You are on page 1of 10

Exp Fluids (2008) 45:537546

DOI 10.1007/s00348-008-0499-z

RESEARCH ARTICLE

Experimental investigation into wing span and angle-of-attack


effects on sub-scale race car wing/wheel interaction aerodynamics
S. Diasinos A. Gatto

Received: 29 October 2007 / Revised: 9 March 2008 / Accepted: 22 March 2008 / Published online: 13 April 2008
Springer-Verlag 2008

Abstract This paper details a quantitative 3D investigation using LDA into the interaction aerodynamics on a subscale open wheel race car inverted front wing and wheel.
Of primary importance to this study was the influence of
changing wing angle of attack and span on the resulting
near-field and far-field flow characteristics. Results
obtained showed that both variables do have a significant
influence on the resultant flow-field, particularly on wing
vortex and wheel wake development and propagation.

1 Introduction
Current open wheel race car performance depends heavily
on the effectiveness and efficiency of its aerodynamics.
Appendages such as inverted wings, diffusers, barge boards
and splitter plates are currently used to enhance performance, but the physics and mechanisms through which this
is achieved, particularly via various component interactions, remains poorly understood. From a design
perspective, the ultimate goal of the designer is to contribute to the downforce produced by a race car while
minimizing drag. This enhances the mechanical grip of the
tires on the track, allowing the vehicle to corner at speeds
which would otherwise be impossible. On modern open
S. Diasinos
Toyota F1, Koln, Germany
e-mail: s.diasinos@unsw.edu.au
A. Gatto (&)
Department of Mechanical Engineering,
School of Engineering and Design,
Brunel University, UB8 3PH Uxbridge, England
e-mail: alvin.gatto@brunel.ac.uk

wheeled race cars, approximately 30% of the total downforce produced originates from the front wing (Metz 1987;
Dominy 1992). However, the close proximity of the front
wheels to the front wing can have a significant influence on
this performance metric (Katz 2006; Agathangelou and
Gascoyne 1998).
Surprisingly, very little information relating to open
wheel/inverted wing interaction exists in the open source
literature. This is most probably due to the sensitive nature
of this information in the extremely competitive and lucrative businesses at the pinnacle of motor racing, particularly
Formula 1. There are however, an ever increasing number of
investigations appearing in the literature concerned specifically with investigating an inverted wing or wheel in
isolation (Zerihan and Zhang 2000; Katz 1986; McManus
and Zhang 2006). Two such investigations for the isolated
wing case were conducted recently by Zerihan and Zhang
(2000) and Ranzenbach and Barlow (1996). Independently,
they both found that a downforce loss phenomenon can
occur with a small, subsequent reduction beyond a critical
ride height. This was later found to be fundamentally caused
by the development of severe adverse pressure gradients on
the wing undersurface as the ride height was reduced. Zerihan and Zhang (2003) followed up this work with an
additional 3D experimental analysis, using LDA, to determine the effect of varying ride height and front wing angle
of attack (AOA) on the overall wake structure, together with
the formation and displacement of tip vortices. Results from
this analysis showed that as the wing approached the
ground, the region of separated flow and the size of the wake
increased, moving towards the ground plane as it travelled
downstream. With subsequent further reductions in ride
height, the vortex strength was found to increase up until a
critical ride height at which a loss of downforce, and consequently wing tip vortex strength, occurred.

123

538

As one of the first investigations to consider the realistic


flow conditions over an isolated rotating wheel, Fackell and
Harvey (1975) published both qualitative and quantitative
experimental results using a moving ground at a Reynolds
Number (based on wheel diameter) of 5.3 9 105. In this
work, it was demonstrated that spinning the wheel and
having it in contact with the ground is of vital importance
to obtain appropriate lift and drag values. In previous
studies this was not, necessarily, considered critical
(Stapleford and Carr 1970). Perhaps, the most striking of
the results presented by Fackrell (1974) was the detection
of surface pressure coefficients greater than two in the
upstream region of the contact patch (i.e the cross-sectional
area of the wheel in contact with the ground) due primarily
to the effects of viscous jetting (Fackrell 1974). This jetting
effect was found to be a bi-product of the interaction
between the oncoming airstream and the wheel/road juncture, producing an effective and significant pumping of
the air out into the freestream from either side of the wheel.
Considering the specific case of a front wing and wheel
operating in close proximity, Thisse (2004) conducted a
small number of computational investigations using Fluent on differing endplate designs with the front wheel
included. The analysis was compiled at fixed wing span,
chord, ride height, and AOA with results obtained indicating as much as a 36% reduction of wing downforce
with the wheel present over the isolated wing case. This
was thought to be due to lower wing surface exposure to
the high pressure region formed forward of the wheel
contact patch. Curiously, results also showed that the flow
separation position on the wheel occurred aft of the
position of maximum wheel height with the wing installed. Mortel (2003) and Cumming (2002) also presented
results for the wing/wheel interaction case with the former
investigating improved endplate designs and the later, the
effect of wing sweep. Unfortunately, neither offered any
great insight into the aerodynamic interaction of a wing
and wheel, but they did indicate that the inclusion of the
wheel made the flow over the wing significantly more
three-dimensional.

Exp Fluids (2008) 45:537546

position of the wing surface was also used as a reference


for ride height which was defined as the distance from this
position to the ground plane. For this investigation, ride
height was fixed at h = 0.13 c. The overall wheel diameter
and width were 1.17 and 0.631 c, respectively, with both
wheel edges incorporating shoulder radii of 0.067 c to
more accurately depict a real tire cross-section. The minimum distance between the endplate and the front of the
wheel was maintained at 0.087 c for all cases investigated,
with the wing span adjustable from O = 0100% overlap
of the wheel width. The range of AOA investigated was
from 0 to 12 with increments of 4. A picture of the setup
used is shown in Fig. 1.
2.2 Experimental setup
All experimental results were obtained using a 225 mm by
340 mm open circuit, closed test section, wind tunnel. A
moving ground was also built and installed in the wind
tunnel with a belt span of 200 mm (93% of the tunnel
width) and an overall length of 990 mm. A purpose built
duct, located upstream of the belt (Fig. 2a), was used to
remove the boundary layer formed over the tunnel floor to
accurately simulate moving model conditions. An example
of the velocity profile at a location -2 c upstream of the
model station, under operating conditions without the
model installed, is shown in Fig. 3. The nominal freestream
turbulence level in the wind tunnel was measured at 0.15%,
with all experiments conducted at a freestream velocity
(Vo) of 10 ms-1 0.05 ms-1. This gave a test Reynolds
number, based on wing chord, of 5.11 9 104.
The wing, wheel, and endplate used for all experimental
testing was manufactured from polished acrylic to ensure a
smooth, transparent surface (Fig. 1). The wheel was
mounted in position via a faired sting connected to the test
section wall and incorporated deep groove ball bearings to
allow free rotation from contact with the moving ground

2 Experimental setup and apparatus


2.1 Inverted wing and wheel geometry
For all experimental investigations, the wing comprised a
single element, constant chord (c = 0.075 m), NACA 4412
aerofoil profile. A rectangular endplate with a rounded
leading edge and a tapered trailing edge was fitted at the
wing tip. For all AOA conditions considered, the base of
the endplate was fixed at a position 0.04c below the lowest
vertical position of the wing surface. The lowest vertical

123

Fig. 1 Inverted wing and wheel setup

Exp Fluids (2008) 45:537546

539

Fig. 2 a Overall layout of test


rig, b LDA measurement planes

belt. The scale of the experimental wind tunnel model was


approximately 1:7.5.
The support system for the wing/endplate combination
allowed an AOA adjustment resolution of 0.05, with the
span of the wing, wheel track, ride height of the endplate,
and distance between the endplate and the wheel all positioned to better than 0.1 mm. Under experimental test
conditions, the belt of the moving ground, at no stage, was
observed to lift up and was operated at 10 ms-1 0.2 ms-1.
The maximum test section blockage for the wing/wheel
combination was calculated at 9.5%. At this level, it is
envisaged that blockage effects will cause a certain degree
of flow constrainment (i.e reduced vortex expansion and

artifical delay of flow separation) on the flowfield. It should


be noted, however, that the main focus of this work is
considered comparative, thereby reducing somewhat the
significance of blockage effects on the overall trends and
correlated results.
To obtain all experimental data, LDA surveys were
performed over four planes located, between, and aft of the
wing and wheel combination. As indicated in Fig. 2b, the
location of these planes were referenced to the wheel center
plane and staggered at x = -0.63 c, x = 0.75 c, x =
1.50 c, and x = 3.00 c, respectively. All LDA results were
taken in an area of interest 1.63 c high by 2.1 c wide referenced 2 mm from the left test section wall (Fig. 1) and

123

540

Fig. 3 Velocity profile -2 c upstream of wheel centreline; empty


test section, mid-span

1 mm off the belt surface. LDA measurements within these


planes, were taken at approximately 400 points within
these boundaries, with measurement points selected to be
more densely packed in close proximity to the wing and
wheel. For the plane located at x = -0.63 c, access of the
LDA beams was inhibited by the model, allowing therefore, only partial data collection at this measurement
station.
The 3D Dantec LDA system used to measure all
velocities comprised a 5 W ArIon laser and was configured to operate in coincident, backscatter mode. The LDA
probes were mounted on a computer controlled traverse
which was adjustable to within 0.01 mm over identical
translation ranges in the X, Y, and Z directions of
1,010 mm. Procedures outlined by Benedict and Gould
(1996) were used to estimate the 95% confidence interval
for the LDA results with the accuracy of non-dimensional
velocity magnitude found to be better than 0.03. A single,
Laskin type atomizer, using vegetable oil to generate
seeding particles, was incorporated into the test setup to
seed the flow and obtain all velocity data. At each measurement point analyzed, within each measurement plane,
the final value for the three components of velocity
obtained was acquired from averaging more than 2,000
instantaneous samples.

3 Results and discussion


3.1 Wheel without wing, comparison of stationary
and rotating wheel
Results obtained from the rotating and stationary isolated
wheel cases are presented in Fig. 4. As with all following
results, to aid in both qualitative and quantitative analyses, both constant non-dimensional streamwise velocity

123

Exp Fluids (2008) 45:537546

iso-contours (Vx/Vo) as well as the in-plane flow direction


and relative magnitude (arrows indicate Vyz) are presented.
On first inspection, the two main wheel vortex structures
found in previous investigations are clearly evident in all
but the foremost measurement plane (Fackrell 1974; Axon
1998). These vortices are setup in the recirculation zone
behind the wheel from flow entrained from the freestream
(McManus and Zhang 2006; Zhang et al. 2006). Clearly
visible also at the measurement plane X = -0.63, is a
stagnation region located at approximately mid-height for
both wheel configurations. Flow physics dictate that flow
stagnation also occur at the junction of the wheel/road
interface (i.e the contact patch), however the proximity of
the measurement plane to this region gives very little
indication of its existence. The high pressure region generated at this location is known to accelerate and eject the
flow cross-stream, where a complex interaction with the
freestream occurs around the wheel edge (McManus and
Zhang 2006).
Considering more closely the development of the wheel
vortices, from Fig. 4ch, the general trend is for the vortex
size and separation distance to increase with further
propagation downstream. The wheel vortex separation
distance is particularly evident when comparing Fig. 4c
(0.9 \ Y \ 1.55) to Fig. 4g (0.75 \ Y \ 1.9) and Fig. 4d
(0.85 \ Y \ 1.65) to Fig. 4h (0.5 \ Y \ 1.9). These
results show that through rotating the wheel, decreases in
wheel vortex separation distance by up to 20% (at X = 3)
can be achieved over the stataionry wheel configuration.
Also clear from comparing Fig. 4c, e to d, f is that the
general wheel vortex shape appears significantly flatter
in the non-rotating case than that observed for the rotating
configuration. This result has been inferred by past publications (Fackell and Harvey 1975; McManus and Zhang
2006) and is thought to result from the differing amounts of
the viscous jetting flow and the different flow separation
positions between the two configurations.
Further comparisons between the rotating and stationary
cases in Fig. 4 also show subtle differences in the complex
near wake structure directly behind the wheel (Fig. 4c, d).
First, the results indicate that the wheel wake for the rotating
case is thinner and marginally higher (0.95 \ Y \ 1.6,
Z = 1.5) than that found for the stationary configuration
(0.8 \ Y \ 1.65, Z = 1.4), indicating delayed separation
over the top of the wheel. This finding is further supported by
differences found in the degree of flow entrainment down the
centre of the rear of the wheel at this measurement location.
From direct comparison of the in-plane velocity results
between the two cases, velocity magnitudes of up to 10%
greater where found for the stationary case. Supplimental to
this result is that this increased degree of flow entraiment is
seen to extend closer to the ground plane for this case
(indicated in Fig. 4d). In previous work, (Fackell and Harvey

Exp Fluids (2008) 45:537546

541

Fig. 4 Comparison of
stationary and rotating isolated
wheel results

1975; McManus and Zhang 2006), this increased entrainment for the stationary case was found to be due to the
decelerating flow on the rear surface of the wheel being
energized by the entrained main streamwise flow around the
sides. This promotes the central region of attached flow as it

is directed down the rear face of the wheel and toward the
ground. Additionally, this characteristic of flowfield, and
the subsequent bi-furcation of the flow as proximity to the
ground increases, clearly implies the existence of a downstream stagnation line on the ground plane (Fig. 4f).

123

542

3.2 Wheel and wing, wing span 100%


From Figs. 5 and 6, it is immediately apparent that the flow
characteristics have become much more complex with the
addition of the front wing at full span (silhouette indicated).
This is particularly the case in the flowfield directly behind
and around the wheel boundaries. In this region, the overall
flowfield is clearly more asymmetric with an additional
primary wing vortex clearly visible at the wingtip (i.e
Fig. 5g at Y = 1.4, Z = 0.15). This vortex is generated via a
pressure differential setup around the wing tip/endplate
structure through wing undersurface suction (Zerihan and
Zhang 2003). This vortex rotates in an anti-clockwise
direction and when comparing Fig. 5e, g, as expected,
clearly becomes larger with increasing angle of attack.
Together with these characteristics, increasing wing angle
of attack also produces more significant crossflow across
the wheel face as can be most vividly shown when comparing Fig. 5a, g. For these two particular cases, a change in
angle of attack of 12 has increased the local in-plane nondimensional velocity magnitude Vyz/Vo, from essentially
zero at Y = 1.25, Z & 0 to Vyz/Vo = 0.4. As the angle of
attack is reduced however, the ability of this vortex to
promote crossflow velocity and overcome the high pressure
region within the contact patch clearly diminishes with
Vyz/Vo at Y = 1.25, Z & 0 for AOA = 4 and 8 being
Vyz/Vo & 0 and Vyz/Vo = 0.1, respectively. The primary
vortex, under these conditions, would posses an enhanced
ability to move around the outside of the wheel rather than
along the front face, and finally, on its inside.
Further to the previous discussion, increasing the AOA
of the wing up to 12 is seen to fundamentally shift the
aerodynamics in the vicinity of the contact patch. At these
extreme conditions, the enhanced crossflow generated
creates conditions where the initially symmetric viscous
jetting (Fig. 4a, c) becomes highly asymmetric (Fig. 5ch).
The primary effect of this resulting asymmetry is seen both
at this measurement station (X = 0.75) as well as further
downstream. From comparing Figs. 6 to 4, a significantly
increased degree of wake asymmetry is also clearly evident. The resultant wake, with the wing installed, leads to a
general bias toward the inside of the wheel, entraining the
flow to this side, and ultimately setting up the predominant,
final anti-clockwise flow rotation condition as shown in
Fig. 6b, d, f, h. In Fig. 6b, d, f, h, the flowfield has become
much less distinct to individual flow characteristics,
effectively being now, only a mixture of the more specific
flow features shown upstream.
Another interesting aspect of note with the addition of
the front wing is the movement of the stagnation position at
mid-wheel height as angle of attack is increased. From an
AOA = 012, the height of the stagnation position was
found to move from Z = 0.5 to Z = 0.55. Adjacent to this

123

Exp Fluids (2008) 45:537546

region, at mid-wheel height, evidence also exists for a


substantial degree of flow acceleration (Vyz/Vo = 0.70.9)
over the top of the endplate and into the mainstream flow
along the outside of the wheel (Fig. 5a, c, e, g). This flow
characteristic, which intensifies with AOA increase, is also
prominent when the AOA = 0o (Fig. 5a), which would be
expected due to the proximity of the stagnation region at
mid-wheel height.
On further comparison of results from Figs. 4 and 5, the
wake behind the wheel with the addition of the front wing
was also found to be significantly more compact than that
shown for the isolated wheel case (both stationary and
rotating) shown in Fig. 4. Additionally, the wake height
with the addition of the front wing seems to have reduced
slightly to approximately the top of the wheel between
1.2 \ Z \ 1.3 (Fig. 5b, d, f, h) when compared to both
wheel alone cases (1.4 \ Z \ 1.5). In this instance, the
wing clearly assists the flow in overcoming the obstacle of
the wheel thereby tending to reduce the likelihood of premature flow separation over the top of the wheel and the
subsequent higher wheel wake. As can be seen in Fig. 5b, d,
f, h, this decrease in wake height with the wing installed was
also found to reduce the flow velocity entrained to the floor
by upwards of 30% compared to the isolated wheel cases.
3.3 Wheel and wing, wing span 0%
Results for the wing/wheel combination as the front wing
angle of attack is increased for the minimum span condition are shown in Fig. 7. At first inspection, there exists a
much larger vortex located at Y = 0.5, Z = 0.2 (Fig. 7c)
which travels on the inside of the wheel and towards the
wind tunnel wall (symmetry plane). This larger vortex,
which is now augmented by the high pressure region of the
contact patch, is also shown to translate vertically upwards
from Z = 0.2 to Z = 0.3 (Fig. 7d) with change in wing
AOA. This result was thought to be due mainly to the
added contribution of flow upwash with increase in wing
AOA. Additionally, as can be seen from Fig. 7eh, this
flow feature also remains one of the dominant flow features
for both angles of attack, and together with the left wheel
vortex, produces a general clock-wise rotation on the
overall downstream wake field (Fig. 7g, h). This final result
is in direct opposition to the result found at the same
measurement locations for the full span condition shown in
Figs. 5 and 6.
Another interesting contrast with the results presented
for the longest wing span tested, is that for the shortest
wing span configuration, for the largest angle of attack,
there is a clear absence of any dominant crossflow within
the contact patch region (Figs. 5g, 7b). Additionally, the
mid-wheel height stagnation position remains relatively
unaffected with AOA change at this decreased wing span

Exp Fluids (2008) 45:537546

543

Fig. 5 Comparison of results


for maximum wing span;
X = -0.63 and X = 0.75

condition. However, the acceleration of the flow over the


end plate clearly remains, albeit, now moving inboard
instead of outboard as described in Sect. 2. Interestingly,
for the short wing span, the maximum in-plane velocity
magnitude within this region was found to be measurably
smaller with Vyz/Vo = 0.8 at AOA = 12 compared to
Vyz/Vo = 0.9 found in Fig. 5. Under these conditions, the
high pressure regions at both the mid-height of the wheel

and on the upper wing surface now tend to react against


each other, leading to lower flow velocities.

4 Conclusion
An investigation into the effect of changing wing span and
angle of attack on the interaction aerodynamics of a sub-

123

544

Exp Fluids (2008) 45:537546

Fig. 6 Comparison of results


for maximum wing span;
X = 1.5 and X = 3

scale inverted wing and wheel, in ground effect, is presented. Experimental results obtained using 3D LDA are
presented and were used to show various component

123

interaction aerodynamic phenomenon in an attempt to


broaden understanding of this complex interaction. It was
demonstrated that when the wing span is adjusted across

Exp Fluids (2008) 45:537546

545

Fig. 7 Comparison of results


for minimum wing span

123

546

the width of a wheel at different AOA, the primary wing


vortex plays a significant role in the resulting flow physics.
Primarily, at low wing spans the primary wing vortex
travels on the inside of the wheel producing a complex
asymmetric wake structure with general rotational characteristics in the clockwise direction when viewed from in
front of the combination. At larger spans the reverse was
found. The primary mechanism for changing wing AOA
was to change the size, strength and degree of movement of
the tip vortex generated from the main wing element. It is
hoped that while the conditions specified in this paper are
specific and sub-scale, there are clearly defined trends and
characteristics in the results presented that may give the
designer an initial, useful, and macroscopic insight into this
highly complex flow phenomenon.

References
Agathangelou B, Gascoyne M (1998) Aerodynamic considerations of
a formula 1 racing car. Society of automotive engineers (980399)
Axon L (1998) The aerodynamic characteristics of automobile
wheels-cfd prediction and wind tunnel experiment. Ph.D.
dissertation, College of Aeronautics, Cranfield University
Benedict L, Gould RD (1996) Towards better uncertainty estimates
for turbulence statistics. Exp Fluids 22(2):129136
Cummings R (2002) Cfd investigation of front wing geometry of a
formula 3 car. M.Sc., Cranfield College of Aeronautics, Cranfield University
Dominy RG (1992) Aerodynamics of grand prix cars. In: Proceedings
of the institution of mechanical engineers, Part D. J Automob
Eng 206(D4):267274

123

Exp Fluids (2008) 45:537546


Fackell JF, Harvey JK (1975) The aerodynamics of an isolated road
wheel. AIAA proceedings of the second symposium on aerodynamics of sports and competition automobiles
Fackrell JE (1974) The aerodynamics of an isolated wheel in contact
with the ground. Ph.D. dissertation, Faculty of Engineering,
University of London
Katz J (1986) Aerodynamic model for wing-generated downforce on
open-wheel-racing-car confiurations. Society of automotive
engineers (860218)
Katz J (2006) Aerodynamics of race cars. Annu Rev Fluid Mech
36:2763
McManus J, Zhang X (2006) A computational study of the flow
around an isolated wheel in contact with the ground. J Fluids Eng
128(3):520530
Metz LD (1987) Aerodynamic properties of indy cars. Society of
Automotive engineers (870726)
Mortel F (2003) The front wing. M.Sc., Cranfield College of
Aeronautics, Cranfield University
Ranzenbach R, Barlow J (1996) Cambered airfoil in ground effect
an experimental and computational study. Society of Automotive
engineers (960909)
Stapleford WR, Carr GW (1970) Aerodynamic characterisitcs of
exposed rotating wheels. Technical report 1970/2
Thisse E (2004) Influence of end plates on tip vortices in ground
effect for a 2004 formula one front wing. M.Sc., Cranfield
College of Aeronautics, Cranfield University
Zerihan J, Zhang J (2000) Aerodynamics of a single element wing in
ground effect. J Aircr 6(37):10581064
Zerihan J, Zhang J (2003) Off-surface measurements of a wing in
ground effect. J Aircr 40(4):716725
Zhang X, Toet W, Zerihan J (2006) Ground effect aerodynamics of
race cars. Appl Mech Rev, 3349

You might also like