing the vehicle (see Fig. 1). The mean flow is

    assumed to be steady and incompressible, and gravity and

    turbulence effects are considered negligible compared to

    the pressure effect (Cousteix 1989). These simplifications

    lead to the Onorato expression of the aerodynamic drag,

    given in Eq. 5.

    Fx qV20

    2

    ZS

    1 Vx

    V0

    !2dr

    qV20

    2

    ZS

    Vy

    V0

    2

    Vz

    V0

    2" #dr

    ZS

    Pi0 Pidr 5

    where Pi0 is the reference total pressure, V0 is the external

    flow velocity, q is the density, Pi is the total pressure and

    Vx, Vy, Vz are the components of the velocity vector. The

    expression (5) is then used to define the aerodynamic drag

    of a motor vehicle according to velocity and total pressure

    fields measured in the wake cross section S, downstreamfrom the base (Ardonceau and Amani 1992).

    The first term of the expression (5) represents the drag

    associated with the longitudinal velocity deficit measured

    inside the near-wake zone. Far downstream, in the wake,

    the longitudinal velocity component Vx becomes approxi-

    mately equal to Vo and hence this term is equal to zero.

    This term is related to the development of transversal

    vortices at the base (Onorato et al. 1984). The second term

    corresponds to the vortex drag, associated with the devel-

    opment of longitudinal vortices on the geometry. Finally,

    the third term expresses the drag induced by the total

    pressure loss between the upstream and the downstream of

    the motor vehicle, associated with the formation and

    maintenance of separated swirling structures in the wake.

    According to the Onorato expression (5), the aerody-

    namic drag of a motor vehicle is mainly due to the for-

    mation of separated flow on the geometry, and on the

    formation of transversal and longitudinal swirling struc-

    tures in the wake. Therefore, the drag reduction can be

    obtained by reducing, or even eliminating, the longitudinal

    vortices (second term, 20% of a total drag), by reducing the

    wake cross section Sor by limiting the total pressure loss in

    the wake (third term, 80% of a total drag). The separation

    locations depend on the local wall curvature, the longitu-

    dinal static pressure gradient (Cousteix 1989), the inflow

    turbulence intensity (Arnal et al. 1976) or the roughness

    (Granville 1985).

    3 Experiment conditions

    The experiments were performed in the wind tunnel at the

    Paris ENSAM [Ecole Nationale Superieure dArts &

    Exp Fluids (2010) 48:116 3

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    4/16

    Metiers] Aerodynamics Laboratory. This Prandtl-type

    closed wind tunnel has a working section of 2.00 m long,

    1.65 m wide and 1.35 m in height. It is an open work

    section wind tunnel (Fig. 2). The maximum turbulence

    level of this wind tunnel is less than 1%, and the maximum

    inflow velocity Vo may reach 40 m/s. This wind tunnel is

    equipped with a three-component aerodynamic balance

    allowing the measurement of the drag, the side force and

    the yawing moment. The experimental model is placed on

    a circular cylinder raised from the wind tunnel floor

    (Fig. 3). This element allows to reduce the boundary layer

    thickness up to 80% at the wind tunnel roof and to model

    the lateral wind (side wind).

    The experiments were performed on a simplified auto-

    motive geometry (Fig. 4), namely, Ahmed body at scale

    0.75 (Ahmed et al. 1984; Gillieron and Chometon 1999).

    The Ahmed body lengths LA and lA, width wA and height

    HA used are, respectively, equal to 783, 158, 216 and

    292 9 10-3 m. In these conditions, the blockage ratio is

    less than 3%. The model is positioned on the aerodynamic

    balance with the help of three circular cylinders (diameter

    2 9 10-3 m), and the distance between the model and the

    plateau representing the road is 7 9 10-3 m. For this

    geometry, the angle a represents the rear window inclina-

    tion relatively to the horizontal plane. Tests were per-

    formed at two angles a = 0 and 25 (Fig. 4). With the

    angle a equal to 0, the flow is separated at the periphery of

    the base to form a tore vortex (Gillieron and Chometon

    1999; Spohn and Gillieron 2002). The Ahmed body is also

    representative of a square-back vehicle. When the angle a

    equals 25, the separation at the end of the roof interacts

    with the two contra-rotating longitudinal vortices issued

    from both sides of the rear window. The separation on the

    rear window reaches the wake. Inside this separation,

    Fig. 2 Sketch of wind tunnel

    Fig. 3 Circular cylinder to control roof boundary layer thickness

    (indicated by an arrow)

    Fig. 4 Geometry, scale 0.75 compared to the Ahmed body reference

    (Ahmed et al. 1984): the line A represents the end of the roof and the

    start of the rear window

    4 Exp Fluids (2010) 48:116

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    5/16

    appear two flow rotations centred on two focuses which

    contribute to the base separation. Detailed description of

    these phenomena was presented by Spohn and Gillieron

    (2002). The Ahmed body can be also likened to a simpli-

    fied vehicle with tailgate-type rear section.

    The splitter plates are attached to the experimental

    model with circular cylinder having 4 9 10-1 m length

    and 10-2 m diameter. To fix correctly these connectionsand stop translation and rotation motions, sharp pressure

    screws are used. The error on the measurement of the

    distance between the splitter plate and the Ahmed body is

    less than 10-3 m.

    4 Results

    Experiments are performed with splitter plates placed at the

    rear of the base and in the front of the Ahmed body.

    Analysis is performed only by comparing the drag forces.

    In the case where the splitter plates are placed behind thebase, analysis is performed only for the rear window angle

    equal to 0 (a = 0). When the splitter plates are positioned

    in front of the model, three different shapes of the model

    front are examined by varying or not the model orientation

    (rotation of 180). The skew angle effect on the standard

    Ahmed body with square back and by using splitter plates

    in the front and in the back of the experimental model is

    then examined.

    For all these configurations, the goal consists on

    reducing the surface of the wake that contributes to the

    drag force (see Eq. 5). To limit the side force on the splitter

    plates, the working velocity does not exceed 30 m s-1.

    Taking into account the flow velocity (20 and 30 m s-1)

    and the reduced size of the used model, the aerodynamic

    coefficient cannot be directly compared to the Ahmed

    study performed at 60 m s-1 and using one model at size 1.

    In this study, for the square-back geometry without adding

    a splitter plate, the aerodynamic drag coefficient is equal to

    0.305 at 30 m s-1. For the configuration with a rear win-

    dow inclined at 25 (used in the front side), it is equal to

    0.448. In the following, only the percentages of the devi-

    ation from the aerodynamic coefficients obtained on the

    reference geometry are specified.

    The height and width of splitter plates were between

    0.6 and 0.9 times the height and width of the Ahmed

    body. The analysis was performed by comparing the

    aerodynamic drag coefficient values measured with and

    without a splitter plate. If Cd and Cdref are the aerody-

    namic drag coefficients measured, respectively, with and

    without a splitter plate, the relative drag reduction

    100[Cdref - Cd/Cdref] were plotted and analysed below for

    various skew angles b and various splitter plate orienta-

    tions k, as defined in Fig. 5.

    4.1 Vertical splitter plates effects

    The experiments were performed with splitter plates posi-

    tioned downstream and upstream of the Ahmed body, and

    with a zero skew angle (b = 0) (see Fig. 6). The analysis

    is performed by varying the distance between the base andsplitter plate (Mair 1965). The Reynolds number used in all

    the experiments is based on the model length L.

    4.1.1 Downstream vertical splitter plates

    The vertical splitter plates were positioned on a square-

    back-type model (a = 0). The origin of the coordinates in

    this case is in the plane of the base, and the position of the

    plate in the x direction is kept non-dimensional by dividing

    x by the base height HA. The results plotted in Fig. 7 are

    obtained for three different plate sizes and for an upstream

    velocity V0 = 30 m s-1. With each splitter plate size, the

    results show that the drag reduction increases with the

    distance from the base x/HA, reaches a minimum and then

    decreases. It can be observed that the maximum aerody-

    namic drag reduction, close to 12%, is obtained with

    a splitter plate measuring 0.9HA 9 0.9wA, placed at

    x/HA = 0.5. An enhancement of the drag is observed for

    the splitter plate 0.6 when the non-dimensional distance

    x/HA becomes higher than 1.3. This result suggests an

    amplification of the shear layer instabilities (issued from

    V0

    Skew angle

    x/H

    < 0

    Reduced abscissa

    Plate orientation

    angle

    +-

    Fig. 5 Definitions of skew angle b and orientation angle k

    Fig. 6 Vertical splitter plate downstream the rear of the square-back

    Ahmed body (a = 0)

    Exp Fluids (2010) 48:116 5

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    6/16

    the separation at the end of the roof) that increases the

    pressure force on the front side of the splitter plate. This

    phenomenon is like the one observed in a deep cavity. For

    the cavity flow, when the aspect ratio (width over deep) is

    small, the upstream-separated shear layer from the corner

    does not interact with the downstream corner and hence

    does not affect the farfield flow. When this aspect ratio

    increases, the instability of the shear layer amplifies and

    interacts with the downstream vertical wall of the cavity

    inducing retroactive pressure wave. These pressure oscil-

    lations amplify the shear layer instabilities, increase the

    parietal pressure distribution along the downstream cavity

    vertical wall (Rossiter 1964; Kourta and Vitale 2008) and

    contribute to increase the splitter plate drag. So the drag of

    the global geometry (body ? plate) increases. The results

    show clearly that the efficiency of the plate increases when

    the relative distance x/HA is reduced. Considering h the

    angle between the upper part of the plate and the rear plane

    of the model, the higher drag reduction is obtained for

    h = 5.7, 7.1 and 11.3 (Fig. 8). These angles are lower

    than critical angle a = 12 corresponding to the appear-

    ance of rear window separation (Ahmed et al. 1984).

    Figures 9 and 10 show, respectively, the vertical and the

    horizontal planes of the flowfield for different longitudinal

    plate positions. The formation and the evolution of two

    vortical tore structures between the base of the model and

    the splitter plate are clearly observed in the vertical planes

    (Levallois and Gillieron 2005). When the longitudinal

    position x/HA increases, the upper vortex centre moves

    downstream, near to the splitter plate. At the same time, the

    lower vortex moves from the model base to the bottom side

    towards the wind tunnel wall. When the maximum drag

    reduction is reached, the upper vortex centre is near the

    cavity centre. Moving the splitter plate downstream, this

    vortex centre moves towards the splitter plate and at the

    same time the aerodynamic drag increases. In all cases, the

    drag reduction is correlated with the transversal wake

    section diminution (the surface S in Eq. 5).

    From the horizontal planes (Fig. 10), for x/HA\ 0.6,

    between the base and the splitter plate, no vortex appears.

    For x/HA higher or equal to 0.6, two vortices are observed.

    When x/HA increases these vortices move from the splitter

    plate towards the base and the drag increases. For

    x/HA = 0.7, the vortex centres are located at the mid dis-

    tance between the base and the splitter plate. Behind the

    splitter plate, the wake exhibits the classical tore structure.

    The Reynolds number effect was also analysed at two

    inflow velocities, V0 = 20 m s-1 and V0 = 30 m s

    -1,

    corresponding, respectively, to a Reynolds numbers of

    1.0 9 106 and 1.6 9 106. The obtained results are plotted

    in Fig. 11. It can be observed that the Reynolds number has

    small effect on the drag reduction obtained by using ver-

    tical splitter plates. The value and the maximum of drag

    reduction position are not affected by the Reynolds num-

    ber. This result confirms the weak influence of the Rey-

    nolds number on the aerodynamic drag beyond 25 m s-1.

    The influence of a second splitter plate positioned

    downstream of the first one measuring 0.9HA 9 0.9wA at

    x/HA = 0.5 was also analysed. This configuration allows to

    reduce the surface S of Eq. 5 and to decrease the aerody-

    namic drag for the same wake topology. The fact that this

    result is not obtained suggests the existence of cavity

    instability producing additional aerodynamic drag due to

    the front side of the last or of the both splitter plates.

    -1%

    1%

    3%

    5%

    7%

    9%

    11%

    0,0 0,3 0,5 0,8 1,0 1,3 1,5

    Reduced Abscissa x/HA

    DragReduction%

    splitter plate 0.6

    splitter plate 0.8

    splitter plate 0.9

    Fig. 7 Aerodynamic drag coefficient reduction versus x/HA with

    splitter plates of 0.9HA 9 0.9wA, 0.8HA 9 0.8wA and 0.6HA 90.6wA (V0 = 30 m s

    -1)

    x

    y

    z

    Fig. 8 Vertical splitter plate downstream the rear of the square-back

    Ahmed body (a = 0)

    6 Exp Fluids (2010) 48:116

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    7/16

    Fig. 9 Mean velocity

    distribution and streamlines for

    different longitudinal splitter

    plate positions: a x/HA = 0.4.

    b x/HA = 0.5. c x/HA = 0.6.

    d x/HA = 0.7 (splitter plates of

    0.9HA 9 0.9wA, vertical plane

    at y = -27.5 mm)

    Exp Fluids (2010) 48:116 7

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    8/16

    Fig. 10 Mean velocity

    distribution and streamlines for

    different longitudinal splitter

    plate positions: a x/HA = 0.4.

    b x/HA = 0.5. c x/HA = 0.6.

    d x/HA = 0.7 (splitter plates of

    0.9HA 9 0.9wA, horizontal

    plane at z = 27.5 mm)

    8 Exp Fluids (2010) 48:116

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    9/16

    4.1.2 Upstream vertical splitter plates

    Significant drag reductions could also be achieved by

    positioning vertical splitter plates upstream of the model

    (Roshko and Koenig 1978). The experiments were per-

    formed with various front section forms and various splitter

    plate sizes and positions. In each test, the reference height h

    and width w were related to the front section geometry (see

    Fig. 12).

    The first test series were performed on the Ahmed body,

    for which the height h and width w are defined in Fig. 12,

    configuration (a) where h = HA and w = wA. When the

    rear window inclination a is equal to zero (see Fig. 4), the

    geometry is a square back. When this angle equals 25, it

    corresponds to a rear window for a simplified vehicle with

    tailgate rear window. The second and third test series were

    obtained by swinging round the Ahmed body by 180 on itsown yaw axis (vertical). For the second configuration, the

    rear window angle of inclination from horizontal was 25

    (Fig. 12, configuration (b)). The reference height h and

    width w can therefore be likened to the height and width of

    the straight base underneath the rear window. This con-

    figuration is representative of an MPV (Espace) or utility

    vehicle front (h = HA and w = wA). For the third test

    series, the base was still facing the wind, but with a rear

    window angle of inclination of zero (a = 0, Figs. 4 and

    12, configuration (c)). This configuration corresponds to a

    front part of truck or bus. In this case, the reference height

    h and width w can be likened to the Ahmed body height HAand width wA.

    For these three configurations, the experiments were

    performed using two splitter plates with a height and width

    values of 0.8 and 0.9 times the reference height h and width

    w. The origin of the abscissas in this case belonged to the

    splitter plate plane. As above, the effect of each splitter

    plate is analysed when the reduced distance x/h between

    the plate and the front of the model increased. The aero-

    dynamic drag coefficient reduction percentages obtained

    with these three configurations are plotted in Figs. 13, 14

    and 15, respectively.

    0%

    2%

    4%

    6%

    8%

    10%

    12%

    0,00 0,25 0,50 0,75 1,00 1,25 1,50

    Reduced Abscissa x/HA

    DragReduc

    tion%

    splitter plate 0.6 - 20m/s

    splitter plate 0.8 - 20m/s

    splitter plate 0.9 - 20m/s

    splitter plate 0.6 - 30m/s

    splitter plate 0.8 - 30m/s

    splitter plate 0.9 - 30m/s

    Fig. 11 Reynolds number effect on aerodynamic drag coefficient

    reduction: Inflow velocity of 20 and 30 m s

    -1

    , Re = 1.044 9 10

    6

    and 1.566 9 106

    Fig. 12 The various

    configurations and definitions

    of the Ahmed body (height h

    and width w)

    Exp Fluids (2010) 48:116 9

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    10/16

    With vertical splitter plates measuring 0.9h 9 0.9w and

    0.8h 9 0.8w positioned upstream of the rounded part of the

    Ahmed body (Fig. 12, configuration (a)), the experimental

    results show that aerodynamic drag decreases, reaches a

    minimum and then increases rapidly as the reduced dis-

    tance x/h increases (Fig. 13). The results obtained with a

    farfield velocity of 20 and 30 m s-1 demonstrate that the

    drag reduction decreases as the Reynolds number increa-

    ses. At these two velocities, the maximum drag reduction

    values observed in the vicinity of x/h = 0.15 are 2.00 and

    0.15%, respectively. These reductions are insignificant, but

    the influence of the relative position of the splitter plate

    appears to be important.

    The experiments performed on MPV- and/or utility-type

    fronts (front end inclined 25/horizontal, Fig. 12 configu-

    ration (b)) are more interesting. With splitter plates mea-

    suring 0.8h 9 0.8wA, drag reductions of nearly 28% were

    obtained with the reduced position x/H equal to 0.3

    (Fig. 14). Drag reduction remains greater than 25% over

    the reduced position interval [0.30.6]. The Reynolds

    number has very little influence on the results. Also,

    through a geometry adaptation designed to recover the flow

    for engine cooling, this type of solution could be a mean of

    progress for reducing automotive drag.

    With a straight front end (a = 0, Figs. 4, 12 configura-

    tion (c)), drag reduction is particularly significant (Fig. 15).

    This reduction reaches 45% with a vertical splitter plate

    measuring 0.8HA 9 0.8wA placed at the reduced position

    x/HA = 0.3. It remains greater than 40% as x/HA increases

    from 0.3 to 1.0. The sensitivity to the x/HA relative position

    in this case appears to be less significant than above. By

    enabling a reduction in the surface area of the transverse

    section separated from the upstream end of the model, the

    results obtained in this case confirm the advantage of

    rounding vehicle front end connection surfaces (transversal

    wake surface reduction).

    -8%

    -6%

    -4%

    -2%

    0%

    2%

    0,0 0,2 0,4 0,6

    Reduced Abscissa x/h

    DragReduction%

    splitter plate 0.8 - 20m/s

    splitter plate 0.9 - 20m/s

    splitter plate 0.8 - 30m/s

    splitter plate 0.9 - 30m/s

    Fig. 13 Percentage of aerodynamic drag coefficient reduction with

    splitter plates measuring 0.9h9

    0.9w and 0.8h9

    0.8w as a functionof relative x/h positions. Splitter plates upstream of Ahmed body

    (Fig. 12, configuration (a))

    0%

    5%

    10%

    15%

    20%

    25%

    30%

    0,00 0,25 0,50 0,75 1,00 1,25 1,50

    Reduced Abscissa x/h

    DragRe

    duction%

    splitter plate 0.6 - 20m/s

    splitter plate 0.8 - 20m/s

    splitter plate 0.9 - 20m/s

    splitter plate 0.6 - 30m/s

    splitter plate 0.8 - 30m/s

    splitter plate 0.9 - 30m/s

    Fig. 14 Percentage of aerodynamic drag coefficient reduction

    with splitter plates measuring 0.9h 9 0.9wA, 0.8h 9 0.8wA and 0.6h

    90.6wA upstream, as a function of reduced abscissas x/h. Upper frontpart inclined 25, estate or MPV configuration (Fig. 12, configuration

    (b))

    0%

    10%

    20%

    30%

    40%

    50%

    0,00 0,50 1,00 1,50

    Reduced Abscissa x/HA

    DragReduc

    tion%

    splitter plate 0.6 - 20m/s

    splitter plate 0.8 - 20m/s

    splitter plate 0.9 - 20m/s

    splitter plate 0.6 - 30m/s

    splitter plate 0.8 - 30m/s

    splitter plate 0.9 - 30m/s

    Fig. 15 Percentage of aerodynamic drag coefficient reduction withsplitter plates measuring 0.9HA 9 0.9wA and 0.8HA 9 0.8wA as a

    function of reduced position x/HA. Straight front geometry (Fig. 12,

    configuration (c))

    10 Exp Fluids (2010) 48:116

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    11/16

    For the configuration (b) with front splitter plate, PIV

    measurement can provide explanation of the drag reduc-

    tion. Figures 16 and 17 show the vertical and horizontal

    planes, respectively, for three different plate positions. In

    Fig. 16, for the vertical plane, the separation zone is at the

    beginning near to the body, and when the plate moves

    upstream, this separation zone grows and moves far from

    the body so it does not influence the aerodynamic

    Fig. 16 Mean velocity distribution and streamlines (configuration (a)) for different longitudinal front splitter plate positions:a x/h = 0.24.

    b x/h = 0.4. c x/h = 0.56 (splitter plate of 0.9h 9 0.9wA, vertical plane at y = 26.8 mm)

    Exp Fluids (2010) 48:116 11

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    12/16

    Fig. 17 Mean velocity distribution and streamlines (configuration (b)) for different longitudinal front splitter plate positions:a x/h = 0.24.

    b x/h = 0.4. c x/h = 0.56 (horizontal plane at z = 26 mm)

    12 Exp Fluids (2010) 48:116

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    13/16

    performance. The bubble is somewhere between the splitter

    plate and the front body. The flow is no more separated on

    the body. For the horizontal plane, the flow configuration

    near the front body does not change when the splitter plate

    moves upstream. The velocity flowfields show that the

    local evolution at the bottom (Fig. 16a, c) and lateral sides

    (Fig. 17a, c) of the body changes the upstream velocity

    direction. The transversal size decreases with the increaseof the relative distance x/h. Hence, the drag related to the

    transversal wake surface S (Eq. 5) decreases.

    4.2 Splitter plates with non-zero skew angle

    The influences of skew angle b and the effects of orien-

    tation angle k on aerodynamic drag evolution were ana-

    lysed using splitter plates applied on type (b) and (c)

    geometries; see Figs. 12, 18 and 19. All the results,

    expressed as percentages, were determined relatively to

    values measured without a splitter plate for the same skew

    angles b.

    4.2.1 Downstream vertical splitter plates

    The experiments were performed on a straight base model

    (configuration (a) representative of the rear of a Renault

    Espace-type vehicle, with a = 0), fitted with a vertical

    splitter plate, with height and width values of 0.90HA and

    0.86wA, respectively. The aerodynamic drag measured

    with a splitter plate was always less than its value observed

    without a splitter plate, but the splitter plates influence

    decreases as the skew angle b increases (Fig. 20). Drag

    reductions greater than those obtained with splitter plates

    positioned parallel to the base demonstrate the advantage

    of adapting the splitter plate orientation k (orientation/

    vertical) to the skew angle b. Finally, the changes observed

    as a function of the orientation angle with a skew angle

    b = 5 suggest the existence of strong interactions between

    the splitter plate and the base.

    The influence of longitudinal spacing between the ver-

    tical splitter plate and the base was analysed with a skew

    angle b = -15, varying the reduced position x/HA at

    various orientation angle k values (Figs. 21, 22). With

    orientations angle k greater than -15, the aerodynamic

    drag observed with a splitter plate at skew angle b = -15

    was always less than its value observed without a splitter

    plate, and the optimum aerodynamic drag reduction posi-

    tion was poorly influenced by the orientation angle k on the

    angular domain [-20, ?20]. This optimum position is

    closer to the base (from x/HA = 0.3 to x/HA = 0.4) at non-

    zero, when positive value of the orientation angle k is

    increased (leeward part nearer the base than the windward

    part), and is further from the base when its negative absolute

    value is increased (from x/HA = 0.4 to x/HA = 0.5). At

    skew angle b = -15, the maximum drag reduction

    (10.6%) is observed at x/HA = 0.4 with orientation angle of

    k = 5 (Figs. 21, 22).

    In addition, at all x/HA positions less than 0.4 (maximum

    drag reduction with k = 0, Fig. 21), the drag observed at

    positive orientation angle k values was less than its value

    observed with vertical splitter plate (k = 0). The maxi-

    mum drag reduction was therefore obtained by moving the

    windward part (the leeward part respectively) of the splitter

    plate away from (respectively closer to) the base. All these

    V0

    Skew angle

    x/H = 0.6

    < 0

    Reduced abscissa

    Plate orientation

    angle

    +-

    Fig. 18 Skew angle b and orientation angle k of downstream splitter

    plate, configuration (a) with a = 0 (Fig. 2)

    V0

    Skew

    anglex/h = 0.4

    < 0

    Reduced abscissa

    -

    +

    Plate orientation

    angle

    Fig. 19 Skew angle b and orientation angle k of upstream splitter

    plate, type (b) geometry

    0%

    2%

    4%

    6%

    8%

    10%

    12%

    14%

    -20 -10 0 10 20

    Orientation angle ()

    Dragre

    duction(%)

    -20 skew -15 skew -10 skew

    -5 skew 0 skew

    Fig. 20 Rear splitter plate 0.90HA 9 0.86wA with x/HA = 0.6: Per-

    centages of aerodynamic drag coefficient reduction as a function of

    orientation angle k with various skew anglesb

    Exp Fluids (2010) 48:116 13

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    14/16

    results demonstrate the advantage of adapting splitter plate

    orientation to external conditions.

    4.2.2 Upstream vertical splitter plates

    The vertical splitter plate was positioned on the front of a

    type (b) model, front configuration of a RenaultEspace

    (Square-back)-type vehicle (Fig. 19). Its height and width

    values were 0.8 times the height h and width w = wA defined

    in Fig. 12. Theset-up geometry restricted thex positioning of

    the splitter plate, and the lowest possible value for the x/h

    ratio increased with the orientation angle a. The measure-

    ments were performed at a skew angle b = -15.

    At a skew angle b = -15, and beyond a maximum

    positive orientation angle k, the value of which increases

    with reduced longitudinal spacing x/h, the aerodynamic

    drag rapidly decreases, and its value may fall below its

    value observed without a splitter plate (Fig. 23). In this

    configuration, unlike the results observed with the down-

    stream splitter plate, the maximum drag reduction was

    obtained by moving the leeward part (the windward part

    respectively) of the splitter plate away from (respectively

    closer to) the vehicle, see Figs. 19 and 23. Drag reductions

    of approximately 7.1% were obtained with the orientation

    angles k = 15 and 20.

    At a skew angle b = -15 and zero orientation

    (k = 0), the splitter plate always has an adverse effect on

    the aerodynamic drag coefficient value (Figs. 23, 24),

    orientation curve k = 0. Analysis of the results demon-

    strated that the reductions are greater when the splitter

    plate was placed as close as possible to the model (Fig. 24).

    At a given reduced position less than 0.35, the maximum

    reductions were observed at maximum positive orientation

    angle values (with negative b).

    5 Conclusion

    In the worldwide context strongly constrained by the

    climatic consequence of CO2 emission and the fossil

    -4%

    -2%

    0%

    2%

    4%

    6%

    8%

    10%

    12%

    0,2 0,3 0,4 0,5 0,6

    Reduced abscissa x/HA

    Dragreduction(%)

    -20 orientation

    -15 orientation

    -10 orientation

    -5 orientation

    0 orientation

    5 orientation

    10 orientation

    15 orientation

    20 orientation

    Fig. 21 Rear splitter plate 0.90HA 9 0.86wA: Percentage aerody-

    namic drag coefficient reduction as a function of reduced positionx/HA with various orientation angles k and skew angle b = -15

    -4%

    -2%

    0%

    2%

    4%

    6%

    8%

    10%

    12%

    -20 -10 0 10 20

    Orientation angle ()

    Dragred

    uction(%)

    x/HA = 0,2

    x/HA = 0,3

    x/HA = 0,4

    x/HA = 0,5

    x/HA = 0,6

    Fig. 22 Rear splitter plate 0.90HA 9 0.86wA: Percentage of aerody-

    namic drag coefficient reduction as a function of orientation anglek

    with various reduced abscissas x/HA, skew angle b = -15

    -8%

    -6%

    -4%

    -2%

    0%

    2%

    4%

    6%

    8%

    -20 -10 0 10 20

    Orientation angle ()

    Dragreduction(%)

    x/h = 0,23

    x/h = 0,3

    x/h = 0,4

    x/h = 0,5

    x/h = 0,6

    HA = 0,6

    Fig. 23 Front splitter plate 0.8 h 9 0.8wA: Percentage of aerody-

    namic drag coefficient reduction versus the orientation angle k atvarious reduced abscissas x/h, and skew angle b = -15

    14 Exp Fluids (2010) 48:116

    123

  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    15/16

    combustible rarefaction, automotive industry has to search

    for a new solution to increase the loading energy efficiency

    (fossil energies, hydrogen or electric). In this situation,

    many works on passive and active separation control have

    been initiated both in industry and academic research

    laboratories.

    The obtained results confirm the interest of using splitter

    plates to reduce the aerodynamic drag for the road vehicle.

    By choosing adapted positions and orientations with respect

    to the flow conditions, drag reduction can be obtained. By

    using the splitter plate alone or with another control solution

    (vortex generators or active control systems), these solu-

    tions allow to reduce for at least 10% the drag of the road

    vehicles. For vehicles with square back, the gas consump-

    tion will be reduced by 0.8 L for 100 km at stabilized

    vehicle velocity of 130 km h-1. If the added mass related to

    the splitter plates is compensated by the reduction of engine

    mass due to the engine size reduction (downsizing), at this

    velocity, the CO2 will be diminished by 20 g/km and by

    3.5 g/km on the NEDC (New European Driving Cycle)

    reference. The final goal is to integrate these solutions on

    road vehicle at the horizon of 2015. The solutions obtained

    on different models studied here show the complexity of

    physical phenomena and the necessity to continue the

    development of knowledge in this case. Noise, instabilities

    or at least interactions with vortex structures coming from

    front part of the model or from the rear window have to be

    carefully analysed to improve the control performances.

    With vertical splitter plates positioned downstream of a

    straight base, drag reductions of nearly 12% were obtained.

    The reductions increased with increasing splitter plate

    transverse dimensions and decreasing distance between the

    splitter plate and the base. The influence of the Reynolds

    number remained low, with apparently no influence on the

    physical phenomena, whereas drag reduction increased

    with increasing Reynolds numbers.

    The influence of vertical splitter plates positioned

    upstream was also analysed. Drag reductions of nearly 27

    and 45% were obtained, respectively, using splitter platespositioned upstream of an angled front face with or without

    an inclination from vertical. The Reynolds number influ-

    ence appears to be less significant than for splitter plates

    positioned on the rear of the models.

    For splitter plates mounted downstream, the optimum

    longitudinal position of the splitter plates was poorly

    influenced by the skew angle and the angular variations of

    the splitter plate from vertical. For orientation angle k less

    than -15, the aerodynamic drag observed with a splitter

    plate at a skew angle b = -15 was always less than its

    value observed without a splitter plate. The greatest

    reductions were obtained with optimum x/H positionobserved with zero skew angle by moving the windward

    part of the splitter plate away from the base. Drag reduc-

    tions of nearly 10.6% were observed. With splitter plates

    positioned upstream in parallel to the front face of a

    straight base-type model, the skew effect had an adverse

    effect on aerodynamic drag, and the results demonstrated

    the need to adapt the splitter plate orientation to the skew

    angle. Drag reductions of nearly 7% were observed, and

    the maximum drag reductions were obtained by moving the

    leeward part of the splitter plate away from the vehicle.

    Finally, the results presented herein confirm the advan-

    tage of splitter plates in reducing aerodynamic drag, and

    demonstrate the need to develop systems capable of

    adapting their position and angular orientation to external

    conditions.

    References

    Ahmed SR, Ramm R, Faltin G (1984) Some salient features of the

    time averaged ground vehicle wake. SAE technical Paper Series

    840300

    Ardonceau P, Amani G (1992) Remarks on the relation between liftinduced drag and vortex drag. Eur J Mech B 11(4):455460

    Arnald D, Cousteix J, Michel R (1976) Couche limite se developpant

    avec gradient de pression positif dans un ecoulement exterieur

    turbulent, La Recherche Aerospatiale, n1976-1

    Baudoin JF, Aider JL (2008) Drag and lift reduction of a 3D bluff

    body using flaps. Exp Fluids 44:491501. doi:10.1007/

    s00348-007-0392-1

    Cousteix J (1989) Aerodynamique, Turbulence et Couche Limite.

    Cepadues-Editions, Toulouse, pp 1114

    Gad-el-Hak M (1996) Compliant coatings: a decade of progress.

    In: Workshop on flow controlfundamentals and practices,

    Cargese, Corse, 1996

    -8%

    -6%

    -4%

    -2%

    0%

    2%

    4%

    6%

    8%

    0,2 0,3 0,4 0,5 0,6

    Reduced abscisse x/h

    Dragreduction(%)

    -20 orientation

    -15 orientation

    -10 orientation

    - 5 orientation

    0 orientation

    5 orientation

    10 orientation

    15 orientation

    20 orientation

    Fig. 24 Front splitter plate 0.8: Percentage of aerodynamic drag

    coefficient reduction versus x/h at various orientation angles a andskew angle b = -15

    Exp Fluids (2010) 48:116 15

    123

    http://dx.doi.org/10.1007/s00348-007-0392-1http://dx.doi.org/10.1007/s00348-007-0392-1http://dx.doi.org/10.1007/s00348-007-0392-1http://dx.doi.org/10.1007/s00348-007-0392-1
  • 7/28/2019 Art-1. Exp Fluids-2010, Vol.48 (Pages 116)

    16/16

    Gillieron P, Chometon F (1999) Modelling of stationary three

    dimensional detached airflows around an Ahmed reference body.

    In: Proceedings of the third international workshop on vortex,

    ESAIM, vol 7, pp 173182

    Granville PS (1985) Mixing length formulation for turbulent boundary

    layers over arbitrarily rough surfaces. J Ship Res 29(4):223233

    International Energy Agency (IEA) (2007) World energy outlook

    2007China and India insights (executive summary). Interna-

    tional Energy Agency, ISBN: 978-92-64-02730-5

    Kourta A, Vitale E (2008) Analysis and control of cavity flow. Phys

    Fluids 20:0771041 (10 pp)

    Levallois E, Gillieron P (2005) Controle des ecoulements en

    aerodynamique automobile & reduction de tranee par elements

    separateurs - Analyse par PIV, Colloque National de Visualisa-

    tion et de Traitement dImages en Mecanique des Fluides,

    Fluvisu11, Lyon, 69 juin

    Mair WA (1965) The effect of a rear mounted disc on the drag of a

    blunt based body of revolution. Aeronaut Q 16:350360

    Onorato M, Costelli A, Garonne A (1984) Drag measurement through

    wake analysis, SAE, SP-569. In: International congress &

    exposition, Detroit, MI, 27 February2 March 1984, pp 8593

    Passenger Cars, University of Sheffield and University of Michigan,

    Produced by SASI Group (Sheffield) and Mark Newmann

    (Michigan), Map031, 2006

    Roshko A, Koenig K (1978) Interaction effects on the drag of bluff

    bodies in tandem. In: Sovran G, Morel T, Mason WT (eds)

    Aerodynamic drag mechanisms of bluff bodies and road

    vehicles. Plenum, New York, pp 253286

    Rossiter JE (1964) Wind tunnel experiments on the flow over

    rectangular cavities at subsonic and transonic speeds. Aeronau-

    tical Research Council Reports and Memo n 3438

    Spohn A, Gillieron P (2002) Flow separations generated by a

    simplified geometry of an automotive vehicle. In: IUTAM

    symposium on unsteady separated flows, Toulouse, France, 812

    April 2002

    16 Exp Fluids (2010) 48:116

    123