Professional Documents
Culture Documents
Robert Cousseau
School of Engineering
MSc
Cranfield University
School of Engineering
MSc
2007
Robert Cousseau
Cranfield University, 2007. All rights reserved. No part of this publication may be reproduced without
the written permission of the copyright holder
Cranfield University
Aerospace Dynamics
Abstract
An experimental investigation into the impact of a mounting rear wing within the flow
structure in the near wake of a saloon race car has been carried out using a scale glass
fibre model in the Cranfield University Aerodynamics Laboratories wind tunnels.
Surface flow visualisation involving the oil-dot technique on the car with and without
the wing has been performed in order to gain an understanding of why the near wake
has a particular structure involving trailing edge vortices and contra-rotating vortices
and how these features are affected by the presence of the wing. The effect of sideslip
has also been investigated. Force measurements were also carried out using an internal
balance in order to support the results obtained.
The flow visualisation results showed that the presence of sideslip or the wing has a
significant effect on the flow over the backlight whereas the flow over the trunk was
virtually unaffected. Sideslip had an effect on the balance of the contra-rotating vortex
wake structure and determined which one of the pair is dominant. The presence of the
wing and its location had an effect on the whole contra-rotating vortices structure,
making it smaller as the wing was close to the backlight. These results showed some
differences with what has been noted in previous studies.
The force measurements results showed that some wing/body interactions were
involved and produced some favourable and unfavourable effects which significantly
influenced the lift and drag experienced by the model.
Cranfield University
Aerospace Dynamics
ii
Cranfield University
Aerospace Dynamics
Acknowledgements
I would like to thank Prof KP Garry for his help and support during thesis project.
I would also like to thank John, Jenny and Linton for their help during the experiments
in wind tunnels.
iii
Cranfield University
Aerospace Dynamics
iv
Cranfield University
Aerospace Dynamics
Contents
ABSTRACT ...............................................................................................................................................I
ACKNOWLEDGEMENTS .........................................................................................................................III
TABLE OF CONTENTS ............................................................................................................................. V
LIST OF FIGURES ................................................................................................................................. VIII
LIST OF TABLES ......................................................................................................................................X
NOMENCLATURE ..................................................................................................................................XI
1
INTRODUCTION ............................................................................................................................ 1
1.1
AIMS ................................................................................................................................................ 1
1.2
OBJECTIVES ........................................................................................................................................ 2
2.1.1
2.1.2
2.1.3
Sideslip .................................................................................................................................. 7
2.2
2.2.1
2.2.2
Wings .................................................................................................................................... 9
2.3
2.3.1
2.3.2
2.4
3
MODEL ........................................................................................................................................... 25
3.1.1
Car ...................................................................................................................................... 25
3.1.2
Wing ................................................................................................................................... 25
3.1.3
3.2
3.2.1
3.2.2
3.2.3
Cranfield University
TEST METHOD............................................................................................................................. 29
4.1
4.1.2
4.2
4.3
4.3.1
4.3.2
4.4.2
RESULTS ...................................................................................................................................... 39
SIDESLIP TESTS .................................................................................................................................. 39
5.1.1
5.1.2
5.2
5.2.1
5.2.2
5.2.3
5.3
5.3.1
5.3.2
5.4
6.1.1
6.1.2
6.2
6.3
6.4
6.4.1
6.4.2
6.4.3
6.5
vi
4.4.1
5.1
4.1.1
4.4
Aerospace Dynamics
Cranfield University
Aerospace Dynamics
6.5.1
6.5.2
6.5.3
6.6
6.7
7.2
7.3
7.4
EFFECT OF THE WING ON THE OVERALL VEHICLE LIFT AND DRAG CHARACTERISTICS .......................................... 89
7.5
7.6
7.7
REFERENCES ......................................................................................................................................... 91
vii
Cranfield University
Aerospace Dynamics
Figures
FIGURE 2-1: REAR-END FORMS: NOTCHBACK (LEFT); HATCHBACK (CENTRE); SQUAREBACK (RIGHT) ................................... 4
FIGURE 2-2: RAKE ANGLE OF A HATCHBACK CAR ...................................................................................................... 4
FIGURE 2-3: NOTCHBACK REAR END PARAMETERS ................................................................................................... 5
FIGURE 2-4: TRANSVERSE VORTEX ........................................................................................................................ 6
FIGURE 2-5: C-PILLAR VORTICES ........................................................................................................................... 6
FIGURE 2-6: THE ARCH VORTEX ........................................................................................................................... 7
FIGURE 2-7: THE EFFECT OF A REAR SPOILER ........................................................................................................... 9
FIGURE 2-8: EFFECT OF A REAR WING ON THE STREAM LINES NEARBY A GENERIC BODY.................................................. 10
FIGURE 2-9: EFFECT OF DISTANCE TO THE BODY AND ASPECT RATIO ON THE WING EFFECTIVENESS ................................... 11
FIGURE 2-10: EFFECT OF WING PROXIMITY TO THE GROUND ON THE DOWNFORCE ...................................................... 11
FIGURE 2-11: COMPARISON OF THE CP DISTRIBUTIONS OF A MOUNTED WING WITH THAT OF THE WING ALONE ................ 12
FIGURE 2-12: THE EFFECT OF END PLATES ........................................................................................................... 13
FIGURE 2-13: BOUNDARY LAYER AND SCALE EFFECT .............................................................................................. 15
FIGURE 2-14: MOVING ROAD PROBLEM .............................................................................................................. 16
FIGURE 2-15: VARIOUS METHODS FOR SIMULATING A MOVING GROUND IN A WIND TUNNEL ......................................... 17
FIGURE 2-16: BLOCKAGE EFFECT ........................................................................................................................ 18
FIGURE 2-17: RESULTS OBTAINED WITH THE OIL DOT TECHNIQUE ............................................................................. 21
FIGURE 3-1: GLASS FIBRE MODEL ....................................................................................................................... 25
FIGURE 3-2: MODEL MOUNTED IN THE 86 WIND TUNNEL WITH THE WING ............................................................. 26
FIGURE 3-3: WING SUPPORT ............................................................................................................................. 27
FIGURE 4-1: COMPARISON BETWEEN THE TWO MIXTURES USED (POSTER PAINT ON THE TOP PICTURE, PARAFFIN ON THE
BOTTOM PICTURE).................................................................................................................................. 31
FIGURE 4-2: REPEATABILITY TESTS WITH DOTS OF MEDIUM SIZE ON THE TOP RIGHT CORNER........................................... 32
FIGURE 4-3: REPEATABILITY TESTS WITH BIG DOTS PLACED ON THE TOP LEFT CORNER.................................................... 33
FIGURE 4-4: ORIENTATION OF THE MODEL IN THE WEYBRIDGE WIND TUNNEL ............................................................ 34
FIGURE 4-5: AXES USED TO LOCATE THE WING ...................................................................................................... 34
FIGURE 4-6: WING DRAG COEFFICIENT VARIATION WITH WING VERTICAL LOCATION WITH ERROR BARS............................. 37
FIGURE 5-1: FLOW COMING FROM THE SIDES ....................................................................................................... 39
FIGURE 5-2: PATTERNS OBTAINED ON THE BACKLIGHT FOR = 0 ............................................................................. 40
FIGURE 5-3: PATTERNS OBTAINED ON THE BACKLIGHT AND THE TRUNK FOR = 0 ....................................................... 41
FIGURE 5-4: REATTACHMENT LINE...................................................................................................................... 41
FIGURE 5-5: DOTS PLACED ON THE TRUNK ........................................................................................................... 42
FIGURE 5-6: FEATURES OF THE PATTERNS ............................................................................................................ 43
viii
Cranfield University
Aerospace Dynamics
ix
Cranfield University
Aerospace Dynamics
Tables
TABLE 4-1: MIXES OF POSTER PAINT AND WATER USED ........................................................................................... 29
Cranfield University
Aerospace Dynamics
Nomenclature
Roman
Drag coefficient
Lift coefficient
Backlight length
Trunk length
Reynolds number
Trunk width
Sideslip angle
Backlight
Trunk
xi
Cranfield University
Aerospace Dynamics
xii
1 - Introduction
1 Introduction
Increasingly designers and engineers work together. When a new project is being
developed, two main requirements must be fulfilled; the mechanical aspect, dealt with
by the engineers, must obviously be conceived with great care since it gives the product
its main function; and the aesthetical aspect, dealt with by the designers, which must not
be neglected either as the exterior look is often a very important parameter for
customers.
The automotive industry is not an exception. One very important feature to take into
account when conceiving an automobile is its drag coefficient since it influences engine
requirements, fuel consumption and the overall aerodynamic performance. The major
component of the drag force is called the form drag and mainly arises from the flow
separation. However, some appealing designs happen to increase flow separation.
In addition, race cars must generate some downforce to improve tire adhesion and as a
result, vehicle acceleration and turning rate. The addition of spoilers such as a rear
mounted wing is a very good way to increase the downforce. But to understand properly
the effects of such a wing and its shape, location or inclination on a race cars
aerodynamic properties, it is necessary to have a good understanding of the features of
the wake of the car without spoilers and of how the flow separates.
Two contra-rotating vortices are located on in the near wake of a notchback car. Fisher
(1) found that as the wing is moved forward, one of these vortices becomes dominant
and eventually, when the wing is at its foremost position, only one bigger vortex exists.
1.1 Aims
The aim of this study is to investigate the effect that adding a rear mounted wing to a
saloon car has on the structure of the wake and understand why one of the contrarotating vortices becomes dominant as the wing is moved forward.
1 - Introduction
It will be important to get an understanding of what gives this particular structure to the
wake and how its structure is affected by external parameters. The influence of side
winds on the structure on the wake will be investigated.
Finally, the interaction of the wing with the two contra-rotating vortices located near the
backlight of the car will be investigated. More particularly, the influence of axial and
vertical position will be of interest.
1.2 Objectives
To achieve this, wind tunnel testing will be carried out using a quarter scale model of
the car. A particular technique, surface oil-dot flow visualisation, will be used to
precisely visualise the structure in the wake. The technique will first have to be adjusted
for this particular case. The analysis of the effect of wing location on the wing lift and
drag and on the car total lift and drag will also enable a better understanding of the flow
structure.
2 - Literature review
2 Literature review
As mentioned earlier, it is very important to understand the structure of the wake behind
a car when no spoilers are mounted on it. In the first part of this section, the basic
feature of the wake will be described. The second part of this section will consist of a
short description of rear spoilers and wings and the effect they have on the main
aerodynamic characteristics. This will be a good starting point for further investigation
which will be carried out in the frame of this thesis project. Finally, the last part of this
section will describe some issues encountered in wind tunnel testing as well as some
flow visualisation techniques useful for this thesis.
2 - Literature review
As a consequence, the drag also starts to increase from 10 and a peak in drag
coefficient has been revealed for rake angles around 30 (2).
Figure 2-1: Rear-end forms: notchback (left); hatchback (centre); squareback (right)
2 - Literature review
Due to separation, a shear layer is created which may reattach on the decklid or not
depending on the backlight angle (), the height (d), the decklid length (t) (Figure 2-3)
and the downwash created by the transverse vortex. In 1990, Nouzawa, et al. (6)
determined that = 25 is the critical angle above which the flow does not reattach and
bellow which it does reattach. If the flow reattaches, a separation bubble is created as a
part of the flow is engulfed in the recirculation area.
Quasi-two-dimensional separation also occurs at the decklid trailing edge and another
separation bubble behind the base is formed when the flows from below and above the
car merge.
2 - Literature review
2 - Literature review
One major effect of these vortices is the downwash they produce in the centre plane
since it pushes the shear layer created by the rooftop separation towards the decklid and
therefore helps the flow to reattach and delays the decklid trailing edge separation (5).
However this effect is not so pronounced on notchback cars as their aspect ratios are
larger and consequently the produced downwash is weaker.
2.1.2.3 Arch vortex
If the angle defined in Figure 2-3 is below 25 but close or equal to this value, the
recirculation within the separation bubble forms an arch vortex as show in Figure 2-6.
The base of the arch vortex forms two contra-rotating vortices on the surface of the
trunk between which some reverse flow goes upwards.
2.1.3 Sideslip
The effect of crosswinds can have a strong influence on the vehicles behaviour. If the
effective yaw angle is high, the rear positive lift coefficient can be dramatically
increased, inducing a decrease in tyre grip and in this way a decrease in acceleration and
turning rate capabilities. The drag coefficient undergoes a similar change with sideslip
(9).
2 - Literature review
However, the effect of sideslip on saloon cars and the efficiency of spoilers in the case
of a cross wind has yet to be investigated since the wake structure may be affected and
as a consequence, the loading of the wing would be significantly different. This is all
the more important as in a real situation it is very unlikely that there will be a fully
streamwise air flow.
2 - Literature review
It was also noted that the use of rear lip spoilers does not necessarily come with an
increase in drag. In some cases, fitting a car with a rear spoiler can actually decrease the
drag. But even though in the majority of cases rear lip spoilers increase the drag and
decrease the top speed, the overall performance such as acceleration and turning rate are
increased and as a consequence, the lap times decrease (10).
Another way to reduce the overall lift coefficient is to produce some downforce using
an inverted wing at the rear of the car. Wings are lifting surfaces usually used to lift
aircraft off the ground, so by using wings upside down, the force generated is logically
pointed towards the ground.
2.2.2 Wings
2.2.2.1 Wing/body interactions
It is interesting to note that the flow over a saloon car rear-mounted wing can be very
different from that of the wing alone. Indeed, there are strong interactions between the
wing and the car body. The effects of adding a rear wing are described bellow.
2.2.2.1.1 Effect on the cars own downforce
The addition of a wing can increase the downforce of the body itself. Indeed, the flow
going between the wing and the body is deflected upwards. This means that the flow
going under the body is also deflected upwards and its velocity is increased. And as a
2 - Literature review
higher velocity implies a lower pressure, more downforce is produced by the body itself,
independently of the wings own downforce, as shown in Figure 2-8.
Figure 2-8: Effect of a rear wing on the stream lines nearby a generic body (11)
layer generated on the trunk blocks the flow under the wing. Therefore, the flow mainly
goes over the wing and the downforce produced is highly reduced. Note that this
10
2 - Literature review
Figure 2-9: Effect of distance to the body and aspect ratio on the wing effectiveness (12)
Figure 2-10: Effect of wing proximity to the ground on the downforce (11)
11
2 - Literature review
mounted wing and the wing alone are very different in shape and magnitude. The
mounted wings lower surface experiences a much larger suction than the wing alone
while the upper surface distribution is roughly the same in both cases. Consequently,
the wing produces more downforce when it is mounted on the car. This is explained
firstly by the fact that, as mentioned previously, the flow velocity is increased over the
lower surface when the wing is close to the car. The second reason is the fact that the
upstream flow changes direction because of the body. The flow is deflected in the
downward direction which effectively increases the angle of attack and, as stated above,
the lift is increased.
Figure 2-11: Comparison of the CP distributions of a mounted wing with that of the wing alone (11)
12
2 - Literature review
their thickness should be chosen carefully as too thick end plates will eat too much
plane area for a fixed span and too thin end plates will not be rigid enough (10).
2.2.2.3 End plates
End plates are not just used to mount the wing on the car, they are also used to increase
the downforce generated by the wing and reduce the drag. Without end plates, the
difference of pressure between the upper and the lower surface of the wing makes the
air from the high pressure surface move to the low pressure surface. This has the effect
of decreasing the difference of pressure between the two sides and as a result the
downforce is reduced. By adding end plates to the wing, the difference of pressure is
maintained and no loss of downforce is experienced. The migration of the air from one
side to the other also generates tip vortices and therefore a large amount of drag. The
addition of end plates prevents the formation of tip vortices and in this way the drag is
decreased (13).
Due to the small amount of published data, it is difficult to predict with precision how
the addition of a rear wing may affect the flow structure. This issue will be investigated
within the framework of this thesis project.
13
2 - Literature review
In contrast, wind tunnel testing makes the tests a lot easier. The most obvious advantage
is that the car stays stationary while the air is moving. This also implies that it is
possible to use full scale or even reduced scale models. The aerodynamic loads can be
measured by a stationary scale, or balance. The test conditions can be controlled.
However there are, here as well, some issues in the use of wind tunnels such as scale or
Reynolds number effect, simulation of the moving road problem and errors due to
blockage (2).
2.3.1 Wind tunnel testing issues
2.3.1.1 Reynolds number effect
Scale effects can be expected due to the difference in model and full-scale Reynolds
number. Figure 2-13 shows a thin plate placed in a in a stream of air and its scale model
placed in the same stream of air and in the same conditions. It can be seen that the
transition point is located at the same distance from the leading edge. However the
boundary layer transition occurs at about 25% of the length from the leading edge for
the full scale plate but it occurs at more than 50% of the length from the leading edge
for the scale model. The drag per unit area is therefore lower for the scale model and is
not representative of the full scale phenomenon.
14
2 - Literature review
To prevent Reynolds number effects occurring, it is necessary to run the tests at the
same Reynolds number. To do so, the easiest way is often to increase the air velocity
( =
), however if the scale ratio is too high the velocity could have to be
increased so much that it would go supersonic and the flow over the model would be
completely different from the full size. To tackle this problem, the density can be
increased using a pressurized wind tunnel or the viscosity can be decreased by using
cryogenic cooling but even though these methods exist, they are extremely expensive.
For automotive wing tunnel testing, the solution is often to perform full scale tests, the
price of which is not excessive for major manufacturers (2). If full scale testing is
impossible, to avoid compressibility effects, the Mach number must not exceed 0.4.
This implies that, as the density and the viscosity are unlikely to be changed, the scaling
factor must be kept under a certain value depending on the full scale velocity that must
be tested (14).
2.3.1.2 Ground simulation
A simulation issue arises from the relative velocity of the wind and the ground in wind
tunnel testing. Indeed, in real conditions shown in Figure 2-14A, the vehicle is moving
relatively to the ground and the wind but the wind is not significantly moving relatively
to the ground. Therefore there is no road boundary layer. However in a wind tunnel
Figure 2-14B, the air is moving relatively to the vehicle and the ground, thus developing
a boundary layer. As a consequence, the ground plane boundary layer velocity profile is
not the same and the results will be significantly affected (2).
15
2 - Literature review
There are many way to tackle this problem as shown in Figure 2-15.
Figure 2-15A shows the ground board method. By mounting the model above an
elevated board, the boundary layer developed will be a lot thinner and the errors will be
minimized. This is a very simple method that can be used in small wind tunnels.
The method shown in Figure 2-15B consists in boundary layer suction ahead of the
model. The boundary layer developed under the model is much thinner. This method is
also very simple. This method can be improved by applying the suction under the whole
model (see Figure 2-15C). This method is complex and expensive (11).
A similar method is to blow air into the boundary layer to reenergize it and make it
thinner (see Figure 2-15D). This is a quite efficient method but also expensive (11).
Another method is to use a mirror image underneath the model (see Figure 2-15E).
Since there is symmetry, the symmetry line between the two models is a stream line.
Therefore, no boundary layer effects are experienced. However, the models must be
exactly identical and every changes made on one model during the tests must be done
on the other one. Moreover, the test section size must be increased to contain the two
16
2 - Literature review
models. The costs and complexity of this method are thus very high which makes it
unused nowadays (2) & (11).
The last method is the moving belt technique, popular among race car designers, shown
in Figure 2-15F. It consists in removing the relative motion between the ground and the
model. Despite the good results this method gives, it is not simple to perform. The
model has to be mounted by above using a sting, which can interfere with the flow.
Suction must be applied before the belt and under the whole model and air must be
reintroduced behind it. The sting may also interfere with the measurements of lift and
drag. The last issue is the limited speed of the belt which is usually less than the
maximum wind tunnel speed (11) & (2).
Figure 2-15: Various methods for simulating a moving ground in a wind tunnel (11)
17
2 - Literature review
appears concerning the model size; it is always preferable to perform full scale testing in
order to keep the same Reynolds number as explained in Section 2.3.1.1 and also to
include some small details of the design whereas the model should be kept as small as
possible so that the blockage effect can be minimized. So there must be a compromise
the wind tunnel facility and design requirements since the wall interferences are not
negligible (11).
The correction that must be applied mainly concerns the velocity, the dynamic pressure
and the Reynolds number. The real values of these parameters are greater than the ones
calculated with the wind tunnel instrumentation. They can be obtained as follows:
= 1 +
= 1 + 2
= 1 +
Where V, q and are the air velocity, the dynamic pressure and the Reynolds number
of the test section respectively and is the blockage correction factor.
Several ways to obtain the blockage correction factor exist. However, for unusual or
complicated shapes it can be very difficult to obtain it. It those cases, Barlow, Rae and
Pope (15) defined the following approximate blockage correction factor:
18
2 - Literature review
Another way is the tuft wand technique. It consists in a flow field description using a
tuft wand. The direction of the flow field is revealed by introducing a wool tuft attached
to a rod into the airstream. The rod must be long enough to allow the use of this
technique from a sufficient distance so that the aerodynamic interferences are
minimized. The wool tuft used must also be long enough to examine the flow
phenomenon in question. This technique is easy and quick to deploy. Large scale or
19
2 - Literature review
small scale phenomena can be observed by changing the tuft length. However, in low
speed flow, the weight of the tuft modifies the indication of the flow and if the tuft
length is not properly adjusted, it may be difficult to observe small scale phenomena
(16).
2.3.2.2 Oil flow visualisation
This technique consists in applying a film of oil on the surface of the model. The oil
must be mixed with a pigment. The mixture can either be kerosene and a fluorescent
powder or liquid paraffin, titanium dioxide and oleic acid.
The airflow over the surface will create shear stresses which will move the pigment
particles and eventually show the flow lines. If the first mixture is used, then ultraviolet
lamps must be used to visualise the flow lines. This technique gives some detailed
qualitative information about the flow lines direction and speed of attached flow on the
surface of the model. The zone of separation as well as the vortices close to the surface
can also be observed.
The main advantages of this technique are that it is very simple and inexpensive. The
fact that the flow lines can still be observed after the wind tunnel has stopped is also an
advantage. However, it is quite messy to use as the fluorescent particles can deposit on
the floor downstream of the model and also on the clothes and hand of the tunnel staff
when handling the model. It is often difficult to differentiate forward from reverse flow
in the zones where the flow has separated. Another disadvantage is the time it takes for
the flow patterns to fully develop and gravity can affect the results on non horizontal
surfaces (16).
2.3.2.3 Oil dot flow visualisation
Oil dot flow visualisation involves the use of small dots of oil (or ink) to visualise the
surface stream lines (see Figure 2-17). Some droplets are placed on the surface of the
model so that the air flow blows them and in this way the stream lines appear. The oil
contains a pigment so as to have a good contrast with the surface colour. To help the oil
dots to move more easily, it is often necessary to cover the surface with a thin layer of
clear oil before applying the dots (17).
20
2 - Literature review
Figure 2-17: Results obtained with the oil dot technique (16)
This technique gives indications about the surface flow direction, vortex patterns and
the location of flow separation.
This technique is widely used due to its numerous advantages. As the oil flow
visualisation technique described in Section 2.3.2.2, the flow patterns can be more
closely observed and studied after the wind tunnel has stopped. It is also convenient to
be able to develop the full scale patterns by successively applying the droplets and
blowing the tunnel. This technique also provides a precise indication of flow separation
points and enables to establish flow directions in the turbulent wake, which is usually
difficult with other flow visualisation methods.
This technique unfortunately comes with its limitations. It is difficult to use it on quasivertical surfaces. The droplets can not flow past model joint lines if they are not taped.
The model must be cleaned after using this technique. The flow condition can not be
changed during the test; if several test conditions are to be tested such as several angles
of incidence or several wind tunnel speeds, then several tests have to be run otherwise
21
2 - Literature review
the patterns would be confused. Finally, a high freestream flow velocity is required to
observe the stream line in low surface velocity areas.
This study confirmed the general wake structure suggested by Nouzawa et al (6), more
particularly the two contra-rotating vortices on the backlight and the trailing edge
vortices.
It was noted that the lift (downforce) increased as the wing was moved rearwards. The
lift also increased with the height of the wing.
The results obtained for the drag were quite surprising since for all wing locations in the
wake, the drag was negative. This result was not in accordance with the flow
visualisation and more detailed flow visualisation around the wing would have been
necessary to get a better understanding of how the flow is circulating around it.
The addition of the wing had the effect of increasing the general model downforce for
all locations and decreasing its drag for low and forward locations. For high and
rearward locations of the wing, the wake size and therefore the drag increased.
An increase of angle of attack caused the drag and lift to increase. An optimum angle of
5, leading to an increase in lift but a small increase in drag, was determined. The
downward flow direction of the wake increased the angle of attack and caused the wing
to stall at 10.
22
2 - Literature review
The experiments also showed that the addition of the wing makes one of the two contrarotating vortices dominant on the other one and that this effect is more apparent as the
wing is moved forwards. At the wings foremost position, only one single region of
recirculation was observed. However the cause of this phenomenon is still not well
understood and more detailed flow visualisation is required. This will be done as a part
of this thesis project.
The effects of yaw and side winds were not studied by Fisher (1), even though these
factors would affect the wake structure and consequently the aerodynamic forces on the
car. This will also be studied in the frame of this thesis project.
23
Cranfield University
Aerospace Dynamics
24
3 - Experimental set up
3 Experimental set up
3.1 Model
3.1.1 Car
The model used was a scale glass fibre model based on the MIRA Variable Geometry
model for a notchback vehicle and created for the purpose of a previous MSc project by
Fisher (1) (Figure 3-1). This model was chosen due to its geometric similarities with
several saloon race cars. Its main glass fibre body is connected to four non rotating
wheels. The scale was chosen to limit the blockage effect by having a sufficiently
small frontal area.
3.1.2 Wing
The rear spoiler used for the experiments was an inverted wing with a Clark-Y aerofoil
section (Figure 3-2). The Clark-Y section was chosen for its good performance at low
Reynolds number and for its flat bottom making easier the manual settings of angle of
attack.
25
3 - Experimental set up
The wing is 415 long and 40 wide which make it no longer than the model is
wide as specified by the Touring Car Championship regulations. Thus, it has an aspect
ratio of 10.4 and a planform area of 0.0166 2 .
Figure 3-2: Model mounted in the 86 wind tunnel with the wing
26
3 - Experimental set up
27
3 - Experimental set up
layer suction. The model can be accurately located at a specified height with an active
driven servo strut system.
Two balances were used to measure aerodynamic forces. The Aerotech 6 component
Internal Balance was used to measure the total lift and drag of the car with the wing
mounted on it. It contains one strain gauge for each aerodynamic component.
The calibration, which was carried out by the manufacturer, takes into account all
interaction between the 6 components measured.
An additional balance was used to measure the lift and drag of the wing and its support
only. The balance used for the wing was composed of two 50 load cells connected
together in such way that one would measure lift and the other rotated through 90 to
measure drag without interacting with each other.
The calibration was carried out by Fisher (1) during her MSc project.
28
4 - Test Method
4 Test Method
4.1 Oil-dot technique
The oil-dot technique was used to visualise the surface flow in the near wake. This
technique had to be optimised for the case of low flow velocity on an inclined surface
since the optimum mixture to use is really dependent the flow conditions.
4.1.1 Mixtures tested
Two mixtures have been tested. The tests were carried out in the Weybridge wind
tunnel in Cranfield University. The wing was not mounted on the model and no sideslip
angle was set.
4.1.1.1 Poster paint
The first mixture tested was a mix of poster paint and water. Several were tested (Table
4-1).
Test number
1
2
3
4
5
volume of water
1
2
3
4
5
The mixtures used in tests 1 and 2 were too thick and the dots would not flow. The
mixture used in test 3 was still too thick and the dots just moved a few centimetres. The
mixture used in test 4 was thin enough to draw the patterns but the dots would not move
in the areas of very low velocity such as the reverse flow or within the contra-rotating
vortices. Moreover the water would not evaporate fast enough and the dots would flow
without leaving clear traces. The mixture used in test 5 was too thin and the dots could
not be placed on the inclined backlight as they would flow with gravity.
29
4 - Test Method
The mix used in test 4 gave the best results was still not suitable to visualise properly
the surface flow on the backlight.
4.1.1.2 Paraffin and invisible blue
The second mixture tested was a mix of paraffin and a fluorescent pigment, invisible
blue. Again, several proportions were tested. Since a very high precision balance would
have been needed to measure the quantity of invisible blue added to the paraffin and
considering that such a balance was not available, the quantity of invisible blue was
measured using the number of full teaspoons of powder added.
With high viscosity (one full teaspoon for 2 of paraffin) the dots would not flow.
With low viscosity (one full teaspoon for 7 of paraffin) the dots were difficult to
place on the inclined backlight and started to flow with gravity as soon as they were
placed so the wind tunnel had to be run as quickly as possible. However the results were
very good: the dots would flow even in the low speed area, enabling to visualise the
reverse flow and a good part of the contra-rotating vortices. The patterns were also very
clear due to the fluorescent pigment (Figure 4-1).
30
4 - Test Method
Figure 4-1: comparison between the two mixtures used (poster paint
on the top picture, paraffin on the bottom picture)
As said before, the limitation of this mixture is that to visualise the reverse flow, the
mixture must not be too viscous, therefore it starts to flow with gravity as soon as the
dot is placed and the tunnel must be run quickly. Therefore the first centimetres of the
patterns are not due to the flow but to gravity.
31
4 - Test Method
This mixture was chosen for the experiments since in spite of its limitations, it gave
good results for the visualisation of the contra-rotating vortices and the surface flow on
the trunk.
4.1.2 Repeatability tests
To assess the reliability of this technique, repeatability tests have been performed. They
involved placing one dot, running the wind tunnel, taking a picture and starting again
placing another dot at the exact same position. It was also important to wind up the
tunnel the same way so that the wind accelerates the same way every time.
Figure 4-2 shows the patterns left during three different runs with dots of medium size
placed on the top right corner. The patterns are very similar.
Figure 4-3 shows the patterns left during three different runs with big dots placed on the
top left corner. Again the patterns are very similar.
This shows that the oil-dots technique is reliable.
Figure 4-2: Repeatability tests with dots of medium size on the top right corner
32
4 - Test Method
Figure 4-3: Repeatability tests with big dots placed on the top left corner
33
4 - Test Method
34
4 - Test Method
done without the wing in order to compare and study the effects of adding the wing,
regardless of its position.
4.3.2 With sideslip
A few runs were done with a sideslip angle of 2. The freestream flow velocity was
30 . 1 .Only the effect of axial position was assessed. The vertical position was
= 0.52
and
the
axial
position
tested
were
the
same
as
before,
= 0.15, 0.13, 0.55 & 0.97. A first run was done to visualise the flow with a 2
sideslip angle and without the wing on since the results may be different from the one
obtained previously as the tests were not carried out in the same wind tunnel.
tested
were
and
= 0.30, 0.41, 0.52, 0.74, 0.95 & 1.17 . However the position corresponding to
To be able to work out the lift and drag on the wing alone, two runs were done for each
wing set up: one with the struts and the wing and one with just the struts but without the
wing attached on it.
35
4 - Test Method
For all the tests the wind tunnel speed was set to 40 . 1 which corresponds to a
Reynolds number based on model length of 2.60 106 .
4.4.2 Force measurements repeatability errors
The force measurements in the Cranfield University 8 6 Automotive Wind Tunnel
were all repeated several times in order to assess the repeatability.
The maximum repeatability errors as a percentage coefficient and are the following:
= 1.7%
= 125%
= 0.6%
= 0.2%
The maximum repeatability errors for the Car Lift Coefficient and the Car Drag
Coefficient are very good and the maximum repeatability error for the Wing Lift
Coefficient is quite good as well. The maximum repeatability error for the Wing Drag
Coefficient seems extremely high, however it is due to the values of the Wing Drag
Coefficient which are very close to zero for some wing locations. Figure 4-6 shows the
error bars on the curve of wing drag coefficient variation with wing vertical location. It
is now clear that the errors are very low for the majority of the points. Even for the
points with the highest errors, the trend of the curve is not affected.
36
4 - Test Method
Figure 4-6: Wing drag coefficient variation with wing vertical location with error bars
37
Cranfield University
Aerospace Dynamics
38
5 - Results
5 Results
5.1 Sideslip tests
5.1.1 Structure of the near wake
Figure 5-1 shows that the dots placed on the sides of the backlight are blown away from
the centre of the backlight (a) and the droplets placed on the front of the car flow
towards the rear and are sucked onto the backlight (b) and then blown away from the
centre and towards the sides (c). This suggests the presence of the two trailing edge
vortices.
39
5 - Results
The dots placed closer to the centre start to flow down, turn towards the centre of the
backlight and then go upwards (Figure 5-2). This indicates the presence two contra
rotating vortices inducing some reverse flow around the centre of the backlight. It was
also observed that the right vortex is the dominant of the pair. It is rounder and bigger
and its centre is close to the backlights centre whereas the left vortex is not round but
oval and it is difficult to locate its centre. The separation lines between the trailing edge
vortices and the contra-rotating vortices will be referred to as backlight trailing edge
vortex limits. The two trailing edge vortices seem to have the same size since the
backlight trailing edge vortex limits are at the same distance from the side edge of the
backlight.
Figure 5-3 shows the patterns left by the dots placed on the trunk. It can be seen that the
dots placed near the centre and close to the backlight, flow towards the centre and then
upwards whereas then other dots flow downwards. This indicates that near the backlight
the flow is still separated the lower parts of the contra-rotating vortices are on the trunk
(Figure 5-4). The dots going downwards reveal the presence of attached flow, implying
that the flow reattaches on the trunk. The line separating the dots going upwards and the
dots going downwards will be referred to as reattachment line.
40
5 - Results
Figure 5-3: Patterns obtained on the backlight and the trunk for = 0
Figure 5-3 also shows that the dots placed on the sides of the trunk flow away from the
centre whereas the dots placed closer to the centre flow towards the centre. This
indicates again the presence of the trailing edge vortices (Figure 5-5). The limits
between the trailing edge vortices and the reattached flow on the trunk can be found and
prolongs the backlight trailing edge vortex limits. These limits seem to be straight lines
41
5 - Results
parallel to the axis of symmetry of the model. They will be referred to as trunk trailing
edge vortex limits
The droplets flowing towards the centre and downwards on the trunk meet on a line
which seems to continue the line separating the two contra-rotating vortices (Figure 5-3).
This line will be called centre limit.
the size of the contra-rotating vortices. (A) shows the size of the right vortex. It
was defined as the distance between the first dot on the side going inwards and
the lateral centre of the reverse flow region.
the positions of their centres. (B) shows the position of the right vortexs centre.
It was defined as the centre of the smallest circle drawn by the dots.
the backlight trailing edge vortex limits (C). They were obtained by tracing
vertical lines on the backlight between the dots turning inwards and the dots
turning outwards.
the trunk trailing edge vortex limits (D). They were obtained by tracing vertical
lines on the trunk between the dots turning inwards and the dots turning
outwards. These lines always prolonged the trailing edge vortex limits.
42
5 - Results
the centre limit (E). It was obtained by tracing a vertical line on the trunk where
the dots going inwards meet.
the reattachment line (F). It was obtained by tracing a line on the trunk between
the dots going upwards and the dots going downwards.
The location of these features could not be determined precisely, the error on the
locations of these lines is estimated to 1 on the model which is not excessive
compared to the model width (40.1 ). Moreover, only the trends were of interest in
this thesis.
Note that since the backlight trailing edge vortex limits are always prolonged by the
trunk trailing edge vortex limits, it is not useful to study both their locations. Therefore
only the trunk trailing edge vortex limits will be looked at.
43
5 - Results
The position of the trunk trailing edge vortex limits, the limit between the contra
rotating vortices on the backlight's lower edge and the centre limit will be given as the
ratio of the distance from the centre line on the trunk ( ) over the trunks width ( );
the reattachment lines position will be given as the ratio of the distance from the
trunks rear edge ( ) over the trunks length ( ); the size of the contra-rotating vortices
and the position of their centres will be given as the ratio of the distance from centre
line on the backlight ( ) over the roofs width ( ) or/and the ratio of the distance
from the backlights lower edge ( ) over the backlights length ( ) (Figure 5-7).
44
5 - Results
Figure 5-8: Patterns obtained with the oil-dot technique for different sideslip angle
45
5 - Results
Figure 5-10 shows the effect of changes in sideslip angle on the positions of the contrarotating vortices. The axial position seems more affected than the lateral position. Again,
as the non-dominant vortex dos not have a round shape, is was not possible to determine
the position of both vortices for all sideslip angles.
For sideslip angles below zero, the right vortex was centred near the centre of the
backlight. Approaching the critical angle, the right vortex starts to move up to the top
46
5 - Results
right corner of the backlight. After the critical angle, the left vortex moves towards the
centre line with increasing sideslip angle but stays at the same axial position.
Even when the left vortex is fully dominant, it is not centred on the centre of the
backlight.
Figure 5-11 shows the position of the non-dominant vortex for extreme angles. It can be
seen that for extreme angles the non-dominant vortex, either left or right, eventually
moves up. It is also notable that the non-dominant right vortex is significantly bigger
than the non-dominant left vortex.
47
5 - Results
48
5 - Results
Figure 5-13 shows effect of sideslip on the trailing edge vortices and the centre limit.
The change of sideslip angle seems to have very little effect on the trailing edge vortices.
The right trailing edge vortex limit does not undergo any significant changes with .
The left trailing edge vortex limit moves slightly towards the centre to reach
= 0.27 of the trunk for positive angles whereas for negative angles it stays close to
= 0.3. This can be another indication that the flow structures for the two extreme
Figure 5-13: Effect of sideslip on the trailing edge vortices and the centre limit
49
5 - Results
left vortex of the contra-rotating pair was often not observable precisely enough to study
its size and position. The right one was always observable since it was often bigger than
the other one.
5.2.1 Effects of wing axial location
5.2.1.1 Effects on the Contra-rotating vortices
Figure 5-14 shows the effect of changing the wing axial position on the size of the right
one of the pair of contra-rotating vortices. The general trend is an increase in size as the
wing is moved rearwards until = 0.5 and then a stabilisation. For = 0.20 & 0.41, at
the wings rearmost positions, the vortex is even bigger than without the wing. At
= 0.85, the vortexs size remains significantly bellow the value obtained without the
wing.
Figure 5-14: Effect of wing axial position on the right contra-rotating vortexs size
Figure 5-15 shows the effect of changing the wing axial position on the position of the
centre of the right sided vortex within the contra-rotating pair. The vortex seems to
move down slightly and go away from the centre line as the wing is moved rearwards
until
50
= 0.5. From
5 - Results
lowest location, the vortexs centres lateral position does not follow this trend and
seems to stay close to the value obtained without the wing.
At every wing location, the vortex centre is close to the location it is without the wing.
Figure 5-15: Effect of wing axial position on the right contra-rotating vortex position
Even though it was not possible to precisely determine the size and location of the left
vortex, the tests showed that the left vortex was following the same trend as the right
one. As the wing was moved forwards, the left vortex became smaller and its centre
seemed to move up and towards the centre line.
5.2.1.2 Effects on the flow over the trunk
Figure 5-16 shows the effect of wing axial location on the reattachment line position. It
is clear that as the wing is moved forward, the flow reattaches sooner. It can also be
noted that for most forward positions of the wing, the flow reattaches to the backlight
The addition of the wing makes the flow reattach sooner than without it, whatever its
position.
51
5 - Results
Figure 5-16: Effect of wing axial position on the reattachment line position
Figure 5-17 shows the effect of wing axial position on the trailing edge vortices and the
centre limit. Changes in wing axial location seems to have very little effect on the
trailing edge vortices limits. These limits remain very close to their location when no
wing is used.
The general trend for the centre limit position is to slightly move left as the wing is
moved rearwards. For almost all the positions, the centre limit is located on the left of
the centre line. Again, the results obtained with the wing on are very close to that
obtained without the wing.
52
5 - Results
Figure 5-17: Effect of wing axial position on the trailing edge vortices and the centre limit
right one of the pair of contra-rotating vortices. At = 0.55 & 0.97 the right vortexs
size first stays stable above the value obtained without the wing as the wing is moved
upwards and then decreases as the wing is moved further upwards. At
= 0.15 & 0.13, the vortexs size decreases as the wing height is increased.
At every wing location, the vortex centre is close to the location it is without the wing.
53
5 - Results
Figure 5-18: Effect of wing height on the right contra-rotating vortexs size
Figure 5-19 shows the effect of changing the wing vertical position on the position of
the centre of the right sided vortex within the contra-rotating pair. At the wings
rearmost positions = 0.55 & 0.97 and as the wing is moved higher, the right vortexs
centre first stays at its in initial location but when the wing reaches its highest position
the vortex centre slightly moves right. At = 0.15 & 0.13, the vortexs centre also
stays more or less at the same axial location, however at the wings lowest position, the
vortex centre moved away from the centre line.
54
5 - Results
Figure 5-19: Effect of wing height on the right contra-rotating vortex position
Although it was not possible to precisely determine the size and location of the left
vortex, the tests showed that at = 0.55 & 0.97, the wings height had an influence on
the balance of the contra-rotating vortices. At these axial positions, at low wing heights,
the right vortex was slightly dominant whereas at larger wing heights, the left vortex
was slightly dominant. This was not observed at = 0.15 & 0.13. This may explain
why the curves for = 0.15 & 0.13 have different trends than the curves for
= 0.15 & 0.13 the flow reattaches on the backlight for all the wing vertical
= 0.55 & 0.97 the flow is not influenced by the wings vertical position.
The addition of the wing makes the flow reattach sooner than without it, whatever its
position.
55
5 - Results
Figure 5-20: Effect of wing axial position on the reattachment line position
Figure 5-21 shows the effect of wing axial position on the trailing edge vortices and the
centre limit. With increasing wing height, the right trailing edge vortex limit slightly
moves to the right and the left trailing edge vortex limit slightly moves to the left. The
centre limit moves from the left and towards the centre line with increasing wing height.
The results obtained with the wing on show little difference than that obtained with the
wing off.
56
5 - Results
Figure 5-21: Effect of wing height on the trailing edge vortices and the centre limit
general trend is an increase in vortex size as the wing is moved rearwards until = 0.6
and then a stabilisation at the value obtained with the wing off.
57
5 - Results
Figure 5-22: Effect of wing axial position on the right contra-rotating vortexs size with sideslip
Figure 5-23 shows the effect of changing the wings axial position on the position of the
centre of the right sided vortex within the contra-rotating pair with a 2 sideslip angle.
The vortex seems to move down and away from the centre line as the wing is moved
rearwards until = 0.6. From = 0.6, the vortex adopts its location with the wing off.
58
5 - Results
Figure 5-23: Effect of wing axial position on the right contra-rotating vortex position with sideslip
59
5 - Results
Figure 5-24: Effect of wing axial position on the reattachment line position with sideslip
Figure 5-25 shows the effect of wing axial position on the trailing edge vortices and the
centre limit with a 2 sideslip angle. Changes in wing axial location seems to have very
little effect on the trailing edge vortices limits. These limits remain very close to their
location when no wing is used.
The general trend for the centre limit position is to move left as the wing is moved
rearwards. Again, the results obtained with the wing on are very close to that obtained
without the wing.
60
5 - Results
Figure 5-25: Effect of wing axial position on the right contra-rotating vortex position with sideslip
As described in Section 4.4, the wing was placed in the wake of the model at locations
in the range 0.15 0.97 and 0.30 1.17 at an angle of attack of 5 and the
aerodynamic forces were measured.
5.3.1 Effect of wing axial location
5.3.1.1 Effect on the wing
Figure 5-26 shows the variation in wing lift coefficient with axial location in the wake.
At all vertical location tested, the wing lift coefficient is decreasing (the downforce is
increasing) as the wing is moved rearwards until the location = 0.40 above which the
lift coefficient is constant.
61
5 - Results
Figure 5-27 shows the variation in wing drag coefficient with wing axial location. At
= 0.30, 0.74, 0.95 & 1.17 the wing drag is clearly decreasing as the wing is moved
The wing drag is decreasing until = 0.70 and then increasing again. It is also to be
noted that the values of drag are higher for low values of / than for high values of
/.
62
5 - Results
Figure 5-28 shows the variation in car lift coefficient increment with wing axial location
in the wake. The lift is clearly decreasing as the wing is moved rearwards at all vertical
locations.
Figure 5-29 shows the variation in car drag coefficient increment with wing axial
location in the wake. The car drag increment is stable as / increases but increases
from
63
5 - Results
Figure 5-28: Car lift coefficient increment variation with wing axial location
Figure 5-29: Car drag coefficient increment variation with wing axial location
Figure 5-30 shows the variation in the ratio of car lift over car drag with wing axial
location. The car lift over drag ratio decreases as the wing is moved rearwards at all
vertical locations.
64
5 - Results
Figure 5-30: Car lift over drag ratio variation with axial location
between = 0.4 and = 0.4 which does not appear at = 0.69 and = 0.97.
Figure 5-32 shows the variation in wing drag coefficient with wing vertical location. At
= 0.15, = 0.13 and = 0.41, an increase in wing height initially results in a drag
to increase, then decrease for higher positions and finally increase again for even higher
locations whereas at = 0.69 and = 0.97, the drag only increases with wing height. It
is also to be noted that at
higher for low values of / than for high values of /. At = 0.69 and = 0.97, the
values of drag increase when / in decreased lower than 0.5.
65
5 - Results
Figure 5-31: Wing lift coefficient variation with wing vertical location
Figure 5-32: Wing drag coefficient variation with wing vertical location
66
5 - Results
Figure 5-34 shows the variation in wing drag coefficient increment with wing vertical
location. The general trend is an increase in drag as the wing vertical location is
increased.
Figure 5-33: Car lift coefficient increment variation with wing vertical location
67
5 - Results
Figure 5-34: Car drag coefficient increment variation with wing vertical location
Figure 5-35 shows the variation in the ratio of car lift over drag with wing vertical
location. The ratio of car lift over drag decreases as the wing is moved upwards and
stabilises from
68
5 - Results
Figure 5-35: Car lift over drag ratio variation with wing vertical location
Figure 5-36 shows the surface flow over the rear window and boot of the model
supported by the strut. At point (a) the flow direction is towards the edges, confirming
the presence of the trailing edge vortices.
The line of separation is at the upper edge of the backlight, as shown by point (b),
however at the centre of this edge, the flow is not separated (c).
At point (d) the flow is attached and goes downwind as opposed to the case without a
strut where some reverse flow was observed on the same area. There are no contrarotating vortices.
At point (e) the flow near the centre line turns towards the edges whereas at point (f) the
flow turns towards the centre. At point (g) the mixture was accumulated so it is difficult
to tell the flow direction however given the flow direction at points (e) and (f), it is
probable that there is some reverse flow.
69
5 - Results
Figure 5-36: Oil flow over the rear of the model with the strut
70
6 - Discussion of results
6 Discussion of results
6.1 Structure of the wake
The results presented in Section 5.1.1 showed that the structure of the wake is
essentially the same as that described by Nouzawa (6) and explained in Section 2.1.2.3.
It is mainly composed of two trailing edge vortices arising from the C-pillar and an arch
vortex in between (Figure 6-1). The main difference compared to the structure described
by Nouzawa relates to the arch vortex which is located on the backlight whereas
Nouzawa located it on the trunk. This difference is assumed to be due to the fact that the
model used for this thesis project was not the same than the one used by Nouzawa
whose model featured a smaller backlight, a larger trunk and a higher backlight angle.
71
6 - Discussion of results
the normal mixture used for all the other tests, by placing the dots on the backlight.
They show the contra-rotating vortices.
It can be seen that the dots placed near the roofs side edge dont stop at the backlights
top edge but keep flowing on the backlight and start to draw the contra-rotating vortices
patterns (a) whereas the dots placed lower on the side of the car go on the backlight but
are then pushed away by the trailing edge vortices (b). This shows that even though the
flow coming from the roof separates at the backlights top edge, it remains attached near
the backlights top corners. Then the air flows on the backlight, curves inwards and
when the flows from the two sides meet, the air goes up and produces some reverse
flow, thus forming the arch vortex (Figure 6-3). The base of the arch vortex is actually
the two contra-rotating vortices. The arch vortex is probably causing the flow to reattach
on the trunk by curving downwards the flow coming from the roof.
72
6 - Discussion of results
The attached flow coming from the backlights top corners will be referred to as corner
flow.
6.1.2 Flow over the trunk
Figure 6-4 shows the surface flow pattern obtained without sideslip and without the
wing in the G13 wind tunnel. The dots placed on the trunk and far enough from the
edges flowed towards the centre line. One of the reasons for that may be that the flow
coming from the roof of the car curves inwards (Figure 6-5). However it is unlikely that
the flow from the roof would curve inward to such an extent. It is also possible that the
flow coming from the sides is deflected onto the trunk by the trailing edge vortices
(Figure 6-6). It is more likely a combination of those two possible phenomena that
causes the dots to flow inwards. The deflected flow coming from the sides would also
help the flow coming from the roof reattach on the trunk. This flow will be referred to
as side flow.
73
6 - Discussion of results
74
6 - Discussion of results
This happens because as the sideslip from one side increases, the pressure on this side
also increases and the corner flow is favoured whereas the pressure on the other side is
decreased and the corner flow tends to be reduced. This makes the one of the contrarotating vortices which is formed by the favoured corner flow become dominant while
the other one becomes smaller (Figure 6-8).
It was also observed that without sideslip, the right vortex is already dominant and a 2
sideslip angle was necessary for the two contra-rotating vortices to be the same size.
The reason for this is not yet understood but this was not observed in the G13 wind
tunnel so it is possible that it was due to a poor alignment of the model in the
Weybridge wind tunnel even though this had been checked carefully. More experiments
would be needed to find out the cause of this.
75
6 - Discussion of results
Figure 6-8: Formation of the contra-rotating vortices with a positive sideslip angle
The centre limit is also affected by sideslip and moves away from the windward side
(Section 5.1.2.3). Figure 6-6 showed how the side flow affects the flow over the trunk.
With sideslip, the difference in pressure on the two sides made the flow coming from
one side be deflected by the trailing edge vortices more easily whereas the flow from
the other side will not be deflected as easily as before. As a result, the flow over the
trunk will not be symmetric and the centre limit will move towards the side where the
flow is less deflected (Figure 6-9). This confirms that the flow on the backlight is not
linked to the flow in the separated area and explains why for negative angles, the centre
limit kept moving to the left even though the contra-rotating vortices sizes had stopped
changing as described in Section 5.1.2.3.
Figure 6-9: Flow over the trunk with a positive sideslip angle
76
6 - Discussion of results
The trailing edge vortex limits were not affected by sideslip. However it would be rather
surprising that the trailing edge vortices should not be affected by sideslip. It would be
necessary to investigate the way the trailing edge vortices form in more details and how
they are affected by sideslip by using 3D flow visualisation such as tuft wand
visualisation or smoke visualisation.
77
6 - Discussion of results
To conclude, the strut cannot be used when studying the structure of the wake as it
significantly affects the contra-rotating vortices structure and therefore the results
obtained using flow visualisation are very different from what actually happens in real
conditions.
The fact that the strut has such a big effect on the flow structure also raises other
question such as whether or not it has an effect on the forces measured. The effect of the
strut on lift and drag measurements should be investigated in more details by doing
force measurements with and without the strut, without the moving belt in both cases.
78
6 - Discussion of results
before the wing. Figure 6-13 shows the corner flow and the formation of the contrarotating vortices when the wing has been moved forwards.
Figure 6-11: Corner flow when the wing is close to the backlight
The trailing edge vortex limits and the centre limit stay at the same locations for all
wing axial positions. The stream lines on the trunk were not affected by changes in
wing axial position. This shows again that the arch vortex structure and the flow over
the trunk are interdependent.
6.4.2 Effect on downforce
Figure 5-26 show that the wing downforce increased (the wing lift decreased) as the
wing was moved rearwards for low values of / (Figure 6-13). This indicates that the
flow velocity over the wing increases as the wing is moved rearwards which shows that
the influence of the side flow on the wing progressively increases as the wing is moved
rearwards. From
starts to stall. As wings angle of attack was 5, the distance between the wing and the
trunk is larger at the wings trailing edge than at the wings leading edge; this structure
acts like a diffuser and therefore the flow is more likely to separate.
79
6 - Discussion of results
Figure 5-28 showed that the car downforce increases as the wing is moved rearward
even after the location = 0.4 from which the wing downforce stops increasing. The
car downforce increment is therefore not only linked to the wing downforce. This could
be due to wing/body interactions (Section 2.2.2.1.1). When the wing is located near the
rear edge of the trunk, the flow from under the model is deflected upwards and its
velocity is increased, thus increasing the bodys downforce independently of the wings
own downforce.
6.4.3 Effect on drag
Figure 5-27 showed that the wing drag is decreasing until = 0.4 and then increasing
again. The values of the drag were higher when the wing was close to the backlight (low
values of /) than over the trunk. This is surprising since the velocity is assumed to be
higher over the trunk. This could be due to some other interactions between the wing
and the flow structure on the backlight that would artificially increase the value
recorded by the load cell.
At higher values of / , the increase in drag as the wing is moved rearwards could be
due to an increase in velocity next to the rear edge of the trunk. However the lift does
80
6 - Discussion of results
not increase with the drag, which means that the flow separates. As explained before,
when the wing is close to the trunk, it acts like a diffuser which favours flow separation.
Figure 5-29 shows that the total car drag increment stayed constant for low values of
/ but increased for values of / higher than 0.4. As the wing is moved rearwards,
the model wake size increases and leads to an increase in car drag (Figure 6-14). The
constant drag obtained for low values of / is due to the decrease in wing drag when
the wing is very close to the backlight and is moved rearwards. For low values of / ,
this would nullify the increase in car drag due to the wake size.
81
6 - Discussion of results
contra-rotating vortices structure became less wide with increasing wing height but its
length did not change, as a result the vortices centres moved towards the centre line. At
= 0.55 & 0.97 the whole contra-rotating vortices structures size did not significantly
change with increasing wing height but the left vortex became slightly dominant which
explains the right vortexs decrease in size. The vortices centres both moved away from
the centre line as the left vortex became dominant.
This behaviour is still not clearly understood and no explanation was found for the
phenomena described above based on the data available.
At every wing position, even the lowest, the trailing edge vortices limits were clearly
observable and the air flowing on the side of the trunk was still pushed away from the
trunk which suggests that the trailing edge vortices are always under the wing.
The flow over the trunk was not significantly affected by changes in wing height. This
again shows that the arch vortex structure and the flow over the trunk are interdependent.
6.5.2 Effect on downforce
Figure 5-31 shows that the downforce decreases as the wing is moved upwards which
again can be explained by the fact that as the wing moves away from the model, the
flow velocity over it increases. At high /, the downforce increases more slowly and
almost stabilises which indicates that at these locations, the wing is far enough from the
model for the flow velocity to be effectively uniform (Figure 6-15).
82
6 - Discussion of results
Figure 5-33 shows that the car downforce increment increases as the wing is moved
upwards, which is due to the higher flow velocity away from the model. At the rearmost
locations, the car downforce starts to decrease when the wing is moved higher than
= 0.8. This may again be due to the wing/body interactions described in Section
2.2.2.1.1. For the rearmost locations of the wing and when it is close to the model, the
flow coming from under the model is deflected upwards thus increasing its speed and
the overall car downforce. This is not happening when the wing is too high above the
model (Figure 6-16). Therefore when the wing is close to the rear edge of the model and
moved upwards until = 0.8, the flow velocity increases and the wing body interaction
described in Figure 6-16 decreases; these are two opposite effects but their combined
effect is an increase in the overall car downforce. When the wing is moved upwards
above = 0.8, the flow velocity stays the same but the wing body interaction described
in Figure 6-16 keeps decreasing and the overall car downforce starts to decrease.
83
6 - Discussion of results
= 0.15,
= 0.13 and
sharp increase and then a sharp decrease as the wing height is increased (from = 0.3
to = 0.8) and then slowly increases again as the wing is moved further up. The drag is
a lot higher for low values of / than for high values. These results, and particularly
the high value of the drag experienced, show again that there may be some interference
between the flow in the separated area and the wing. This interference seems to stop
from = 0.8 and higher which indicates that at these positions, the wing is far enough
from the backlights near wake structure. The slow increase in drag as the wing is
moved upwards from the location = 0.8 is due to the increase in flow velocity with
the distance from the model.
At = 0.69 and = 0.97, the curves are smoother which confirms that there is not any
interference when the wing is far enough from the backlights near wake structure. At
these axial locations, the drag first decreases and then increases with wing height. This
is probably due to the fact that when the wing is very close to the trunk, the flow
velocity between it and the model increases, this is called ground effect (see Section
2.2.2.1.3). This and the fact that, as explained before, the angle of incidence of the wing
make it act like a diffuser when it is close to the trunk, make the wing stall, thus
increasing the drag. Consequently, as the wing is moved away from the trunk, the flow
around the wing progressively reattaches and as a result the drag decreases; as the wing
is moved further up the flow velocity increases to its freestream value and the drag
increases to reflect this.
Figure 5-34 showed that the car drag increment increases as the wing is moved upwards.
This is due to the fact that that the models wake size increases as the wing is moved
upwards therefore increasing the drag (Figure 6-17).
84
6 - Discussion of results
for rearmost positions of the wing. The most efficient height seems to be = 0.8.
85
6 - Discussion of results
Fishers experiments conditions, a sideslip angle has been set to make the right vortex
dominant while investigating the effects of wing axial position.
However the phenomenon observed by Fisher was still not observed and even though
the right vortex was dominant, the left one remained even at the wings foremost
position. As described in Section 5.2.3, the contra-rotating vortices structure and the
overall wake structure reacted to changes in wing axial position very similarly as
without sideslip. As the wing moved forwards, the contra-rotating vortices both became
smaller, moved up and got closer to the centre line while the centre limit moved left.
The trailing edge vortex limits were still not affected by the wing position.
The reasons why Fisher observed only one vortex at the wings foremost position
remain unexplained.
86
The separated flow from the rear end of the roof was curved downwards by the arch
vortex and reattached right after it.
The flow from the sides was sucked by the trailing edge vortices and deflected onto the
trunk.
87
and moved up. In the Weybridge wind tunnel, the right vortex was observed to be
already dominant without sideslip whereas in the G13 wind tunnel the contra-rotating
vortices were the same size when the sideslip angle was set to 0. It was not clear if this
was due to the difference between the wind tunnels or to the model alignment with the
tunnels.
The pressure difference between the two sides also resulted in the side flow from one
side being deflected more easily than the flow from the other side. As a result, the line
on the trunk where the flows from the two sides meet, the centre limit, moved away
from the sideslip side.
Surprisingly, the trailing edge vortices size and position did not seem to be affected by
sideslip.
The flow over the trunk was not affected by changes in wing axial or vertical location.
The trailing edge vortices remained under the wing, even when the wing was at its
lowest positions.
88
7.4 Effect of the wing on the overall vehicle lift and drag
characteristics
The vehicles drag increased as the wing was moved rearwards and upwards relative to
the car. The downforce also increased as the wing was moved rearwards and upwards
except at the wings rearmost position for which the downforce reached a maximum
value at 80% of the trunk to roof distance above the trunk.
Fisher also observed that the flow did not reattach on the trunk whereas during this
studys tests, the flow always reattached right after the arch vortex.
Fisher observed that both the downforce and drag increased as the wing was moved
rearwards and upwards whereas during this study, the maximum downforce experienced
by the car was reached at a moderate wing height at the rearmost wing location. Such
differences remained unexplained
89
The effect of the strut on the measured lift and drag should be investigated to determine
the reliability of the obtained results. Since the moving belt is used to improve the
results by artificially reproducing the relative velocity between the car and the ground
and considering that the strut used with the moving belt possibly reduces the reliability
of the measurements, a comparison between the results obtained with and without the
moving belt would be necessary to know the best experimental arrangement.
This study only looked at one type of rear-end vehicle shape. The wake structure greatly
depends on the cars geometry therefore the influence of different rear-ends should be
studied.
90
Cranfield University
Aerospace Dynamics
References
1. Fisher, K. S. Aerodynamic Effectiveness of Rear Wings for Saloon Race Cars.
Cranfield University. 2004. MSc Thesis Project.
2. Barnard, R. H. Road Vehicle Aerodynamic Design: An Introduction. s.l. : Longman,
1996.
3. Hucho, W-H. The Aerodynamic Drag of Cars - Current Understanding, Unresolved
Problems, and Future Prospects. New York : G.Sovran, T. Morel and W.T. Mason, Jr.,
Plenum Press, 1978.
4. Jenkins, L. An Experimmental Investigation of the Flow Over the Rear End of a
Notchback Automobile Configuration. SAE Paper 2000-01-0489, 2000.
5. Carr, G. W. Influence of Rear Body Shape on the Aerodynamic Characteristics of
Saloon Cars. MIRA Report 1974/2, 1974.
6. Nouzawa, T., et al. Analysis of the Wake Pattern for Reducing Aerodynamic Drag of
Notchback Models. SAE Paper 900318, 1990.
7. Hucho, W. H. Aerodynamic Drag of Passenger Cars, Aerodynamics of Road
Vehicles. s.l. : edited by W. H. Hucho, SAE, Pennsylvania, 1989.
8. Morel, T. Effects of Base Slant on Flow in the Near Wake of an Axisymmetric
Cylinder. Aeronautical Quarterly. May 1980, Vol. 31.
9. Goetz, H.; Society of Automotive Engineers Crosswind Facilities and Procedures
st. Crosswing Facilities and Procedures. [d.] H. Goetz. s.l. : SAE International, 1995.
10. McBeath, S. Competition Car Dowforce: A Practical Handbook. s.l. : Haynes
Publishing, 1998.
11. Katz, J. Race Car Aerodynamics: Designing for Speed. s.l. : Robert Bentley, 1995.
12. Katz, J. et Dykstra, L. Effect of Wing/Body Interaction on the Aerodynamics of
Two Generic Racing Cars. SAE Paper 920349, 1992.
13. McBeath, S. Competition Car Aerodynamics: A practical Handbook. s.l. : Haynes
Publishing, 2006.
14. Gross, D. S. et Sekscienski, W. S. Some Problems Concerning Wind Tunnel
Testing of Automotive Vehicles. SAE Paper 660385, 1966.
15. Barlow, J. B., Rae, W. H. et Pope, A. Low-Speed Wind Tunnel Testing. [d.] John
Wiley. 3rd edition. 1999.
91
Cranfield University
Aerospace Dynamics
92