AERODYNAMIC CFD STUDY

ON A PRODUCTION VEHICLE

EXPERIMENTAL VALIDATION
ON A WIND TUNNEL

-
FINAL DISSERTATION
AUTOMOTIVE ENGINEERING

Iván Esteban Macho – 43466115W

UPC Manresa – Tutored by: Jordi Vives

 ABSTRACT

The automotive industry has been driven by the need of creating faster, safer and more
fuel-efficient cars. In recent times however, and with the disruption of electric cars, a new
player has come into place: aerodynamics. Over the past years, new powerful ways of
studying and improving the way a vehicle interacts with the air around it have been gaining
relevance, up to a point where it is feasible to determine just how well designed a car is
regarding this term. This dissertation presents a complete virtual aerodynamic study on a
production vehicle from 20 years ago via Computer Fluid Dynamics simulation software to
analyse just how different a modern car design study is to what it was then. The output results
are compared with the ones given by the manufacturer at the time of development, in order to
see how powerful and accurate modern software technology is when studying the
aerodynamic performance of a vehicle. After that, a real 3d scale model of the geometry will
be created and put into a real wind tunnel to further validate the extracted results. In order to
accomplish what is being described, a full virtual model of the vehicle in case will manually be
designed in a Computer Aided Design program. The geometry will then be optimized and put
under study. By doing so, two main objectives are being met. Analyse how powerful and
accurate these new simulation software are when designing and testing a production vehicle,
and further explore the possibility of fully designing and studying a vehicle aerodynamically
solely with virtual tools by comparing the way the results evolve from two simulations two
decades apart.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 2
 RESUMEN

La industria de la automoción siempre ha estado conducida por el deseo de crear

vehículos más rápidos, más seguros y más eficientes. En estos tiempos, sin embargo, y con
la irrupción de la electrificación, un nuevo jugador se ha sumado a la partida: la aerodinámica.
A lo largo de estos años, nuevas maneras de estudiar y mejorar la forma en la que un vehículo
interactúa con el aire que le rodea han estado ganando fuerza, hasta llegar al punto en el que
es totalmente factible determinar realmente cómo de bien está diseñado dicho vehículo en lo
que a este término concierne. Este trabajo presenta un análisis virtual completo sobre un
vehículo de producción de hace 20 años mediante un software de simulación de Fluidos
Dinámicos Computacionales para poder analizar lo diferente que es la forma de analizar y
desarrollar el diseño de un coche actual, comparado con lo que por entonces era nuevo. Los
resultados extraídos serán analizados y comparados por aquellos facilitados por el fabricante
en el momento de desarrollo, para comprobar realmente la precisión de los programas de
simulación modernos al estudiar el campo a tratar. Posteriormente, se materializará un
modelo 3D a escala de la geometría, y será introducido en un túnel de viento real para validar
los resultados extraídos previamente. Para poder conseguir lo descrito, se creará un modelo
completo virtual del vehículo mediante un programa de diseño por ordenador. La geometría
final será optimizada y puesta a estudio. Mediante lo mencionado, se cumplirán los dos
objetivos principales de este trabajo. Analizar la potencia y precisión de los nuevos softwares
de análisis de fluidos a la hora de diseñar y testear un futuro vehículo de producción, y
explorar la posibilidad de diseñar y estudiar aerodinámicamente dicho vehículo de manera
puramente virtual, comparando la forma en que los resultados evolucionan a través de dos
simulaciones separadas por dos décadas.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 3
 ACKNOWLEDGEMENTS

Hace cuatro años me fui de casa y me separé de mi familia. Me mudé a cientos de

kilómetros de distancia para poder perseguir un sueño presente desde que tengo uso de
razón. Ahora, cuatro años después he llegado al final de este camino que empecé por mí,
pero con la mente llena de gente a la que estoy increíblemente agradecido.

Ahora, como (casi) Ingeniero de Automoción, quisiera dedicar unas líneas a algunas
de las personas que han contribuido a que estos últimos cuatro años, sean los mejores de mi
vida.

Por delante de todos, quisiera dar las gracias a mi padre, mi madre y mi hermana
pequeña, por apoyarme en todos y cada uno de mis pasos, por no dudar nunca de mí. Por
ser la fuerza desde la distancia, y ser ese lugar seguro al que acudir cuando las cosas no van
bien.

Un Trabajo de Final de Grado al final es un reflejo de todo lo aprendido a lo largo de

una carrera universitaria, y como tal, encuentro que es el momento para también agradecer
a aquellas personas que tanto te han enseñado por el camino. Andrea y Paola, con vosotras
he podido crecer, caerme, levantarme de nuevo, saber lo que valgo, y lo que no merezco. De
vosotras he aprendido como se ve realmente la amistad que uno quiere, y que uno merece.
Gracias por formar parte de la mejor versión de Iván.

Guillem, gracias por enseñarme que las amistades van mucho más allá de cuatro
paredes de hojalata. En Manresa, Barcelona o Mallorca, a través de nuestras aventuras,
noches de charla y sustos con TIBs, de ti he aprendido lo que es de verdad tener una conexión
especial con alguien, y como es realmente un lazo que no quiere romperse.

A toda la gente que he dejado en Mallorca, y que con cada vuelta a “casa” me hacen
sentir como si ni un minuto hubiera pasado desde que me fui. Gracias.

En una página es difícil incluir a toda la gente especial por la que estar agradecido,
sin embargo, espero en algún momento u otro, haberos podido transmitir este sentimiento.
Gracias por hacer de la universidad, un verdadero sueño cumplido.

Gracias también a la EPSEM por brindarme la formación que buscaba para

convertirme en quien quería ser, y a IDIADA por darme la oportunidad de ponerla en práctica.

Agradecer finalmente, a mi tutor, que ha servido como guía y ayuda durante todo el
proceso del que es mi gran último trabajo como universitario.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 4
 ABBREVIATIONS

CFD: Computer Fluid Dynamics

CAD: Computer Aided Design

CAE: Computer Aided Engineering

Cl: Lift Coefficient

Cd: Drag Coefficient

Cm: Pitching Moment Coefficient

Cs: Side Force Coefficient

Cy: Yawing Moment Coefficient

Cr: Rolling Moment Coefficient

RANS: Reynolds-Averaged Navier-Stokes

LGA: Lattice Gas Automata

LBM: Lattice Boltzmann

SA: Spalart-Allmaras

DOF: Degrees of Freedom

MCS: MINI Cooper S

NVH: Noise, Vibration, Harness

OEM: Original Equipment Manufacturer

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 5
 TABLE OF CONTENTS

TABLE OF FIGURES ....................................................................................................................... 8

1. INTRODUCTION .................................................................................................................. 10
1.1. AERODYNAMICS ........................................................................................................................ 11
1.1.1. EVOLUTION OF AERODYNAMICS .......................................................................................... 12
1.2. COMPUTATIONAL FLUID DYNAMICS PROGRAMS .................................................................... 14
1.2.1. CFD: CONSERVATION ............................................................................................................ 15
1.2.2. CFD: REYNOLDS-AVG AND NAVIER-STOKES EQUATIONS ..................................................... 16
1.2.3. CFD: LATTICE BOLTZMANN ................................................................................................... 17
1.2.4. CFD: TURBULENCE ................................................................................................................ 17
1.2.4.1. TURBULENCE: SPALART-ALLMARAS .................................................................................. 19
1.2.5. CFD: BOUNDARY LAYER ........................................................................................................ 20
1.3. STUDY SUBJECT: R53 MINI COOPER S ....................................................................................... 21
1.3.1. THE MINI COOPER S R53: DESIGN OVERVIEW ...................................................................... 24
2. AERODYNAMIC CONCEPTS .................................................................................................. 27
2.1. AERODYNAMIC CONCEPTS: SEPARATION AND WAKE .............................................................. 28
2.2. AERODYNAMIC CONCEPTS: FORCES AND MOMENTS .............................................................. 28
2.3. AERODYNAMIC CONCEPTS: FLOW COEFFICIENTS .................................................................... 29
2.4. AERODYNCAMIC CONCEPTS: AERODYNAMIC DESIGN ............................................................. 30
3. GEOMETRY DEVELOPMENT ................................................................................................. 31
3.1. DESIGN ENVIRONMENT ............................................................................................................ 31
3.2. GEOMETRY DEVELOPMENT: INITIAL DESIGN............................................................................ 35
3.3. GEOMETRY DESING: SECOND ITERATION ................................................................................. 37
3.4. GEOMETRY DESIGN: THIRD ITERATION .................................................................................... 40
3.5. GEOMETRY DESIGN: DETAILED VERSION .................................................................................. 42
4. CFD VALIDATION ................................................................................................................ 44
4.1. SIMULATION: GEOMETRY SETUP .............................................................................................. 44
4.2. SIMULATION: MESHING ............................................................................................................ 46
4.3. SIMULATION: BOUNDARY CONDITIONS ................................................................................... 48
4.4. SIMULATION: RESULTS.............................................................................................................. 49
5. VALIDATION OF RESULTS .................................................................................................... 52
5.1. VALIDATION OF RESULTS: REAL VALUES................................................................................... 52
5.2. VALIDATION OF RESULTS: ANALYSIS AND COMPARISON ......................................................... 53
6. EXPERIMENTAL VALIDATION ............................................................................................... 59
6.1. EXPERIMENTAL VALIDATION: EXPLANATION ........................................................................... 59

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 6
6.2. EXPERIMENTAL VALIDATION: THE GEOMETRY ......................................................................... 60
6.3. EXPERIMENTAL VALIDATION: THE TEST ................................................................................... 63
6.4. EXPERIMENTAL VALIDATION: RESULTS AND COMPARISON..................................................... 66
7. CONCLUSIONS .................................................................................................................... 69
REFERENCES .............................................................................................................................. 71
APPENDIX 1: FIGURES AND IMAGES............................................................................................ 75

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 7
TABLE OF FIGURES
Figure 1: Turbulent flow around a “bad” shape. Source: Wildstar-Schleswig-Holstein.

Figure 2: Laminar flow around a “good” shape. Source: Wildstar-Schleswig-Holstein.

Figure 3: Paul Jaray T6 blueprints. Source: Conciertos y Desconciertos (blogspot).

Figure 4: 1934 Chrysler Airflow. Source: Chrysler. Stellantis.

Figure 5: 2013 Volkswagen XL1. Source: Volkswagen AG.

Figure 6: 2015 Mercedes-Benz Concept Intelligent Aerodynamic Automobile. Source:

Mercedes-Benz

Figure 7: Mesh example on a race car geometry. Source: FetchCFD.

Figure 8: (Table 1): Spalart-Allmaras parameters.

Figure 9: MINI Cooper S R53 (front). Source: Bring a Trailer

Figure 10: John Cooper with the original MINI Cooper (left) and the R50 MINI Cooper
(right). Source: MINI.

Figure 11: Original MINI (far back) and R50 MINI Cooper (up front) in Solid Gold Met.
Source: MINI.

Figure 12: Visual representation of “Three-Tiered cake” MINI. Source: Frank

Stephenson

Figure 13 and 14: Fender joint and bonnet bulge: Source: Bring a Trailer.

Figure 14: MINI Cooper S sketch. Source: Frank Stephenson.

Figure 13: MINI Cooper (left) and Cooper S (right) front: Source: EVO Magazine and
Bring a Trailer.

Figure 14: MINI Cooper (left) and Cooper S (right) back. Source: MINI.

Figure 15: Elements of an aerodynamic design. Source: Volkswagen AG.

Figure 16: (Table 2): MINI R53 CAD Measures.

Figure 17: R53 MINI Cooper S blueprints used for CAD design.

Figure 18: Layout of blueprints in CATIA environment.

Figure 19: Original Master Geometry Lines.

Figure 20: Final version of Master Geometry Lines.

Figure 21: Initial Design – Body parts.

Figure 22: Second Iteration – Full body.

Figure 23: Second Iteration – Undertray design.

Figure 24: Second Iteration – Full body with undertray assembly.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 8
 Figure 25: Third Iteration – Full enclosed (solid) body.

Figure 26: Third Iteration – Full solid body section.

Figure 27: Keyshot Render – CAD detailed version (front)

Figure 28: Keyshot Render – CAD detailed version (back)

Figure 29: Keyshot Render – CAD detailed version (details)

Figure 30: Geometry Setup – Repair CAD window.

Figure 31: “Wind Tunnel” bloc domain.

Figure 32: Result of Subtract operation.

Figure 33: Mesh result.

Figure 34: Cutting plane.

Figure 35: Console Output – Drag report.

Figure 36: Console Output – Downforce report.

Figure 37: Manufacturer simulation results.

Figure 38: Siemens STAR CCM+: Simulation Results Representation.

Figure 39: Siemens STAR CCM+: Simulation Results Representation (front).

Figure 40: Siemens STAR CCM+: Simulation Results Representation (back).

Figure 41: (right) manufacturer simulation and (left) dissertation simulation.

Figure 42: HM170 Open Wind Tunnel. Source: EPSEM

Figure 43: (CATIA) CAD of the geometry to print.

Figure 44: preview (measures) of the printed body. Source: 3DStore.

Figure 45: 3D printed model with aluminium supporting stick before painting.

Figure 46: EPSEM’s HM170 Open Wind Tunnel.

Figure 47: Simulation Chamber – HM170 Wind Tunnel.

Figure 48: Initial-Zero adjustment – HM170 Wind Tunnel.

Figure 49: Speed sensor (at 20 m/s) – HM170 Wind Tunnel.

Figure 50: CAD Model prepared for simulation – HM170 Wind Tunnel.

Figure 51: Experimental validation – HM170 Wind Tunnel.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 9
 1. INTRODUCTION
We have all heard the expression “It moves through the air like a brick” when talking
about a car that looks square-ish or old-designed. But really, where does this come from? And
how does this design affect a car’s performance?

Through these past two decades the automotive industry has come a long way, so has
vehicle design, so has the knowledge around it, but most of all, so has the tools that make
possible a better and more efficient design.

The way air moves around a moving object is basically the reason why cars do not
look like giant boxes with wheels. It is well known that if someone were to throw a smashed
piece of paper, or a paper plane, the latter one would be the one which would go further.

The way this vision translates into the automotive world is tied to the science of
aerodynamics and the concept of streamlining. This is achieved by giving the vehicle an
“aerodynamic shape”, concept which will further be discussed. Nowadays, this shape is
analysed with new powerful tools, in the shape of simulation software that allow complete
studies of vehicles without the need of materialising them.

In the following pages the relevant concepts of aerodynamics will be presented and
put under application on a production vehicle from two decades ago. The aerodynamic
simulation performed upon it will provide a clear view of the beneficial and prejudicial decisions
that were made when it was designed and compared with the computational power of the
1990s. In conjunction with a real-world representation of the performed simulation, a complete
overview can be had of the importance of computer-based simulation, and how their results
translate (or not) to reality.

An aerodynamic simulation process is based upon what is called a virtual wind tunnel.
So, parting from this, on this dissertation an experimental part is to be found, where a real
scale-model, will be put inside a real wind tunnel, to validate those results extracted from the
computational work.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 10
 1.1. AERODYNAMICS
Fluid mechanics defines the science in charge of studying the behaviour of fluids both
at rest and in motion. External fluid flow exerts determined forces and moments onto a vehicle,
which in essence affects not only its performance on how much power it needs to cut through
the air, but also in its cooling ability, wind noise, and direction of dirt and water away from
windows and lights.

According to NASA1, aerodynamics is the way air moves around things. Anything that
moves through air reacts to aerodynamics. Aerodynamics acts on cars, since air flows around
them. That way, aerodynamics is a branch of fluid mechanics which deals with the
understanding of how solid objects (such as cars) affect fluid flow and how that fluid in motion
induces forces upon that same body.

Car aerodynamics, just like with other things, come from the aerodynamics of planes,
but here, there is a key difference that must be noted. Planes cannot exist without
aerodynamics, their ability to fly relies on them. On the other hand, a car can and will work
without any aerodynamic design.

The first and most important factor when considering aerodynamics is the shape of the
car itself, since it determines how air flows over and around it. It is not rocket science that
smooth rounded edges let air flow over them better than boxy shapes with sharp edges. With
that clear, one said car with no aerodynamic design taken into consideration, although with
greater effort, could still work just like any other. In fact, aerodynamics on passenger vehicles,
has not always been a priority, and has greatly evolved through time.

Image 2: Turbulent flow around a “bad” shape. Image 1: Laminar Flow around a “good” shape.

Source: Wildstar-Schleswig-Holstein Source: Wildstar-Schleswig-Holstein

1
See reference [1] in Bibliography section.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 11
 1.1.1. EVOLUTION OF AERODYNAMICS
The first proper interest in car aerodynamics can be found as early as the 1900s. As
cars began to gain both popularity (tied to affordability) and speed (tied to powertrain
evolution), engineers started to experiment with their designs in search of further boosting
their performance. May the most famous example of this initiation era be Paul Jaray2 and the
first teardrop body. This revolutionary design made to reduce the air resistance the car made
led to a new interest in improving fuel efficiency and boosting speed through optimizing a car’s
body.

Image 3: Paul Jaray T6 blueprints.

Source: Conciertos y Desconciertos (blogspot)

Through the 1930s, car manufacturers started adopting more streamlined designs,
inspired by the Jaray prototypes and others. All of them featured curved, flowing lines. One
key example, which is widely considered as the first mass produced car focused on
aerodynamics is the 1934 Chrysler Airflow.3

In the 60s and 70s, aerodynamics took a turn to focus more on improve a vehicle’s
handling, rather than as a way of making them more efficient. This is the time at which we find
the first modern applications of spoilers, air dams and diffusers. These elements were adopted
as a way not only to make the car slip through the air better, but also to take advantage of the
same air to make it handle better.

2
See reference [3] in Bibliography section.
3
See reference [4] in Bibliography section.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 12
 With the introduction of computer design and simulations, wind tunnels and new
materials such as carbon fibre, vehicle aerodynamics has evolved a long way, now being a
primordial focus point for car manufacturers.

In essence a vehicle needs power in order to push through the air. Power is fuel and
fuel is money. If a vehicle’s drag is reduced, it makes it easier for it to push through the air,
which, in essence, makes it more efficient. This way, the engine of an aerodynamic vehicle
does not need to work as hard to push the car through that wall of air.

In a modern world where being as fuel efficient as possible is what matters, the need
of a car which can cut through the air, not like a brick, but rather like a drop of water is essential.
Then, aerodynamics become fundamental.

Image 4: 1934 Chrysler Airflow.

Source: Chrysler. Stellantis

Image 5: 2013 Volkswagen XL1.

Source: Volkswagen A.G.

Image 6: 2015 Mercedes-Benz Concept Intelligent Aerodynamic Automobile.

Source: Mercedes-Benz

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 13
 1.2. COMPUTATIONAL FLUID DYNAMICS PROGRAMS
Computational Fluid Dynamics (CFD) programs are computer-based tools to analyse and
predict how liquids and gases will perform under some defined situations. In the automotive
sector, this is used to intelligently predict and further optimize the aerodynamics of cars. CFD
simulations show the behaviour of airflow around a vehicle’s body (and its components).

With the introduction of this technology, now it is possible to test different configurations
and designs, and as such, make as many changes as one desires, without the need of the
expense on real vehicle models prototypes. These programs usually adopt the form of a
“virtual wind tunnel” where a Computed Aided Design (CAD) model is put and analysed,
mimicking the one in real life, with the benefit of it being able to provide detailed information
on a wide range of configurations and situations.

CFD is based upon non-linear partial differential equations which attempt to

computationally model theoretical and experimental models. As it is expected, the accuracy
of CFD results rely on its requirements. Since the goal of this dissertation is to obtain an overall
knowledge of the airflow around the vehicle, as well as drag and lift coefficients, qualitative
simulations will be required. The simulations conducted in this dissertation are going to be
done using the continuum approach method and applying conservation laws; then, ordinary
(or partial) equations can be realized for an infinitesimal control volume.

Finite volume (or finite element) methods, along with initial and boundary conditions
for the model are then used to convert the differential equations into a system of algebraic
equations for them to be solved by a number of succeeding iterations, until convergence, and
as such, a solution is achieved. Next, that same domain is discretized into volume (or
elements). This conjuncture of small elements, which in themselves contain all the information
of that determined differential part of the geometry form what is known as the mesh, which
creates a differential solution that approximates to that one from the original geometry.

Image 7: Mesh example on a race car geometry.

Source: FetchCFD

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 14
 1.2.1. CFD: CONSERVATION

Inside a virtual wind tunnel, there are some basic laws of conservation that applies for
the simulation to work. These are being laid out here.

Conservation of mass: The amount of mass entering the control volume must be the
same leaving it. The amount of mass in the system remains constant.

𝜕𝑢 𝜕𝑢 𝜕𝑢 (1)
+ + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧

Conservation of Linear Momentum: Referring to the relationship between the

pressure, viscous forces and momentum acting in the system.

𝜕𝑝 𝜕2𝑢 𝜕2𝑢 𝜕2 𝑑𝑢 (2)

𝜌𝑔𝑥 − + 𝜇 ( 2 + 2 + 2) = 𝜌
𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝑑𝑡

𝜕𝑝 𝜕2𝑢 𝜕2𝑢 𝜕2 𝑑𝑣 (3)

𝜌𝑔𝑦 − + 𝜇 ( 2 + 2 + 2) = 𝜌
𝜕𝑦 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝑑𝑡

𝜕𝑝 𝜕2𝑢 𝜕2𝑢 𝜕2 𝑑𝑤 (4)

𝜌𝑔𝑧 − + 𝜇 ( 2 + 2 + 2) = 𝜌
𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝑑𝑡

Conservation of energy: energy cannot be created nor destroyed in the system.

𝑑𝑇
𝜌𝑐𝑝 = k∇2 𝑇 = Φ (5)
𝑑𝑡

Where:
𝝆 : Density
𝒈 : Gravity
𝝁 : Dynamic Viscosity
𝚽 : Viscous Dissipation
𝒌 : Thermal Conductivity
𝑻 : Temperature
𝑪𝒑 : Specific heat

𝒕 : Time

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 15
 1.2.2. CFD: REYNOLDS-AVG AND NAVIER-STOKES EQUATIONS
The Reynolds-averaged Navier-Stokes (RANS) equations are time-averaged equations
referring to the motion of a fluid. They are widely used when representing turbulent flow
behaviour. The properties of said flow fives approximate time-averaged solutions to the
Navier-Stokes equations [Cangel and Cimbala, 2009]. A shortcoming of the most RANS
solvers is their inherent incapability to deal with greatly separated flows apparent in many
logical structures [Ferziger and Perié, 2002]

The Reynolds-averaged Navier-Stokes non-linear equations are not analytically solvable. The
RANS approach splits the instantaneous speed and pressure into fluctuating and average
components.

1 𝑇 (6)
𝑢̅ = ∫ 𝑢𝑑𝑡
𝑇 0

𝑝 = 𝑝̅ + 𝑝′ (7)

𝑢 = 𝑢̅ + 𝑢′ (8)

(9)
𝑣 = 𝑣̅ + 𝑣 ′

̅ + 𝑤′
𝑤=𝑤 (10)

From the Navier-Stokes equation, inserting Reynolds decomposition4 will result in the
continuity equation, giving the new fluctuating terms:

𝜕𝑢̅ 𝜕𝑣̅ 𝜕𝑤
̅ (11)
+ + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧

𝜕𝑝̅ 𝜕 𝜕𝑢̅ 𝜕 𝜕𝑢̅ 𝜕 𝜕𝑢̅ 𝑑𝑢̅

𝜌𝑔𝑥 − + (𝜇 − 𝜌𝑢𝑟̅ 2 ) + (𝜇 − 𝜌𝑢′̅ 𝑣 ′ ) + (𝜇 − 𝜌𝑢′̅ 𝑤 ′ ) = 𝜌
𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑧 𝜕𝑧 𝑑𝑡
(12)

New unknown terms known as Reynolds stresses [Cengel and Cimbala, 2009] now exist.
There will cause a closure problem, resulting in these stresses having to be modelled in order
to get a close system on equations. This can only be done by introducing turbulence models.

4
Science Direct: Reynolds decomposition allows the simplification of the Navier-Stokes equations by substituting
in the sum of the steady component and perturbations to the velocity profile and taking the main value. The
resulting equation contains the term known as the Reynolds stresses which gives rise to turbulence. See
reference [5].

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 16
 1.2.3. CFD: LATTICE BOLTZMANN
Another form in which fluid flow analysis can be performed is by considering the
medium to be microscopic.5 At each time step, it is required to identify every particle location
and trajectory. This makes this method impossible to be used for large scale simulations
(moreover, it does not consider viscosity, nor temperature, nor pressure). There are some
other mesh-less numerical methods to solve CFD, such as Lattice Gas Automata (LGA) and
Lattice Boltzmann (LBM), which use algorithms based on a mesoscopic framework. These
two methods have been extensively researched in recent years, especially in the automotive
industry, due to their affinity to computational ability. LBM has the advantage of being easily
applied to complex domains, plus having no needs or requirements to solve Laplace equations
at each time step in order to satisfy the incompressible, unsteady flow continuity (as it is with
RANS).

However, both LGA and LBM schemes require more computing memory compared with a
RANS solver. Then, its main disadvantage is the complexity to theoretically analyse the
emergent behaviour of the system at a macroscopic level, from the laws imposed at a
mesoscopic level [McNamara and Zanetti, 1988], as well as simulations of the incompressible
Navier-Stokes equations, especially for unsteady flows.

1.2.4. CFD: TURBULENCE

Following the dimensional analysis proposed by Kolmogorov at high Reynolds6 numbers,
the flow tends to break into smaller eddies in order to transform the kinetic energy into internal
energy.

This process is known as the Kolmogorov cascade, and it explains the turbulence
phenomenon.

The time necessary to break an eddy in the flow follows the following expression:

𝐿𝑒𝑑𝑑𝑦 (13)
𝑡𝑏𝑟𝑒𝑎𝑘~
𝑣𝑒𝑑𝑑𝑦

The time necessary to dissipate the kinetic energy through viscosity is expressed as:

𝐿2 𝑒𝑑𝑑𝑦 (14)
𝑡𝑑𝑖𝑠𝑠~
𝑣

5
As in made up of small particles which collide with each other.
6
For cars Reynolds is high due to large measures. Turbulent model is applied here.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 17
 Then, for large eddies and high Reynolds numbers, the break time is smaller than the time
employed to dissipate the energy. This is what creates the Kolmogorov cascade.

The kinetic energy of a turbulent structure can be estimated by:

𝐸𝑐 𝑒𝑑𝑑𝑦 ~𝑉 2 𝑒𝑑𝑑𝑦 (15)

The specific kinetic energy dissipation ratio, then:

𝑉 2 𝑒𝑑𝑑𝑦 𝑉 3 𝑒𝑑𝑑𝑦 (16)

𝜀𝑏𝑟𝑒𝑎𝑘 ~ =
𝑡𝑏𝑟𝑒𝑎𝑘 𝐿𝑒𝑑𝑑𝑦

The smallest eddies, present in the flow of size ~𝐿𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 have a break time equal to the
time necessary to transform their kinetic energy to viscous energy (𝑡𝑏𝑟𝑒𝑎𝑘 ~𝑡𝑑𝑖𝑠𝑠𝑖𝑝 ). Then the
kinetic energy dissipation ratio can be estimated by:

𝑉 3 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝑣3 𝑉 3 𝑒𝑑𝑑𝑦
𝜀𝑣𝑖𝑠𝑐𝑜𝑢𝑠 ~ ~ 4 ~ ~𝜀𝑏𝑟𝑒𝑎𝑘 (17)
𝐿𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝐿 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝐿𝑒𝑑𝑑𝑦

Thus,

𝐿𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 −3/4 (18)

~𝑅𝑒𝑒𝑑𝑑𝑦
𝐿𝑒𝑑𝑑𝑦

Knowing this relationship, in order to explicitly solve each and every eddy in a three-
dimensional flow, the number of elements ends up being in the order of:

𝐿𝑒𝑑𝑑𝑦 3 9/4 (19)

𝑁𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 ~ ( ) ~𝑅𝑒𝑒𝑑𝑑𝑦
𝐿𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙

The RANS approach models turbulences in a global way, making it the most widely
adopted approach for this type of application. As it calculates averaged values in time, the
approach removes the time dependence of the solution. Although calculating averaged results
is computationally less expensive, new terms appear in the Navier-Stokes equations which
must be modelled by new transport equations. Moreover, there are several RANS models,
which suit different specific problems, whose parameters need to be adjusted empirically.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 18
 1.2.4.1. TURBULENCE: SPALART-ALLMARAS

The Spalart-Allmaras (SA) turbulence model was used in the Reynolds Averaged Navier
Stokes Equations model. This model has been found to be suitable for flow over structures,
shear layer detail, as well as attached wall bounded flows.

Its disadvantages include however, an important inaccuracy for greatly separated flows,
degenerating turbulence and internal flows [Ashgriz and Mostaghimi, 2002].

The transport equation for the kinematic viscosity (𝑣̃) is the working variable for the
Spalart-Allmaras model, given by:

𝜕(𝜌𝑣̃) 𝜕(𝜌𝑣̃𝑢𝑘 ) 1 𝜕 𝜕𝑣̃ 𝜕 𝜕𝑣̃

+ = 𝑃𝑣 + [ { (𝑣 + 𝑣̃) } + 𝐶𝑏2𝜌 ( ) ] − 𝐷𝑣 (20)
𝜕𝑡 𝜕𝑥𝑘 𝜎𝑣 𝜕𝑥𝑘 𝜕𝑥𝑘 𝑥𝑘 𝑥𝑘

The turbulent viscosity and viscous damping are given by:

(21)
𝑣𝑡 = 𝜌𝑣̃𝑓𝑣1

𝑋3
𝑓𝑣1 = (22)
𝑋 3 + 𝑐 3 𝑣1

𝑣̃
By having 𝑋 ≡ 𝑣 , the turbulent production term in Equation (20) is given by:

𝑃𝑣 = 𝐶𝑏1 𝜌𝑆̃𝑣̃ (23)

Where the modified vorticity is defined as:

𝑣̃
𝑆̃ = 𝑆 + 𝑓 (24)
𝑘2 + 𝑑2 𝑣2

𝑋
𝑓𝑣2 = 1 − (25)
1 + 𝑋𝑓𝑣1

With 𝑆 ≡ √2Ω𝑖𝑗 Ω𝑖𝑗 , being the magnitude of the vorticity, the vorticity is found from the
mean rate-of-rotation tensor:

1 𝜕𝑢𝑖 𝜕𝑢𝑗 (26)

Ω𝑖𝑗 = ( − )
2 𝜕𝑥𝑗 𝜕𝑥𝑖

The variable 𝑑 is the distance referent to the nearest wall applying in the standard RANS
model. That way, the turbulent wall destruction term in Equation (20) is defined by:

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 19
 𝑣̃ 2 (27)
𝐷𝑣 = 𝐶𝑏1 𝜌𝑓𝑤 [ ]
𝑑

6 1/6
1 + 𝐶𝑤3 𝑣̃
𝑓𝑤 = 𝑔 [ 6 ] 𝑔 = 𝑟 + 𝐶𝑤2 (𝑟 6 − 𝑟) 𝑟=
𝑔6 + 𝑐𝑤3 𝑆̃𝑘 2 𝑑2
(28)
𝐶𝑏1 (1 + 𝐶𝑏1 )
𝐶𝑤1 = 2 +
𝑘 𝜎𝑣

The Spalart-Allmaras parameters input into FLUENT by the user do not include 𝑘 or 𝜎𝑣 but the
rest of the mentioned parameters can be modified with the values used given inTable 1.

𝐶𝑏1 0.1335
𝐶𝑏2 0.622
𝐶𝑣1 7.1
𝐶𝑤2 0.3
𝐶𝑤3 2.0
𝜎𝑣 2/3
𝑘 0.4187
𝑃𝑟 0.667
Table 1: Spalart-Allmaras parameters.

1.2.5. CFD: BOUNDARY LAYER

When there is a fluid flow next to a geometry, that is, a body, a boundary (limit) layer of a
slower moving fluid is created near the surface. That way, the fluid velocity closest to the body
is equal to that of the solid itself. This is known as the non-slip condition.

The velocity of the fluid increases toward the freestream velocity as it moves further from
the surface of the body.

That way, the boundary layer thickness is defined as the distance from the surface to
where the velocity of the fluid reaches 99% of the freestream speed. Just like with fluid stream,
a boundary layer can be categorised into three types of flow: laminar, turbulent and a
combination of both. The fluid flow can be said to start as laminar, and as it progresses along
the length of the body, the flow becomes turbulent.

Wall functions assume that the flow near the wall behaves in a fully turbulent manner and
use an algorithm to resolve the gradients in the boundary layer [Fluent, 2012].

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 20
 1.3. STUDY SUBJECT: R53 MINI COOPER S

The subject of study on this dissertation is the 2002-2006 Mini Cooper. Specifically, the
Mini Cooper S, under the chassis code R53.

Image 8: MINI Cooper S R53 (front).

Source: Bring a Trailer

The R50 MINI was introduced to the market in 2002 as a successor of the original Mini.
Now under the parent company BMW, this new B-segment compact car had the task of
making a modern reinterpretation of the original formula, with retro styling that harked back to
the 50s model but adopting a more up-market premium position.

The R53 MINI Cooper S presented itself as a sporty and performance version of the
“standard” MINI. The first generation of what is known as the “new MINI” had three versions:
the base model ONE and the COOPER, which shared the chassis code R50, and then, the
COOPER S, under the chassis code R53.

Frank Stephenson7 was the man in charge of reinventing the original MINI design which
had not been altered for over 40 years. The idea was simple, to maintain that same identity of
the MINI yet it had to look modern and feel as an evolution.

7
See reference [6] in Bibliography.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 21
 Image 9: John Cooper with the original MINI Cooper (left) and the R50 MINI Cooper (right).

Source: MINI

Image 10: Original MINI (far back) and R50 MINI Cooper (up front) in Solid Gold Met.

Source: MINI

In order to maintain that ethos of a very compact car on the outside, the front and rear
overhangs8 of the car were reduced to the minimum and studying its proportions from a side
point of view, it was designed to look sort of like a “Three-tiered cake”, formed by the roof, the
windows, and the body itself.

Image 11: Visual representation of "Three-Tiered cake" MINI.

Source: Frank Stephenson

8
According to Autolexicon: overhang is the distance from the wheel centreline to the furthest point on the
front and rear of the car.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 22
 One characteristic element of the original MINI is a diagonal “joint” line on the front fender.
This line was maintained on the so called New MINI, which caused the bonnet to become what
is known as a clam shell, in which the bonnet itself is the front of the car. From an aerodynamic
point of view, it is important to rake elements backwards (or lean them to a certain degree),
so, in order to maintain the rounded headlights with the two power bulges from the original
vehicle, this were tilted backwards and made flush with the rest of the bodywork which became
soft and flowing on purpose, including a power dome in the middle to make clearance.

Image 12 and 13: Fender joint and bonnet bulge.

Source: Bring a Trailer

A curious aspect about the first-gen New MINI is that if someone were to extrapolate and
extend all the side lines that conform the bodywork, all of them would cross at the same
vanishing point.

From the back, one trait of the MINI that sets it apart from nearly any other car on the
road is the position of the taillights. Normally, the placement of the taillights is always done in
such a way that they somehow line up with the shut line of the tailgate. In the MINI however,
this element is pushed aside and isolated and put inside the rear fender, giving the car a wider
look.

Image 14: MINI Cooper S Sketch.

Source: Frank Stephenson

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 23
 Specifically talking about the Cooper S, a set of modifications were made to the
powertrain, suspension, gearbox, and body. Due to the introduction of a supercharger and an
intercooler to the engine, a hood scoop was introduced, and with it, the whole bonnet has
raised to make room for it.

The detailed design overview and explanation from an aerodynamic point of view will
be described in further down.

1.3.1. THE MINI COOPER S R53: DESIGN OVERVIEW

In this following chapter a detailed overview of the design characteristics of the R53
Cooper S are to be explained. Important traits needed to take into consideration when
designing and replicating the body into a CAD model.

As explained above, the MCS is the high-performance version of the regular Cooper,
and apart from performance upgrades, it also included visual differences and personal traits
to make it stand out compared to the standard model.

Image 13: MINI Cooper (left) and Cooper S (right) front.

Source: EVO Magazine and Bring a Trailer

Image 14: MINI Cooper (left) and Cooper S (right) back.

Source: MINI

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 24
 At the front, the R53 Cooper S presents a very short nose-end, with a clam-shell bonnet
dominated by the two power bulges emerging from the wheel arches and extending to the
very front of the vehicle. This bulges act as the housing where the headlight units are.
Aerodynamically speaking, this trait is to be highlighted. The original MINI also presented this
type of front-end, but it did so in a rather vertical way. In order to make the car more
aerodynamic, all this design characteristic, while being carried over, it was adapted, tilted
backwards and made flush with the rest of the bodywork. By doing so, the lines end up flowing
better around the bonnet, integrating them into the design.

The MCS also added a central hood scoop to let air into the intercooler for the
supercharger. This caused the bonnet to be raised up, so the overall attack angle of the front
end changed. It can be perceived even by the naked eye that the bonnet of the S variant
seems higher and bigger, that is, because it raises more vertically before reaching the slant,
flat zone where it meets the windscreen.

The grill of the vehicle was included into the bonnet, flush with it and with horizontal
rounded openings that led air into the radiator. There is a second fake grill embedded into the
bumper with no openings. This, in fact, is aerodynamically beneficial as the air will flow around
the front of the body better than if it had to go through a meshed grill into the engine bay.

The frontal area of the MINI presents itself as a rather big one for a vehicle it size. This
is due to a quite steep windshield. Apart from this, it is also quite squared when projecting its
frontal area into a plane (1688x1408 mm9). This trait, added to that steep windshield probably
is going to affect quite negatively the overall aerodynamic performance (simulation pending).
The fluid stream flowing around the front of the vehicle above the bonnet is going to suddenly
find a “nearly vertical wall” (windshield), which is going to make its flowing speed decrease
significantly, needing more time to travel around it, causing a boost in pressure, which in
essence is going to increase the overall drag of the geometry.

The roof of the R53 Cooper S is nearly horizontal, having only a slight negative angle
(descending) with the horizontal plane. The fact of it being this flat is probably going to, again,
harm the drag performance of the vehicle. Even though the air would easily attach to the
surface of the roof thanks to it being so flat, it will probably show to problems. The first one
where the top of the windscreen and the roof meet, and the second, where the roof ends, and
meets with the boot. This problem will be graphically shown and proven further down in this
dissertation.

9
Measures extracted directly from manufacturer (BMW-MINI)

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 25
 The R53 incorporated a roof spoiler with an opening in the middle, which gave the car
a sportier look. The brand never gave word of whether this component was aerodynamically
functional or if it just had cosmetic purposes.

All the previously mentioned aerodynamic assumptions on the vehicle are going to be
later contrasted and validated with the corresponding simulation.

When analysing the vehicle’s design from an aerodynamics point of view, it must be
said that the rear of it plays a rather small paper in the overall results it will probably output. In
the R53 MCS this concept is further accentuated by the flat shape of the roof, and by the
verticality and abrupt ending of the rear of the vehicle. What this will bring is for the fluid stream
to suddenly detach from the surface of the body when the roof ends since it is not able to find
a continuous adjacent surface to follow, and due to the speed it carries, it follows tangent,
creating separation from the geometry.

At the rear, there was a new bumper, with now dual centre exhausts, and a set of fake
vents.

The Cooper S sat lower to the ground compared to the standard Cooper due to a
lowered suspension.

The rest of the body remained the same, and no further changes were made that could
make a change in its aerodynamic behaviour.

The decision of choosing of this vehicle has not been at random, but rather taking into
account a series of specific traits.

First, the vehicle is 21 years old, even older if taken into consideration the years of
design and product development needed to make a production vehicle. In more than two
decades, CFD programs have come a long way, providing way more complete and precise
information about a car design concerning aerodynamics. This gives the opportunity of
experimenting just how much of a difference there is in the information given by such programs
in 20 years, plus trying to explain some design choices made then, that now could have been
made different.

The second motive of choosing this specific car is that it is the real access to it in real
life, as an actual personal vehicle, which gives the chance of being able to fully explore and
analyse every corner of the subject without any compromise. This is very important since an
accurate 3d model of the whole car is going to be made.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 26
2. AERODYNAMIC CONCEPTS
It has already been established that streamlining would seem to be very important
when designing a production vehicle, after all it is wanted that the car moves more easily
through the air. In order to fully understand the aerodynamics of a car, this section describes
the basis upon from which the study conducted in this dissertation is set.

All experimental observation of a fluid in motion show that it becomes stationary the
moment it comes into contact with a solid surface. This is known as the non-slip condition.
The flow areas surrounding the body are in fact affected by the fluid’s viscosity, causing as a
result a velocity gradient formed close to the surface. This thin layer where the velocity
gradient is notable, is known as the boundary layer and increases from zero at the surface to
the speed of the input free stream fluid. This layer’s thickness increases away from the surface
of the body as it travels along from the front to the rear.

Fluid flow in the boundary layer can be divided and categorized into two main groups:
laminar flow and turbulent flow.

Laminar flow is the one which normally can be found around the front of the vehicle,
where the flow is steady and almost parallel to the “wall”.

As the air flow proceeds to travel along the geometry, the boundary layer increases in
size (due to the increase in kinematic viscosity) and as such, the turbulent flow makes its
presence.

The transition between these two states of flow is determined by what is called the
Reynolds Number, a dimensionless parameter which in fact determines that same point of
change in a fluid by showing a ratio of forces, given a specific length.

For this case, the Reynolds Number is determined by the speed of the vehicle, the
kinematic viscosity of the fluid and the length of the vehicle. It should be noted that if a scaled
model presents the same Reynolds Number as a full-scale one, the results of such simulation
will yield the same aerodynamic coefficients. It is well-accepted (for simulations concerning
commercial vehicles) to not consider compressibility. Since Reynolds Number on a vehicle
are usually high (due to the length of the body), flow regimens are assumed to be turbulent.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 27
 2.1. AERODYNAMIC CONCEPTS: SEPARATION AND WAKE
Flow separation is a phenomenon which can occur at some point along the surface of
a body when the boundary layer detaches from the solid. This can be due to a sudden
deceleration in the flow stream (pressure gradient), causing a large increase in drag on the
vehicle. Traits to pay attention to when trying to avoid flow separation are sharp edges, high
angles of attack and so on. Basically, flow separation happens when inside the laminar layer,
the fluid’s momentum cannot handle the increasing pressure and detaches from the surface
of the body. On the other hand, when there is a turbulent layer, the momentum is far greater,
letting the flow overcome the adverse pressure gradient, avoiding separation.

It is normal however, for the fluid to tend to separate as it drives along the body. So,
the obvious solution is to try and turn the laminar flow into turbulent to prevent (or at least,
delay) that separation.

Drag and Lift forces are influenced by the free stream velocity, that is, the speed of the
fluid at its free-flowing state, when there is no influence of a body. They are also influenced by
the fluid density, the geometry and orientation of the body as well as its position relative to
other bodies.

In this dissertation, the performed study puts the body in isolation, with no other
possible interaction with other objects, but stream flow and boundary layers could still vary if
they are influenced not only by the main body, but by the air entering in contact with the ones
around it.

2.2. AERODYNAMIC CONCEPTS: FORCES AND MOMENTS

Vehicle dynamics refers to the motion of a vehicle and the different forces acting upon
it when moving. A total of six components or degrees of freedom (DOF) add up to making the
resulting forces and moments experienced by the car. In normal conditions of a symmetrical
flow, the vehicle experiments drag along de X axis, and lift along Y, accompanied by a moment
M in the form of pitching in Z. 10

An important factor that should be noted is regarding the centre of gravity. The lower
the vehicle’s centre of gravity is, the more stable will be.

In asymmetrical flow field (conditions with side winds), additional forces and moments
will appear, such as, a side force along Z with a rolling moment R in X, and a yawing moment
N in Y.

10
X axis along the length of the vehicle, Y axis along the height of the car.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 28
 2.3. AERODYNAMIC CONCEPTS: FLOW COEFFICIENTS
Flow coefficients are a way of expressing with numbers how a body reacts to the forces
and moments acting upon it when put under a fluid flow. The most commonly used ones, and
the ones this dissertation will mainly focus on are Drag Coefficient (Cd) and Lift Coefficient
(Cl). Similar to these mentioned, pitching moment (Cm), side force (Cs), yawing moment (Cy)
and rolling moment (Cr) can also be calculated.

Here are described the formulas used to determine Drag (Cd) and Lift (Cl) coefficients
in this dissertation.

2𝐹𝐷
𝐶𝐷 = (29)
𝜌𝐴𝑣 2 ∞

2𝐹𝐿
𝐶𝐿 = (30)
𝜌𝐴𝑣 2 ∞

Where:

𝑭𝑫: Drag Force

𝑭𝑳 : Lift Force

𝝆: Density

𝑨: Frontal Aera of the body

𝒗: Velocity of free-flowing stream

Inside the automotive scope, and when talking about a production vehicle, Drag
Coefficient normally is tried to be reduced to the maximum. A Cd of, for example, 0.5 is quite
easy to achieve without considering next to nothing in aerodynamic terms, however, this would
result in a noisy, unstable and not efficient at all vehicle. In essence, a bad designed vehicle.
A Drag Coefficient of 0.3 however, although could not seem as that big of a difference, requires
a very fine design as well as several months of testing and multiple design iterations. As of
today, the most aerodynamic production vehicle there has ever been is the Volkswagen XL1,
with a Cd of just 0.19.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 29
 2.4. AERODYNCAMIC CONCEPTS: AERODYNAMIC DESIGN
Cause and effect. Relating these two concepts is, in essence, the main purpose of an
aerodynamic study on a vehicle. At the end of the day, if each and every small aspect of a
vehicle could be optimized, the vehicle would achieve the minimum aerodynamic drag
possible. That, however, is not really possible. Due to the high interaction in between the
different parts on a vehicle, it is impossible to simply analyse an isolated element, optimize it
to the max and then expect it to perform just as good when put back together. Furthermore, it
is impossible to then optimize to the maximum again when assembled and then expect nothing
else to change. That same always changing interaction is really what limits the method of
detail optimization.

However, various zones of a vehicle can be defined in such a way to reduce its overall
drag.

For further and better explanation, and example is set below, explaining how these
zones can be worked on for a Hatchback-style vehicle, just like the subject treated on this
dissertation.

1. Round front end

2. Optimised cooling duct
3. Bonnet slope
4. Windscreen slope
5. Roof camber
6. Rear slope
7. Rear diffuser
8. Covered up wheels
9. Smooth underside
10. Wound wheel-sell
11. Wheel fairing
12. Top view taper
13. Windscreen curve
14. A-pillar round
15. C-pillar inswept
16. Rear-end boat tail

Image 15: elements of an aerodynamic design.

Source: Volkswagen AG

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 30
 In general terms, a low-drag shape for a vehicle starts with a streamlined body, which
in itself contains the main measures of the car’s body, that is, length, height and width. Then
again, this shape should be as it follows: a flat but rounded front, a curved windshield and a
tapered rear. Upon this basic general shape, details and different characteristic of the specific
model are added onto it and further optimised.

It is well known that any shape modification on a vehicle intended to improve drag
resistance, may cause a negative effect on the lift force and yaw moment the car experiments.

3. GEOMETRY DEVELOPMENT
For the sake of proceeding with the main objective of this dissertation, a full-scale 3d
virtual body of the study subject was needed to be modelled. This has been done via a CAD
software called CATIA V5, by Dassault Systems.

3.1. DESIGN ENVIRONMENT

In order to model an accurate representation of the R53 MINI Cooper S, both in shape
and size, some design environment decisions were crucial: obtaining both the general official
measurements of the vehicle, and a faithful blueprint with the principal views (Top View,
Lateral View, Front View, Back View).

R53 MINI COOPER S MEASURES

Length 3655 mm
Width 1688 mm
Height 1408 mm
Wheelbase 2467 mm
Front track 1453 mm
Rear track 1460 mm
Table 2: MINI R53 CAD Measures.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 31
 Image 16: R53 MINI Cooper S blueprints used for CAD design.

The blueprints were then scaled to correct proportions and put inside a CATIA file
environment. These will act as a guide in the background from where draw lines which follow
the ones matching the bodywork.

Image 17: Layout of blueprints in CATIA environment.

It must be noted that only half of the automobile has been design, following the centre
line along the X axis (lengthwise).

This way, only half of the computational power is needed to design it, and half of the
hours are computed. After that half is finished, a simple symmetry of the whole geometry is
performed in order to have the complete vehicle.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 32
 Through 3D-Splines11, which is an operation made for drawing 3d curves in the space,
all the different important lines, curves, and creases of the bodywork have been reproduced.
This way, a “mesh” with all the important lines in order to make the vehicle’s surfaces is
created.

Image 18: Original Master Geometry Lines.

In favour of more faithful surfaces which fit better when compared with the ones from
the original model, additional lines are created and further divisions are made, so the surface
to fill are smaller, which in essence, makes possible perform operations with a tighter
precision.

Image 19: Final version of Master Geometry Lines.

11
Name of the CATIA operation that allows the creation of multidirectional curves on the space with the direct
introduction of points without a positioned sketch limiting the drawing to a plane.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 33
 With all the necessary 3d curves being laid out, the surfaces of the bodywork can start
to be created. It is important for these surfaces to be tangent between each other, that is, that
between one and the adjacent, the difference in angle is zero, so the jump or joint between
them is seamless.

The surfacing process of the bodywork has been divided into sections, each one
corresponding to a different part (or panel) of the vehicle. This way, the design process of the
CAD has been structured as it follows: Front Bumper, Bonner, Side, Back Bumper, Back12,
Windows, Roof and Arches13.

The above shown surface lines correspond to the final iteration of the designed
geometry, from which the simulation has been performed. The final version is the result of
three main iterations (explained below) and an overall simplification of the latter in order to
have a suitable body for the program to be easy to compute.

12
Back consisting of: Trunk door and rear headlight zone emplacement.
13
Arches being the black plastic trim surrounding the vehicle around the wheels and the bottom.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 34
 3.2. GEOMETRY DEVELOPMENT: INITIAL DESIGN
The consideration for an initial design must contemplate a closed overall surface (or a
determined number of joined surfaces). The drawn 3D lines must follow roughly the shape of
the original projected14 body. The surfaces of each small section must not intersect with each
other nor present openings in between them. It is recommended that there is no jump in
between surfaces, imposing a tangency constraint when possible.

Ensuring tangency in between surfaces makes harder to appearance of creases and

small angles at the border of these, these small angles and sharp edges are potentially
problematic when launching a simulation.

The initial design is divided into eight main design sections: Bonnet, Side, Rear,
Windows, Front Bumper, Rear Bumper, Roof, Arches. Each of these main zones (worked
upon separately) is then divided into small sections, delimited by design lines in order to
ensure a more accurate surface treatment.

Image 20: Initial Design - Body parts.

As it can be seen in the images above, all the correspondent surfaces of the body were
successfully created, however due to poor line placement in this first preliminary version, some
parts of the skin present weird, and unsmooth faces. Some of the joints present ridges.

14
Body projected as blueprints with real measures. Always present on the background of the environment.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 35
 As explained before, only one half of the vehicle is actively designed, and then the
other side is performed through symmetry, and here, when performing this action, some gaps
appear in between the left and right side of the figure.

This gap, although being of 0,2 mm at the point where the surfaces are further apart, must be
brought down to zero in order to achieve a complete enclosed.

In this first design, the wheel arches were not a separated group, but instead they were
part of the “Side” group. In the next iteration this is changed due to internal intersections and
lack of tangency between surfaces.

This first design iteration serves as a baseline from which improve the body. The
following conclusions can be extracted from this version: The 3D splines are accurate
enough15 to be able to emplace the surface necessary to materialize the body of the vehicle.
There is a need of more design lines (smaller sections) due to poor surface output. There is a
symmetry gap that needs to be addressed. The body must be completely closed, so grill
openings, exhaust hole, headlights and undertray must be closed out. The roof of the vehicle
precises of special attention due to the impossibility of creating a surface with at the current
state, so this first iteration does not have a roof surface. The way in which the design lines
meet the symmetry plane (y=0) must be as horizontal as possible so when performing the
symmetry this is seamless and no “symmetry ridge line” is visible.

15
In relation to the blueprints of the original vehicle.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 36
 3.3. GEOMETRY DESING: SECOND ITERATION
This first revision of the original design focuses on fixing all the problems explained
above. More specifically, on closing all the openings (grill, headlights…) on the surface of the
vehicle, and creating and enclosed underside.

The main and relevant change this iteration presents, is the inclusion of 37 more 3D
design lines. With these add-ons, surfaces such as the boot lid improve from being just one,
to now being formed by four different surfaces, all tangent between each other. This allows
each surface to not suffer from too much tension in its skin (issue present before). Surface
stress limits the way a surface can develop and follow a determined style. By “relaxing the
surface” making it smaller, it can follow splines and sudden curvature changes in a more
accurate manner, better following the reference blueprints and creating a more realistic model.
There is a noticeable change in some geometries, where in the first iteration, there where lack
of tangencies between the skins, or unsmooth panels, here there is a more consistent
continuous coat all over the body.

Relevant changes the second iteration presents in its design are the segregation of the
wheel arches from the side panel of the vehicle. The rear corner of the vehicle (where the
taillight would be emplaced) has been moved to the side group, previously being in the rear
section.

Image 21: Second Iteration - Full body.

This is due to the need of a more accurate tangency with the side panel that sit on top
of the rear wheels than the one with the tailgate, as this is a movable panel, and the union
here is not as tight and flush as with the other two. Putting them under the same Geometrical
Set16 allows for the Fill operations for each surface to set more precise tangencies when
defining them.

16
Geometrical set is the given name to a folder inside the environment. This CAD is divided into Geometrical
Sets, one for each defined zone of the vehicle.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 37
 As where before every section of the design had the symmetry embedded inside in a
way that when the section was finished, the symmetry was performed, this time around only
one half of the MINI is completely designed and each section (e.g., bonnet) is joined. After all
of them are joined separately, the completed half body is joined under one surface. Then, the
symmetry is performed. This way, there is only two main surfaces to account for, as to before,
there were hundreds of them, which could present microscopic openings that were not
noticeable. The Join operation groups all the surfaces and eliminates those microscopic
openings that the software can compute on its own without modifying the input design.

If the geometry is to be put under simulation, it must include an enclosed underside.

Otherwise, the air hitting the vehicle would enter the “hollow skin” the CAD archive is. Here, a
new Geometrical Set is created where all the operations to create an underside will remain.
The underside of the geometry must be flat and tangent all around the centre of the vehicle.
This first version of the underside contemplates open wheel arches with encased wheel wells.

Image 22: Second Iteration - Undertray design.

At this point, the top surface of the car contemplated both sides, so the underside was
performed on the total width of the R53. A Join operation is performed containing all the
surfaces that conform it, to try and ensure no gaps (open profiles). After this, a final Join
operation is done to performed to put together the underside with the body itself, creating only
one surface for the whole vehicle.

Image 23: Second Iteration - Full body with undertray assembly.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 38
 At this point, the first simulation of this dissertation is performed, as a test, to check if
at this point a suitable geometry has been achieved, and to see if it is one that can create logic
output data to study the vehicle’s aerodynamic and calculate its aerodynamic coefficients. The
results show a deficient geometry, with gaps in its symmetry plane, as well as in the wheel
wells. The exhaust opening at back, which it was believed to not be a problem, as it is not
inside the air inlet influence zone, turns out to be one, so it needs to be closed as well.

The following conclusions can be extracted from this second design iteration: the
overall shape of the bodywork is correct, closed, and accurate; no more design lines need to
be added. There is still a gap at the symmetry plane that needs to be addressed. The exhaust
opening at the back must be closed, although not being in an inlet zone, the air flow in the
simulation does not know how to behave around it. Small openings are still present on the
geometry that cannot be addressed by joins; they need to be closed by hand modifying the 3d
splines. The wheel arches will be closed out on the next iteration due to weird turbulences
inside of them due to being empty (with no wheels).

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 39
 3.4. GEOMETRY DESIGN: THIRD ITERATION
The third iteration of the CAD design is an evolution from the second archive, focusing
on the points and elements creating problems when traying to launch a simulation. So, this
version is the final one (simplified) and the one used to run the simulations the dissertation.
This one is, as explained, a simplified and smoothed out geometry, containing the main design
features of the vehicle, so the CFD software can perform its calculations easier and without a
lot of external elements that really do not affect aerodynamic behaviour but rather, they just
complicate the computations.

After deeply studying the first two design iterations of the prototype, a clear point was
made. In order to make sure the geometry is a complete fully enclosed shell; the body must
become a solid, which means making the now hollow vehicle (made of surfaces with no
thickness) into a chunky solid. In order to do so, and with help from a CATIA tool highlighting
the opening between surfaces that a Join operation cannot eliminate, each zone has been
closed by hand. It must be noted that not only there were openings, but also overlapping. This
occurs when the 3d spline goes over the symmetry plane, so the surface, and its consequent
symmetry extend far too much over the Y axis, interfering with each other, that is, overlapping.

Image 24: Third Iteration - Full enclosed (solid) body.

Going back to the beginning of the design process, the 3d design lines have been
moved, point by point until the surfaces on the symmetry, where the gaps or overlapping were,
disappeared.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 40
 This was checked by performing again ang again a Join and seeing if the line through
the middle of the car (on the symmetry plane) showed up. Once all the errors were addressed,
a Closed Volume operation (creates a solid body if the selected surfaces performed a fully
enclosed volume with no openings or overlapping) is performed on the one surface the Join
has created. It must be noted that the Close Volume operation cannot be carried out if the Join
output is not a perfectly sealed volume.

Image 25: Third Iteration - Full solid body section.

This final iteration is both accurate and suitable for a CFD simulation, as confirmed
later by the SIEMENS STAR CCM+ software17. No more design evolutions are needed in
order to perform the study this dissertation is intended to perform. With that said, there is still
another iteration, or design perfection per say, to be made.

This CAD is later going to be exported in a universal CAD archive format (.stp) and
materialized via 3D printing in order to make an experimental validation of the results obtained
by the performed simulations.

17
See Chapter 5.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 41
 3.5. GEOMETRY DESIGN: DETAILED VERSION
This next part was not initially part of this dissertation’s content, as the simulation process
(main part) does not require a detailed geometry, but rather a simplified one. Considering
however, that there was already a proper geometry from which a detailed CAD version of the
vehicle could be created it only seemed reasonable to take advantage of it.

This way, taking the second iteration of the simplified CAD geometry18 a number of
features have been drawn, designed and incorporated into it to create a new detailed iteration
of the virtual R53 MCS. It must be noted that this exercise is merely to put into practice the
CATIA knowledge and skills acquired until this point, to create a realistic looking vehicle
parting from images as a reference.19

Image 26: Keyshot Render - CAD detailed version (front).

The most important changes performed to this version are the re-introduction of some
characteristic elements of the Cooper S, such as the hood scoop, the chromed front grill and
the rear centre twin exhaust. Both front and rear bumper have been further developed, to
incorporate an opening for the radiator as well as the indicators (front) and the fog lights (front
and rear). Both headlights and taillights have been modelled as well.

The wheels have been modelled separately and incorporated to the body by creating
a product file in CATIA, with the body, and the four wheels as separated independent parts.

18
Chosen because of its opened wheel arches to which wheels can be incorporated.
19
Realistic with the limited time dedicated to it, with more time, the output result would have been better.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 42
 Image 27: Keyshot Render - CAD detailed version (back).

The headlights and taillights have been modelled on their insides to, having different
lenses and an intricate light cluster including all the corresponding parts. They also have a
layer on top covering them (acting as the light glass).

In order to better show the results of the work put on the detailed CAD, a rendering process
has been performed with Keyshot 7, a rendering program which allows the creation of
photorealistic images with applied materials from previously design CAD archives.

Image 28: Keyshot Render - CAD detailed version (details).

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 43
4. CFD VALIDATION
The model chosen for testing and comparison is a simplified version of the CAD full-
scale geometry in its final functional version. The shape of the body is free from wheels and
accessories, but it still preserves the primary flow behaviour of the production vehicle. The
virtual simulation on this dissertation has been done via the CFD specialized program
SIEMENS STAR CCM+, which is a Multiphysics Computational fluid dynamics software for
simulating products operating under real-world conditions developed by Siemens Digital
Industries Software. The simulation has been performed without interruptions and has run for
1000 steps at least. It has been repeated three times in the exact same way in order to ensure
a steady output. The given results have been consistent and plausible, so the simulation is
approved to be well executed.

The following sections describe the common setup steps and process for all the
performed simulations.

4.1. SIMULATION: GEOMETRY SETUP

The first step of the simulation setup is to create an empty simulation environment, to
which a new geometry (empty model) is added. The 3D-CAD Model will then be imported to
it. In this case, the CAD is imported in .CATPart20 form. Once imported, the geometry is
checked up via the Repair CAD option to make sure it is up to the program standards and that
it will not encounter any shape, line, crease or angle that it cannot compute. If this is the case,
the repair tool will highlight the area or defect. For the geometry in case, the output turns out
as an overall “zero”, which is the correct output, meaning no actions are needed to repair the
geometry in order to launch the simulation.

Image 29: Geometry Setup - Repair CAD window.

20
Native CATIA archive extension. Universal .stp or .igs files can also be used as the software detects all of
them, but no difference on the results have been found.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 44
 The 3D-CAD Model is then assigned as a new geometry part, so it is treated as one
single part for the rest of the setup process. Now a new geometry scene is created. A scene
is like a display option to which any input can be added and seen through the simulation. By
adding a geometry scene, all the operations performed on it will appear here visually.

It is important to note one key common aspect of all performed simulations. They are
only considering one half of the vehicle. This is way only half of the computational power is
needed to run the simulation, so the results will be extracted much faster. Although importing
the whole geometry, half of it will later be extracted and the boundary conditions only applied
to the remaining one. This has been done this way because it has been proven to be easier
for the Siemens Star CCM+ software to compute a Subtract operation of half a body, than
encountering only one half of it from the beginning.

The next step is to create the “Wind Tunnel”, that is, a domain to which the boundary
conditions will be applied. In order to do so, a block shape part is created with the following
measures. These measures have been chosen considering the general considerations when
designing a simulation space21, that is, approximately 5 times the reference length of the body
behind and about 2 times at front. Enough space should be left on top for the “non influenced”
air to flow free.

Image 30: "Wind Tunnel" block domain.

As explained before, this domain will run across the symmetry plane of the vehicle
(lengthways), so this way only the left-hand side (driver’s) will be computed in the simulation.
This domain will then be split by patch in order to define each section as it follows: Inlet, Outlet,
Road and Walls.

21
As explained in the Fluid dynamics subject (Q4), Automotive Engineering degree UPC

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 45
 With the domain (having half of the car inside) created, the next operation to perform
is a Subtract of that half of the vehicle which remains outside, from the limit of the domain.
Now, only that left-side surface remains into the scene, and now it is hollow. In order to fully
see how this, to be computable, geometry perform and evolve, a new scene is created with
the Subtract inside of it.

Image 31: Result of Subtract operation.

4.2. SIMULATION: MESHING

The process of meshing on a CFD simulation consists of making the software
understand the geometry as a series of finite elements that can be computed separately,
rather than just a line and a plane. For this simulation, an Automated Mesh is created for the
whole geometry, but this has had its parameters and conditions touched up to create a high-
quality mesh. The mesh has been performed upon the Subtract part.

The Automated Mesh has been included with the following optional elements and
parameters to ensure a better quality of the elements across the surface:

Surface Remesher: re-meshes the initial surface to provide a quality discretized mesh
that is suitable for CFD (triangular based). It is used to retriangulate the surface based on a
target edge length. It can also omit specific surfaces or boundaries preserving the original
triangulation from the imported mesh.

Automatic Surface Repair: Performs automatic Repair operations on the mesh for each
iteration other operations perform to ensure a leak-proof mesh.

Trimmed Cell Mesher: Generates a volume mesh by cutting a hexahedral template

mesh with the geometry surface. It is useful in modelling external aerodynamic flows due to
its ability to refine cells in a wake region.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 46
 Prysm Layer Mesher: The mesher projects the core mesh back to the wall boundaries
to create prismatic cells next to it. This layer of cells is created next to wall boundaries to
improve the accuracy of the flow solution.

As for the conditions the mesh must contain, these go as it follows.

- Base element size: 0,1 m

- Target Surface size: 100%
- Minimum Surface size: 0,4
- Number of Prysm layers: 10
- Prysm Layer total thickness: 10
- Maximum Core / Prysm transition ratio
- Volume Growth rate: Very Slow
- Maximum Cell size: 100
- Post Mesh Optimisation
- Per-Part Meshing
- Mesher Execution Mode: parallel

With all the given conditions applied onto the mesh, the meshing operation is
performed. The output result, that is, the meshed geometry has been checked for errors or
degenerated elements. As seen in the image below, the mesh is formed by 134832 triangles
(elements). After no defects have been found, it has been determined that it is suitable for
simulation.

Image 32: Mesh result.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 47
 4.3. SIMULATION: BOUNDARY CONDITIONS
Boundary conditions allow a simulation design to constrain the system behaviour
based on what is happening in the real system being simulated. In terms of mathematics,
boundary conditions can be split into two groups, constant functions, or terms proportional to
the value of the solution.

The domain (now split by patch) must be assigned the correspondent boundary
conditions so it behaves like a sort of tunnel. This way, the Inlet will be assigned a Velocity
Inlet and the Outlet a Pressure Outlet. These will act like the entering and exiting “walls” where
the sped-up air will go. The Road will be assigned a wall property, with no special conditions.
The Walls of the domain will act as Symmetry Planes.

With the properties assigned to each region, now the Simulation Model to run is
created. This defines what kind of simulation is going to be performed, and how it should be
calculated (which method).

For this dissertation the simulation is going to consider a Gas, with a Segregated
22
Flow . It is going to have a Constant Density and be in Steady form. For the simulation, a
Turbulent model is chosen, specifically a Spalart-Allmaras method.

The previous assigned regions are now defined as Boundaries of the system. As such,
the Inlet domain (velocity inlet) is set to be 20 m/s. The Road Domain is set to have tangential
velocity (in vector form). As such, the road will move as well at 20 m/s, but it will do so in the
opposite direction (x axis negative) compared to the air entering the system, being this way [-
20, 0, 0] m/s. By doing so, a simulated condition is created of the vehicle traveling at that
speed, but instead of moving the geometry itself (not ideal), the road moves at that speed but
in reverse, creating the same effect.

The last setup remaining is to create a visualisation plane. This plane will act as a 2D
visualisation port in order to make easier and faster the progression of the simulation steps
and the given results. In this case, the plane will cut the car in quarters (one half of the half
being simulated). This way, it cuts through the windshield, giving a complete look of how the
air is flowing as the steps get processed.

22
According to Science Direct: That can be treated as taking different flow paths from the matrix to the
hydrofracture network.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 48
 Image 33: Cutting plane.

4.4. SIMULATION: RESULTS

After three simulations have been successfully completed (understanding successful,
as giving converging, feasible and comparable results) each of them running for at least 1000
steps, until the output is stable, the following results and conclusions can be extracted from
them. It must be noted that the results that are shown on this dissertation correspond to the
third (and last) simulation. Due to the repeatability of the given output across all three of them,
this one has been chosen because of it being the last one. The results of all the simulations
however can apply here.

Everything explained in this section can be seen in the images and graphs on this
dissertation. The results will be first presented in numbers, developing and explaining the
given Cd and Cl. Then, a detailed explaining of the airflow graph around the car is going to be
presented.

The output for the Drag Force on the simulation once the corresponding graph has
stabilized comes up to:

𝐷𝑟𝑎𝑔 𝐹𝑜𝑟𝑐𝑒 = 178,41 𝑁

The value shows the final Drag Force on the system, that is, the value the simulation
extracts at the last step once the results have converged. This is the number that should be
chosen when calculating the Cd (Drag Coefficient) as it represents the Drag the vehicle will
have the most accurately. It would not make sense to choose a value as soon as the simulation
starts, as the given values will bounce from one to another until they stabilize. It would not be
representative of how the subject really behaves.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 49
 Image 34: Console Output - Drag report.

Considering the formula for the Drag Coefficient (Cd):

2𝐹𝐷
𝐶𝐷 = (31)
𝜌𝐴𝑣 2 ∞
And knowing the following constant for the vehicle under study (R53 MCS) and the
simulation environment:

𝜌 (𝑎𝑖𝑟 𝑑𝑒𝑛𝑠𝑖𝑡𝑦) = 1,225 𝐾𝑔/𝑚

𝐴 (𝑣𝑒ℎ𝑖𝑐𝑙𝑒 𝑓𝑟𝑜𝑛𝑡𝑎𝑙 𝑎𝑟𝑒𝑎) = 1,98 𝑚2

𝑣 (𝑎𝑖𝑟 𝑠𝑝𝑒𝑒𝑑) = 20 𝑚/𝑠

The output Drag Coefficient (Cd) can be calculated as it follows:

2 ∙ 178,41
𝐶𝐷 = (32)
1,225 ∙ 1,98 ∙ 202

𝐶𝐷 = 0,3677 ≅ 0,37

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 50
 The output for the Lift Force given as an output on the simulation once the graphs show a
stable and feasible result is:

𝐿𝑖𝑓𝑡 𝐹𝑜𝑟𝑐𝑒 = 113,15 𝑁

Once again, this value is representative of the studied system as it is the peak force it
experiments.

Image 35: Console Output - Downforce report.

Keeping in mind the formula for the Lift Coefficient (Cl):

2 ∙ 𝐹𝐿 (33)
𝐶𝐿 =
𝜌𝐴𝑣 2 ∞
And applying the same known constants as before, the overall Lift Coefficient (Cl) of
the R53 Mini Cooper S can be calculated as:

2 ∙ 113,15
𝐶𝐿= (34)
1,225 ∙ 1,98 ∙ 202

𝐶𝐿 = 0,2301 ≅ 0,23

Both the calculated Drag and Lift coefficient are considered correct, as they enter the
scope of what are considered normal values for Cd and Cl for road production vehicles.
Considering the normal23 coefficients for a production vehicle considered as “generic
automobile” oscillate around 0,35 as for Drag Coefficient (Cd) and 0,28 for Lift Coefficient (Cl).
Those values are not far off from the extracted ones on the simulation.

The output values are further confirmed to be feasible by the generated image on how
the different pressure zones appear on the vehicle as the steps succeed (simulating the air
passing by).

23
According to what is explained in the subject Fluid Dynamics (Q4) – Automotive Engineering degree UPC.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 51
5. VALIDATION OF RESULTS
One of the main objectives of this dissertation is to study the power of modern CFD
and CAD software, by checking how close and accurate the results extracted from the
performed simulations are to the real aerodynamic values provided by the OEM at the time of
designing the production vehicle, more than 20 years ago.

5.1. VALIDATION OF RESULTS: REAL VALUES

Although the official values and results by BMW24 are sealed, and as such, cannot be
found on the internet, after getting in touch with an Aerodynamic and Wind Tunnel Engineer
at BMW Group, a permission has been granted to use part of an old internal archive called
R50-3-Project: technische Entwicklung und Validierung vor Procuktion. This internal book
from the year 2000 has compiled the results of a series of different simulations (Structural,
NVH, Aerodynamics) made at the time when the car was in prototype state, in order to prepare
de pre-production state, before its world-reveal in the year 2001.

Here, the official vehicle measures and frontal area values can be found. These are
the ones that have been used to calculate the results of this project. The OEM given Drag and
Lift Coefficient are defined as:

𝐶𝐷 = 0,39

𝐶𝐿 = 0,26

It must be pointed out that the performed simulations on this dissertation follow the
boundary conditions explained on what was accessible on the book as close as possible. This
way, it has been defined at 20 m/s (free air velocity), with no rolling wheels on the vehicle, but
a moving ground (road) at the air speed in reverse.

The archive also provides a ser of pictures showing the results of a CFD simulation
performed on the Y symmetry plane on the vehicle, showing the pressure variation around the
vehicle (colours) as well as the vector direction of the flowing air. Not all the development
report was available to use, just the aspects concerning this dissertation were lent and
intended for academic use.

24
Both the aerodynamic study and the development of the first generation New-MINI were done directly by
BMW Group.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 52
 Image 36: Manufacturer simulation results.

Both the numerical values and the pressure graph are outputs extracted from the
dissertation’s simulation and are going to be compared side by side in order to further validate
its results as valid ones.

5.2. VALIDATION OF RESULTS: ANALYSIS AND COMPARISON

The numerical comparison on the aerodynamic values of the vehicle is done upon what
is given by the R50-3-Project: technische Entwicklung und Validierung vor Procuktion, which
defines a Drag Coefficient of 0,39, and a Lift Coefficient of 0,26 for the R53 Mini Cooper S.

Of the three simulations, the closest achievable output was:

𝐶𝐷 = 0,3677 ≅ 0,37

𝐶𝐿 = 0,2301 ≅ 0,23

Both the Drag and Lift Coefficient are down approximately 0,03 compared to the values
given by the OEM. This can be due to a couple of reasons: complexity of the model, lack of
wheels, evolution of software.

The CAD design of this project is a very simplified version of the R53 MCS, although
being faithful to the overall shape and proportions of the original vehicle, it is simplified to the
maximum in order to minimize the possible errors when launching the simulation. Neither the
simulation performed by MINI nor this one includes the rear wing the production vehicle has,
but the manufacturer states the wind mirrors were included, objects that are missing on the
created CAD. This oversimplification of the geometrical model makes it “smoother” overall, so
it is normal for it to have slightly better results.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 53
 It is generally accepted that the wheels are the most complex element to compute
correctly in a CFD simulation. The performed simulations on this dissertation have no wheels,
as the wheel arches have been sealed. The OEM simulations did have wheels, but they
remained stationary (not rotating). On both cases, there was a rolling road. The effects of not
having an extra object (wheel) have been proved to create slightly off results.

The evolution of Computational Fluid Dynamic simulation software over time has made
the process of running a valid an accurate simulation more forgiving. As the results show, the
generated output is merely 0,03 off from the official value given by the manufacturer. All things
considered, these results have been obtained by creating a complete 3D CAD by only one
person, with little training on the software in a couple of months. So, bearing that in mind, the
output values for the Drag Coefficient and Lift Coefficient can be accepted as correct.

Once the numerical results have been confirmed to be correct, the way they are
projected and distributed around the geometry under study is to be analysed and compared
with the original manufacturer simulation.

Image 37: Siemens STAR CCM+: Simulation Results Representation.

Since the Results Scene of the simulation provided by MINI shows the pressure results
during the different steps, in order to make it comparable, the pressure results have been
extracted from this simulation as well. It makes sense to do so, because the pressure
distribution shows exactly how and where the air hits head-on and where it flows around.
Before comparing results however, an analysis must be done on the results as such, in order
to confirm that they, in fact, make sense.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 54
 At first, the distribution of pressure seems correct. The higher-pressure zone (red
tones) is up front on the front bumper and grill. This is the first zone the air hits, so it is normal
that here higher overall pressures are found. Plus, these elements are quite vertical, which
makes even more evident the fact that they should act like “walls”, stopping the air in its free
stream, and making it start to diverge and flow around the car. The second zone with higher
visible pressure is the lower part of the windshield. This again, is a feasible result. As the air
first hits the front of the vehicle, it moves up the bonnet, following its surface, until it hits another
“wall”, this time, the base of the windshield, which is very steep in comparison with the slope
the air was following (nearly zero from the top of the bonnet). This effect makes the air slow
down again, and as such, it creates a high-pressure zone, which relieves itself as the air moves
up the vehicle, accelerating again.

Image 38: Siemens STAR CCM+: Simulation Results Representation (front).

The horizontal surfaces of the car, that is, the roof and the middle section of the bonnet,
present themselves as neutral pressure zones. They act as transitions in between the high
and low-pressure zones. For example, after the first high pressure zone on the front of the
vehicle, where the air is slowed down, it “releases pressure as it travels along the bonnet and
gains velocity again, before slowing down again at the base of the windshield. The side of the
vehicle also has a neutral-pressure zone because. This is due to the nearly flat surface
treatment the MINI has, with no creases on the bodywork. This causes the air flow to follow a
tangential movement in relation to the one from the free flow stream. As it does not modify its
movement it remains as “neutral” through its pass.

Contrary to the high-pressure zones, the low-pressure zones appear in one of two cases.

When the air is accelerated along the surface (Bernoulli principle) or when the flow
stream encounters a sudden and abrupt change of the surface that it cannot follow; here, the
air detaches from the surface, causing separation from the body. These zones are normally

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 55
found on the edges of the car, if there are no aerodynamic elements to keep the stream
attached to the body (air curtains, spoilers…). The R53 MINI does not have these kinds of
elements, and the simulation faithfully shows a realistic result of what would happen. The roof
is a clear example of the car. The zone where the top of the windshield meets the roof has a
sharp change of direction on the surface. As such, the air will detach from the surface of the
roof, onto the free-flowing state, causing a low-pressure zone around the perimeter of this
whole part.

Image 39: Siemens STAR CCM+: Simulation Results Representation (back).

The back of the vehicle should be mostly a neutral-pressure zone if the simulation is
performed correctly. This is exactly the case. To put it simple, it is a zone of the vehicle that is
not in direct contact with the air flow, but rather it acts as a departure zone or surface, where
the air just detaches from the body into the environment, now, mostly turbulent and
contaminated from traveling from the very front.

Special mention must be made to the rear wheel arches of the vehicle. It is surprising,
and an indicative of how well the simulation is, that they pick a peak in pressure on their frontal
area. It is not really that significative (value-wise), but it can still be seen in the pressure graphs.
It makes total sense that this area experiments a slight pressure increase as the wheel arch
has protrudes outwards from the bodywork, so that air flow traveling tangential through the
side panels across that smooth surface suddenly finds a new curvature it must overcome.

The numerical results extracted from the simulation, are both feasible and accurate to
the ones provided by the manufacturer. The form in which the results are displayed across the
body of the vehicle can be also considered correct as they meet every consideration and
assumption made on the shape of the vehicle, and they are displayed in a feasible way.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 56
In pursuance of fully comprehend the extend of the accuracy of the work performed in this
dissertation, the pressure graph will now be compared with the one from the official simulation
from MINI (explained above).

Image 40: (right) manufacturer simulation and (left) dissertation simulation.

Before diving deep into the comparison, some considerations should be made. The
provided pressure map shows the different pressures not directly applied onto the body of the
vehicle, but rather the ones right next to them (the pressure and direction of the flow stream
barely separated from the surface). So, in order to fully compare the two results, a
comprehension of the surrounding pressures and the air behaviour must be made (the vector
lines on the graph show how the air is pushed), instead of just looking directly at the colours
and assuming the pressures.25

Bearing that in mind, and now comparing the graphs, there are some direct
comparisons about the behaviour that can be made. Starting from the front, the air is clearly
pushed upwards after the first contact with the vehicle, following the hood surface line until it
encounters the base of the windshield, where it is once again deflected vertically onto the roof.
At the rear of the vehicle a clear detachment of the air flow can be seen26 once it approaches
the rear section of the roof where the nearly horizontal line of the roof suddenly stops and
turns into the boot lid (quite vertical). At the speed the air is traveling, it cannot follow the lines
of the car, and it shows a depression zone as it separates from the surface.

Comparing the nose of the car on the two simulations, an understanding of the
pressure maps on the older simulation (the one from MINI) shows the same behaviour in
pressure along the front grill at the edge of the bumper27.

25
Furthermore, the scale of the units is unknown and not provided on the original data, so the pressure values
cannot be directly compared.
26
As shown by the continuous lines which are intended to represent the air flow’s path.
27
See text along Image 34.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 57
It must be taken into consideration that to compare the different zones, it is the area around
that element that must be analysed (and how the air is pushed or dragged) rather than looking
directly at what is next to the surface. So, it can be appreciated how around the frontal part of
the hood (apart from seeing a change of colour indicating an increase in pressure) the air is
pushed way up from the surface of the body, creating as a result a depression from there
backwards.

The zone around the windshield and where it meets the roof also matches the results
extracted from this dissertation’s simulation. Looking at how horizontal the air flow vectors are
along the entirety of the top surface shows, first, a constant pressure behaviour around the
area, and then, how the pressure as the air layer is further from the roof, sees its pressure
increased as its speed decreases. This is why, the layer closest to the surface (colour blue)
being sped up has a decrease in pressure, but as it detaches from the surface and loses its
contact with it, its speed starts to match the one at the free flow, and the contrast (and balance
of properties) makes the pressure equally increase. This behaviour matches the one described
at the performed simulation in this paper, and so, they can be assumed to be relatable between
each other.

Taking everything analysed, explained and compared, the simulation performed on

this simulation can not only be considered valid and feasible for its outputs, but it can also be
deemed comparable to a simulation performed by the official manufacturer of the vehicle.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 58
6. EXPERIMENTAL VALIDATION
The final part of this dissertation consists of an experimental validation of all that it has
been propounded, designed, simulated and studied. The final validation will be based upon
performing a real-life simulation on a wind tunnel under roughly the same boundary conditions
as the virtual one, using a materialised version of the used CAD archive. By doing so, a set of
results (intended to be comparable) will be extracted. The objective is to see how accurate a
virtual simulation performed revolving a student CAD of a real vehicle can get, compared to
the official simulation data from the manufacturer, from more than twenty years ago.

6.1. EXPERIMENTAL VALIDATION: EXPLANATION

The wind tunnel test will be done at the EPSEM (Escola Politècnica Superior
d’Enginyeria de Manresa) using a HM170 Open wind tunnel.

Image 41: HM170 Open Wind Tunnel.

The test will consist of putting a 3D scaled model of the final iteration of the CAD
archive (third iteration) and put it under test by performing an experiment at 20 m/s (the same
inlet velocity used on the simulations). The scaled model will be introduced sideways facing
the tunnel inside a special designed zone (sideways, due to the tunnel design). A set of
sensors will measure the Drag and Downforce (Lift) forces acting upon the model to the be
compared with the ones extracted on the virtual simulation.

Because of the way this instrument is designed, the performed simulations are set up
differently than on a full-scale wind tunnel or a virtual one. Here, the bodies are placed
sideways, suspended on the air and fixed onto a plate which has holes in it. This way the body
must include some sort of fixation stick which fits inside the plate. This design limitation will
make impossible to recreate the condition of a slipping road on the test.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 59
 6.2. EXPERIMENTAL VALIDATION: THE GEOMETRY
Because of the way the HM170 wind tunnel is designed and built, the v3 geometry (the
one used on the simulations) had to be adapted both to fit inside the tunnel, and to be able to
fix it in place during the test.

First, the geometry needs to be scaled down. Originally, the body is made to mimic the
one of the real R53 MINI, and as such, it measures nearly 3,7 meters, which is far too big to
fit inside the tunnel. The measure section (compartment where the objects under study are
placed) limits the objects placed inside of them to be around 200 mm (20 cm) in length (in any
direction). If the body is bigger than that, it would in fact fit inside the measure section, but it
will be far too big to be fully hit by the air stream, since it would take too much space and the
flow would be hitting it inconsistently.

So, a Scaling operation has been performed upon the final geometry with a ratio of
0,045. This way the final measures of the body are 16,3x7,77x5,56 cm. As stated before, on
this experimental validation, and due to the design limitation of the HM170 wind tunnel, the
study object must include some sort of stick coming out from it, that must extend far enough
to be inserted in the holes on the base of the measure section. This way the body remains
suspended sideways mid-air, facing the “inlet wall”.

A 4mm (of diameter) circle has been drawn on the side of the vehicle, trying to make it
on the middle of the geometry, so it is balanced, and extruded 190mm. This is the height the
vehicle needs to be suspended at, so it remains on the centre of the inlet wall, and the air fully
hits it in a symmetric way. A further 30mm have been added to the end of this cylinder, which
is the depth of the hole on the base plate where the body must stick into. At the end, the scale
model now presents a 4x220 mm “stick” coming out from the side of it. With these
modifications, the final overall measure of the body (with the support element) comes out to
25,8x16,3x5,56 cm.

Image 42: (CATIA) CAD of the geometry to print.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 60
 With the confirmation of the dissertation tutor, the geometry is sent to be materialised
via 3D printing. The model is printed in PLA plastic (Polylactic acid) with a 15% infill. This will
make the model more fragile, as there will be less material inside the body (it will be hollow),
but it is necessary to make it as empty on the inside as possible because of weight. The object
will be suspended on the air, subjected by a 4mm diameter cylinder so it must be as light as
possible.

Image 43: preview (measures) of the printed body.

Source: 3DStore

With these characteristics, the final model weighs 575 grams approximately. Once the
body has been printed, a refining process is made upon it, in which it is sanded down to make
the surface as smooth as possible, and the excess of material from the machine printing
material is removed. It is important that the geometry presents a smooth and homogeneous
surface, with no traces of the 3D printing lines. This is because if there are printing lines all
over the geometry, the air stream of the experimental test will not flow around the body
smoothly, but rather, it would jump from ridge to ridge, becoming turbulent and unstable. In
essence, not behaving like when it is flowing around a metal body (real car), like on the virtual
simulation.

As it can be seen, the car model was printed with a supporting stick bonded to it, that
is, it was printed with it included. Due to the flexibility of the plastic from which the model is
printed of (and lack of infill material to make it as light as possible) the supporting stick needs
to be removed as it is not able to support the weight of the geometry and stay still.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 61
 This was experimentally proven on the first validation session after it was clearly seen
that the supporting stick was flexing, and the model moved from one side to another once
suspended inside the simulation chamber of the Wind Tunnel. This way, it has been cut off,
and an aluminium profile with similar section (4mm in diameter) has been introduced in its
place (inserted inside the geometry) by drilling a hole with a power drill. The aluminium profile
has been bonded to the plastic body with a special bi-component concrete-glue used for
joining aluminium profiles to carbon fibre infusions (borrowed from Dynamics Formula Student
racing Team).

Image 44: 3D printed model with aluminium supporting stick before painting.

The 3d Printed Body needed some preparation before putting it to test in order to
ensure valid results. Due to the poor print quality the model had ridges and line marks of the
different layers of material the machine was depositing when elaborating the figure. These in
essence cause the surface to be uneven and unsmooth, so the whole body has been sanded
down with a 240-grit sandpaper. The model has also been spray painted in matt finish red28.
In between the three layers of paint the scale model ended up having, it has been sanded
down again to provide a smoother and even finish.

28
The Dissertation Tutor recommended painting it in red to make it easier to see the results inside the wind
tunnel with the water steam flowing along the geometry.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 62
 6.3. EXPERIMENTAL VALIDATION: THE TEST
The experimental Wind Tunnel test has been performed in the HM170 Wind Tunnel
inside the EPSEM across two private laboratory session.

Image 45: EPSEM's HM170 Open Wind Tunnel.

Following instructions provided by the machine’s user’s manual, the test is fairly simple
and quick to perform, but it needs a couple of preparation steps for it to be correctly performed.

The first step is to introduce the model inside the simulation chamber, as explained
before it is introduced sideways, with the supporting aluminium stick holding the weight of the
model. The stick is introduced into a base which acts both as an anchor and, most importantly
is the force measurement instrument. As the geometry (suspended mid-air and centred in
relation to the air outlet chamber) moves and vibrates due to the force the wind imposes upon
it, the supporting stick will do the same. This vibration is captured by the measuring base
which translates it into the force the body is experimenting by trying to move but not being
able to.

Image 46: Simulation Chamber - HM170 Wind Tunnel.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 63
 After the model is introduced, the simulation chamber is closed, and the measuring
scale is set to zero to balance the system from the start. This way, there is a steady zero from
which elaborate results.

Now, the HM170 Wind Tunnel is turned on and the inlet air speed is set to 20 m/s. The
simulation speed is achieved by slowly and steadily turning a knob which accelerates the fan.
A measuring device shows the speed inside the measuring chamber in real time (in a similar
way to a thermometer). Once the desired 20 m/s have been achieved, the measuring process
can be started.

Image 47: Initial-Zero adjustment - HM170 Wind Tunnel.

Image 48: Speed sensor (at 20m/s) - HM170 Wind Tunnel.

The test itself had a duration of three minutes in which the body was suspended mid-
air and a steady 20m/s flow acted as the inlet air stream. During these three minutes, the
output registered forces did not change, so it was concluded the experiment to be a success
and the extracted values to be representative of the geometry put under test. The output forces
can be seen both on a side screen attached to the wind tunnel itself or in the Tunnel’s software
on the adjacent computer.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 64
 Image 49: CAD Model prepared for simulation - HM170 Wind Tunnel.

After the main test was completed, a second additional experiment was conducted,
this time, introducing a device which blows hot air at 100ºC in the shape of water steam. When
introduced inside the simulation chamber (with air flowing) the steam is rapidly pushed onto
the stream direction, creating a famously known “white line” effect that can be seen on the full-
scale wind tunnels when testing a vehicle. This steam line follows the shape of the vehicle and
provides a clear and visual look at the path the air follows around the surface of the body.

Image 50: Experimental validation - HM170 Wind Tunnel.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 65
 This experiment, however, did not bring the expected results due to the difficulty of
seeing the air stream. This test is normally conducted in dark chambers with a contrast back
lighting to further boost the contrast of the dark ambiance with the light-toned air stream. Here,
the test was performed during daytime, in a room with windows that cannot be covered, so
even though the lines they can be sort of perceived, they cannot be considered real outputs.
Plus, this test can only be performed for a very short amount of time because of the high
temperature of the air stream. If that short amount of time was exceeded, it could melt the
paint of the model and even the plastic itself. So, this second experiment, although interesting
because of its illustrative nature, it cannot be considered as real results to be compared with
the virtual ones.

6.4. EXPERIMENTAL VALIDATION: RESULTS AND COMPARISON

The output results of this dissertation’s experimental validation are given by the HM170
Wind Tunnel as measured Drag and Lift forces. It should be noted that these values differ a
lot from the virtual ones out of the full-scale designed vehicle due to the much smaller size of
the real-life model. By calculating the projected frontal area of the scale model, the
correspondent Drag and Lift coefficients are calculated. Ideally, these non-dimensional
coefficients should match the ones given by the manufacturer (and the ones from the virtual
simulations).

The extracted values from the simulation are as follows:

𝐷𝑟𝑎𝑔 𝐹𝑜𝑟𝑐𝑒: 0,45 𝑁

𝐿𝑖𝑓𝑡 𝐹𝑜𝑟𝑐𝑒: − 0,13 𝑁

As it can be seen, the Lift force here shows up negative, indicating the car would indeed
experiment Lift rather than Downforce. The reality however is that this should be inverted due
to the position of the geometry inside the simulation chamber. With that, the real Lift Force is
0,13 Newtons in positive.

The original frontal projected area of the R53 Mini Cooper S is 1,98 square meters. To
express the results as realistic as possible, rather than simply applying a reduction of the
original frontal projected area corresponding to the reduction applied to the geometry (0,045),
the new projected frontal area has been measured directly on the real model, knowing its
measures (proportioned by 3DStore)29 and then contrasting them in CATIA.

𝐴𝑠𝑐𝑎𝑙𝑒 : 0,0043 𝑚2

29
7,77 cm in width and 5,56 cm in height.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 66
 Applying the same formulas for the Drag and Lift coefficients used in the elaboration
of the virtual results:

2 ∙ 𝐹𝐷 (35)
𝐶𝐷 =
𝜌𝐴𝑣 2 ∞

2 ∙ 𝐹𝐿 (36)
𝐶𝐿 =
𝜌𝐴𝑣 2 ∞

𝐶𝐷 being Drag Coefficient.

𝐶𝐿 being Lift Coefficient.

The rest of the values and constants remain untouched, being the equal to the ones
chosen in the first part of the dissertation. With these, the extracted results from the
experimental validation can be calculated as:

2 ∙ 0,45 (37)
𝐶𝐷 =
1,225 ∙ 0,0043 ∙ 202

2 ∙ 0,13
𝐶𝐿 = (38)
1,225 ∙ 0,0043 ∙ 202

From these, the following results can be extracted:

𝐶𝐷 = 0,4271 ≅ 0.43

𝐶𝐿 = 0,1233 ≅ 0.12

Numerically, the values are close to the ones extracted from the virtual simulation, and
by extrapolation, to the ones from the real vehicle.

Drag Coefficient on the production vehicle comes out at 0,39. Since the experimental
validation is done upon the geometry made for this dissertation, it is the 0,36 (CD) that needs
to be compared. Here, it is 0,03 above the previously mentioned value. The value being higher
makes total sense due to a couple of factors.

There is a notable difference on the surface treatment between the printed scale model
and the virtual simulated geometry. The former one (due to its nature) does not present a
perfectly smooth surface, the different layers of material can be clearly seen 30. This does not
happen on the virtual model, where the surface of the body is perfectly smooth (as well as on
the production vehicle). Those ridges cause the air stream to find small jumps along the

30
Even after three layers of paint and after being sanded (and wet sanded).

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 67
direction of travel, causing small eddies (and turbulences) which accumulate as the air flow
further back. So, it is normal and expected to find a slightly higher value in Drag on the scale
model compared to the virtual one due to the imperfections on the surface of the geometry.
The supporting stick is an element that although it may seem relevant, it is not. Being only
4mm in diameter its presence does not alter the output results due to its relatively small size
compared to the geometry put under study, so it is correct to assume that the peak in Drag
Force measured comes from the body and not from the aluminium profile.

The Lift Coefficient on the production R53 MINI is 0,26. The result output on the virtual
simulation comes out at 0,23. There is a big difference between these and the 𝐶𝐿 = 0,12 from
the experimental validation. It must be noted that this result is not really representative
because of the different conditions in which the mentioned experiments/simulations have been
performed, and thus, there is a logical explanation for such a big difference on Lift
(Downforce). The principal responsible for the discrepancy on the results is the (lack of)
ground on the experimental validation. It can then, be proven, that the only accurate simulation
technique is the moving ground simulation. [B. Fago, H. Lindner, 1991]. Having this premise
in mind, it is correct to assume that the big difference between values is normal, and they must
not be treated as equal.

The effect of a close moving ground accentuates the presence of a low-pressure zone
underneath the vehicle, which boosts significantly an increase of overall Downforce. The
complete loss of this effect, as it can be seen, makes a huge difference in the output result.

All in all, the results given by this dissertation’s experimental validation can be
considered both correct and accurate for several reasons. First, the calculated coefficients are
somehow feasible31; not only that but they are close to the confirmed correct values for the
studied production vehicle, which is a clear indication that the experiment was in fact well
conducted. The alteration between the different results can be logically explained, that is,
there’s an explanation on seeing an increase in drag, but a reduction in lift on the output result.

About the graphical comparison of the results between the simulation chapter and the
experimental one, this part cannot be addressed as intended due to the impossibility of
properly seeing the air stream evolution in person when the experiment was conducted.

31
Understanding feasible as being inside what is considered normal values for a production road vehicle.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 68
7. CONCLUSIONS
The main objectives of this dissertation were to analyse the power and accuracy of the
new CFD simulation software when developing a production vehicle, and with the extracted
results, further explore the possibility of fully carrying the design and study process relying
solely on virtual tools, by comparing them with the way the manufacturer did that same process
over two decades ago, and with an experimental validation.

For the objectives to be successfully met, it was key to create an accurate CAD model
of the vehicle intended to be put under study. The model was designed by hand with the
original vehicle’s blueprints as a reference. This model needed to be as realistically as close
to the original vehicle as possible, as the extracted results from the simulation relied on it.

Having the original manufacturer simulations (from the time the vehicle was in
development) has made it possible to replicate a modern simulation under the same
characteristics.

A series of simulations have been performed, whose results have shown to be very
similar to the ones provided by the original manufacturer of the vehicle put under study. This
fact serves as a proof that the hand-designed CAD geometry is correct, as if it were not the
case, a discrepancy on the results would have been evident.

Putting everything down, and judging by the extracted results, both numerically and
graphically, it has been proven that the power of new simulation software is much greater than
it was 20 years ago, to a point that a student designing on its own a simplified model of a
vehicle, can approximate and replicate results extracted by a big company. Not only that, but
it must be noted that this case studies an overall simplification of the original geometry, which
means that these new programs are in fact way more forgiving.

The conducted experimental validation has brought the confirmation that it is totally
feasible to study a production vehicle “from home” since a real working tunnel has brought
results close to the ones from the virtual simulations, and from the official values32. This has
served to prove that in fact the results brought by the simulation were not a match to the official
values by mere chance, since a real experiment using the same geometry has extracted in
fact similar results which are, then again, a match to the aerodynamic coefficients provided by
MINI.

32
Plus, having an evidential explanation on the little deviation in results.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 69
 After having completed both a virtual and a real experimental aerodynamic study on
the same vehicle it can be assured, by the repeatability of results, that it is totally feasible to
aerodynamically develop (up to a certain point) a production vehicle with virtual tools. It must
be stated, however, that real tests must be conducted in order to study certain aspects CFD
cannot, such as, wind noise or vibrations.

There are, however, some considerations to be made. Further improvements could be

made to this dissertation to explore the chance of closing the gap between the initial and
boundary conditions of the conducted experiments: a second iteration of simulations and
experimental validations with a geometry with open wheel wells and glued simplified wheels;
some sort of ground in the HM170 Wind Tunnel to analyse the effects of the ground effect on
the Lift Coefficient. Finally, a series of simulations at higher speed (for the air stream) in which
the Drag and Lift of the vehicle really make themselves present and relevant, and thus, the
difference between them increases significantly.33

Summing everything up, after successfully completing all that is necessary to carry this
dissertation, it can be confirmed that these objectives have been met. The power of new
Computational Fluid Dynamic simulation software has been proven to be powerful enough to
a point of a student replicating results of an automotive manufacturer.

33
The latter cannot be conducted at the moment due to the limitation of the HM170 Open Wind Tunnel.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 70
REFERENCES

[1] “What is Aerodynamics?,” NASA, 04-Jun-2011. [Online]. Available:

https://www.nasa.gov/audience/forstudents/k-4/stories/nasa-knows/what-is-
aerodynamics-k4.html. [Accessed: 05-2023].

[2] “Part 2: Aerodynamics,” Yamahapart.com. [Online]. Available:

https://www.yamahapart.com/part2aerodynamics. [Accessed: 31-May-2023].

[3] M. Panciutti and V. T. mi Perfil, “Conciertos y desconciertos,” Blogspot.com. [Online].

Available:
http://desconciertos3.blogspot.com/2016/10/a-proposito-del-escarabajo-5.html.
[Accessed: 31-May-2023].

[4] “What Is It?,” The Augusta Chronicle, 17-Apr-2014. [Online]. Available:

https://eu.augustachronicle.com/story/lifestyle/2014/04/17/what-it/14415241007/.
[Accessed: 31-May-2023].

[5] N. D. Katopodes, “Boundary-Layer Flow,” in Free-Surface Flow, N. D. Katopodes, Ed.

Oxford, England: Elsevier, 2019, pp. 652–708.

[6] F. Stephenson, “How I Designed the MINI COOPER,” 09-Apr-2020. [Online]. Available:
https://www.youtube.com/watch?v=KoQP5qjSYvw. [Accessed: 31-May-2023].

[7] Mini2.com. [Online]. Available:

https://www.mini2.com/threads/per-mini-drag-coefficient-0-39.134713/. [Accessed: 31-
May-2023].

[8] G. S. Samy, S. Thirumalai Kumaran, M. Uthayakumar, M. Sivasubramanian, and K.

Bhagavathi Sankar, “Numerical analysis of drag and lift coefficient of a Sport Utility
Vehicle (SUV),” J. Phys. Conf. Ser., vol. 1276, no. 1, p. 012013, 2019.

[9] L. Axon, K. Garry, and J. Howell, “An evaluation of CFD for modelling the flow around
stationary and rotating isolated wheels,” in SAE Technical Paper Series, 1998, vol. 107,
pp. 205–215.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 71
[10] BMW GROUP, R50-3-Project: technische Entwicklung und Validierung vor Procuktion.
2000.

[11] “Drag Reduction of a Mini Cooper,” 1000sciencefairprojects.com. [Online]. Available:

https://www.1000sciencefairprojects.com/Aerodynamics/Drag-Reduction-of-a-Mini-
Cooper.php. [Accessed: 09-Jun-2023].

[12] B. Park, “Winging it: Mini’s fuel-efficient future design tricks,” Wheels, 17-Jul-2018.
[Online]. Available: https://www.whichcar.com.au/news/winging-it-minis-fuel-efficient-
future-design-tricks. [Accessed: 09-Jun-2023].

[13] V. Ferrand, “Forces and flow structures on a simplified car model exposed to an
unsteady harmonic crosswind,” J. Fluids Eng., vol. 136, no. 1, p. 011101, 2014.

[14] “Aerodynamics for a MINI,” North American Motoring, 16-Oct-2019. [Online]. Available:
https://www.northamericanmotoring.com/forums/general-discussion/343124-
aerodynamics-for-a-mini.html. [Accessed: 09-Jun-2023].

[15] Wikipedia contributors, “Mini Hatch,” Wikipedia, The Free Encyclopaedia, 06-Jun-2023.
[Online]. Available:
https://en.wikipedia.org/w/index.php?title=Mini_Hatch&oldid=1158761202.

[16] M. Speed, “Gen 1 MINI Wind Tunnel Test at Aerodyne March 2018,” 25-May-2018.
[Online]. Available: https://www.youtube.com/watch?v=PizalOrkBwo. [Accessed: 09-
Jun-2023].

[17] “M7 Wind Tunnel Test - pretty cool!,” Mini Cooper Forums - Mini Cooper Enthusiast
Forums, 18-Apr-2013. [Online]. Available: https://minicooperforums.com/forum/sights-
n-sounds-9/m7-wind-tunnel-test-pretty-cool-6146/. [Accessed: 09-Jun-2023].

[18] J. Ramsey, “Mini shows Cooper Works GP goes for a smoke, cheats the
wind,” Autoblog, 11-Aug-2012. [Online]. [Accessed: 09-Jun-2023].

[19] “YAMAHA WERKSTATT / ALLGEMEIN,” Wildstarfreunde-Schleswig-Holstein. [Online].

Available: https://www.wildstar-schleswig-holstein.de/yamaha-werkstatt-aerodynamik/.
[Accessed: 10-Jun-2023].

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 72
[20] FetchCFD, “Racing car mesh for simulation,” FetchCFD. [Online]. Available:
https://fetchcfd.com/view-project/1257-. [Accessed: 10-Jun-2023].

[21] “Turismos Mercedes-Benz,” Mercedes-benz.es. [Online]. Available:

https://www.mercedes-benz.es/passengercars.html. [Accessed: 10-Jun-2023].

[22] “Volkswagen España: encuentra tu nuevo coche,” Volkswagen.es. [Online]. Available:

https://www.volkswagen.es/es.html. [Accessed: 10-Jun-2023].

[23] “Chrysler official site - cars and minivans,” Chrysler. [Online]. Available:
https://www.chrysler.com/. [Accessed: 10-Jun-2023].
[24] “No Reserve: 41k-Mile 2006 Mini Cooper S 6-Speed for Charity,” The best vintage and
classic cars for sale online | Bring a Trailer. [Online]. Available:

https://bringatrailer.com/listing/2006-mini-cooper-s-62/. [Accessed: 10-Jun-2023].

[25] “9k-Kilometer 2004 Mini Cooper S 6-Speed,” The best vintage and classic cars for sale
online | Bring a Trailer. [Online]. Available: https://bringatrailer.com/listing/2004-mini-
cooper-s-32/. [Accessed: 10-Jun-2023].

[26] “No Reserve: 2005 Mini Cooper S 6-Speed,” The best vintage and classic cars for sale
online | Bring a Trailer. [Online]. Available: https://bringatrailer.com/listing/2005-mini-
cooper-s-51/. [Accessed: 10-Jun-2023].

[27] A. Ingram, “Mini Cooper (R50) and Cooper S (R53) – history, prices and specs,” evo.
[Online]. Available: https://www.evo.co.uk/mini/cooper/19635/mini-cooper-r50-and-
cooper-s-r53-history-prices-and-specs. [Accessed: 10-Jun-2023].

[28] N. Wu, “Clean fully loaded 2006 MINI Cooper S with JCW engine upgrade,” North
American Motoring, 26-May-2013. [Online]. Available:
https://www.northamericanmotoring.com/forums/minis-and-minis-for-sale/251691-
clean-fully-loaded-2006-mini-cooper-s-with-jcw-engine-upgrade.html.
[Accessed: 11-Jun-2023].

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 73
[29] Linke, Dipl. -Geogr Uta, G.U.N.T., “Manual de experimentos,” Jul. 2018.

[30] G. U. N. T. Dipl. -Geogr. Uta Linke, “Manual de instrucciones - HM170 - Túnel de viento
abierto inclusive técnica de medición,” Jul. 2018.

[31] “Como escribo mi TFG / TFM,” Upc.edu. [Online]. Available:

https://bibliotecnica.upc.edu/es/estudiants/6-passos-que-teu-tfg/tfm-sigui-exit/escric-
meu-tfg/tfm. [Accessed: 26-Jun-2023].

[32] “Como uso esta información y como la cito en mi TFG / TFM,” Upc.edu. [Online].
Available: https://bibliotecnica.upc.edu/es/estudiants/6-passos-que-teu-tfg/tfm-sigui-
exit/uso-aquesta-informacio-cito-meu-tfg/tfm. [Accessed: 26-Jun-2023].

[33] J. Howell and D. Hickman, “The influence of ground simulation on the aerodynamics of
a simple car model,” in SAE Technical Paper Series, 1997.

[34] B. Fago, H. Lindner, and O. Mahrenholtz, “The effect of ground simulation on the flow
around vehicles in wind tunnel testing,” J. Wind Eng. Ind. Aerodyn., vol. 38, no. 1, 1991.

[35] “Ground proximity simulation,” Dnw.aero. [Online]. Available:

https://www.dnw.aero/services/in-ground-effect-2/ground-simulation/.
[Accessed: 28-Jun-2023].

[36] J. Wiedemann, “The influence of ground simulation and wheel rotation on aerodynamic
drag optimization - potential for reducing fuel consumption,” in SAE Technical Paper
Series, 1996, vol. 105.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 74
 APPENDIX 1: FIGURES AND IMAGES

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 75
 Appendix figure 1: R53 Mini Cooper S (inspiration for dissertation) front.

Appendix figure 2: R53 Mini Cooper S (inspiration for dissertation) ¾ front.

Appendix figure 3: R53 Mini Cooper S (inspiration for dissertation) side.

Appendix figure 4: R53 Mini Cooper S (inspiration for dissertation) ¾ back.

Appendix figure 5: Original used blueprints for CAD design.

Appendix figure 6: (CATIA) Master 3D Lines for geometry surfacing.

Appendix figure 7: (CATIA) Comparison (in order) of the three design iterations of the
CAD model.

Appendix figure 8: (Keyshot 7) Detailed CAD version render – front.

Appendix figure 9: (Keyshot 7) Detailed CAD version render - 3/4 front.

Appendix figure 10: (Keyshot 7) Detailed CAD version render - 3/4 rear.

Appendix figure 11: (SIEMENS STAR CCM+) Meshed geometry for simulation.

Appendix figure12: (SIEMENS STAR CCM+) Results - Pressure output result 3/4
front.

Appendix figure 13: (SIEMENS STAR CCM+) Results - Pressure output result side.

Appendix figure 14: 3D printed geometry before sanding and painting.

Appendix figure 15: Experimental Validation - Prepped geometry inside HM170 Wind
Tunnel.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 76
 Appendix Image 1: R53 Mini Cooper S (inspiration for dissertation) front.

Appendix Image 2: R53 Mini Cooper S (inspiration for dissertation) 3/4 front.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 77
 Appendix Image 3: R53 Mini Cooper S (inspiration for dissertation) side.

Appendix Image 4: R53 Mini Cooper S (inspiration for dissertation) side.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 78
 Appendix Image 5: Original used blueprints for CAD design.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 79
 Appendix Image 6: (CATIA) Master 3D Lines for geometry surfacing.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 80
 Appendix Image 7: (CATIA) Comparison (in order) of the three design iterations of the CAD model.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 81
 Appendix Image 8: (Keyshot 7) Detailed CAD version render – front.

Appendix Image 9: (Keyshot 7) Detailed CAD version render - 3/4 front.

Appendix Image 10: (Keyshot 7) Detailed CAD version render - 3/4 rear.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 82
 Appendix Image 11: (SIEMENS STAR CCM+) Meshed geometry for simulation.

Appendix Image 12: (SIEMENS STAR CCM+) Results - Pressure output result 3/4 front.

Appendix Image 13: (SIEMENS STAR CCM+) Results - Pressure output result side.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 83
 Appendix Image 14: 3D printed geometry before sanding and painting.

Appendix Image 15: Experimental Validation - Prepped geometry inside HM170 Wind Tunnel.

Aerodynamic CFD Study on a production vehicle and experimental validation on a wind tunnel – Final Dissertation 84