ACKNOWLEDGEMENTS by mikeholy

VIEWS: 4 PAGES: 52

									ACKNOWLEDGEMENTS
The Author would like to thank the following:

      Dr. Matt Stickland, for project guidance
      Mr. Chris Cameron, for technical assistance
      Mr. Alistair Duff, for technical assistance
      Mr. Andrew Crockett, for strain gauge assistance
      Mr. Jim Doherty, for supplies assistance




                                                          1
ABSTRACT
This paper studies the aerodynamic properties of a Range Rover L319 wing
mirror, both through experimental means in a wind tunnel and through the
use of CFD software. The results from each method are compared and
consequently the accuracy of the CFD analysis is validated. From there, two
design alterations are applied to the CFD model in an aim to reduce both the
wing mirror drag and the sources of aerodynamically created noise on the
wing mirror surfaces. The design alterations are proven to have a beneficial
impact on the wing mirrors aerodynamic performance with regards to many
features and therefore the usefulness of CFD software in engineering design
is underlined. A simplified CFD analysis of the flow over the car is also
performed and therefore a better understanding of the flow over the A-Pillar
region is achieved. The possibilities of further optimization of the wing mirror
design and potential alterations to the A-Pillar geometry are then discussed,
along with the potential for further study with regards to aeroacoustic
modelling.




                                                                                   2
Contents
ACKNOWLEDGEMENTS ........................................................................................ 1
ABSTRACT ............................................................................................................. 2
NOMENCLATURE................................................................................................... 5
1. INTRODUCTION ................................................................................................. 6
2. THE WING MIRROR ........................................................................................... 7
   2.1 HISTORY ....................................................................................................... 7
   2.2 DESIGN IMPROVEMENT .............................................................................. 8
3. THE A-PILLAR..................................................................................................... 9
4. AERODYNAMIC NOISE .................................................................................... 10
   4.1 SIGNIFICANCE & SOURCES ...................................................................... 10
   4.2 NOISE REDUCTION .................................................................................... 11
5 DEBRIS/DROPLET SHEDDING ......................................................................... 11
6. COMPUTATIONAL FLUID DYNAMICS ............................................................. 12
   6.1 FUNCTION................................................................................................... 12
   6.2HISTORY ...................................................................................................... 13
   6.3 ENGINEERING APPLICATIONS ................................................................. 13
7. WIND TUNNEL TESTING.................................................................................. 14
   7.1 CLOSED RETURN....................................................................................... 14
   7.2 TESTING METHOD ..................................................................................... 15
   7.3 WING MIRROR & BAR CONNECTION........................................................ 15
   7.4 RIGGING TO DATA ACQUISITION ............................................................. 16
   7.5 DATA ACQUISITION SOFTWARE .............................................................. 17
   7.6 STRAIN GAUGE BAR CALIBRATION ......................................................... 17
   7.8 WIND TUNNEL SAFETY & OPERATING PROCEDURE ............................. 22
8. CFD – WING MIRROR ON FLAT PLATE .......................................................... 24
   8.1 GAMBIT: MODELLING ................................................................................ 24
   8.2 GAMBIT: MESHING & EXPORTING ............................................................ 26
   8.3 FLUENT: ANALYSIS SETUP ....................................................................... 28
   8.4 FLUENT: SOLVING ..................................................................................... 29
9. WIND TUNNEL RESULTS & CFD VALIATION.................................................. 31
   9.1 RESULTS COMPARISON ........................................................................... 31
   9.2 ERRORS ...................................................................................................... 32
   9.3 VALIDATION ................................................................................................ 32

                                                                                                                      3
14. DESIGN PERFORMANCE ASSESMENT ........................................................ 34
   14.1 PRESSURE DRAG .................................................................................... 34
   14.2 AEROACOUSTICS .................................................................................... 35
15. OPTIMIZATION (I) ........................................................................................... 36
   15.1 MODELLING .............................................................................................. 36
   15.2 RESULTS................................................................................................... 37
16. OPTIMIZATION (II) .......................................................................................... 39
   16.1 MODELLING .............................................................................................. 39
   16.2 RESULTS................................................................................................... 40
17. A-PILLAR MODELLING ................................................................................... 42
   17.1 A-PILLAR INFLUENCE .............................................................................. 42
   17.2 GAMBIT: MODELLING............................................................................... 42
   17.3 GAMBIT: MESHING & SOLVING ............................................................... 43
   17.4 FLAT PLAT & A-PILLAR COMPARISON ................................................... 44
   17.5 FLOW OVER THE A-PILLAR ..................................................................... 44
   17.6 WATER DROPLET & DEBRIS SHEDDING ............................................... 45
   17.7AERO ACOUSTIC NOISE SOURCES ........................................................ 46
18. DISCUSSION AND FURTHER STUDY ........................................................... 48
   18.1 WIND TUNNEL TESTING .......................................................................... 48
   18.2 OPTIMIZATION EFFORTS ........................................................................ 48
   18.3 A-PILLAR ................................................................................................... 49
   18.4 AEROACOUSTICS .................................................................................... 50
   19. CONCLUSION ............................................................................................. 50
REFERENCES ...................................................................................................... 52




                                                                                                                     4
NOMENCLATURE
Air Density (kg/m3) ………………………………………………………….........ρ

Area (m2) ……………………………………………………………………...…...A

Constant …………………………………………………………………………….k

Drag Coefficient………………………………………..…………………………CD

Drag Force (N)……………………………………………………………………D

Lift Coefficient …………………………………………………………………….CL

Normal Force on Strain Gauge Bar (N)………………………………………...Fs

Skin Friction Drag Coefficient ………………………………………………… Cd0

Velocity (m/s)……………………………………………….……………………..V




                                                       5
1. INTRODUCTION
In automotive engineering and design the application of wing mirrors can
affect the performance of motor vehicles in numerous ways. The most
significant effects being: the vehicle aerodynamics, cabin comfort and driver
and passenger safety.

The wing mirror in most motor vehicle designs is, in essence, an exposed
bluff body, and thus produces high levels of pressure drag. A wing mirror
typically represents about 2.5% of the vehicle frontal area but has been
found to contribute up to 5% of the total vehicle drag [1] which can be
considered significant. Furthermore, a typical modern production vehicle
                                                              [2]
usually has a drag coefficient value of around 0.3 to 0.5           , and the wing
mirrors of the vehicle can make a contribution to this value at around the
order of 0.01[3].

This drag contribution has a detrimental effect on vehicle acceleration and
top speed, with the most noticeable reduction in acceleration occurring at
speeds of around 60 miles per hour and higher (motorway cruising speed).
For example, studies have shown that a particular vehicle with an overall
drag coefficient of 0.45 can reach a speed of 75 miles per hour in 20
seconds. However, if this value of drag coefficient is improved to 0.25, the
                                                      [4]
time taken to reach 75mph is reduced by 3 seconds           , which is a noticeable
improvement in the vehicle’s acceleration performance. Due to the
detrimental effects of drag on vehicles at motorway cruising speeds, another
negative impact on the vehicle’s performance is fuel economy.

Wing mirrors can also influence cabin comfort for passengers and drivers
within the vehicle due to the aeroacoustic effects produced by the airflow
over the A-Pillar and wing mirror. The aerodynamic noise created is more
significant now in modern cars due to the mechanical noise present within
the cabin being reduced as a result of enhanced quality engines.

                                                                                     6
Poor wing mirror/A-Pillar design can also result in debris and water droplets
being shed from the wing mirror onto the front side windows resulting in
impairment of visibility for the driver and therefore a reduction in vehicle
safety.

This report will study the aerodynamic characteristics of a Range Rover L319
wing mirror (Figure 1) through wind tunnel testing as part of a small A-Pillar
configuration. The wing mirror will then be analyzed both in isolation on a flat
plate and connected to the Range Rover car body through use of
Computational Fluid Dynamics software.




                             Figure 1 - L319 Wing Mirror


The results from wind tunnel testing and CFD modelling on the flat plate will
then be compared to validate the accuracy of the CFD results. With the
results validated the software will then be utilised in design optimization
efforts, with the aim of reducing the wing mirror pressure drag. Changes will
also be made in an effort to minimize the sources of aerodynamically created
noise on the surfaces of the wing mirror.

The flow over the A-Pillar will also be studied to better understand the nature
of the flow over this region and to assess whether applying and testing the
design optimizations on the wing mirror in isolation is a valid approach to
design improvement.


2. THE WING MIRROR

2.1 HISTORY

                                                                                 7
The shape and aesthetics of wing mirrors has evolved throughout the years
of automotive design. In the early periods of automotive development the
wing mirror was typically a flat, bluff body shape and was one of many
components on a vehicle to be exposed to the free stream along with others
such as headlights and wheel fairings. As the aerodynamics of motor
vehicles became better understood the shape of most wing mirrors were
adapted to be more streamlined. However, the majority of car designs still
feature the wing mirror exposed to the free stream whilst other components
such as the headlights and wheel fairings have been sunk into the body work
to reduce drag.

2.2 DESIGN IMPROVEMENT
A select number of car manufacturers are addressing the issue of wing
mirror drag by replacing them with small mounted cameras. Images taken by
these cameras are then relayed on small screens on the inside of the car
door. This setup may improve the vehicle’s performance with regards to
acceleration and top speed. However, the beneficial effects on fuel economy
may be questionable due to the extra fuel consumption required for the
operation of both the camera and screen.

This setup may be more effective when applied to hybrid cars such as the
Toyota 1/X concept car which omits wing mirrors from its design; as seen in
Figure 2.




                      Figure 2 - Toyota 1/X (omitted wing mirrors)


It may also prove effective in future applications if the use of hydrogen fuel
cell power plants are utilised in the automotive industry.




                                                                                 8
At present, the best practise for automotive design remains to be the
improvement and optimization of the wing mirror design. Some methods
exercised in automotive design with the aim of improving wing mirror
performance are:

      Streamlining the mirror shape, resulting in lowered pressure drag by
       minimizing flow stagnation on the leading faces and lowering the
       amount of recirculation downstream of the mirror. Some designs tend
       to resemble a cross section similar to a typical aerofoil, however, care
       must be taken to limit the lift force created as this can create vortex
       shedding downstream of the mirror and thus increase drag, when
       considering:

                                    Cd  C D 0  kCL
                                                          2
                                                                …E1

      Reducing the wing mirror frontal area, which should reduce the drag ,
       given that:
                                         1
                                    D     . . A.V 2Cd         … E2
                                         2

       This should work in efforts to reduce the pressure drag. However,
       using a small mirror can have a negative effect on the cars
       performance with regards to safety due to reduced visibility for the
       driver.

      Smoothing the surfaces of the mirror housing to minimise any possible
       skin friction drag created, although the contribution of skin friction drag
       towards the total overall drag is a great deal smaller than that of
       pressure drag.
      To smooth out any small gaps or exposed features that may
       encourage localised stagnation and increase pressure drag: such as
       joints in the housing and areas around the hinges between the mirror
       housing and the mounting


3. THE A-PILLAR



                                                                                 9
The A-Pillar is defined as the region where the windshield joins the side
window and is a source of large vortex shedding which consequently
produces sizeable levels of induced drag. This flow behaviour can be
attributed to the accelerated flow over the A-Pillar meeting the lower velocity
flow travelling along the side of the car. This difference in flow velocity - and
thus static pressure - results in vortex shedding similar to that found on the
wing tips of aeroplanes. The presence of wing mirrors mounted on the A-
Pillar encourages this vortex shedding due to the wing mirror decelerating
the flow in this region and thus increasing the difference in velocity, and
consequently pressure (as shown in Figure 3).




      Figure 3 - A-Pillar Vortex shedding (www.exa.com/pages/pflow/pflow_main.html)


Intelligent design and placement of the wing mirror can help reduce this
vortex shedding and therefore can have a significant influence on the overall
drag characteristics of a vehicle.


4. AERODYNAMIC NOISE

4.1 SIGNIFICANCE & SOURCES
As aforementioned, the significance of aerodynamically produced noise has
increased in recent times due to the development of quieter power plants
and transmission. Furthermore, with the introduction of hybrid vehicles and
the prospect of completely silent fuel cell engines, the importance of limiting
any aerodynamic noise generation will prove even greater.



                                                                                      10
The creation of aerodynamic noise from wing mirrors results primarily from
the vortex shedding at the A-Pillar location which travels downstream from
the wing mirror and strikes the side window. This vortex shedding on the
window creates pressure fluxes and consequently produces noise which can
be heard from inside the cabin, causing the driver or passenger inside some
degree of discomfort.

Predicting or understanding the aero acoustic characteristics of wing mirrors
can often be extremely difficult due to their complex geometry. It is also
made difficult due to the fact that it is the interaction between the A-Pillar, the
wing mirror and the door window that creates the noise recognisable to the
driver/passenger inside the vehicle. It is for this reason that modelling or
measuring the aerodynamic noise created by the wing mirror alone, will only
help to identify the sources of noise from the wing mirror.

4.2 NOISE REDUCTION
Although predicting, modelling and understanding the aeroacoustic
characteristics of wing mirrors can be difficult, there are some proven
methods of design that help reduce the creation of noise. It has been found
that abrupt changes in surface curvature, at the corners of the mirror casing
for example, can encourage the production of noise through the creation of
span wise pressure fluctuations [5]. For this reason it is advantageous for
designers to keep the surface transitions on the mirror casing as smooth as
possible.


5 DEBRIS/DROPLET SHEDDING
In wet or hostile driving conditions when there is a large amount of water and
debris coming into contact with a vehicle’s surface, problems can arise from
poor wing mirror/A-Pillar Design. The shape of most wing mirrors dictates
that any water droplets or debris which comes into contact with the mirror will
travel along its surface in the downstream direction and will detach at any
sharp edges or angles such as the flat edge on the back of the mirror. For a
poorly designed wing mirror/A-Pillar configuration, this water and debris



                                                                                11
detaching from the mirrors trailing edges can result in the shedding of these
particles onto the front side window.

Water and dirt build-up on the glass can pose a risk to passenger and driver
safety due to the locality of the build-up near the mirror. Water accumulation
in particular can significantly reduce the clarity at which the driver can view
the wing mirrors through the window.

Some automobile manufacturers have devised methods and designs with the
aim of preventing the accumulation of water and debris. One such method is
to alter the shape and design of the mirror mount in such a way that there is
a gap created through the wing mirror mount centre. This results in a wing
mirror design that is, in essence, fork mounted. The rationale behind this
type of design is that it reduces the ‘bluffness’ of the shape and allows high
velocity flow to pass through it, thus resulting in a higher velocity flow aft of
the wing mirror. This higher velocity flow should then, in theory, transport any
water droplets or debris shed from the wing mirror further downstream where
it can then strike the side of the vehicle in a safer location away from the
mirror. This setup can also contribute to reducing the wing mirror’s pressure
drag due the reduction of flow stagnation on the leading face of the support
and consequently the reduction in stream-wise pressure difference across
the support.


6. COMPUTATIONAL FLUID DYNAMICS

6.1 FUNCTION
Computational Fluid Dynamics software (CFD) offers a method of solving
simple to very complex fluid flow problems through the use of computational
processing power. CFD works by dividing up the fluid domain into numerous
smaller control volumes, linked together in a mesh. It then employs the
Navier Stokes equations to relate the fluid properties (such as flow velocity
and pressure) from each control volume to its neighbouring control volume.
These equations are then solved by an iterative process, until a level of
convergence is achieved and thus an accurate solution is provided.



                                                                                    12
The solution obtained (and therefore the behaviour of the flow) can then be
shown in graphic visualisation on screen in numerous types of plots,
displaying the various fluid properties involved.

6.2HISTORY
The application of CFD technology within the engineering industry has
increased rapidly over the last twenty to thirty years; with the science behind
its method having existed long before this time. This increase has been
brought about primarily by the rapid growth and affordability of computational
processing power transferring the software codes from large corporate and
governmental supercomputers in the 1970’s down to the desktop PC’s of
even the smallest firms today.

6.3 ENGINEERING APPLICATIONS
CFD software allows designers to produce and test new designs on a
timescale that is significantly shorter than that of physical prototyping and
testing. The results and visualisations can show the flow characteristics in
localised parts of the design which could not be identified as accurately
through wind tunnel testing. These visualisations can then in turn be used to
identify the strengths and weaknesses of the design and thereafter
amendments and improvements can be applied. The altered design can then
be analyzed to study the influence and effectiveness of any change, and thus
an optimized design can be obtained.

The use of CFD technology can also be applied to the study of aeroacoustics
as the values of noise generation can be directly derived from the equations
of fluid flow. The software can help determine sources of high aerodynamic
noise production, and can be used to predict the nature of sound propagation
in a system.




                                                                                13
7. WIND TUNNEL TESTING

7.1 CLOSED RETURN
For the purpose of physically testing the wing mirror a closed return wind
tunnel with an open working section was used. The diagram in Figure 4
illustrates a typical closed return wind tunnel layout and is similar to the one
used in this study. The system works by continually circulating the air around
the tunnel, by drawing the air through a diffuser after travelling through the
working section.




         Figure 4 - Closed Return Wind Tunnel (http://www.mi.uni-hamburg.de/uploads/)


The fan is typically placed in a smaller cross-sectional area region in the
tunnel to increase the flow velocity over the blades and thus increase the
efficiency. This is also beneficial when considering that the costs associated
with fans is proportional to the diameter squared [6]. Some wind tunnels
employ counter rotating fans to reduce the rotational flow behaviour imparted
on the airflow by the standard single fan type. However, it is more common
to use a single fan and then introduce anti-swirl vanes downstream to reduce
swirl.

The air propelled by the fan then flows through two sets of 90˚ turning vanes
which direct the air into the settling chamber. The airflow then passes
through a contraction cone, which imposes a reduction in cross-sectional


                                                                                        14
area on the flow and thus results in flow acceleration into the working
section.

When the air has passed through the working section it then flows into the
downstream diffuser which is used to decelerate the airflow as quickly as
possible in an aim to recover the static pressure and reduce power losses in
                                                                  [6]
the boundary layer which are proportional to the velocity cubed     . Most wind
tunnels also feature a second diffuser in the section downstream of the fan,
parallel to the working section.

7.2 TESTING METHOD
The purpose of testing the wing mirror in the wind tunnel was to establish its
drag characteristics at speeds ranging from 10 to 60mph, with the aim of
using these results to validate values produced through the use of CFD.

In order to measure the drag force, the use of a small bar with strain gauges
attached to either side was used, rather than employing the under floor
balance method. A diagram of the apparatus is shown in Figure 5.




                             Figure 5 - Strain Gauge Bar


The bar was to be calibrated against load, the wing mirror was then rigidly
attached and the apparatus placed in the wind tunnel for testing.

7.3 WING MIRROR & BAR CONNECTION
Before any preparatory wind tunnel work could begin, a means of rigidly
connecting the bar to the wing mirror had to be developed. The strain gauge
bar provided had three pre-existing holes cut through its top half and so this
provided a starting place for a rigid connection. All that was needed was the


                                                                              15
fabrication of a bracket to utilise these holes and enable connection to the
wing mirrors metal base.

A small triangular shaped rigid metal plate (T-Plate) was manufactured,
which featured three countersunk holes that matched up with the strain
gauge bar holes and two further holes situated in the corners for the
attachment of the plate to the wing mirror base. A small washer plate was
also produced to fit on the other side of the wing mirror fixings to ensure a
completely rigid connection.

When tightly secured, this setup proved to be sufficiently rigid (Figure 6).




               Figure 6 - Wing mirror and Strain gauge bar rigid connection


7.4 RIGGING TO DATA ACQUISITION
To calibrate the strain gauge bar it first had to be wired up correctly to the
data acquisition hardware but before this the strain gauges were tested with
a multi-meter and found to be standard 120Ω strain gauges. The wires were
then connected up to a SCX1-1121 Font Panel as shown in Figure 7.




                                                                                 16
            Figure 7 - Data acquisition front panel, Wheatstone bridge configuration


This setup created two full Wheatstone bridge circuits. One of which sensed
the direct load applied, and the other sensed the moments applied. This
configuration would therefore supply the desired output for which the loading
applied could be related to.

The front panel was then connected to the data acquisition chassis, and the
machine was switched on.

7.5 DATA ACQUISITION SOFTWARE
With the necessary circuit created and connected to the data acquisition
hardware, it was then possible to configure the data acquisition software to
process and display the output from the strain gauge bar.

The data acquisition software used was Labview, version 8.2. The
programme was set up to receive the voltage outputs from the strain gauges,
and then calculate the mean of each output (which were named ‘mean 1’ and
‘mean 2’.

Through this setup, it was then possible to obtain an on-screen display of the
mean voltage output against time. It was then checked that through loading
the strain gauge bar the voltage output displayed on the plot was effected;
and this was indeed the case.

7.6 STRAIN GAUGE BAR CALIBRATION



                                                                                       17
Before the calibration of the bar could begin, the wing mirror and T-Plate
were detached so that it could be loaded in isolation without the effects of
any other components attached. The bar was then firmly clamped at its base
onto a steel shelf, therefore ensuring no movement at the base was possible.

It was decided to load the bar by applying hanging weights at the closest
hole to the strain gauges (Figure 8).




                    Figure 8 - Point of Calibration load application


Hanging weight rods were pushed through the hole so that with the
application of weights onto the hanging weight support, the load would be
applied directly to the area around the hole.

With the hanging weight rod and support in place, the readings from the
strain gauges produced in Labview were observed and recorded. These
values would represent ‘zero’ loading in the calibration process. In essence
the bar was loaded, but the aim was to find the relationship between voltage
difference and the application of load.

With these zero values recorded, the application of weights could begin. This
process was performed through applying load in increments of 200 grams
from 0 to 2kg. At each increment the voltage outputs from both ‘mean 1’ and
‘mean 2’ were observed and recorded in a spreadsheet. This process was
repeated four times so that the mean averages of voltage output at each

                                                                               18
loading stage could be determined, and thus a more accurate relationship
between load and voltage difference could be reached.

With all the desired readings obtained and recorded, the process of
determining the relationship could be carried out. Through examining the
results obtained through both outputs of ‘mean 1’ and ‘mean 2’ it was clear to
see that although both sets of results displayed a similar linear trend for each
of the four loading runs, it was the values from ‘mean 1’ that gave the most
consistent results. For this reason it was decided that the data obtained from
‘mean 1’ would be used as the basis for determining the relationship
between load and voltage difference.

Through manipulation of the spreadsheet used for recording the values, a
trend line of average voltage difference against load was produced. The
software was used to produce an equation to represent this trend line, and
the relationship was found to be:

                        Voltage _ Difference
                 Fs                                       …E3
                          (7.44607  10  6 )

However, this equation represents loading normal to the bar, whereas in
reality the bar would be facing the airflow at an angle (about its length) in the
test section due to the connection between the bar and wing mirror not being
normal to the flow. For this reason some trigonometry would have to be
applied to the force values obtained from this relationship. Inspection of the
wing mirror base indicated that the strain gauge bar would be mounted at an
angle of 14˚ incidence to the flow. From this, it was then determined that the
relationship between the force on the strain gauge bar and the overall drag
force was:


                      D  Fs cos(14 o )
                                                         …E4

Where ‘D’ represents the drag force and ‘Fs’ represents the component of
drag force acting normal to the strain gauge bar.



                                                                               19
With these equations obtained it was then possible to place the strain gauge
bar (with wing mirror attached) into the wind tunnel for testing.

1.2 TESTING SET UP

The first step towards mounting the wing mirror in the wind tunnel test
section was to secure a ground board in place to act as a base support to the
setup. The ground board selected was a large rectangular board of wood
with a large circular hole cut in its centre in which a purpose sized circular
mounting board could be placed.

To support the ground board in place in the test section, two rectangular
wooden beams were slotted into supports at either side of the test section in
the flow-wise direction. On top of these two wooden beams, two rectangular
metal beams were placed running across the test section and clamped in
place. This four beam lattice formed the support for the ground board which
was placed on top and clamped tightly.

The mounting board used was circular in shape with a square hole cut in the
middle through which the wing mirror could be placed. In order to allow the
wing mirror to be clamped rigidly onto the mounting board, a small ‘L’ shaped
metal bracket was screwed tightly onto the underside of the board next to the
hole through which the mirror was to be placed. This made it possible to
clamp the strain gauge bar onto the bracket with the attached wing mirror
protruding through the hole to the upper surface of the mounting board
(Figure 9).




                   Figure 9 - Mounting Board and 'L' clamp apparatus



                                                                                 20
Care had to be taken to ensure that both the strain gauge bar (except at the
base) and the wing mirror were not in direct contact with any other surfaces
as this could restrict the movement of the bar and thus affect the strain
gauge readings. The mounting board setup was then placed into the purpose
made hole in the ground board.

With the solid support structure in place, the imitation A-Pillar around the
wing mirror could be created. This was created simply with thick card and
strong duct tape to make an approximation of the Range Rover A-Pillar.
Some reinforcement was needed under the construction to prevent sagging
and reduce vibration due to the airflow. Care had to again be taken to ensure
that none of the cardboard structure touched the strain gauge bar or the wing
mirror.

This addition of the cardboard structure represented the last addition to the
test model (Figure 10).




                       Figure 10 - Complete Wind Tunnel Model


Some additional work had to be carried out on the underside of the ground
board as the L-Bracket attached to the mounting board was exposed to the
airflow. Therefore, with the wind tunnel operating at speed, the L-Bracket
would create pressure drag and consequently bending in the L-bracket would
occur which would affect the strain gauge readings. To remedy this, a small
wooden board (of equal width to the bracket) was secured upstream by
clamping it to the under floor balance below and taping it at the point of
contact with the ground board. This board would reduce the velocity of the air

                                                                                21
flow travelling over the L-Bracket and therefore reduce the pressure drag
acting on it to a negligible level. The cable from the strain gauge bar was
also wrapped round its securing G-Clamp to ensure that it did not blow in the
wind and therefore pull on the bar.

7.8 WIND TUNNEL SAFETY & OPERATING PROCEDURE
Whilst operating the wind tunnel a lab coat had to be worn, however, eye
protection was not necessary as there were no moving parts in the
experimental apparatus. Before the wind tunnel could be operated it was
essential to clean and clear any loose objects in the test section that may be
blown into the wind tunnel.

To avoid any possibility of blowing the fuses in the wind tunnel’s circuitry,
both the ‘coarse’ and ‘fine’ velocity control dials were always set to zero
before starting up the wind tunnel.

To power up the wind tunnel the lever on the motor located behind the fan
was pulled fully back and held. The lever was held in this position until the
motor speed levelled out, at which point the lever was sharply pushed
forward and released. With the motor up and running, the wind speed could
then be adjusted accordingly.

To raise the air speed in the tunnel the coarse dial was turned in the
clockwise direction whilst using the velocity indicator situated behind the test
section as an indication of wind speed. Attention also had to be paid to the
ammeter located next to the wind speed indicator as it was recommended
that the wind tunnel should not be operated at levels exceeding 50 amperes,
to minimise the risk of burning out the fuses in the circuits. To verify the wind
speed in the tunnel, readings were taken from a manometer that was
connected to a Pitot tube located at the mouth of the wind tunnel. These
readings were then input into a pre-made spreadsheet along with the lab air
temperature and atmospheric pressure and the spreadsheet then produced
the true air speed.

The first step in the testing process was to use the data acquisition system to
determine the voltage output from the strain gauge with the tunnel air speed

                                                                                22
set at nil. From there, the air speed was then increased in increments of
10mph up to a speed of 60mph, with the voltage readings being taken at
each point. This procedure was repeated several times to obtain enough
data to assess the consistency of the outputs.

With all the necessary result obtained, the speed dials were both set to zero
and the stop button on the motor was pushed, which shut down the wind
tunnel.




                                                                            23
8. CFD – WING MIRROR ON FLAT PLATE

8.1 GAMBIT: MODELLING
The pre-processing software package used for the wing mirror model was
Gambit version 2.4.6. The purpose of using this software was to take the
wing mirror geometry and place it in a virtual wind tunnel, this model (or
mesh) could then be exported into the CFD package.

To create this virtual wind tunnel the most commonly applied approach is to:

1. import the model geometry, which can come in various formats from
   several software types;
2. create a real volume of suitable dimensions around the imported model
   geometry;
3. subtract the model geometry from this brick geometry, thus leaving one
   volume which will represent the flow domain;
4. apply geometry clean-up measures to the geometry, to reduce any
   complexities in the model that could represent problems when meshing;
5. mesh all the faces in the model including the domain walls;
6. mesh the volume of the model;
7. set the boundary conditions and the continuum type;
8. export the final mesh for use in a CFD package.

For the purpose of this project, the wing mirror geometry of the L319 wing
mirror was provided in ‘dbs.’ format which was a suitable format to open in
Gambit directly. With the geometry loaded in Gambit, the model was
examined for any noticeable differences between it and the real wing mirror
provided. It was recognised that the wing mirror computer geometry
consisted of two ‘real’ volumes, one representing the wing mirror mount and
the other representing the wing mirror casing. This was concurrent with the
real wing mirror as it consisted of these two components hinged together.
Upon further inspection, it was also noticed that on the underside of the wing
mirror casing model there was a recess in the surface, no such featured
existed on the real wing mirror.



                                                                              24
To fix this discrepancy, a ‘real’ face was created over the recess, essentially
closing it in. This enclosed region was then transformed into a real volume
using the ‘stitch faces’ command, and the resultant volume was then merged
into the rest of the wing mirror casing using the ‘merge volumes’ command.
These operations resulted in a flat surface over the area in which the recess
was found (Figure 11).




                             Figure 11 - Recess Filling


With this discrepancy resolved, the model was then deemed an accurate
virtual representation of the actual wing mirror.

The next step was to create the volume around the wing mirror geometry
from which the wing mirror geometry would then be extracted. This was
carried out by using the ‘create real brick’ command, and entering the values
of length, height and width as 4000x1500x1500mm respectively (a rough
approximation of the wind tunnel test section). This volume was then aligned
(using the ‘move/align’ function) with the wing mirror base, with the wing
mirror centrally positioned on one of the 4000x1500mm faces of the brick.

With all the volumes in their desired position, the volume subtraction could
then be performed. The first volume subtracted was that of the wing mirror
mount, Gambit performed this ‘subtract volume’ operation successfully and
as a result created a new volume which represented the brick volume with a
cavity in the form of the wing mirror mount. However, when attempting to


                                                                               25
subtract the wing mirror housing from this newly formed volume, an error
message of ‘coincident face_face_ints with different body vertices’ was
displayed. This problem arose because around the area where the wing
mirror casing and mount are hinged, some of the faces on each volume were
coincident or overlapping. To rectify this issue, the wing mirror casing volume
was moved away from the mount by a few millimetres on each axis to
prevent face overlapping or coincidence. The volume subtract operation was
then tried again and was successful, thus the desired single volume was
created.

With the flow volume defined the geometry could then be ‘cleaned’. This
process consisted of eliminating any sharp angles or short edges on the
geometry and merging small faces together to reduce the complexity of the
model. To eliminate short edges on the model, the ‘connect edges’ tool was
used to highlight the shortest edge present in the geometry and this was
consequently merged with its neighbouring edge to form a larger edge. This
process was repeated until the shortest edge highlighted was no longer
judged too short to cause meshing problems. The ‘merge faces’ tool was
then used to merge any small or awkward shaped faces into its neighbouring
face, thus reducing the geometry complexity through producing larger more
easily meshed faces.

8.2 GAMBIT: MESHING & EXPORTING

The first surfaces to be meshed were that of the wing mirror. When applying
a face mesh it is beneficial to apply a quad based mesh as this will deliver
the most accurate results. However, due to the complexity of the wing mirror
geometry, applying a quad mesh would have proven extremely difficult. For
this reason it was decided to apply a triangular based mesh as this would be
applied more easily to the complex geometry. Therefore, a tri-pave mesh
with an interval size of 3 was selected and applied.

Upon completion of this meshing process, the mesh was then checked using
the ‘mesh > check’ function which checked all the applied face meshes for
any flaws. The check highlighted that there were several skewed elements,


                                                                               26
which meant that the mesh applied was not an accurate representation of its
host face.

Using the list of skewed elements obtained from the check, further face
merging was applied to these faces. Once completed, the mesh was again
applied and a new check performed. This process of meshing, checking and
face merging was repeated until the mesh was successfully applied to all the
selected faces and no skewed elements were present.

The six faces of the domain boundary were then meshed using a tri-pave
mesh with an interval size of 40. This meshing process was successful on
the first attempt, and the mesh around the area at which the wing mirror was
placed was noticeably denser. This dense mesh concentration would ensure
that when modelling the flow in CFD a greater degree of accuracy and detail
could be found at the wing mirror and surrounding region. The volume was
then meshed with a tetrahedral scheme with an interval size of 80. This
mesh application was also immediately successful and therefore the
meshing of the domain was complete (Figure 12).




                         Figure 12 - Meshed Flow Domain


The boundary conditions of the domain were then set but only after selecting
the Fluent 5/6 option from the solver list to determine the boundary


                                                                           27
conditions available. The face that represented the flow inlet was set as a
Velocity Inlet, and the opposite outlet face set as a Pressure Outlet. The flat
plate upon which the wing mirror was placed was set as a Wall and given the
identity ‘Flat Plate’. Similarly the wing mirror was set as a wall but given the
identity of ‘Mirror’. This assignment of two different identities was to enable
the two different components to be analyzed and viewed independently of
one another when modelled in CFD. The three remaining domain walls were
then specified as Symmetry faces; which means that flow can pass by them
unaffected. The volume continuum type was selected as Fluid and named
‘air’.

This was the last step in pre-processing and therefore all that was required
was to export the mesh for use in Fluent.

8.3 FLUENT: ANALYSIS SETUP

For the CFD analyses Fluent version 6.3.26 was used. The software was
used to perform pressure based steady state analyses with the aim of
modelling the flow over the wing mirror at the same 10mph increments from
0-60mph as the wind tunnel testing.

With the mesh obtained from the pre-processing in Gambit, the initial
conditions were set before the analysis could begin.

The first issue that was addressed was the scale applied to the grid, as the
mesh was created in millimetres in Gambit, but the default unit for length in
Fluent is metres. Therefore the scale command was used to scale the model
down to the correct size as created in Gambit.

As already mentioned, the solver was set as steady state and pressure
based. It may have been useful to model the flow as transient but due to time
and computational restrictions this was not feasible.

The turbulence model selected was the standard k-ε model. This model is
one of the most commonly used turbulence models in industry, due to its
speed and simplicity of use REFERENCE. The model employs two transport


                                                                                  28
equations: one for the kinetic energy of the turbulent flow (k) and another for
the rate of dissipation of the turbulent flow (ε).

For the purpose of aeroacoustic analyses, the Broad Band Noise Sources
model was selected. This was the only option available for aeroacoustics due
to the analysis being run as steady state. This model uses the values
obtained from the turbulence model for such things as the mean velocity
components, mean pressure, turbulent kinetic energy and so on. It then uses
these values to determine the broadband noise present in the model. The
Broad Band Noise model is limited in its usefulness as it only determines the
sources of broad band noise but cannot determine sound propagation. As
such, it is often employed as a qualitative method of assessing the
‘noisiness’ of designs.

The boundary conditions had then to be set and this involved setting the
wind speed through defining the velocity at the inlet. For the purposes of this
explanation, the velocity at the inlet was set to 13.4112m/s (or 30mph) and
the analysis was then initialised from the inlet with the speed defined as
13.4112m/s. This process produced a known value as a starting point for the
solving process to work from.

The convergence criterion for the solution was set to the default of
convergence to 10-3. This meant that during the iterative process of solving
the flow over the wing mirror, the residuals had to converge to 10-3; which
was deemed a suitable level of accuracy. The real-time convergence of the
residuals were set to be displayed during the solution process; this would
make it possible to observe if the solution was nearing completion or in the
worst case diverging.

Before the analysis could begin, reference values had to be set for the
calculation of forces and drag coefficient. The wing mirror was used as the
reference and values of 0.0375m2 and 0.1m were set for the cross sectional
area and object length respectively. The velocity of 13.4112m/s was also
entered for reference.

8.4 FLUENT: SOLVING

                                                                               29
As mentioned previously, Fluent uses an iterative process to solve problems,
so accordingly the software was set to perform 200 iterations; with the
solution expected to converge before this number was reached. The
software was then instructed to iterate, and thus the solving process began.
A plot of the residuals converging is shown in Graph 1.




                         Graph 1 – Residuals Convergence


With the solution achieved Fluent was then used to print the pressure and
viscous forces and their sum total acting on the wing mirror plus the pressure
and viscous coefficients. These results were then recorded, along with the
values obtained from the analyses of all other speed increments.




                                                                             30
9. WIND TUNNEL RESULTS & CFD VALIATION

9.1 RESULTS COMPARISON

To validate the accuracy of the results obtained from Fluent, a comparison
between these results and the data from the wind tunnel was made. Graph

2 shows the slopes of drag against velocity for both the wind tunnel derived
values and the results supplied by Fluent.




                    Graph 2 - Wind Tunnel & CFD, Drag v Velocity


Reviewing Graph 2, it is possible to see that the trends produced from each
set of data correlate well. The gradient of both slopes compare closely and
from a speed of around 30mph and upwards the slopes can be seen to run
almost parallel.

However, there is a difference in drag between each set of data at each of
the recorded speed increments. The maximum difference can be identified at
a wind speed of 40mph where the drag measured in the wind tunnel is just
under 2N less than that of the drag found through the use of CFD. Although




                                                                              31
this difference is not sizeable, it was still necessary to identify the source of
any errors which may be accountable.

9.2 ERRORS
The most probable error that occurred may have been through setting the air
speed in the wind tunnel. As was detailed before, the true air speed was
determined through taking readings from a manometer connected to a pitot
tube, located in the upper region of the inlet mouth. The problem with this
method is that the velocity profile across the test section is not uniform, and it
is most likely that the values of velocity determined through the manometer
readings, were consistently higher than the actual airflow velocity over the
wing mirror. This lowered value of velocity over the wing mirror could be
directly related to the lower values of drag obtained (compared to the CFD
results), due to the fundamental relationship of drag stating that the drag
increases proportionally with the velocity squared (E).

                               1
                          D     ACDV 2             … E5
                               2

The relationship between drag and velocity may also explain why the two
sets of data are more analogous at the lower speeds of 10 and 20mph where
the velocity is less and therefore the effects of the velocity squared impact
less on the values of drag. Furthermore, the fact that the wing mirror drag is
almost negligible at low speeds such as 10mph makes the margin of error in
taking readings larger.

Another contribution of error may be attributed to the fact that the CFD
analyses were modelled with the wing mirror mounted on a flat surface,
whereas in the wind tunnel the wing mirror was placed in a small A-Pillar
mock-up. The effects of the A-Pillar on the wing mirror drag are hard to
quantify without the need of further testing, however, it may be a contributing
factor.

9.3 VALIDATION
With the possible errors recognised and taken account of, it is apparent that
the results from the two methods of testing correlate well. This provides the


                                                                                    32
evidence that the use of CFD can provide an accurate representation of fluid
flow in the application of testing such things as wing mirrors. With this CFD
validation accomplished, it was then possible to assess the performance of
the wing mirror in greater detail in Fluent, with the aim of performing design
optimization.




                                                                                33
14. DESIGN PERFORMANCE ASSESMENT
One of the many advantages of using a CFD package such as Fluent is that
with the analysis solved, various plots and graphs of the fluid behaviour and
properties can be produced. These features can then be utilised to assess
the performance of the model and identify any strengths or weaknesses.
Using this method, it was possible to assess the strengths and weakness of
the L319 wing mirror in Fluent.

14.1 PRESSURE DRAG
By producing a contour plot of the static pressure levels on the wing mirror
surfaces, it was possible to identify the areas of high pressure and low
pressure as can be seen in Figure 13; with blue representing the lowest
pressure and red representing the highest.




                Figure 13 - Contours of Static Pressure (Pascal) at 60mph


From this plot, it can be seen that on the leading surfaces of the wing mirror
there are high levels of static pressure. It can also be observed that on the
trailing faces the static pressure is comparatively low, and thus this pressure
difference is a major contributor to the wing mirror’s pressure drag and
therefore its overall drag.


                                                                                34
The high static pressure on the leading faces can be attributed to the fact
that a sizeable amount of flow stagnation takes place on this area resulting in
low velocity air flow and thus a high static pressure. This flow behaviour is
especially evident on the wing mirror mount.

14.2 AEROACOUSTICS
The aeroacoustic properties of the wing mirror could also be studied through
plotting the surface acoustic power levels on the wing mirror (Figure 14)




           Figure 14 - Contours of Surface Acoustic Power Level (dB) at 60mph


From the plots it could be seen that the highest sources of noise were
present at areas of sharp changes in surface curvature, such as at the edges
of the wing mirror mount, at the stepped faces on the wing mirror’s underside
(also a region of high static pressure) and the trim around the mirror casing;
as shown by the red ring-like feature on the wing mirror.

With the performance assessment of the wing mirror complete, efforts to
improve the design could be carried out. The first design changes applied
would be to alter the shape of the wing mount with the intent of lowering the
wing mirror’s drag. If successful this optimized design would then be further
altered through smoothing any sharp changes in surface curvature, with the
aim of decreasing the amount of high acoustic power level sources on the
wing mirror surfaces.



                                                                                35
15. OPTIMIZATION (I)

15.1 MODELLING
As assessed, one of the main contributing factors towards the wing mirrors
drag, results from the stagnation of flow on the leading faces; particularly
around the area of the mount. In an effort to reduce this source of drag, it
was decided to create a small square channel, passing from the leading face
of the mount to its trailing face in the flow-wise direction (Figure 15).




                              Figure 15 - Optimization (i)


This design alteration was performed in Gambit, by deleting the mount
volume but not the lower topology (faces, edges and vertices) and creating a
face in the shape of the channel on the mount front surface. This face was
then subtracted from the mount face and thus a hole was created. This
process was then repeated on the opposite side of the mount to produce
another hole. Each of the vertices at the corners of the two square holes
were then linked to their opposing vertex with an edge, therefore allowing the
creation of the internal faces of the channel. With all the necessary faces
created, a face stitch command was ordered and the new mount volume was
created. The model was then meshed with the same procedure as the
original wing mirror geometry and the same boundary conditions were again
applied before exporting the mesh for use in Fluent.

This new wing mirror design was then run through the same six analyses as
the original design so that a comparison between the drag results could be
made.



                                                                               36
15.2 RESULTS
The information in Table 1 displays the drag results from both the original
and the optimized design; it also details the percentage reduction in drag
resulting from the design optimization.

                                     Drag (N)
       Velocity (mph)                                           Drag Reduction (%)
                        Original Design    Optimization (i)
            10              0.383                0.375            1.995090615
            20              1.360                1.283             5.69159497
            30              2.816                2.698            4.193871862
            40              4.790                4.572            4.554051177
            50              7.276                6.949            4.494191928
            60              10.293               9.940             3.42960622
                           Table 1 - Optimization (i) Results


As shown in the table, it would appear that the optimization effort towards
reducing the wing mirror drag were successful with a resulting average
reduction in drag of just under 5% from a speed of 10mph and upwards.

To check that the design alteration did not have a detrimental effect on the
wing mirrors surface acoustic power levels, a comparison between the
acoustic power level plots for the original and optimized design was
performed. Upon reviewing these plots there were no clear signs that the
design changes made had any negative impact on the wing mirrors
aeroacoustic performance with regards to surface broadband noise sources;
even around the area of the new channel.

As mentioned previously, some motor vehicle manufacturers design their
wing mirrors to be fork mounted in an effort to increase the wind speed aft of
the wing mirror. This increased speed results in any water droplets or dirt
shed from the wig mirror being deposited on the side of the car further
downstream thus reducing the safety risks associated with water or dirt
accumulation on the front side window. The altered wing mirror mount in the
first optimization attempt can be assumed to be a crude approximation of a
fork shape, and therefore should in theory result in a higher velocity flow aft
of the wing mirror thus delivering the associated safety benefits. Through

                                                                                     37
studying a velocity vector plot on the surface of the flat plate, a comparison
between the original design and the optimized design of the downstream
velocity magnitude and distribution could be made (Figure 16).




                 Figure 16 - Velocity Vector Plot on Flat Plate at 60mph


From these plots it can be identified that the alterations of the wing mirror
mount design resulted in a narrowing of the low velocity wake on the flat
plate (as seen in blue). This narrowing of the low velocity wake could result
in improved water droplet and debris shedding performance. However this
can only be taken as an indication of improved performance and further
physical testing (or complex CFD analysis) would be required to validate this
assumption.



                                                                                 38
16. OPTIMIZATION (II)

16.1 MODELLING
The second optimization effort was made with the aim of minimizing the
sources of high acoustic power levels on the wing mirror surfaces. As
mentioned previously, the primary source of high levels of acoustic power in
aerodynamics arise from sharp changes in surface curvature. For this reason
it was decided that any recesses or obtrusion on the wing mirror surfaces
would be smoothed over.

The first changes to the geometry were applied to the joint between the trim
piece on the trailing edge of the wing mirror casing and the main wing mirror
casing itself. This recess was smoothed over by deleting the recessed faces,
creating a new face over the hole and then stitching the faces together to
create the new volume. This process was then also applied to the indentation
that ran along the front face of the wing mirror casing. Due to the underside
of the wing mirror casing featuring several sharp changes in surface
curvature, it was decide to create a new face to smooth over the entire
underside as can be seen in Figure #.




                            Figure 17 - Optimization (ii)




                                                                             39
As with the first design optimization, this new design was meshed with the
same properties of the original geometry and the same boundary conditions
were set before exporting for use in Fluent.

The model was then run through the same six analyses as before.

16.2 RESULTS
With the analyses complete, it seemed sensible to once again examine the
drag readings obtained (Table 2).

                                     Drag (N)
       Velocity (mph)                                            Drag Reduction (%)
                        Original Design     Optimization II
               10           0.383                0.379              1.08361101
               20           1.360                1.271             6.566291639
               30           2.816                2.630             6.598306222
               40           4.790                4.458              6.93378089
               50           7.276                6.764             7.035519126
               60           10.293               9.555             7.171114054
                           Table 2 - Optimization (ii) Results


As can be seen from the readings, the second optimization attempt resulted
in a further decrease in the wing mirror drag, with an average reduction of
just under 7% at speeds of 20mph and upwards (compared to the original
geometry). This reduction in drag is most likely resultant from the general
streamlining of the wing mirror shape in the efforts to reduce the sharp
angles in surface curvature. Plotting the contours of static pressure on the
wing mirror’s underside shows that the smoothing of this area resulted in a
significant decrease in static pressure due to the reduction of flow stagnation
(Figure 18).




                                                                                      40
                  Figure 18 - Optimization (ii) Flow Stagnation Reduction


The reduction of the wing mirror drag is most likely a result of this change in
flow behaviour.

Studying a plot of the surface acoustic power levels on the original geometry
and the optimized design, shows a significant reduction of the presence of
high acoustic power levels on the wing mirror surface (Figure 19).




             Figure 19 - Optimization (ii) Contours of Surface Acoustic Power


As can be seen, the optimization effort was successful in reducing the
sources of aerodynamic noise on the wing mirror, especially at the mirror
housing trim and on the edges of the underside.



                                                                                41
However, as previously stated, the Broad Band Noise Sources model is
limited in its level of accuracy and for this reason the optimization efforts can
only be used as an indication of the reduction in wing mirror noise. To
establish more accurately the noise produced by the wing mirror, further CFD
testing would be required, using unsteady flow models and more
sophisticated aeroacoustic models.


17. A-PILLAR MODELLING

17.1 A-PILLAR INFLUENCE

Although the design optimization efforts proved successful with regards to
the wing mirror being positioned on a flat plate it is difficult to determine
whether or not the same benefits would be produced with the wing mirror
mounted on the A-Pillar.

For this reason it seemed constructive to model the wing mirror mounted on
the A-Pillar, with the aim of better understanding the flow characteristics over
the area and to make a comparison between the values of drag obtained
from each CFD model.

17.2 GAMBIT: MODELLING
The pre-processing of the A-Pillar was again performed using Gambit. For
the purposes of this study, a model representing half of the whole Range
Rover car geometry was provided. However this model did not feature the
wing mirror and therefore the wing mirror geometry was imported into the
model and attached to the car body through stitching the lower topology of
the volumes together. Similar to the flat plate model, a brick volume was then
created around the car geometry, with the dimensions of 10x4x4m and the
volumes were aligned with brick volume wall. The car and wing mirror
volume were then subtracted from the brick volum, thus creating the flow
domain.

Extensive geometry cleaning measures had to then be taken due to
complexity of the car geometry; this was performed with the use of the


                                                                                42
automatic geometry clean up tools and through manual manipulation of the
model. Multiple face merging operations were carried out and the effects of
these processes are illustrated in Figure 20 which shows certain areas on
the car and wing mirror surfaces smoothed over (door handles, trim,
headlights etc.).




                         Figure 20 - Simplified Car Geometry


This reduction in geometry complexity was essential due to the limited
processing power available and thus the limited meshing and solving
capabilities. Too fine a mesh would prove difficult to apply to such a large
and complex model; furthermore, the capability of the available processing
power would fall short of solving the resultant flow simulation.

17.3 GAMBIT: MESHING & SOLVING
Meshing of the car and wing mirror was again an iterative process, similar to
the process employed for the flat plate model. Several mesh sizes were
applied before a successful configuration was found.

The wing mirror and wing mirror mount were meshed with a triangular mesh
with an interval size of 5 which is a coarser mesh than that applied to the
wing mirror on the flat plat setup. A triangular mesh with an interval size of 50
was then applied to the rest of the car surfaces. The flow domain walls were
also meshed with a triangular scheme, with an interval size of 300. The flow
domain volume was then meshed with a tetrahedral scheme with an interval

                                                                               43
size of 350.These mesh sizes represented the best compromise between
model accuracy and realistic solvability.

The same boundary conditions applied to the flat plate were then set, with
the exception of the domain wall situated under the car geometry which was
set as a wall to represent the ground. The continuum was then set as a fluid
and the mesh was exported for use in Fluent.

The exported mesh was opened in Fluent and the six analyses running from
10 to 60mph were carried out - using identical initial conditions and reference
values to the ones set in the flat plate analyses

17.4 FLAT PLAT & A-PILLAR COMPARISON
With the solutions obtained, a comparison between the drag readings
obtained through the two fluent models was made. Graph 3 shows a plot of
drag against velocity for each Fluent model across the velocity range.




               Graph 3 – Flat Plate & A-Pillar comparison, Drag v Velocity


From Graph 3 it can be seen that the presence of the A-Pillar brought about
very little change in the drag produced by the wing mirror, with a maximum
difference in drag of just over 1N at 60mph.

17.5 FLOW OVER THE A-PILLAR




                                                                             44
With the aim of better understanding the flow behaviour over the A-Pillar and
to find the possible sources of the difference in drag between the Flat Plate
and Car model, it was possible to use Fluent to provide visualisations of the
flow over the A-Pillar region.

Figure 21 shows a Path Lines plot with ribbons being placed on the car and
wing mirror surface.




                    Figure 21 - Pathlines Plot over whole car, 60mph


From this plot it is possible to identify that lateral flow occurs over the
windshield and as a result it then passes over the A-Pillar. The flow over the
A-Pillar is then accelerated onto the side of the car and onto the wing mirror,
thus increasing the local velocity over the wing mirror. This increase in
velocity may be accountable for the increased drag results when compared
to the wing mirror on the Flat Plate. However the difference in drag is small
and therefore the decreased accuracy of the car and wing mirror mesh may
also contribute to the difference.

17.6 WATER DROPLET & DEBRIS SHEDDING
As detailed previously, poor positioning of the wing mirror on the A-Pillar can
result in water and debris shedding on the front side window. The presence
of the wing mirror and its mount causes flow deceleration around the area of



                                                                                45
the window and this can lead to water and debris accumulation. Figure 22
shows a velocity vector plot at the area of the wing mirror on the car.




                   Figure 22 - Velocity Vector Plot on A-Pillar, 60mph


Similar to the flow on the Flat Plat in the previous analyses, the plot shows a
region of low velocity flow on the side window aft of the mirror (as shown in
blue). The plot also details the area of flow acceleration over the A-Pillar and
the flow stagnation present at the bottom of the windshield.

17.7AERO ACOUSTIC NOISE SOURCES
The sources of broad band noise production on the car surfaces were plotted
as shown in Figure 23.




                                                                                46
        Figure 23 - Contours of Surface Acoustic Power Levels on whole car, 60mph


As expected the sources of the largest levels of broad band noise creation
occur at the sharp changes in surface curvature such as at: the intersection
of the front of the car and the bonnet, the wheel fairings, the A-Pillar and the
wing mirror. It must again be stated that these plots only serve as an
indication of the noise created as the broad band noise model accuracy is
limited. The model also does not determine the strength of the noise
propagation, which could be determined through transient analyses.




                                                                                    47
18. DISCUSSION AND FURTHER STUDY

18.1 WIND TUNNEL TESTING
The wind tunnel testing conducted for the purpose of this study provided
useful information for the comparison and validation of the CFD results. The
sources of error were attributed to the fact that the values of velocity
determined through the use of the Pitot tube were consistently greater than
the actual velocity over the wing mirror, due to the position of the Pitot tube
at the flow inlet.

To improve the wind tunnel testing method it may have been beneficial to
place the Pitot tube nearer the position of the wing mirror in the test section.
This would have provided a more realistic value of the flow velocity in this
region and therefore produce more accurate drag readings.

It may have also been useful to test the wing mirror on a flat plate rather than
on the A-Pillar mock-up. This may have resulted in better correlation
between the wind tunnel and CFD drag values. However, without the need
for further wind tunnel testing, the effects of the cardboard A-Pillar structure
on the wing mirror drag cannot be quantified.

18.2 OPTIMIZATION EFFORTS
Through reviewing the effects of the design changes made in Optimization (i)
on the wing mirror’s performance, it appears as though the optimization
efforts made towards reducing the drag were successful. Furthermore, the
results obtained indicate that there may also be an improvement in
performance with regards to water droplet and debris shedding on the side
window. It is also important to note that the changes made to the geometry
also made no significant negative impact on the performance of the wing
mirror with regards to sources of aerodynamic noise creation on the
surfaces.

The second optimization effort was also successful in reducing the drag and
the sources of high acoustic power creation on the wing mirror surface.

                                                                                  48
With more time available it may have been possible to apply further changes
to the wing mirror geometry to further improve and enhance the wing mirror’s
performance. One change that could have been made would have been to
widen the channel running through the wing mirror mount, to further reduce
the flow stagnation on the leading faces. It may have also been beneficial to
apply general streamlining alterations to the wing mirror mount to reduce the
bluffness of the body thus reducing drag. This would have also smoothed out
the sharp changes of surface curvature on the mount and would therefore (in
theory) reduce the amount of high level aero acoustic noise created at the
edges.

With the restrictions on available computational power, it was only really
feasible to model the optimizations accurately on a small scale such as the
flat plate model. If more processing power was available it would have been
useful to model the optimizations on the full car model to gain a more
accurate real life representation of the optimization effects on the wing
mirror/A-Pillar flow interaction. However, the comparison between the wing
mirror drag results for the flat plate model and the whole car model showed
that although the values were not precisely concurrent there was very little
difference. This would suggest that although the effects of the optimization
efforts on the whole car cannot be reviewed when modelling the flat plate
configuration, it still represents an accurate method for the study of
optimizing the wing mirror in isolation.

It was also encouraging to observe similar flow patterns downstream of the
wing mirror in both the flat plat and the whole car models, with both exhibiting
a region of low velocity flow on the surfaces aft of the wing mirror.

18.3 A-PILLAR
With the use of greater computational power it may have been constructive
to perform changes on the A-Pillar design. This may have provided useful
information about the effects of such things as the windshield angle and pillar
curvature on the flow over the A-Pillar region. This information could have
then been utilised to reduce the lateral flow over the windshield and
consequently the A-Pillar, thus reducing the vortex shedding from this region.

                                                                               49
The reduction of lateral flow over the windshield may have also resulted in a
reduction of the acceleration of the flow over the A-Pillar onto the wing
mirror, therefore reducing the flow velocity over the wing mirror and
consequently the wing mirror drag.

18.4 AEROACOUSTICS
Although using the Broad Band Noise model in Fluent proved useful for
assessing the ‘noisy’ features on the wing mirror and the A-Pillar, its
accuracy was fairly limited. Once again, the use of greater computational
power would have provided a means of applying a finer mesh and employing
one of the more sophisticated acoustics models that can be used with
transient analyses in Fluent. The use of such a model could be utilised to
determine the nature of sound propagation from the A-Pillar and could
potentially be used to find the levels of noise audible to the driver/passenger.

19. CONCLUSION
Although the limitations on the work and progress of this project with regards
to available computational power have been detailed and discussed, it is
apparent that the fundamental aims of the project were met.

The wind tunnel testing provided a means of validating the results provided
by the CFD analyses and successful design optimizations were applied to
the wing mirror geometry. An analysis of the flow over the whole car which
detailed the flow over the A-Pillar was also successfully performed.

Upon completion of these initial aims it was then realised that with additional
time, there would be numerous possible avenues of study which could be
focussed on the wing mirror/A-Pillar aerodynamics. The most constructive
path of study would most likely be to alter the A-Pillar and windshield design
in an effort to minimize lateral flow over these regions. A reduction in lateral
flow over the A-Pillar could result in not only a reduction in the wing mirror
drag due to decreased flow acceleration onto the wing mirror, but also a
reduction in the induced drag on the whole car as a consequence of
minimizing the vortex shedding from the A-Pillar.




                                                                                 50
Recognizing the possibilities of further improving the wing mirror and A-Pillar
design emphasizes just how useful CFD software can be when the
necessary computational power is available and the correct expertise is
applied.




                                                                             51
REFERENCES
 1. Hucho, W.H, Aerodynamics of Road Vehicles. 4th Edition, 1998. P196 & 294
 2. http://mayfco.com/tbls
 3. Stickland, M. Aerodynamic Performance Course Notes. The University of
    Strathclyde
 4. Dolek, O., Ozkan, G., Ozdemir, I.B. Structures of flow around a full scael
    side mirror of a car with relevance to aerodynamic noise
 5. Stickland, M. Flight & Spaceflight 2 Course Notes. The University of
    Strathclyde
 6. Modelling Turbulent Flows (Presentation), Introductory FLUENT Training,
    www.fluentusers.com
 7. Sovani, S. Acoustics Modelling, 2004 CFD Summit, Fluent Automotive UGM




                                                                                 52

								
To top