Você está na página 1de 115

Mechanical design of the wheel assembly of

an electric Formula Student prototype

Luís Miguel Marcos de Abrunhosa Vieira de Abreu

Thesis to obtain the Master of Science Degree in

Mechanical Engineering

Supervisor: Prof. Luís Alberto Gonçalves de Sousa

Examination Committee

Chairperson: Prof. João Orlando Marques Gameiro Folgado

Supervisor: Prof. Luís Alberto Gonçalves de Sousa

Member of the Committee: Prof. Paulo Jorge Pires Moita

June 2019
ii
Acknowledgements

To the current and former team members of FST Lisboa, for the friendship and incessant help
given to this work and my time within the team.

To my thesis supervisor Professor Luís Sousa, for the unconditional help during this work,
especially in the closing stages of this work.

To my friends, for their motivation and everlasting friendship, and for not forgetting about me even
when I spent my life at the university.

To my family and girlfriend for the love and support during my entire life and the motivation during
this thesis, with special dedication to my late grandparents, who for sure would be proud of my
achievement.

iii
iv
Resumo

Encontrar o melhor sistema de transmissão para veículos elétricos tem sido o foco de
desenvolvimento de muitas equipas de Formula Student, assim como de equipas profissionais em
outras categorias de automobilismo [1].

Foram modeladas e analisadas oito soluções possíveis. As suas características físicas (incluindo
a massa, inércia e eficiência) foram inseridas num modelo transiente de dinâmica de veículo, no qual o
desempenho de cada solução foi avaliado. A solução que revelou melhor performance foi a que tem o
sistema de transmissão na massa não suspensa, com o motor dentro do porta-cubo, seguido por um
trem de engrenagens com dois estágios de engrenagens planetárias, com a entrada no sol do primeiro
estágio e a saída no anel do segundo estágio, que está ligado à roda através do cubo.

Um modelo em regime estacionário foi usado para estimar as cargas aplicadas em cada roda
durante a vida do sistema. Construiu-se um modelo térmico para estimar a variação de temperatura ao
longo do tempo, em cada componente. As cargas e temperaturas são necessárias para fazer o projeto
mecânico de cada componente. O trem de engrenagens foi projetado com o auxílio do programa
KISSsoft. Otimizações topológicas foram feitas para encontrar as formas para o cubo, porta-cubo e
disco de travão. Essas formas foram modificadas para garantir que todos os componentes podem ser
maquinados e montados com os meios disponíveis na equipa. A regra de Palmgren-Miner foi usada
para estimar o dano em cada componente, garantindo a resistência a toda a vida útil do carro.

Palavras-chave: Transmissão, Formula Student, Modelo transiente, Optimização topológica,


projecto mecânico

v
Abstract

Finding the best drive train concept for electric vehicles has been the focus of research and
development of many teams within the Formula Student community, as well as professional teams in
other motorsport categories [1].

Eight possible concepts were modelled. Their physical characteristics (such as mass, inertia and
efficiency) were inserted into a self-developed transient vehicle model, with which the performance of
each concept was evaluated. From this analysis, the best concept is the one that has an un-sprung drive
train system, with the motor inside the upright, followed by a two-stage planetary gear train, with the
input on the sun of the first stage and the output on the ring of the second stage, which is connected to
the wheel through the hub.

A steady state model was used to estimate the load cases applied on each wheel for the entire
life of the system. A self-developed thermal model was used to estimate the temperature variation
through time, on each component, under operation. The load cases and temperatures are necessary to
perform the mechanical design of each component. The gear train was designed with the aid of
KISSsoft. Topology optimizations were made to find the shapes for the hub, upright and brake disc.
These shapes were modified to assure all components can be assembled, machined and serviced with
the tools available to the team. The Palmgren-Miner rule was used to estimate the damage on each
component, so that the system can withstand the entire service life of the car.

Keywords: Drive train, Formula Student, Transient model, Topology optimization, Mechanical
design

vi
Contents

Acknowledgements ....................................................................................................................... iii


Resumo ..........................................................................................................................................v
Abstract ......................................................................................................................................... vi
List of tables ...................................................................................................................................x
List of figures ................................................................................................................................. xi
List of symbols ............................................................................................................................. xiv
List of abbreviations ..................................................................................................................... xx
1 Introduction .......................................................................................................................... 1
1.1 Formula Student .............................................................................................................. 1
1.2 State of the art and Motivation ......................................................................................... 2
1.3 Objectives ........................................................................................................................ 4
1.4 Thesis outline................................................................................................................... 4
2 Possible concepts ................................................................................................................ 5
2.1 Initial concepts ................................................................................................................. 6
2.1.1 Concept 1 ................................................................................................................ 6
2.1.2 Concept 2 ................................................................................................................ 7
2.1.3 Concept 3 ................................................................................................................ 8
2.1.4 Concept 4 ................................................................................................................ 9
2.1.5 Concept 5 ................................................................................................................ 9
2.1.6 Concept 6 .............................................................................................................. 10
2.1.7 Concept 7 .............................................................................................................. 11
2.1.8 Concept 8 .............................................................................................................. 12
2.2 Transmission efficiency estimation ................................................................................ 12
2.3 Data for simulation ......................................................................................................... 13
3 Vehicle dynamics modelling .............................................................................................. 15
3.1 Transient model ............................................................................................................. 15
3.1.1 Overview ................................................................................................................ 15
3.1.2 Tires ....................................................................................................................... 16
3.1.3 Motor ...................................................................................................................... 17
3.1.4 Vertical model ........................................................................................................ 18
3.1.5 Planar model .......................................................................................................... 20
3.1.6 Aerodynamics ........................................................................................................ 22
3.1.7 Sensors .................................................................................................................. 22
3.1.8 Driver inputs ........................................................................................................... 23
3.1.9 Validation ............................................................................................................... 24
3.1.10 Simulation results .................................................................................................. 26
3.1.10.1 Accelerating and braking ................................................................................. 26
3.1.10.2 Skid-pad .......................................................................................................... 27

vii
3.1.10.3 Chicane ........................................................................................................... 28
3.1.10.4 Hairpin ............................................................................................................. 29
3.1.10.5 Slalom.............................................................................................................. 30
3.1.10.6 Simulation conclusions .................................................................................... 31
3.2 Stead state model .......................................................................................................... 31
3.2.1 Software overview ................................................................................................. 31
3.2.2 Software inputs ...................................................................................................... 32
3.2.2.1 Motor ................................................................................................................. 33
3.2.2.2 Track layout ....................................................................................................... 34
3.2.3 Results conversion ................................................................................................ 35
3.2.4 Temperature estimation ......................................................................................... 38
3.2.5 Simulation results .................................................................................................. 42
3.2.5.1 Converted results .............................................................................................. 42
3.2.5.2 Temperature ...................................................................................................... 43
4 Load cases for structural analysis ..................................................................................... 44
4.1 Drive train load cases .................................................................................................... 44
4.2 Suspension load cases .................................................................................................. 45
4.3 Fatigue ........................................................................................................................... 47
4.4 Stiffness targets ............................................................................................................. 49
5 Final solution...................................................................................................................... 52
5.1 Transmission design ...................................................................................................... 53
5.1.1 Gear material selection .......................................................................................... 53
5.1.2 Gear design ........................................................................................................... 54
5.1.3 Lubrication ............................................................................................................. 58
5.1.4 Shafts ..................................................................................................................... 60
5.1.5 Connection between stages .................................................................................. 60
5.1.6 Bearings ................................................................................................................. 61
5.1.7 Transmission FEA ................................................................................................. 62
5.2 Hub (transmission housing) ........................................................................................... 64
5.3 Upright ........................................................................................................................... 68
5.4 Brake system ................................................................................................................. 71
5.5 Full system FEA analysis .............................................................................................. 72
6 Conclusions ....................................................................................................................... 75
6.1 Summary and conclusions ............................................................................................ 75
6.2 Future work .................................................................................................................... 76
7 References ........................................................................................................................ 77
Appendix A .................................................................................................................................... 1
Appendix B .................................................................................................................................... 2
Appendix C .................................................................................................................................... 4
Appendix D .................................................................................................................................... 6

viii
Appendix E .................................................................................................................................... 8
Appendix F .................................................................................................................................. 10
Appendix G .................................................................................................................................. 13

ix
List of tables
Table 3-1 – Car characteristics used in Optimum Lap simulations ....................................................... 32
Table 4-1 – First four entries of the drive train load spectrum for a rear wheel .................................... 44
Table 4-2 – Maximum reference load cases for each wheel of the front axle ....................................... 46
Table 4-3 – Maximum reference load cases for each wheel of the rear axle ....................................... 46
Table 4-4 – Temperature of components considered in the structural analyses .................................. 46
Table 4-5 – Skid-pad results for Formula Student Spain 2018 - Electric .............................................. 50
Table 5-1 – Mechanical and thermal properties of the considered materials, at 20ºC [29] .................. 53
Table 5-2 – Results obtained for the second stage of the transmission ............................................... 57
Table 5-3 – Endurance limit and maximum allowable stress for each material .................................... 66
Table 5-4 – Accumulated damage for the main components of the system ......................................... 74
Table 5-5 – Accumulated damage with updated materials ................................................................... 74
Table A-1 – Characteristics of each concept........................................................................................... 1
Table B-1 – Car characteristics used in the transient model. .................................................................. 3
Table D-1 – Thermal constants used in the thermal model .................................................................... 6

x
List of figures
Figure 1-1 – FST 08e .............................................................................................................................. 1
Figure 1-2 – Current drive train concept used by FST Lisboa ................................................................ 2
Figure 2-1 – Diagram of the possible positions of the motor/drivetrain system ...................................... 5
Figure 2-2 – CAD of concept 1 (left) and drive train schematics (right) .................................................. 6
Figure 2-3 – CAD of concept 2 (left) and drive train schematics (right) .................................................. 7
Figure 2-4 – CAD of concept 3 (left) and drive train schematics (right) .................................................. 8
Figure 2-5 – CAD of concept 4 (left) and drive train schematics (right) .................................................. 9
Figure 2-6 – CAD of concept 5 (left) and drive train schematics (right) .................................................. 9
Figure 2-7 – CAD of concept 6 (left) and drive train schematics (right) ................................................ 10
Figure 2-8 – CAD of concept 7: gear train (top-left); wheel, uptight and hub (top-right); assembly (bottom-
left); drive train schematics (bottom-right). ............................................................................................ 11
Figure 2-9 – CAD of concept 8: gear train (top-left), assembly (bottom-left) and drive train schematics
(right). .................................................................................................................................................... 12
Figure 3-1 – Car model for transient simulation .................................................................................... 15
Figure 3-2 - Tire coordinate system [13] ............................................................................................... 16
Figure 3-3 – Example of a tire curve, modelled with the Magic Formula of Pacejka. Represents 𝐹𝑦 vs 𝛼
for multiple 𝛾, for the tires used by FST Lisboa ..................................................................................... 16
Figure 3-4 – Example of a tire curve, modelled with the Magic Formula of Pacejka. Represents 𝐹𝑦 vs 𝛼
for multiple 𝛾, for the tires used by FST Lisboa. .................................................................................... 17
Figure 3-5 - Motor controller model [19] ................................................................................................ 17
Figure 3-6 - Motor torque (top) and power (bottom) curves [19] ........................................................... 18
Figure 3-7 – 7-dof vertical model ........................................................................................................... 18
Figure 3-8 – Illustration of the planar model .......................................................................................... 21
Figure 3-9 – Speed references for simulations ..................................................................................... 24
Figure 3-10 – Steering references for simulations ................................................................................ 24
Figure 3-11 – Comparison between measured and simulated lateral acceleration .............................. 25
Figure 3-12 – Plot of longitudinal acceleration (top), with detail for accelerating (middle) and braking
(bottom) ................................................................................................................................................. 26
Figure 3-13 – Plot of lateral acceleration in the Skid-pad scenario, with detail of the beginning of the left
corner..................................................................................................................................................... 27
Figure 3-14 – Plot of lateral acceleration in the Chicane scenario (top), with detail of corner-entry
(middle) and corner-exit (bottom) .......................................................................................................... 28
Figure 3-15 – Plot of lateral acceleration in the Hairpin scenario (top), with detail of corner-entry (middle)
and corner-exit (bottom) ........................................................................................................................ 29
Figure 3-16 – Plot of lateral acceleration in the Slalom scenario, with detail of one corner (bottom) ... 30
Figure 3-17 – Illustration of the point-mass model ................................................................................ 31
Figure 3-18 – Efficiency map provided by the motor manufacturer ...................................................... 33
Figure 3-19 – Efficiency map provided by the motor manufacturer [19] ............................................... 33

xi
Figure 3-20 – Example of the 4th degree approximation, of the first row of the efficiency map, of Figure
3-18........................................................................................................................................................ 33
Figure 3-21 – Output torque and power curves of the equivalent motor ............................................... 34
Figure 3-22 – Illustration of how the track of Formula Student Spain was obtained (Background from
Google Maps) ........................................................................................................................................ 35
Figure 3-23 – Speed plot of an autocross lap in Formula Student Spain. Example of an output from the
simulation in Optimum Lap .................................................................................................................... 35
Figure 3-24 – Plot of total braking power, electrical regeneration and mechanical braking ................. 37
Figure 3-25 – Diagram of the thermal model in Simulink ...................................................................... 38
Figure 3-26 – Setup to obtain the value of K. The infra-red sensor is underneath the brake disc........ 40
Figure 3-27 – Comparison between measured and simulated temperature of the brake disc ............. 41
Figure 3-28 – Measured temperature variation of 𝑇𝑠𝑢𝑟𝑓 and 𝑇𝑟𝑒𝑓 ...................................................... 41
Figure 3-29 – Forces Fx, Fy and Fz on each wheel, for an autocross lap in Formula Student Spain .... 42
Figure 3-30 – Temperatures at the front-left (top) and rear-left (bottom) wheels, during an autocross lap
in Formula Student Spain. It represents the fourth lap in a row, which justifies the high temperatures at
the beginning of the plot ........................................................................................................................ 43
Figure 4-1 – G-G diagram of the car (with Concept 4), with concentric inner and outer rings .............. 45
Figure 4-2 – Fatigue diagram for various criteria of failure. For each criterion, points “above” the
respective line indicate failure [5] .......................................................................................................... 48
Figure 5-1 – Illustration of the iterative process used in the design ...................................................... 52
Figure 5-2 – Illustration of how the involute is constructed [34] ............................................................ 56
Figure 5-3 - Comparison between “nominal” tooth profile with negative shifted (left), and positive shifted
(right) profiles [38] .................................................................................................................................. 56
Figure 5-4 - Typical regimen determination based on surface finish and EHD oil-film thickness[34] ... 59
Figure 5-5 – Cross-section of P3G-profile (left) and P4C-profile (right) [44] ......................................... 61
Figure 5-6 – Transmission system of Concept 4 ................................................................................... 62
Figure 5-7 – Transmission FEA model .................................................................................................. 63
Figure 5-8 – Analysis results: displacement magnitude [m] (left) and Von-Mises stress [Pa] (right), for a
load case in sector O1 of Figure 4-1 ..................................................................................................... 64
Figure 5-10 – Design space of the hub ................................................................................................. 65
Figure 5-10 – FEA model of the hub, used in the topology optimization (Part of the rim is removed to
show the position of the design space relative to the rim) .................................................................... 65
Figure 5-11 –Optimization results for aluminium (left), magnesium (middle) and titanium (right) ........ 67
Figure 5-12 – Final design of the hub .................................................................................................... 67
Figure 5-13 – Design space for the upright of the front wheels (left, red) and the rear wheels (right, white)
............................................................................................................................................................... 68
Figure 5-14 – FEA model of the rear upright, used in the topology optimization (Part of the rim is removed
to show the position of the design space relative to the rim) ................................................................ 69
Figure 5-15 – FEA model of the front upright, used in the topology optimization (Part of the rim is
removed to show the position of the design space relative to the rim) ................................................. 69

xii
Figure 5-16 – Final design of the front upright ...................................................................................... 70
Figure 5-17 – Optimization results for the front upright (top) and rear upright (bottom) ....................... 70
Figure 5-18 – Final design of the rear upright ....................................................................................... 71
Figure 5-19 – Model of the brake disc, with the design space for the structural part in blue ................ 71
Figure 5-20 – Topology optimization result (left) and final design of the brake disc (right) .................. 72
Figure 5-21 – Full-system FEA model (front wheel) .............................................................................. 73
Figure 5-22 – Displacement results for the front wheel full-system FEA model, for the first point of the
load spectrum (pure acceleration) ......................................................................................................... 73
Figure C-1 – Expanded efficiency map. .................................................................................................. 4
Figure C-2 – Map of output power of each motor. .................................................................................. 4
Figure C-3 – Map of power supplied to each motor. ............................................................................... 5
Figure C-4 – Map of total power supplied to the all motor. ..................................................................... 5
Figure D-1 – Identification of each zone considered in the thermal model of Concept 4. ...................... 6
Figure D-2 – Position of the temperature sensors mounted on FST 08e. .............................................. 7
Figure E-1 – Track layouts used in the steady-state simulations. ........................................................... 9
Figure F-1 – KISSsoft “Basic data” interface for planetary gears.......................................................... 12
Figure F-2 – KISSsoft “Rating” interface for planetary gears ................................................................ 12
Figure G-1 – Import geometry in Hypermesh ........................................................................................ 14
Figure G-2 – 3D mesh generation in Hypermesh ................................................................................. 14
Figure G-3 – Position of the contact, material and load sub-menus ..................................................... 14

xiii
List of symbols

𝑎 Distance between the front wheels to the CoG


𝐴 Amplitude ratio
̅
𝒂 Acceleration vector
𝑝𝑟𝑜𝑗
𝐴𝑥 Projected section of the car, perpendicular to the X axis
𝑝𝑟𝑜𝑗
𝐴𝑧 Projected section of the car, perpendicular to the Z axis
𝛼𝑖 Slip-angle of wheel 𝑖 (𝑖 = 1,2,3,4)
𝛼𝑡ℎ𝑒𝑟𝑚𝑎𝑙 Coefficient of thermal expansion
𝑎𝑥 Longitudinal acceleration
𝑎𝑦 Lateral acceleration
𝑏 Distance between the rear wheels to the CoG
𝑏𝐺 Facewidth (m)

𝛽 Slip-angle of the car


𝛽𝐺 Helix angle
𝛽𝑖 Projection of the slip-angle of the car on wheel 𝑖 (𝑖 = 1,2,3,4)
𝐶 Centre distance
𝑐 Distance between the left wheels and the CoG

𝐶0 Undeformed centre distance


𝑐𝑑𝑖𝑠𝑐 Thermal capacity of the brake disc
𝑐𝑝𝑎𝑑 Thermal capacity of the brake pad
𝐶𝑑 Drag coefficient
𝐶𝑖 Adjusted damping coefficient of wheel 𝑖 (𝑖 = 1,2,3,4)

𝐶𝑙 Lift coefficient
𝑐𝑝 Thermal capacity
𝐶𝑃𝑖 Damping coefficient of the tire of wheel 𝑖 (𝑖 = 1,2,3,4)
𝐶𝑠𝑖 Damping coefficient of wheel 𝑖 (𝑖 = 1,2,3,4)
𝐷 Accumulated damage
𝑑 Distance between the right wheels and the CoG

𝑑𝑏𝐺 Gear base diameter


𝑑𝐺 Gear pitch diameter
𝛿 Steering input
Δ𝐶 Centre distance variation
𝛿𝑖 Steering angle of wheel 𝑖 (𝑖 = 1,2,3,4)
Δ𝑇𝑖−𝑗 Temperature difference between component 𝑖 and 𝑗

xiv
𝑑𝑟 Radial displacement

𝑑𝑡 Tangential displacement
𝐸 Young modulus
𝐸′ Effective elastic modulus
Vertical displacement of the connection between sprung and un-sprung mass of
𝑒𝑖
wheel 𝑖 (𝑖 = 1,2,3,4)
𝐹 Facewidth (in)
𝑓 Coefficient of friction between gears
𝐹𝑑𝑜𝑤𝑛𝑓𝑜𝑟𝑐𝑒 Aerodynamic downforce
𝐹𝑑𝑟𝑎𝑔 Aerodynamic drag
𝜙 Pressure angle

𝜙𝑛 Normal pressure angle


𝜙𝑡 Transverse pressure angle

𝐹𝑥 Total longitudinal force


𝐹𝑋𝑠 Total longitudinal force in steady-state
𝐹𝑥𝑖 Longitudinal force on wheel 𝑖 (𝑖 = 1,2,3,4)
𝐹𝑦 Total lateral force
𝐹𝑌𝑠 Total lateral force in steady-state
𝐹𝑦𝑖 Lateral force on wheel 𝑖 (𝑖 = 1,2,3,4)
𝐹𝑧 Total vertical force
𝐹𝑍𝑠 Total vertical force in steady-state
𝐹𝑧𝑖 Vertical force on wheel 𝑖 (𝑖 = 1,2,3,4)
̅
𝒈 Gravity acceleration vector
𝛾 Wheel camber
GE Transfer function for the electrical controller system
𝐺𝑖 Ground height
GM Transfer function for the mechanical controlled system
𝐺𝑅 Total transmission ratio
𝐻 Applied heat power
ℎ𝐶𝑜𝐺 Height of centre of gravity
ℎ𝑓𝑜𝑟𝑐𝑒𝑑 Forced convection coefficient
ℎ𝑚𝑖𝑛 Minimum oil film thickness
ℎ𝑛𝑎𝑡𝑢𝑟𝑎𝑙 Natural convection coefficient
𝐻𝑏𝑟𝑎𝑘𝑒 Power dissipated in the braking system
𝐻𝑑𝑖𝑠𝑐 Power dissipated on the disc
𝐻𝐷𝑇 Power dissipated on the drive train

xv
ℎ𝑖 Vertical position of wheel 𝑖 (𝑖 = 1,2,3,4)
ℎ𝑖𝑠 Vertical position of wheel 𝑖 in steady-state (𝑖 = 1,2,3,4)

ℎ𝑖̇ Vertical velocity on wheel 𝑖 (𝑖 = 1,2,3,4)


𝐻𝑝𝑎𝑑𝑠 Power dissipated in the brake pads
𝐻𝑡𝑜𝑡𝑎𝑙 Total dissipated power
𝐼𝑡𝑜𝑡𝑎𝑙𝑖 Total inertia of wheel 𝑖 (𝑖 = 1,2,3,4)
𝑖1 Transmission ratio of the first stage
𝑖2 Transmission ratio of the second stage

𝐼𝐷𝑇 Inertia of the drive train


𝐼𝑖 Inner sector 𝑖 of the G-G diagram (𝑖 = 1, … ,16)
𝐼𝑚𝑜𝑡𝑜𝑟 Inertia of the motor
𝐼𝜑 Roll inertia of the sprung mass
𝐼𝜃 Pitch inertia of the sprung mass
𝐼𝑤ℎ𝑒𝑒𝑙 Inertia of the rim and tire
𝐾 Total kinetic energy
𝑘 Thermal conductivity
𝑘𝑑𝑖𝑠𝑐 Thermal conductivity of the brake disc
𝐾𝑖 Wheel-rate of wheel 𝑖 (𝑖 = 1,2,3,4)
𝑘𝑝𝑎𝑑 Thermal conductivity of the brake pad
𝐾𝑎𝑟𝑏𝐹 Stiffness of the front anti-roll bar
𝐾𝑎𝑟𝑏𝑅 Stiffness of the rear anti-roll bar
KI Speed controller integration gain

KIQ Current controller integration constant


Kp Speed controller gain
𝐾𝑝𝑖 Spring-rate of the tire of wheel 𝑖 (𝑖 = 1,2,3,4)

KPQ Current controller gain


𝐾𝑠𝑖 Spring-rate of wheel 𝑖 (𝑖 = 1,2,3,4)

Kt Motor torque constant


𝑙𝑎 Length of approach path
𝜆 Lubricant pressure-velocity coefficient
𝑙𝑓 Length of recess path
𝐿𝑖 Distance between the instantaneous corner centre and wheel 𝑖 (𝑖 = 1,2,3,4)

𝑚 Module
𝑀 Total mass
𝑚𝑛 Normal module

xvi
𝑚𝑐ℎ Sprung mass
𝑀𝑒𝑞 Total equivalent mass
𝑚𝐺 Gear ratio between mating gears
𝑚𝑖 Un-sprung mass of wheel 𝑖 (𝑖 = 1,2,3,4)
𝑀𝜑 Roll moment
𝑀𝜑𝑖 Roll moment of the tire of wheel 𝑖 (𝑖 = 1,2,3,4)
𝑀𝜓 Yaw moment
𝑀𝜓𝑠 Yaw moment in steady-state
𝑀𝜓𝑖 Self-aligning moment of the tire of wheel 𝑖 (𝑖 = 1,2,3,4)
𝑀𝑅 Motion ratio
𝑀𝜃 Pitch moment
𝑀𝜃𝑖 Rolling resistance of the tire of wheel 𝑖 (𝑖 = 1,2,3,4)
𝜇0 Lubricant viscosity at operating temperature
𝑛 Factor of safety
𝑁𝑟𝑒𝑓1 Reference speed of the first stage
𝑁𝑟𝑒𝑓2 Reference speed of the second stage
𝑁𝑓𝑖 Number of cycles without failure
𝑁𝑖 Number of cycles experienced
𝜈 Poisson ratio
𝑛𝑝 Pinion speed

𝑛𝑟𝑒𝑎𝑙 Actual speed value


𝑛𝑟𝑒𝑓 Speed setpoint

𝑂𝑖 Outer sector 𝑖 of the G-G diagram (𝑖 = 1, … ,16)


̅
𝝎 Angular velocity vector
𝜔𝑖 Angular velocity of wheel 𝑖 (𝑖 = 1,2,3,4)

𝑝 Gear pitch
̅
𝒑 Position vector of the car

𝑝ℎ𝑒𝑎𝑡 Heat partition coefficient


𝜓 Yaw angle

𝜓̇ Yaw rate
𝑅 Stress ratio
𝑟 Tire radius

𝒓̅𝒊 Position vector of wheel 𝑖 relative to the CoG (𝑖 = 1,2,3,4)


𝑅𝑐𝑜𝑟𝑛𝑒𝑟 Instantaneous corner radius
𝑅1 Pitch radius of the driving gear

xvii
𝑅2 Pitch radius of the driven gear

𝑟𝑒 Effective radius of curvature at pitch diameter


𝜌 Material density
𝜌𝑎𝑖𝑟 Air density
𝜌𝑑𝑖𝑠𝑐 Brake disc density
𝜌𝑝𝑎𝑑 Brake pad density
𝑅𝑖−𝑗 Thermal resistance between component 𝑖 and 𝑗
𝑆 Sensitivity of infra-red sensor

𝑠1 , 𝑠2 , 𝑠3 Constants characterizing the sectors of the G-G


𝑆𝑎 Limiting value of the amplitude stress
𝑆𝑒 Endurance limit
𝑆𝑓∗ Fatigue strength (ksi)
𝑆𝑓 Fatigue strength (MPa)
𝜎𝑎 Amplitude stress
𝜎𝐹 Gear root bending stress
𝜎𝐻 Gear contact stress
𝜎𝑚 Mean stress
𝜎1 , 𝜎2 , 𝜎3 Stresses characterizing the sectors of the G-G
𝜎𝑚𝑎𝑥 Maximum stress
𝜎𝑚𝑖𝑛 Minimum stress

𝑆𝑢 Ultimate tensile strength


𝑆𝑦 Yield strength

𝑡 Time
𝑇 Input torque
𝑇𝑁𝑇𝐶𝑖 Temperature measured by NTC 𝑖 (𝑖 = 1,2,3)
𝑇𝑟𝑒𝑓 Reference temperature
𝑇𝑟𝑒𝑓1 Reference torque of the first stage
𝑇𝑟𝑒𝑓2 Reference torque of the second stage
𝑇𝑠𝑢𝑟𝑓 Surface temperature
𝜃 Pitch angle of the sprung mass

𝜃̇ Pitch rate of the sprung mass


𝑇𝑖 Torque on wheel 𝑖 (𝑖 = 1,2,3,4)
𝑇𝑚𝑎𝑥,𝑛𝑒𝑔 Maximum negative torque
𝑇𝑚𝑎𝑥,𝑝𝑜𝑠 Maximum positive torque
𝑢 Rolling velocity at pitch line

xviii
𝑈𝐼𝑅 Measured voltage of the infra-red sensor

𝑈𝑁𝑇𝐶 Measured voltage of the NTC of the infra-red sensor


𝑈𝑖 Measured voltage measured in each NTC sensor
𝑈𝑖𝑛 Voltage supplied to each NTC sensor
𝑈𝑖𝑛𝐼𝑅 Voltage supplied to the infra-red sensor
𝑣 Velocity magnitude
̅
𝒗 Velocity vector of the CoG
̅𝒊
𝒗 Velocity vector of wheel 𝑖 (𝑖 = 1,2,3,4)

𝜑 Roll angle of the sprung mass


𝜑̇ Roll rate of the sprung mass
𝑣𝑖 Velocity magnitude of wheel 𝑖 (𝑖 = 1,2,3,4)
𝑣𝑥 Longitudinal velocity of the CoG
𝑣̇𝑥 Longitudinal velocity derivative of the CoG
𝑣𝑦 Lateral velocity of the CoG
𝑣̇𝑦 Lateral velocity derivative of the CoG
𝑊𝑡 Gear tangential load (lb)
𝑋 Longitudinal axis
𝑌 Lateral axis
𝑍 Vertical axis

𝑧𝐺 Number of teeth of a gear

𝑍̇ Vertical velocity of the sprung mass


𝜂 Transmission efficiency

𝐹𝑡 Gear tangential load (N)

xix
List of abbreviations

ABS Anti-locking Braking System


ARB Anti-Roll bar
CFD Computational Fluid Dynamics
CNC Computational Numeric Control
CoG Centre of Gravity
dof Degrees of freedom
EHD Elastohydrodynamic lubrication
FEA Finite element analysis

G-G Diagram of lateral vs longitudinal accelerations


IMU Inertia Measurement Unit
NTC Negative Temperature Coefficient

W-EDM Wire Electric Discharge Machining

xx
xxi
1 Introduction

1.1 Formula Student

Formula Student is an engineering competition. Students are pushed to put in practice the
engineering and management capabilities they have learned throughout their academic path, as well as
further developing them in a practical way. Teams are encouraged to design, build and race a formula
prototype vehicle, built according to a specific set of rules renewed and published every year. Over time
the competition has grown and became a perfect testing ground for new concepts and ideas, allowing
students to assess real world problems in an academic environment.

During the competition events, design decisions involved in the project and the performance of
the cars are evaluated, allowing the participants to showcase their skills to renowned judges in the
engineering, motorsport and automotive industries. The running prototypes are evaluated dynamically
in four separate events that try to push the vehicles to the limits of their capabilities:
• Acceleration: consisting in a straight line with a length of 75m;
• Skid-pad: consisting in a figure-of-eight track;
• Autocross: consisting of a single lap around a handling track with a typical length of 1km;
• Endurance & Fuel Economy: consisting of 22 laps around a closed circuit, usually similar to the
one of autocross.

FST Lisboa team represents the University of Lisbon. Founded in 2001, has already 18 years of
history and is the longest-active and most successful team in Portugal. Through those years the team
built 8 cars, three with internal combustion engines and five fully electric cars. The most recent one has
an all-wheel drive electrical powertrain, with hub mounted motors and a maximum (peak) power of
130kW (although it’s limited to 80kW imposed by the competition rules). It’s also equipped with a carbon
fibre monocoque chassis, suspension arms and aero package, all to increase total grip and handling
performance of the car.

Figure 1-1 – FST 08e


1.2 State of the art and Motivation

Since the rules of the competition started allowing cars with electrical powertrains (from 2009),
teams started showing different solutions for their drive train systems and how they are mounted on the
car. In the early days of the electric category, teams used a single electric motor connected to a
mechanical differential, both mounted in the chassis behind the driver, with two half drive shafts
connecting the differential to the wheels. This is the easiest solution as it only requires a single motor
(and a single inverter/controller) which for unexperienced teams starting to work on electrical
powertrains, means a lower risk solution with a higher chance of success when participating in the
events.

Since the speed and torque of electric motors are easy to control by electronic systems, the
natural evolution was to remove the mechanical differential, replacing the solution with two electric
motors (each one directly connected to each rear wheel). Implementing a torque vectoring system,
which is more effective than the mechanical differential, we can control the lateral and longitudinal grip
of the car in an advanced way [2].

From then on, the natural evolution was to add traction to the front wheels, which would be very
difficult and complex with an internal combustion powertrain. This adds complexity to the car’s design,
since in the rear the motors can be easily placed in the chassis behind the driver, whereas in the front
the chassis is heavily restricted by the rules (because of the required space for driver’s legs and feet).
This led to the emergence of many different solutions when teams started implementing all-wheel drive
systems. The most common is the one implemented in FST Lisboa cars since 2017, which has the
motors mounted directly on the upright, with the drive train inside it. This is the easiest approach to an
all-wheel drive Formula Student electric car, as it represents a solution that is reliable and is rather easy
to design and manufacture. It also allows an overall better packaging of the chassis.

Figure 1-2 – Current drive train concept used by FST Lisboa

As seen in Figure 1-2, the motor hangs on one side of the upright, between the suspension arms,
which is the main downside of this solution, as it restricts significantly the design freedom of the

2
suspension geometry. Also, since the motor and drive train are mounted on the un-sprung part of the
car, it means this solution may not be the best from the dynamic performance point of view, as the
positive aspect of improving chassis packaging may not compensate the downside of losing suspension
performance, when getting the best performance of the car. It is then interesting to evaluate the
performance of this solution and compare it with others, to find which is in fact the one that gives the
best overall performance.

Finding the best drive train concept has been the focus of research and development of many
teams within the Formula Student community as well as professional teams in other motorsport
categories [1]. The most common approach has been using steady state vehicle dynamics simulations,
to understand which basic design features are the best, such as gear ratios, motor characteristics and
tires selections [3]. By using transient simulations, Santos [4] has shown how the variation of suspension
stiffness and un-sprung mass affect the overall performance of a Formula Student prototype. However,
his work had the main purpose of helping in the design of the suspension system and it is not enough
to take an objective conclusion about the best drive train concept.

Structural design is necessary to assure that all parts of the system are resistant and stiff enough
to withstand the load cycles they experience [5]. To find the expected load spectrum for the system,
steady-state vehicle simulations can be made, that can estimate the forces acting on each wheel in
absolute maximum performance conditions. However Jacobson [6] shows that this approach can result
in load cases that are higher than the ones experienced in real life and suggests a transient vehicle
simulation to estimate the loads. This requires a driver model, which may respond to changes to the
car’s parameters differently to a driver in real life. For that reason Jacobson [6] recommends using
steady state simulation to find load cases and use transient vehicle simulation only when strictly
necessary.

Finite element analyses are the fastest and accurate way of performing structural design of
components. They are based in the finite element method, a numerical method, similar to the finite
difference method, but more general and more suitable for analysing complex physics problems,
geometries and boundary conditions [7]. For this thesis, Hypermesh, a commercial software, was used
to perform these analyses

Fatigue has been a matter of continuous research for more than 150 years, although only in the
late 19th century did its development increase. With the industrial revolution, the focus of scientific
interest shifted towards the mechanical behaviour of materials, due to frequent accidents with
continuous operation of stagecoaches and steam engines, namely a railroad accident in 1842, when a
train with 1500 occupants, moving from Versailles to Paris, crashed because of a fractured axle. This
tragedy became a milestone in fatigue research, as it led to the conclusion by Wilam Rankine, who later
would become a leading scientist in thermodynamics, that the fracture did not occur due to time-
dependent changes in the material, but it was a consequence of the continuous reduction of the load-
carrying cross section, due to crack propagation [8]. Experimental studies introduced by Wöhler were

3
the foundation for systematic approaches to fatigue-life predictions. His results led to the creation of the
Wöhler curve (S-N – stress amplitude vs number of cycles to fracture). More detailed studies followed
Wohler’s work, such as those of Gerber and Goodman on predicting the effects of mean stress [9].

1.3 Objectives

The purpose of this thesis is to find the best drive train concept for the next Formula Student
prototype to be built by FST Lisboa.

Simple CAD models of several different drive train concepts must be created, to estimate their
physical characteristics, which are then inserted in a transient simulator developed on purpose to find
which concept provides the best performance. At this point, the selected concept is further developed
into a design that can be machined, assembled and serviced, that will be studied using FEA (Finite
Element Analysis), to guarantee the necessary strength and stiffness.

1.4 Thesis outline

Following this introductory section, chapter 2 shows the design constraints and the possible
conceptual solutions for the system, with the respective characteristics, like mass, inertia and efficiency.

Chapter 3 shows the simulation models that were used. In Section 3.1, the transient model is
described, through the mathematical formulations for the vertical and planar vehicle models and other
sub models that affect the vehicle performance. Section 3.2 shows how the results from a steady state
simulation can be used to estimate the loads and temperatures on each wheel at each moment, that will
be used for the structural design.

Chapter 4 presents how the loads estimated in Section 3.2 can be used in the structural design,
as well as the approach to fatigue analyses. Stiffness targets are also presented in this chapter.

Chapter 5 shows how the mechanical design was performed, how the shapes and materials for
each component were selected.

Finally, chapter 6 presents conclusions and some suggestions for future work.

4
2 Possible concepts

There is almost an infinite amount of possible solutions for any kind of system. For that reason,
design constraints must be defined in order to restrict the number of possibilities. The main ones are:

• To have one DD5-14-POW-19000-Formula Student motor per wheel, made by AMK®;


• To have the same single transmission gear ratio (around 1:15) on all motors;
• To comply with the Formula Student rules, Rules 2019, published on the Formula Student
Germany website;
• No interference with any other parts of the car on the wheel travel;
• To withstand the loads and external solicitations applied to the system;
• To be stiff enough to not compromise dynamic handling performance;
• To be compatible with the next car produced by FST Lisboa, FST09e.

Where in the
car?

Sprung Un-sprung

Parallel to the Aligned with the Perpendicular to


Longitudinal Transversal
wheel axis wheel axis the wheel axis

Drive train Drive train Drive train Drive train


1 stage, aligned aligned with the aligned with the around the around the
2 stages
with the motor motor, inside the motor, outside motor, outside motor, outside
upright the upright the upright the upright

Figure 2-1 – Diagram of the possible positions of the motor/drivetrain system

All areas of a car are linked in some way or another. For that reason, designing a car requires an
iterative process, so that the final product may represent the best combination among all different areas.

The diagram in Figure 2-1 shows where and how the motor and drive train parts can be
assembled. From those possibilities, two were immediately excluded: the motor mounted longitudinally
on the car, or the motor mounted directly on the wheel, with its axis parallel to the wheel axis. The former
was excluded because it would interfere with the positioning of the car’s battery and the latter would
interfere with the suspension geometry, restricting its design freedom. They are both feasible solutions,
however such heavy interferences would require a much deeper analysis to the entire car’s design, as

5
the new position for the battery or the new suspension geometry could significantly change the dynamics
of the car.

From the diagram, 6 possibilities can be observed, which were transformed into 8 different design
concepts, presented in section 2.1, representing all the possible designs that comply with the
constraints. They were modelled in the CAD software SolidWorks, so the values for physical
characteristics such as mass and inertia can be easily extracted. To be able to compare all concepts,
the same materials are used in the CAD models, steel for gears, shafts and brake disc, and aluminium
for all other parts, with the typical physical properties of those kinds of materials.

The simulation models evaluate the performance of the entire car, so a base car is defined, to
which the masses and inertias relative to each concept are added. The base car is the current one used
by FST Lisboa, without any components of its drive train system.

2.1 Initial concepts

Some overall design concepts are presented in the next subsections.

2.1.1 Concept 1

4 (fixed)

2
2

5 (output)

1 (input) 4 (fixed)
1 (input) 5 (output)
3

Figure 2-2 – CAD of concept 1 (left) and drive train schematics (right)

This solution represents the most common one used by most teams, including FST Lisboa (the
same seen in Figure 1-2). The electric motor is mounted on the upright, aligned with the axis of the
wheel. Inside the upright, there´s a set of compound planetary gears, with input in the sun (1 in Figure
2-2) and output in the carrier (5 in Figure 2-2). In this solution the carrier is also the hub, which is divided
into two parts. Two bearings allow rotation and ensure the alignment of the hub, while a retainer prevents
lubricant from leaking out. Due to manufacturing reasons, the wheel rim is divided into two parts, an

6
aluminium centre and a carbon fibre rim and is secured by a single central nut. The braking system is
made of a floating disc mounted on the hub, actuated by a calliper on the outside of the disc, mounted
on the upright.

2.1.2 Concept 2

4 (output)
2
2
3

5 (fixed)
4 (output)
1 (input)
1 (input) 5 (fixed)
3

Figure 2-3 – CAD of concept 2 (left) and drive train schematics (right)

This concept is similar to Concept 1, but the motor and drive train parts are all shifted outwards,
with the motor now lying inside the upright and the drive train on the outside, with a similar set of
compound planetary gears, with input in the sun (1 in Figure 2-3) but now with the output in the ring (4
in Figure 2-3). In this solution the ring is connected to the hub, which rotates around the gear train. Two
bearings allow rotation and ensure the alignment of the hub, while two retainers, assembled next to
each bearing, prevent lubricant from leaking out. The wheel rim no longer needs to be divided into two
parts, as now it’s bolted to the hub using 5 M6 bolts. The braking system is made of a floating disc
mounted on the hub, actuated by a calliper on the outside of the disc, mounted on the upright.

7
2.1.3 Concept 3

4 (output)

1 (input)
1 (input)
2
3

3
4 (output)
2

Figure 2-4 – CAD of concept 3 (left) and drive train schematics (right)

In this case, the motor is placed inside the upright, with its axis perpendicular to the one of the
wheel. The gear train consists of two consecutive stages, the first with cylindrical gears and the second
with bevel gears. The input is in the first cylindrical gear (1 in Figure 2-4) and the output is in the second
bevel gear (4 in Figure 2-4). That bevel gear rotates around the motor and is connected to the hub,
which rotates around the motor, transmission and upright. Two large bearings allow rotation and ensure
the alignment of the hub, while two retainers, assembled next to each bearing, prevent lubricant from
leaking out. The wheel is bolted to the hub using 8 M4 bolts. The braking system is made of a floating
disc mounted on the hub, actuated by a calliper on the inside of the disc mounted on the upright.

8
2.1.4 Concept 4

6 (output) 6 (output)
3 (fixed)
3 (fixed)

4 2 5

1 (input)

2 7
(fixed)
5

7 (fixed) 1 (input) 4

Figure 2-5 – CAD of Concept 4 (left) and drive train schematics (right)

Similar to Concept 2, this concept has also the motor lying inside the upright, followed by the drive
train on the outside. In this case, however, a two-stage planetary system is used, with input in the sun
of the first stage (1 in Figure 2-5), and output in the ring of the second stage (6 in Figure 2-5). The ring
is connected to the hub, which rotates around the gear train. Two bearings allow rotation and ensure
the alignment of the hub, while two retainers, assembled next to each bearing, prevent lubricant from
leaking out. The wheel rim is bolted to the hub using 5 M6 bolts. The braking system is made of a floating
disc mounted on the hub, actuated by a calliper on the outside of the disc, mounted on the upright.
Comparing the geometry of this solution with the one of Concept 2, this one is longer, since it has two
complete stages of planetary gears, but it’s also leaner, as it has a smaller outside diameter.

2.1.5 Concept 5

4 (output)

2
1 (input)
3
3

2
4 (output)
1 (input) 5 (fixed)
5 (fixed)

Figure 2-6 – CAD of concept 5 (left) and drive train schematics (right)

9
This concept is characterised by having the motor aligned with the wheel, followed by a compound
planetary gear train. To reduce the overall length of the assembly compared to Concept 1, Concept 2
and Concept 4, the diameter of the gears was increased so that the space between the smaller planets
would be larger enough to push the motor slightly inwards of the gears and bearings. The input remains
in the sun (1 in Figure 2-6) and the output in the ring (4 in Figure 2-6), connected to the hub, which
rotates around the motor, gear train and upright. Two large bearings allow rotation and ensure the
alignment of the hub, while two retainers, assembled next to each bearing, prevent lubricant from leaking
out. The wheel is bolted to the hub using 8 M4 bolts. The braking system is made of a floating disc
mounted on the hub, actuated by a calliper on the inside of the disc mounted on the upright.

2.1.6 Concept 6

6 (output)
1 (input)
1 (input)
2

4
4
3
5 5

3 6 (output)

Figure 2-7 – CAD of concept 6 (left) and drive train schematics (right)

As in Concept 5, the overall diameter of the gear train was increased so the motor could be
pushed inwards, to reduce the overall length of the system. In this case however, the gear train is
comprised of three stages, being the first and third made of bevel gears, while the intermediate is made
of cylindrical gears. The input is in the first bevel gear of the first stage (1 in Figure 2-7), and the output
in the second bevel gear of the third stage (6 in Figure 2-7). Two large bearings allow rotation and
ensure the alignment of the hub, while two retainers, assembled next to each bearing, prevent lubricant
from leaking out. The wheel is bolted to the hub using 8 M4 bolts. The braking system is made of a
floating disc mounted to the hub, actuated by a calliper on the inside of the disc, mounted to the upright.

10
2.1.7 Concept 7

4 (fixed)

1 (input)

5 (output)

4 (fixed)
1 (input) 5 (output)

Figure 2-8 – CAD of concept 7: gear train (top-left); wheel, uptight and hub (top-right);
assembly (bottom-left); drive train schematics (bottom-right).

Unlike the previous concepts, this one has the motor mounted on the chassis. To lower the centre
of gravity height (hCoG) and reduce the possibility of interference with other systems of the car, the motor
was placed on the lowest part of the chassis, slightly tilted so the driveshaft can be aligned with the
motor when the car is in a static position. Directly connected to the motor is a compound planetary gear
train, with input in the sun (1 in Figure 2-8) and output in the carrier (5 in Figure 2-8). The carrier is then
connected to the wheel hub with a driveshaft, with two Rzeppa CV joints, one on each side. This joint
allows the transfer of power from the motor to the wheel in any position of the wheel. On the wheel, a
conventional upright-hub assembly is used, with a single nut securing the rim in place. The braking
system is made of a floating disc mounted on the hub, actuated by a calliper on the outside of the disc,
mounted on the upright.

11
2.1.8 Concept 8

4 (output)

3 4 (output)
1 (input)

2 3

1 (input)

Figure 2-9 – CAD of concept 8: gear train (top-left), assembly (bottom-left) and
drive train schematics (right).

Similar to Concept 7, the motor is mounted on the lowest part of the chassis, but instead of being
tilted, the gear train was sized so the shaft could be aligned with both the axis of the motor and of the
wheel. The gear train is composed by two stages of cylindrical gears, with input in gear 1 of Figure 2-9.
The output (4 in Figure 2-9) is connected to the motor by a driveshaft with two Rzeppa CV joints, one
on each side. On the wheel, a conventional upright-hub assembly is used, the same as in Concept 7.
The braking system is made of a floating disc mounted on the hub, actuated by a calliper on the outside
of the disc, mounted on the upright.

2.2 Transmission efficiency estimation

The CAD models are appropriate for estimating physical properties of the assembly. However, to
better analyse the performance of each concept, the effect of efficiency must be considered. In this type
of system, power losses come from the sliding between gear teeth and friction from the bearings and
retainers. While selecting bearings and retainers, the manufacturer presents the typical values for power
loss, so the efficiency of those parts is easily calculated [10]. For gear meshing, however, it’s a lot harder
to accurately estimate the efficiency value as there is an enormous dependency on the value of friction

12
coefficient f, which in turn depends on elastic deformation of the teeth, machining imperfections,
lubrication and load distribution on the contact surfaces. Typical values of 𝑓 range between 0.04 and
0.06 [11]. To properly compare all the iterations, the same approach is taken to estimate the efficiencies
of all the concepts, by calculating the theoretical efficiency 𝜂, given by:

𝑓 1 1 (𝑙𝑎 2 + 𝑙𝑓 2 )
1 − ( + ) for external gears
2 𝑐𝑜𝑠 𝜙 𝑅1 𝑅2 𝑙𝑎 + 𝑙𝑓
𝜂= (2.1)
𝑓 1 1 (𝑙𝑎 2 + 𝑙𝑓 2 )
1 − ( − ) for internal gears
{ 2 𝑐𝑜𝑠 𝜙 𝑅1 𝑅2 𝑙𝑎 + 𝑙𝑓

𝑙𝑎 = √𝑅𝑎2 2 (𝑅2 𝑐𝑜𝑠 𝜙)2 𝑅2 𝑠𝑖𝑛 𝜙 (2.2)

𝑙𝑓 = √𝑅𝑎1 2 (𝑅1 𝑐𝑜𝑠 𝜙)2 𝑅1 𝑠𝑖𝑛 𝜙 (2.3)

where 𝑙𝑎 is the length of approach path and 𝑙𝑓 is the length of recess path, 𝑅1 is the pitch radius of the

driving gear, 𝑅2 is the pitch radius of the driven gear, 𝑅𝑎1 is the tip radius of the driving gear, 𝑅𝑎2 is the

tip radius of the driven gear , and 𝜙 is the pressure angle . This equation is also valid for helical gears,
as long as the characteristic values of those gears are converted into those of the equivalent spur gear.

2.3 Data for simulation

To easily account for the effect of rotating mass/inertia in the dynamics of the car, Genta [12]
suggests using the equivalent or apparent mass. By combining the kinetic energy of translation and
rotation of the rotating parts of the car, the equivalent energy of the system, 𝐾, can be obtained.

1 1 1
𝐾 = 𝑀𝑣 2 + ∑ 𝐼𝑡𝑜𝑡𝑎𝑙𝑖 𝜔𝑖 2 = 𝑀𝑒𝑞 𝑣 2 (2.4)
2 2 2

For the case under analysis in this work, the equation can be rewritten as:

1 1 1 1 1
𝑀𝑣 2 + 4 (∑ ( 𝐼𝑤ℎ𝑒𝑒𝑙 𝜔𝑤ℎ𝑒𝑒𝑙 2 + 𝐼𝐷𝑇 𝜔𝐷𝑇 2 + 𝐼𝑚𝑜𝑡𝑜𝑟 𝜔𝑚𝑜𝑡𝑜𝑟 2 )) = 𝑀𝑒𝑞 𝑣 2
2 2 2 2 2

13
1 2
𝑣 2 𝑣 × 𝐺𝑅 2 𝑣 × 𝐺𝑅 2 1
𝑀𝑣 + 2 (𝐼𝑤ℎ𝑒𝑒𝑙 ( ) + 𝐼𝐷𝑇 ( ) + 𝐼𝑚𝑜𝑡𝑜𝑟 ( ) ) = 𝑀𝑒𝑞 𝑣 2 (2.5)
2 𝑟 2×𝑟 𝑟 2

where M is the total car mass, Meq is the equivalent car mass, Iwheel and ωwheel are the tire and rim inertias
and angular velocity, IDT and ωDT are the average drive train inertia and angular velocity, and Imotor and
ωmotor are the motor inertia and angular velocity. As shown, the angular velocities of each rotating part
are related to each other through the total gear ratio, GR, and are also related to the car’s velocity by
the tire radius r. The equivalent mass is then:

2
𝐼𝑤ℎ𝑒𝑒𝑙 𝐼𝐷𝑇 𝐺𝑅2 𝐼𝑚𝑜𝑡𝑜𝑟 𝐺𝑅2
𝑀𝑒𝑞 = 𝑀𝑣 + 4 ( 2 + + ) (2.6)
𝑟 4𝑟 2 4𝑟 2

Table A-1 in Appendix A shows the characteristics of each concept, with values of masses and
inertias, as well as gear dimensions and efficiency.

14
3 Vehicle dynamics modelling

The use of mathematical vehicle models to simulate and develop powertrain systems, braking
systems or other vehicle dynamics control systems, has gained significance over recent years, due to
its reduced cost-time compared to on-track testing [3].

Multibody transient models allow the evaluation of vehicle response in the time domain, so the
effects of damping, spring rate and mass can be seen. It is a complex way of modelling the dynamics
of a car, so in many occasions, like finding the absolute possible peak performance of a car, simpler
steady-state models are preferred.

3.1 Transient model

3.1.1 Overview

To simulate the transient response of the car, a numerical model was created using Simulink, a
toolbox software from Mathworks. As a whole, a car is a complex system. To simplify the overall model,
it can be divided into interconnected sub-systems, as shown in Figure 3-1.

𝐹𝑑𝑜𝑤𝑛𝑓𝑜𝑟𝑐𝑒
𝐹𝑑𝑟𝑎𝑔
𝑣̇𝑥

𝑣̇𝑦 𝐹𝑧𝑖

𝑇 ̅
𝒂
𝐹𝑦𝑖 𝜓̇ ̅
𝐩

𝑇𝑖 𝑣𝑥 𝜔𝑖
𝐹𝑥𝑖
𝑣𝑦

Figure 3-1 – Car model for transient simulation

The main elements that affect the car’s performance are the powertrain, tires and aerodynamics.
The loads created by these elements are supplied to a planar model which evaluates the horizontal
movement of the car. The accelerations (𝑣̇𝑥 and 𝑣̇𝑦 ), calculated from the equations of the planar model
in section 3.1.5, generate load transfers that, together with the aerodynamic downforce, are inserted

15
into a vertical model. This simulates the suspension response and calculates the vertical load on each
wheel. These values are then used by the tire model, together with the longitudinal and lateral velocities
(𝑣𝑥 and 𝑣𝑦 ) and the steering angle (𝛿), to calculate the longitudinal and lateral force produced by each
wheel.

3.1.2 Tires

There are three forces and three moments acting on the tire, as shown in Figure 3-2. The vertical
force (𝐹𝑧𝑖 ) represents the vehicle weight on the wheel i. In the case of a Formula Student car, the
longitudinal force (𝐹𝑥𝑖 ) appears while accelerating only on the driving wheels, whereas while braking, it
appears on all the wheels. The lateral force (𝐹𝑦𝑖 ) is responsible for making the car turn and only exists
when the tire slip angle is 𝛼𝑖 ≠ 0. The acting moments are the roll moment (𝑀𝜑 ), pitch moment (𝑀𝜃 ) and
yaw moment (𝑀𝜓 ). The first counteracts the camber angle (𝛾) and contributes to the magnitude of the
lateral force; the second is also called rolling resistance and counteracts the wheel rotation; the third is
also called self-aligning moment, counteracts the steering angle.

𝐹𝑧𝑖

𝑀𝜓

𝐹𝑦𝑖

𝑀𝜃

𝑀𝜑
𝐹𝑥𝑖

Figure 3-2 - Tire coordinate system [13]

Simulating tire behaviour is a difficult task and many models have been created to try to solve this
problem. Among them are Lumped Models [14], Brush Model, Tread Simulation Model [15], TMeasy
[16], Burckhardt [17] and the most commonly used, the Magic Formula from Pacejka [15]. These models
present non-linear equations with coefficients that are adjusted to the real tire response data.

FST Lisboa has been using the Pacejka Magic Formula in recent years, to select the tires to be
used on the car, based on real data supplied by the FSAE TTC (Formula SAE Tire Testing Consortium)
[18]. For this reason, this method was used to model the tire behaviour in the vehicle simulation. The
tires used by FST Lisboa are (brand and model) Hoosier R25B-7.5”x18”-10”.

16
Figure 3-4 – Example of a tire curve, modelled with the Magic Formula of Pacejka.
Represents 𝐹𝑦 vs 𝛼 for multiple 𝛾, for the tires used by FST Lisboa.

3.1.3 Motor

As presented in chapter 2, the car has 4 motors, each connected to one wheel. Each motor
requires a controller and an inverter, both provided by the same manufacturer. The controller is
responsible for calculating the current needed in the motor, so it can follow a desired speed reference,
with the desired torque being represented as a saturation, as shown in the diagram below:

Tmax,pos
nref + + + + nreal
KP 1ൗkt KPQ GE (s) kt GM (s)
- + Tmax,neg - +

KI KIQ

Current controller

Speed controller

Figure 3-5 - Motor controller model [19]

In the actual cars from FST Lisboa, the driver controls the accelerator, which controls de torque
on each motor instead of speed. The speed reference (from the manufacturer) is defined at 20000rpm,
the maximum rated speed of the motor. This way, the motor will always use the saturation torque,
corresponding to the driver input.

There is however, a variation to the torque limit of the motor, as shown in Figure 3-6. Also, the
competition rules state that the maximum power to all motors must be equal or less than 80kW. For
these reasons, in the simulation, the driver’s torque input is compared to the motor torque limit and the

17
power torque limit (obtained by dividing the maximum power of 80kW by the car’s speed at each instant)
and only the minimum among the three will be used as an input to the planar model.

Figure 3-6 - Motor torque (top) and power (bottom) curves [19]

3.1.4 Car vertical model

𝑒4 Z 𝑒3

𝜑
𝐾4 𝐶4 𝐶3
𝐾3
𝜃
𝑚𝑐ℎ , 𝐼𝜑 , 𝐼𝜃 ℎ4 CoG Y ℎ3

𝑚4 𝑚3
𝐾𝑎𝑟𝑏𝑅
𝑒2 X 𝐶𝑝4 𝑒1 𝐾𝑝3
𝐶𝑝3

𝐺4 𝐾𝑝4 𝐺3

𝐾2 𝐶2 𝐾1 𝐶1
ℎ2 ℎ1

𝐾𝑎𝑟𝑏𝐹
𝑚2 𝑚1
𝐶𝑝2 𝐶𝑝1
𝐾𝑝2 𝐾𝑝1
𝐺2 𝐺1

Figure 3-7 – 7-dof vertical model

To accurately calculate the vertical forces acting on each wheel, a 7 degrees of freedom (dof),
full-car, mass-spring-damper vertical model was implemented, as presented in Figure 3-7. This model
represents the best way of evaluating the interaction between all wheels, something that quarter-car or
half-car models don’t account for [20][21].

18
The 7 degrees of freedom are roll (𝜑), pitch (𝜃) and vertical position (𝑍) of the sprung parts of
the car and the vertical position of each un-sprung part (ℎ1 , ℎ2 , ℎ3 , ℎ4 ). The sprung components are
represented in the model by their mass (𝑚𝑐ℎ ) and inertias (𝐼𝜑 and 𝐼𝜃 ) while the un-sprung components
are represented only by their mass (𝑚1 , 𝑚2 , 𝑚3 , 𝑚4 ). The values for these parameters are shown in
Table A-1, in Appendix A.

The suspension is defined by an equivalent linear spring-damper system, mounted in parallel on


each quarter of the car, where 𝐾1 , 𝐾2 , 𝐾3 , 𝐾4 are the wheel-rates and 𝐶1 , 𝐶2 , 𝐶3 , 𝐶4 are the adjusted
damping coefficients, which were calculated using the spring-rates (𝐾𝑠𝑖 ), the damping coefficients (𝐶𝑠𝑖 )
and the suspension motion-ratio (𝑀𝑅), as shown in equation (3.1)[21]. Like the spring rates, the damping
coefficients are also adjusted with the motion ratio of the suspension. The tire can be represented in
similar way with 𝐾𝑝1 , 𝐾𝑝2 , 𝐾𝑝3 , 𝐾𝑝4 being the tire spring-rates and 𝐶𝑝1 , 𝐶𝑝2 , 𝐶𝑝3 , 𝐶𝑝4 being the tire damping
coefficients. Like in the real car, the left and right quarters in each axel are connected by an anti-roll bar,
represented by 𝐾𝑎𝑟𝑏𝐹 and 𝐾𝑎𝑟𝑏𝑅 .

𝐾𝑠𝑖
𝐾𝑖 =
𝑀𝑅 2
(3.1)
𝐶𝑠𝑖
𝐶𝑖 =
𝑀𝑅 2

The inputs for this model are a vertical force 𝐹𝑍 , corresponding to the aerodynamic downforce,
roll moment (𝑀𝜑 ) and pitch moment (𝑀𝜃 ), calculated from the lateral and longitudinal accelerations 𝑣̇𝑥
and 𝑣̇𝑦 ; ground height (𝐺1 , 𝐺2 , 𝐺3 , 𝐺4 ), which will be considered constant and equal to zero, representing
a perfectly smooth and levelled track.

To simplicity reasons the model was considered linear, i.e. the roll and pitch angles (𝜑 and 𝜃) are
considered small enough to use a small-angle approximation. This way, the displacements on the
connection between sprung and un-sprung masses are calculated by:

𝑒1 = 𝑍 + 𝑐𝜑 − 𝑎𝜃 𝑒2 = 𝑍 − 𝑑𝜑 − 𝑎𝜃
(3.2)
𝑒3 = 𝑍 + 𝑐𝜑 + 𝑏𝜃 𝑒4 = 𝑍 − 𝑑𝜑 + 𝑏𝜃

With these displacements, the equilibrium forces can be defined, where ∆𝑖 = 𝑒𝑖 − ℎ𝑖 and ∆̇𝑖 = 𝑒̇𝑖 −
ℎ̇𝑖 , for 𝑖 = 1,2,3,4.

4 4

𝑚𝑐ℎ 𝑍̈ + ∑ 𝐾𝑖 ∆𝑖 + ∑ 𝐶𝑖 ∆̇𝑖 = 𝐹𝑍 (3.3)


𝑖=1 𝑖=1

19
2 2 4 4

𝐼𝜃 𝜃̈ + 𝑎 (∑ 𝐾𝑖 ∆𝑠𝑖 + ∑ 𝐶𝑖 ∆̇𝑠𝑖 ) − 𝑏 (∑ 𝐾𝑖 ∆𝑠𝑖 + ∑ 𝐶𝑖 ∆̇𝑖 ) = 𝑀𝜃 (3.4)


𝑖=1 𝑖=1 𝑖=3 𝑖=3

𝐼𝜑 𝜑̈ − 𝑐 ( ∑ 𝐾𝑖 ∆𝑖 + ∑ 𝐶𝑖 ∆̇𝑖 ) − 𝑏 ( ∑ 𝐾𝑖 ∆𝑖 + ∑ 𝐶𝑖 ∆̇𝑖 ) = 𝑀𝜑 (3.5)


𝑖=1,3 𝑖=1,3 𝑖=2,4 𝑖=2,4

𝑚1 ℎ̈1 − 𝐾1 ∆1 − 𝐶1 ∆̇1 + 𝐾𝑝1 ℎ1 + 𝐶𝑝1 ℎ̇1 + 𝐾𝑎𝑟𝑏𝐹 (ℎ1 − ℎ2 ) = 0 (3.6)

𝑚2 ℎ̈2 − 𝐾2 ∆2 − 𝐶2 ∆̇2 + 𝐾𝑝2 ℎ2 + 𝐶𝑝2 ℎ̇2 + 𝐾𝑎𝑟𝑏𝐹 (ℎ2 − ℎ1 ) = 0 (3.7)

𝑚3 ℎ̈3 − 𝐾3 ∆3 − 𝐶3 ∆̇3 + 𝐾𝑝3 ℎ3 + 𝐶𝑝3 ℎ̇3 + 𝐾𝑎𝑟𝑏𝑅 (ℎ3 − ℎ4 ) = 0 (3.8)

𝑚4 ℎ̈4 − 𝐾4 ∆4 − 𝐶4 ∆̇4 + 𝐾𝑝4 ℎ4 + 𝐶𝑝4 ℎ̇4 + 𝐾𝑎𝑟𝑏𝑅 (ℎ4 − ℎ3 ) = 0 (3.9)

The system of equations is implemented as a state space system, with the expanded equations
𝑇
presented in Appendix B. The state variables are [𝑍, 𝑍̇ , 𝜃, 𝜃̇, 𝜑, 𝜑̇ , ℎ1 , ℎ̇1 , ℎ2 , ℎ̇2 , ℎ3 , ℎ̇3 , ℎ4 , ℎ̇4 ] and the
systems inputs are [𝐹𝑧 , 𝑀𝜃 , 𝑀𝜑 , 𝐺1 , 𝐺2 , 𝐺3 , 𝐺4 ]. The vertical tire load can then be computed as:

𝐹𝑧𝑖 = 𝐾𝑝𝑖 ℎ𝑖 + 𝐶𝑝𝑖 ℎ̇𝑖 , 𝑖 = 1,2,3,4 (3.10)

3.1.5 Planar model

The planar model calculates de car’s responses in the horizontal plane, parallel to the ground.
The car is considered as a rigid body and so the dynamics can be described by the Newton-Euler
equations of motion, using the system’s reference frame coincident with the car’s centre of gravity.

𝐹𝑥 = 𝑀𝑣̇𝑥 − 𝑀𝜓̇𝑣𝑦 (3.11)

𝐹𝑦 = 𝑀𝑣̇𝑦 − 𝑀𝜓̇𝑣𝑥 (3.12)

𝑀𝜓 = 𝜓̈𝐼𝜓 (3.13)

20
Forces 𝐹𝑥 and 𝐹𝑦 are the total longitudinal and lateral forces acting on the car, M is the total mass
of the car with driver; 𝑣𝑥 and 𝑣𝑦 are the longitudinal and transverse velocities, respectively, 𝜓̇ is the
angular velocity (or yaw rate) around the car’s vertical axis (Z as shown in Figure 3-7) and 𝐼𝜓 is the
inertia around the same vertical axis.

𝐿3 𝐿1
𝑅
𝐿4 𝐿2 𝛼1
𝐹𝑦3 𝐹𝑦1 𝛿1
Y
𝐹𝑥3 𝛽1
𝑣𝑦 𝐹𝑥1
𝑐 𝛽3
𝛽 𝑣
X
𝜓̇ 𝑣𝑥
𝛼2
𝑑 𝐹𝑦4 𝐹𝑦2 𝛿2
𝐹𝑥4
𝛽2
𝛽4 𝐹𝑥2
𝑏 𝑎
Figure 3-8 – Illustration of the planar model

Figure 3-8 shows how the forces are applied to the car in the planar model. Only the front wheels
have steering capabilities and each wheel has a different steering angle for the same driver input, due
to the Ackerman geometry [22]. The moment created by the mechanical or electrical differential is
neglected in this model.

Finding the slip-angle 𝛼 is necessary to calculate the lateral load produced by each tire. From
Figure 3-8, it can be seen that:

𝑣𝑦
𝛽 = tan−1 ( ) (3.14)
𝑣𝑥

21
𝛼𝑖 = 𝛽𝑖 − 𝛿𝑖 (3.15)

Where 𝛽𝑖 is the side-slip of the car 𝛽, projected on each wheel i. Knowing 𝛽 and the distance from
each wheel to the CoG, 𝒓̅𝑖 = (𝑥𝑖 , 𝑦𝑖 ), the velocity vector on each wheel, 𝒗
̅𝑖 , can be calculated and, with
it, 𝛼𝑖 as well.

𝑣𝑥 0 𝑥𝑖 𝑣𝑥 − 𝑦𝑖 𝜓̇
̅𝑖 = 𝒗
𝒗 ̅ × 𝒓̅𝑖 = [𝑣𝑦 ] + [ 0 ] × [𝑦𝑖 ] = [𝑣𝑦 + 𝑥𝑖 𝜓̇]
̅+𝜔 (3.16)
0 𝜓̇ 0 0
𝑣𝑦 + 𝑥𝑖 𝜓̇
𝛼𝑖 = tan−1 ( ) − 𝛿𝑖 (3.17)
𝑣𝑥 − 𝑦𝑖 𝜓̇

3.1.6 Aerodynamics

A simple aerodynamics model is used in the simulation, accounting only for the influence of drag
and downforce generated by the aerodynamic elements, which are the resistance the car experiences
while moving through the air and the vertical force, pointing towards the ground, respectively. The
equations that describe these loads are:

1 𝑝𝑟𝑜𝑗
𝐹𝑑𝑟𝑎𝑔 = 𝜌 𝐴 𝐶𝑑 𝑣𝑥 2 (3.18)
2 𝑎𝑖𝑟 𝑥
1 𝑝𝑟𝑜𝑗 (3.19)
𝐹𝑑𝑜𝑤𝑛𝑓𝑜𝑟𝑐𝑒 = 𝜌𝑎𝑖𝑟 𝐴𝑧 𝐶𝑙 𝑣𝑥 2
2

where 𝐶𝑑 and 𝐶𝑙 are the drag and lift (negative downforce) coefficients, respectively, 𝜌𝑎𝑖𝑟 is the air density
𝑝𝑟𝑜𝑗 𝑝𝑟𝑜𝑗
and 𝐴𝑥 and 𝐴𝑧 are the projected section area of the car, perpendicular to the X axis and the Z axis,
respectively.

Aerodynamics is a very tricky and important part of any car. To accurately simulate the
aerodynamics response over time, with the constant change in speed magnitude and direction, as well
𝑝𝑟𝑜𝑗 𝑝𝑟𝑜𝑗
as chassis movement, is extremely complex. To simplify, constant values of 𝐴𝑥 , 𝐴𝑧 , 𝐶𝑑 , 𝐶𝑙 and 𝜌
are used, which were estimated by using CFD (Computational Fluid Dynamics) simulation and wind
tunnel testing performed by FST Lisboa.

3.1.7 Sensors

The outputs calculated in the planar and vertical models are not enough to analyse the car’s
dynamics and take useful conclusions. For this reason, a block was added to the model, to calculate

22
accelerations, car trajectory and instantaneous corner radius with equations (3.20), (3.21) and (3.22),
respectively. It is important to notice that the velocity derivatives, calculated in the planar model, are not
the same as the longitudinal and lateral accelerations of the car (𝑎𝑥 ≠ 𝑣̇𝑥 and 𝑎𝑦 ≠ 𝑣̇𝑦 ). The acceleration
𝑇
̅, can be computed with equation (3.20), where 𝒗
vector 𝒂 ̅ = [𝑣𝑥 , 𝑣𝑦 , 𝑣𝑧 ] is the velocity vector, 𝝎
̅=
𝑇
[𝜑̇ , 𝜃̇, 𝜓̇] is the angular velocity vector and 𝒈
̅ = [0, 0, −9.81]𝑇 is the gravity acceleration vector [20][23].

𝑎𝑥
̅
𝜕𝒗
̅=
𝒂 ̅ ×𝒗
+𝝎 ̅+𝒈 𝑎
̅ = [ 𝑦] (3.20)
𝜕𝑡 𝑎 𝑧

𝑥 ∫(𝑣𝑥 cos 𝜓 − 𝑣𝑦 sin 𝜓) 𝑑𝑡


̅ = [𝑦] = [
𝐩 ] (3.21)
∫(𝑣𝑥 sin 𝜓 + 𝑣𝑦 cos 𝜓) 𝑑𝑡
𝑎𝑦
𝑅𝑐𝑜𝑟𝑛𝑒𝑟 = (3.22)
‖𝒗
̅‖

3.1.8 Driver inputs

The car’s ability of following a path comes from the driver inputs inside it. Several authors have
presented ways of modelling a driver, using predictive methods [24], but they are very complex and
unnecessary for the analysis being performed. To simplify, the inputs for the model, presented in Figure
3-9 and Figure 3-10, represent specific combinations of speed and steering angle, which characterize
well defined on-track situations:

• Accelerating and braking: to simulate longitudinal acceleration and braking event;


• Skid-pad: as per the competition rules and at constant speed, to analyse the transient
behaviour in corner entry and check the evolution towards the steady state phase;
• Chicane: to simulate a pure transient event with steering and velocity change;
• Hairpin: to check all phases of a corner, corner-entry, mid-corner and corner-exit;
• Slalom: at constant speed, to simulate a scenario of repeated change of direction, without
ever reaching the steady state phase.

23
Figure 3-9 – Speed references for simulations

Figure 3-10 – Steering references for simulations

3.1.9 Validation

Any time a simulator is created, it must be validated. Since the model represents the dynamics of
a Formula Student prototype, tests were performed with the current car from FST Lisboa to gather useful
data for the validation. The model was updated to correspond to the configuration of the current car,

24
which is equivalent to the one of Concept 1. Power limitations, aerodynamics coefficients and weight
are the main characteristics that were changed in the model for the validation.

Figure 3-11 – Comparison between measured and simulated lateral acceleration

Sensors installed in the car records the driver inputs, as well as the car’s response, with an Inertia
Measurement Unit (IMU) and speed sensors on the wheels. Figure 3-11 shows the plot of the lateral
acceleration measured in testing, overlaid with the simulated result, for the same inputs.

It’s possible to see that the simulation result doesn’t “follow” precisely the acquired data. Since
the test track used isn’t perfectly levelled and smooth, the driver had to make small corrections in the
throttle and steering inputs, to assure a constant corner radius. In the model, the track is considered
levelled and smooth, which means that when the same inputs are inserted, the simulated result is slightly
different. Despite this, the average error is about 9%, meaning the simulated results are acceptable and
the model can be used.

25
3.1.10 Simulation results

3.1.10.1 Accelerating and braking

Figure 3-12 – Plot of longitudinal acceleration (top), with detail for accelerating (middle) and
braking (bottom)

26
The response plots show that Concept 4 is the one with the highest longitudinal acceleration at
all phases of the Concept 4 event. The average relative difference from the best to the worst solutions,
for this case, is 3.3% while accelerating and 5.6% while braking.

3.1.10.2 Skid-pad

Figure 3-13 – Plot of lateral acceleration in the Skid-pad scenario, with detail of the
beginning of the left corner

27
The response plots show that Concept 4 is the one with the highest lateral acceleration at all
phases of the simulated event. The average relative difference from the best to the worst solutions, for
this case, is 0.6% while turning left and 0.8% while turning right, in the transient phase, and 0.4% in
steady-state.

3.1.10.3 Chicane

Figure 3-14 – Plot of lateral acceleration in the Chicane scenario (top), with detail of corner-entry
(middle) and corner-exit (bottom)

28
Observing the lateral acceleration response, Concept 5 is the best at corner-entry, but in the rest
of the simulation Concept 4 remains the one with higher performance. A chicane manoeuvre never
reaches a steady-state phase, as, while in corner-entry the speed is decreasing and the steering input
is increasing, in corner-exit the opposite happens. This explains why there is such a significant variation
of lateral acceleration over time. The average relative difference from the best to the worst solutions, for
this case, is 5.1% at corner-entry and 2.2% at corner-exit.

3.1.10.4 Hairpin

Figure 3-15 – Plot of lateral acceleration in the Hairpin scenario (top), with detail of corner-entry
(middle) and corner-exit (bottom)

29
Observing the lateral acceleration plot, it’s clear that Concept 5 has greater lateral acceleration
at corner-entry, during the transient phase. After that, when the acceleration progresses into the steady
state phase, Concept 4 becomes the best solution. The average relative difference from the best to the
worst solutions, for this case, is 1.2% at corner-entry, 8.3% in steady-state and 2.1% at corner-exit.

3.1.10.5 Slalom

Figure 3-16 – Plot of lateral acceleration in the Slalom scenario, with detail of one
corner (bottom)

The response plots show that Concept 4 is the one with highest lateral acceleration. The average
relative difference from the best to the worst solutions, for this situation, is 0.9%.

30
3.1.10.6 Simulation conclusions

From the acceleration plots presented, Concept 4 is the one that shows the highest performance
overall. Concept 5 shows higher lateral acceleration while braking and turning, which is when the
steering wheels have a higher vertical load. The car has a complex suspension system, similar to the
one shown in Figure 3-7, so any kind of input somewhere on the system will affect the response of the
entire system. From the response plots, it’s clear that at corner-entry, having a higher un-sprung mass,
and therefore lower natural frequency (as it happens with Concept 5 relative to the others), means there
is less variance in the vertical response of the wheels and consequently in the vertical load on the tire,
which, in turn, means a smaller variance in the total produced lateral force. This only happens in this
phase of the corner (while the car is under braking and turning), since when the response progresses
towards the steady state phase, a lower mass will mean a higher lateral acceleration. This effect is not
seen in the Skid-pad and slalom scenarios, as they are simulated at constant speed, so there isn’t any
braking while cornering situation.

The effect of performance difference at corner-entry is very small when looking at an entire lap of
a track, because the car will only be in that state in an extremely small fraction of the track’s length. For
that reason, this effect will be neglected and the chosen solution for the system is the one of Concept
4.

3.2 Steady state model

3.2.1 Software overview

Optimum Lap is a commercial software created by Optimum G, a motorsport consulting company.


It’s a steady state simulator, that uses a point-mass representation of the car, as shown in Figure 3-17.
It has an accuracy higher than 90% [25], which means it’s suitable for the required analyses.

𝑭𝒔𝒀
𝑭𝒅𝒓𝒂𝒈 𝑭𝒔𝑿

𝑭𝒔𝒁

Figure 3-17 – Illustration of the point-mass model

31
A steady state model shows the absolute peak performance of the car, as it considers the car is
always performing at its limits, being them the tires or the motors. It doesn’t account for any transient
effects or driver inconsistencies. For these reasons, the results extracted from this simulator represent
an upper limit for the performance of the car.

3.2.2 Software inputs

As a simplified model, it requires a small amount of inputs: overall mass, aerodynamic coefficients
(𝐶𝑑 and 𝐶𝑙 ), transmission ratio and efficiency, tire friction coefficients and motor torque vs speed curve
and track layout.

Table 3-1 – Car characteristics used in Optimum Lap simulations

Vehicle Type FSAE


General data Mass 281,8 kg
Driven Type AWD
Type Drag-Lift
Drag Coefficient 1,375
Aero Data Downforce Coefficient 2,875
Frontal Area 1,35 m2
Air Density 1,24 kg/m3
Tire Radius 0,228 m
Rolling Resistance 0,03
Longitudinal Friction 1,6 at 0 kg
Tire Data
Longitudinal Load Sensitivity 0,001
Lateral Friction 1,3 at 0 kg
Lateral Load Sensitivity 0,001
Transmission Type Sequential Gearbox
Gear Ratio 14,96
Transmission Data
Final Drive Ratio 1
Drive Efficiency1 92 %

Unlike in section 3.1.2, this software only uses the tire friction coefficients and load sensitivities.
The equations from the Magic Formula of Pacejka were used to find these constants.

1Accounts for the efficiency of Concept 4 and the average efficiency of the battery and inverters.

32
3.2.2.1 Motor

Since the software considers the car as a simple point-mass, the motor curve associated to that
point must represent all set of the motors existing in the real car. Also, it doesn’t have a way of limiting
the maximum available power in the car, as required by the competition rules. Efficiency is also simplified
and represented by a single efficiency value. So, to consider the real efficiency map shown in Figure
3-18, which was provided by the motor manufacturer, as well as the maximum power requirements and
the amount of motors, an equivalent motor curve was obtained.

Figure 3-19 – Efficiency map provided by the motor manufacturer [19]

The provided efficiency map is quite over-simplified, as there is no information between speed or
torque intervals. To expand this map, a 4th degree polynomial interpolation was used, resulting in the
new map of Figure C-1, presented in Appendix E.

Figure 3-20 – Example of the 4th degree approximation, of the first row of the
efficiency map, of Figure 3-18

A similar map can be created, showing power instead of efficiency, representing the output power
for each combination of speed and torque, Figure C-2. Dividing each entry of that power map by the

33
corresponding entry of the efficiency map, results in the map of Figure C-3, that represents the power
that the battery supplies to the motor (through the inverters).

Figure C-2 and Figure C-3 are relative to a single motor, but the car is an all-wheel drive, so all
the four motors must be considered. In testing, previous FST Lisboa teams have concluded that to have
a good balance between lateral and longitudinal performance, the maximum torque on the front motors
must be 50% of the maximum torque on the rear ones. This occurs because a tire cannot produce the
maximum longitudinal force and lateral force at the same time. The combined map of total input power
to the motors, accounting for the 50% limit at the front wheels is shown in Figure C-4.

It’s clear that there are torque/speed combinations where the total power is higher than the 80kW
limit defined by the competition rules. After removing those combinations, it becomes possible to find
the equivalent motor curve, by multiplying the entries from the map in Figure C-4 by the efficiencies in
Figure C-1, and then selecting the speed/torque combinations with higher power, Figure 3-21.

Figure 3-21 – Output torque and power curves of the equivalent motor

3.2.2.2 Track layout

The track layout represents the trajectory the car takes around the track. Twelve different layouts
were inserted in the software, two corresponding to the acceleration and skid-pad events and the others
corresponding to autocross and endurance events. FST Lisboa only participates in European
competitions, therefore the defined track layouts correspond to the ones of those competitions. Figure
3-22 shows an illustration of how the track layout of Formula Student Spain was obtained, both the
autocross layout as well as the endurance. The other track layouts used in the simulations are presented
in Appendix E.

34
Track limit

Car trajectory
Inward offset, to account for the width of the car
Figure 3-22 – Illustration of how the track of Formula Student Spain was obtained (Background
from Google Maps)

3.2.3 Results conversion

Figure 3-23 – Speed plot of an autocross lap in Formula Student Spain. Example
of an output from the simulation in Optimum Lap

Since the software uses a point-mass model, some calculations must be performed using its
outputs, like the one in Figure 3-23, to find the loads on each wheel. From the car’s longitudinal, lateral
and yaw accelerations, and aerodynamic forces, the total longitudinal, lateral and vertical forces, and
yaw moment, 𝐹𝑋𝑠 , 𝐹𝑌𝑠 , 𝐹𝑍𝑠 and 𝑀𝜓𝑠 , respectively, can be computed as:

𝐹𝑋𝑠 = 𝑀𝑎𝑥 (3.23)

35
𝐹𝑌𝑠 = 𝑀𝑎𝑦 (3.24)

𝐹𝑍𝑠 = 𝑀𝑔 + 𝐹𝑑𝑜𝑤𝑛𝑓𝑜𝑟𝑐𝑒 (3.25)

𝑀𝜓𝑠 = 𝐼𝜓 𝜓̈ (3.26)

The model as a point-mass doesn’t consider any steering input. The lateral friction coefficient
used in the simulation was obtained with the assumption that each tire is always operating at a slip angle
𝛼 that corresponds to the highest lateral load. Knowing the value of 𝛼 , equation (3.15) can be used to
estimate the steering angles, with beta being calculated with equation (3.27).

𝑎 + 𝑅𝑐𝑜𝑟𝑛𝑒𝑟 sin 𝛽
𝛽1 = tan−1 ( )
𝑅𝑐𝑜𝑟𝑛𝑒𝑟 cos 𝛽 − 𝑐
𝑎 + 𝑅𝑐𝑜𝑟𝑛𝑒𝑟 sin 𝛽
𝛽2 = tan−1 ( )
𝑅𝑐𝑜𝑟𝑛𝑒𝑟 cos 𝛽 + 𝑑
(3.27)
−1
𝑏 − 𝑅𝑐𝑜𝑟𝑛𝑒𝑟 sin 𝛽
𝛽3 = tan ( )
𝑅𝑐𝑜𝑟𝑛𝑒𝑟 cos 𝛽 − 𝑐
𝑏 − 𝑅𝑐𝑜𝑟𝑛𝑒𝑟 sin 𝛽
𝛽4 = tan−1 ( )
𝑅𝑐𝑜𝑟𝑛𝑒𝑟 cos 𝛽 + 𝑑

To be able to calculate the longitudinal and lateral forces on the tires, first the vertical forces must
be calculated.

𝑑 𝑏 ℎ𝐶𝑜𝐺 𝑑 ℎ𝐶𝑜𝐺 𝑏
𝐹𝑧1 = 𝐹𝑍𝑠 − 𝐹𝑋𝑠 − 𝐹𝑌𝑠
(𝑐 + 𝑑) (𝑎 + 𝑏) (𝑎 + 𝑏) (𝑐 + 𝑑) (𝑐 + 𝑑) (𝑎 + 𝑏)
𝑐 𝑏 ℎ𝐶𝑜𝐺 𝑐 ℎ𝐶𝑜𝐺 𝑏
𝐹𝑧2 = 𝐹𝑍𝑠 − 𝐹𝑋𝑠 + 𝐹𝑌𝑠
(𝑐 + 𝑑) (𝑎 + 𝑏) (𝑎 + 𝑏) (𝑐 + 𝑑) (𝑐 + 𝑑) (𝑎 + 𝑏)
(3.28)
𝑑 𝑎 ℎ𝐶𝑜𝐺 𝑑 ℎ𝐶𝑜𝐺 𝑎
𝐹𝑧3 = 𝐹𝑍𝑠 + 𝐹𝑋𝑠 − 𝐹𝑌𝑠
(𝑐 + 𝑑) (𝑎 + 𝑏) (𝑎 + 𝑏) (𝑐 + 𝑑) (𝑐 + 𝑑) (𝑎 + 𝑏)
𝑐 𝑎 ℎ𝐶𝑜𝐺 𝑐 ℎ𝐶𝑜𝐺 𝑎
𝐹𝑧4 = 𝐹𝑍𝑠 + 𝐹𝑋𝑠 + 𝐹𝑌𝑠
(𝑐 + 𝑑) (𝑎 + 𝑏) (𝑎 + 𝑏) (𝑐 + 𝑑) (𝑐 + 𝑑) (𝑎 + 𝑏)

The lateral forces 𝐹𝑦𝑖 can be calculated with the vertical forces 𝐹𝑧𝑖 and the tire’s lateral friction
coefficient and load sensitivity. Those lateral forces, together with the diagram presented in Figure 3-8,
the results from equations (3.23), (3.24) and (3.26), and the 50% limit on the front wheels, presented in
section 3.2.2, can be used to calculate the longitudinal forces on each wheel.

𝐹𝑦𝑖 = 𝜇(𝐹𝑧𝑖 ) × 𝐹𝑧𝑖 (3.29)

𝐹𝑥1 cos 𝛿1 − 𝐹𝑦1 sin 𝛿1 + 𝐹𝑥2 cos 𝛿2 − 𝐹𝑦2 sin 𝛿2 + 𝐹𝑥3 + 𝐹𝑥4 − 𝐹𝑑𝑟𝑎𝑔 = 𝐹𝑋𝑠 (3.30)

𝐹𝑥1 sin 𝛿1 + 𝐹𝑦1 cos 𝛿1 + 𝐹𝑥2 sin 𝛿2 + 𝐹𝑦2 cos 𝛿2 + 𝐹𝑦3 + 𝐹𝑦4 = 𝐹𝑌𝑠 (3.31)

36
𝐹𝑥1 (sin 𝛿1 𝑎 − cos 𝛿1 𝑐) + 𝐹𝑥2 (sin 𝛿2 𝑎 + cos 𝛿2 𝑑) − 𝐹𝑥3 𝑐 + 𝐹𝑥4 𝑑
(3.32)
+𝐹𝑦1 (cos 𝛿1 𝑎 + sin 𝛿1 𝑐) + 𝐹𝑦2 (cos 𝛿2 𝑎 − sin 𝛿2 𝑑) − 𝐹𝑦3 𝑏 − 𝐹𝑦4 𝑏 = 𝑀𝜓𝑠

Wheel travel ℎ𝑖𝑠 can be estimated using the vertical loads 𝐹𝑧𝑖 on each wheel and the respective
wheel-rate (𝐾𝑖 ), as shown in equation (3.33).

𝐹𝑧𝑖
ℎ𝑖𝑠 = (3.33)
𝐾𝑖

Knowing the magnitude of 𝐹𝑥𝑖 , the torque on each wheel, 𝑇𝑖 , can be calculated by multiplying the
forces by the tire radius r. The speed of each wheel, 𝑣𝑖 , can be estimated with equation (3.35), aided by
Figure 3-8.

𝑇𝑖 = 𝐹𝑥𝑖 × 𝑟 (3.34)

𝑣𝑖 = 𝜓̇𝐿𝑖 cos 𝛼𝑖 (3.35)

With torque and speed, the output power on each wheel can be calculated. In a braking event,
the power will be negative. Using the limitation imposed by FST Lisboa, regarding the battery, a
maximum of 15kW can be regenerated under braking. Therefore, when the magnitude of braking power
is higher than 15kW, it is considered that the remaining power above those 15kW will come from the
mechanical braking system, Figure 3-24.

Figure 3-24 – Plot of total braking power, electrical regeneration and mechanical braking

There’s always power loss in any type of mechanism. The dissipated power in the motors can be
calculated applying the efficiency map shown in Figure C-1, to the power response plot. This dissipated
power, together with the power dissipated in the drive train parts 𝐻𝐷𝑇 and the power dissipated in the

37
braking system 𝐻𝑏𝑟𝑎𝑘𝑒 , represents the total dissipated power 𝐻𝑡𝑜𝑡𝑎𝑙 , which is useful for estimating the
temperatures of all parts of the system.

3.2.4 Operating temperature estimation

One of the main factors that influences the mechanical properties of materials is temperature. To
assure the proper stiffness and resistance of all components, their operating temperatures must be
estimated. A purpose-developed thermal model was used, created in Simulink, where the dissipated
power losses are the input and the temperatures of the various components are the output.

The model is made of interconnected thermal resistances and masses, which are used to
estimate the temperatures variation over time. These resistances were obtained by making a thermal
finite element analysis of the Concept 4, to which heat powers were applied in different places. In that
FEA model, the applied heat power 𝐻 and the temperature differences between components ∆𝑇𝑖→𝑗 were
used to calculate the resistance 𝑅𝑖→𝑗 , with equation (3.36) [26].

∆𝑇𝑖→𝑗
𝑅𝑖→𝑗 = (3.36)
𝐻

Figure 3-25 – Diagram of the thermal model in Simulink

Faramarz and Salman [27] analysed the temperature distribution and forced convection effect on
brake discs and pads. They concluded that the amount of braking power dissipated through the disc
and pads is:

38
𝐻𝑑𝑖𝑠𝑐 = 𝐻𝑏𝑟𝑎𝑘𝑒 × 𝑝ℎ𝑒𝑎𝑡 (3.37)

𝐻𝑝𝑎𝑑𝑠 = 𝐻𝑏𝑟𝑎𝑘𝑒 × (1 − 𝑝ℎ𝑒𝑎𝑡 ) (3.38)

√𝑘𝑑𝑖𝑠𝑐 𝜌𝑑𝑖𝑠𝑐 𝑐𝑑𝑖𝑠𝑐


𝑝ℎ𝑒𝑎𝑡 = (3.39)
√𝑘𝑑𝑖𝑠𝑐 𝜌𝑑𝑖𝑠𝑐 𝑐𝑑𝑖𝑠𝑐 + √𝑘𝑝𝑎𝑑 𝜌𝑝𝑎𝑑 𝑐𝑝𝑎𝑑

Where pheat is the heat partition coefficient and the term √𝑘𝑖 𝜌𝑖 𝑐𝑖 is the thermal effusivity of the
material.

Vidiya and Singh[28] analysed the typical air flow characteristics in regular Formula Student cars,
through CFD analysis, obtaining a linear forced-convection coefficient ℎ𝑓𝑜𝑟𝑐𝑒𝑑 , dependent on the car’s
speed 𝑣:

ℎ𝑓𝑜𝑟𝑐𝑒𝑑 = 17𝑣 + ℎ𝑛𝑎𝑡𝑢𝑟𝑎𝑙 (3.40)

To validate the model, the thermal constants were modified to correspond to the ones of the most
recent car from FST Lisboa, using the same approach as before to fin the thermal resistances. For the
same test session as the one already presented in section 3.1.9, sensors were mounted on the car to
measure the temperature of the brake disk, pads and calliper, and the upright, in positions shown in
Figure D-1, in Appendix D.

While the car is running, the brake disc is rotating, an infrared sensor (ZTP-135SR) was used to
measure its temperature, as it doesn’t require any contact with the surface. This type of sensor is made
of a thermopile (which are several thermocouples connected in series), and an NTC resistor, which is
used to calculate the reference temperature 𝑇𝑟𝑒𝑓 . Using equations (3.41) and (3.42), the measured
surface temperature can be calculated.

1
𝑇𝑟𝑒𝑓 =
𝑈𝑁𝑇𝐶 (3.41)
ln ( )
𝑈𝑖𝑛𝐼𝑅 − 𝑈𝑁𝑇𝐶 1
+
3920 298.15
1ൗ
𝑈𝐼𝑅 4 4
(3.42)
𝑇𝑑𝑖𝑠𝑐 =( + (𝑇𝑟𝑒𝑓 + 273.15) ) − 273.15
𝑆

where 𝑈𝑖𝑛𝐼𝑅 is the input voltage supplied to the sensor (5V), 𝑈𝑁𝑇𝐶 is the measured voltage of the NTC
and 𝑈𝐼𝑅 is the measured voltage of the infra-red sensor, both in volt [V]. Temperature values in Celsius
[ºC].

39
𝑆 is a constant that represents the sensor’s field of view, the emissivity of the disc surface, the
Stefan-Boltzman constant and the sensor sensitivity, and was defined experimentally, as shown on
Figure 3-26. After heating up a sample brake disc to a temperature of 100ºC, it was placed on 3 supports
with minimum contact over the infra-red sensor, which was used to measure the temperature variation
over time. 𝑆 = 5x10-11 was obtained by comparing the measured values of the infra-red sensor with
those of an external temperature sensor, measuring the surface temperature.

Thermopile voltage

NTC voltage

External temperature sensor

Power supply

Infra-red sensor

Figure 3-26 – Setup to obtain the value of K. The infra-red sensor is underneath the
brake disc

With the voltage measured in the NTC and the surface temperature 𝑇𝑠𝑢𝑟𝑓 , measured in the
external temperature sensor, equation (3.41) can be used to calculate the temperature 𝑇𝑟𝑒𝑓 , and
equation (3.42) can be manipulated into equation (3.43), to find the value of K. The measured 𝑇𝑠𝑢𝑟𝑓 and
𝑇𝑟𝑒𝑓 are presented in Figure 3-28.

𝑈𝐼𝑅
𝑆= 4 4 (3.43)
(𝑇𝑠𝑢𝑟𝑓 + 273.15) − (𝑇𝑟𝑒𝑓 + 273.15)

For the remaining measurement points, NTC resistors (NTCLE100E3) were used. Their
temperatures, 𝑇𝑁𝑇𝐶𝑖 can be estimated with equation (3.44), where 𝑈𝑖𝑛 is 3.3V and 𝑈𝑖 is the voltage
measured in each sensor.

40
1
𝑇𝑁𝑇𝐶𝑖 =
𝑈𝑖 (3.44)
ln ( )
𝑈𝑖𝑛 − 𝑈𝑖 1
+
3977 298.15

Figure 3-27 – Comparison between measured and simulated temperature of the brake disc

Figure 3-28 – Measured temperature variation of 𝑇𝑠𝑢𝑟𝑓 and 𝑇𝑟𝑒𝑓

Figure 3-27 shows the acquired data overlaid with the simulation results. Comparison between
measured and simulated temperatures for other components are presented in Figure 3-27. It’s clear that
the simulation is well designed as the presented result is within an error margin of 6%.

41
3.2.5 Simulation results

3.2.5.1 Converted results

Using equations (3.23) through (3.32), for the track shown in Figure 3-22, the forces on each
wheel are obtained.

Figure 3-29 – Forces Fx, Fy and Fz on each wheel, for an autocross lap in Formula Student Spain

42
3.2.5.2 Temperature

With the thermal constants corresponding to Concept 4 (Appendix E), the following results were
obtained from the thermal model, assuming an ambient temperature of 25ºC.

Figure 3-30 – Temperatures at the front-left (top) and rear-left (bottom) wheels, during an
autocross lap in Formula Student Spain. It represents the fourth lap in a row, which justifies the high
temperatures at the beginning of the plot

43
4 Load cases for structural analysis

The car will be subjected to different types of loading throughout its life. In the earlier days it will
be subjected to mild testing, at low speed and low power, since there are multiple electrical and
mechanical systems that must be checked. After that, the car will be tested at racing speeds, to find the
ideal suspension setup and tune the control systems. Then the car will be subjected to the same types
of loadings, but in the competition environment. Finally, there is a testing phase after the competitions,
to help in the development of the following year’s car.

To simplify the simulations, the initial phase of testing was neglected, as the expected loads are
very small. For the remaining phases, it was assumed that the performed tests represented the dynamic
events of the competitions. For the entire life of the car, the expected number of times each event is
tested is:

• Acceleration: 84 times, 4 laps each time;


• Skid-pad: 70 times, 4 laps each time;
• Autocross: 14 times, 4 laps each time;
• Endurance: 28 times, 22 laps each time.

With this information, together with the results from the steady state simulations, it is possible to
estimate the total expected travelled distance about 760Km, with a total expected operating time about
55h.

4.1 Drive train load cases

The structural design of the gears was performed in KISSsoft (section 5.1.2). The software
accepts a full load spectrum, with relative torque, speed and time, the latter representing the percentage
of time the system is experiencing the specific torque and speed combination. Table 4-1 shows the first
four entries of the spectrum, as an example.

Table 4-1 – First four entries of the drive train load spectrum for a rear wheel
Ner [%] Time [%] Torque [%] Speed
1 0.00241 80.95 45
2 0.00011 80.95 40
3 0.02804 76.19 50
4 0.20479 76.19 45

The load spectrum is the same between the front wheels and between the rear wheels, as
presented in Table 4-1. The values in the spectrum are relative to the maximum torque, speed and time,

44
which apart from time, are different in each stage of the transmission system. In the first stage, the
reference torque and speed are 21 Nm and 20000rpm, corresponding to the limits of the motor. The
references for the second stage depend on the transmission ratio of the first stage. The software uses
this spectrum to apply the Palmgren-Miner rule during the analysis.

4.2 Suspension load cases

All components, except for gears, were analysed using FEA (Finite Element Analysis). Those
analyses require the definition of load cases and constraints applied to the components. From the
calculations made and the results obtained in section 3.2, a load spectrum can be defined, with all the
information regarding the car’s performance throughout its life. This spectrum has 877080 load cases,
each representing a single set of longitudinal, lateral and vertical forces and accelerations, wheel
positions, steering angles and temperatures for each wheel. Naturally this amount of information is too
high to be inserted into a FEA software, therefore only the most significant points of the spectrum were
considered in those analyses.

To find those reference points, the G-G diagram of the car, shown in Figure 4-1, was observed.
Each grey dot represents a point of the spectrum, with a certain longitudinal and lateral acceleration.

O1 I16 I1
O16 O2 I2
O15 O3 I15 I3
O14 O4 I14
I4

O13 O5 I13 I5
I12 I6
O12 O6 I11 I7
I10 I8

O11 O7

O10 O8
O9 I9

Figure 4-1 – G-G diagram of the car (with Concept 4), with concentric inner and outer rings

The diagram is divided into 2 concentric rings, each divided into 16 areas. These areas represent
specific states of the car, such as pure accelerating or braking (areas 1 and 9), pure cornering (areas 5
and 13) and combinations of the two (remaining areas). The outer ring (Oi) represents the absolute peak
performance of the car, so from a structural point of view they correspond to the critical points in terms
of stress, whereas the inner circle (Ii) represents the load cases that the system will experience more
often. Each yellow dot is the selected reference point the respective area, corresponding to a single set
of reference loads. The car is considered to be symmetrical, so both wheels on each axle must have
the same strength and stiffness. For that reason, the same loads were considered for both wheels on

45
the axle. For the initial finite element analyses, presented in sections 5.1, 5.2 and 5.3, only the maximum
load cases from sectors O1, O3, O5, O7, O9, O11, O13 and O15 are considered, which are shown in
Table 4-2 and Table 4-3.

Table 4-2 – Maximum reference load cases for each wheel of the front axle

Maximum load cases on the front axle


ID. 𝑭𝒙 [N] 𝑭𝒚 [N] 𝑭𝒛 [N] 𝑴𝝋 [Nm] 𝜹 [º] 𝒉𝒔 [mm]
O1 832 0 264 -25 0.0 -8.7
O3 360 442 705 10 1.3 -0.1
O5 1 507 741 23 -1.7 0.6
O7 -915 609 1252 65 1.3 10.5
O9 -1755 0 1823 53 0.0 21.6
O11 -1863 -1332 1915 133 1.9 23.3
O13 22 -2750 2156 111 1.9 28.0
O15 557 -477 756 -2 -1.7 0.9

Table 4-3 – Maximum reference load cases for each wheel of the rear axle

Maximum load cases on the rear axle


ID. 𝑭𝒙 [N] 𝑭𝒚 [N] 𝑭𝒛 [N] 𝑴𝝋 [Nm] 𝜹 [º] 𝒉𝒔 [mm]
O1 1359 0 1159 -41 9.4
O3 712 513 755 -17 1.3
O5 2 424 560 4 -2.6
O7 -401 323 448 15 -4.8
0.0
O9 -715 0 742 21 1.1
O11 -890 -1350 1095 15 8.1
O13 43 -2860 1960 -27 25.5
O15 1158 -1629 1606 -49 18.4

Even though the thermal model of section 3.2.4 estimates the temperature variation through time
on each component, the temperatures considered for the structural design are the maximum
experienced by each part, which are the ones presented in Table 4-4.

Table 4-4 – Temperature of components considered in the structural analyses

Temperature [ºC]
Brake disc Calliper Drive train Hub Upright
Front axle 500 100 80 80 70
Rear axle 250 80 80 80 60

46
4.3 Fatigue

Structural components are frequently subjected to repeated loads, and the resulting cyclic
stresses create microscopic physical damage in their materials. Even at stresses significantly below the
ultimate strength of the material, the microscopic damage can accumulate with the continued cyclic
loading, developing cracks or other macroscopic damages that lead to failure of the component [9].

To account for the variance of stress amplitude during the life of components, Palmgren in 1924
and Miner in 1945, introduced the hypothesis of linear damage accumulation, which is a simple tool for
service-life assessment. The resulting Palmgren-Miner rule, characterized by equation (4.1), is used to
estimate the cumulative damage on each component. If the estimated total damage is greater than 1,
then the rule states a failure is highly likely to occur [8][5].

𝑁𝑖
∑ =𝐷 (4.1)
𝑁𝑓𝑖

where 𝑁𝑖 is the number of cycles the component will experience at a certain stress, 𝑁𝑓𝑖 is the maximum
number of cycles the component can withstand at the same stress, without failing, and D is the total
accumulated damage, that must be lower than 1.

To use this rule, it’s necessary to know the applied stresses and the corresponding number of
cycles. Three random points of each sector of Figure 4-1 are selected, each corresponding to a set of
loads that are applied to the FEA model, as presented in section 5.5, resulting in a certain stress
distribution in each component. While analysing the stress result of the several components, the location
of the highest stress (critical) in each component can be found.

The Von-Mises stress corresponding to the 3 random load cases ( 𝜎1 , 𝜎2 , 𝜎3 ), in the most critical
location of each component are inserted into equation (4.2), together with the respective loads of the
set, to find constants 𝑠1, 𝑠2 and 𝑠3, which characterize each sector of Figure 4-1. After applying this
process to every sector, the complete stress spectrum for every component is obtained, to which the
Rainflow method can be applied.

𝑠1 𝐹𝑥1 𝐹𝑦1 𝐹𝑧1 −1 𝜎1


{𝑠2 } = [𝐹𝑥2 𝐹𝑦2 𝐹𝑧2 ] {𝜎2 } (4.2)
𝑠3 𝐹𝑥3 𝐹𝑦3 𝐹𝑧3 𝜎3

where the indices 1, 2 and 3 correspond to the three selected points per sector.

The Rainflow method is a cycle counting algorithm, suitable for fatigue analysis of structures, that
reduces the length of a varying stress spectrum into a set of stress reversals. Originally developed by
Tatsuo Endo and M.Matsuishi in 1689, it was improved by Downing and Socie in 1982, becoming one
of the most used counting techniques [8][29]. It was named after its comparison with the rain “flowing”

47
from the traditional pagoda roofs. This method was applied using a dedicated Matlab function called
“rainflow”, resulting in the number of cycles 𝑁𝑖 , and mean and amplitude stresses (𝜎𝑚 and 𝜎𝑎 ).

Mean and amplitude stresses are necessary to calculate 𝑁𝑓𝑖 . Equation (4.3) corresponds to the
modified Goodman fatigue criterion, which was developed by Smith in 1942, based on an early proposal
by Goodman [9][5]. This criterion is represented by a straight line, as shown in Figure 4-2, which is a
conservative approach, so the consequence of its error is an increase in the factor of safety.

𝑆𝑦

Yield (Langer) line


Alternating stress 𝜎𝑎

𝑆𝑒
Gerber line
Load line

Modified Goodman line


𝑆𝑎

ASME-elliptic line
Soderberg line
0
0 𝑆𝑚 𝑆𝑦 𝑆𝑢

Midrange stress 𝜎𝑚

Figure 4-2 – Fatigue diagram for various criteria of failure. For each criterion, points “above” the
respective line indicate failure [5]

𝜎𝑎 𝜎𝑚 1 (4.3)
+ =
𝑆𝑒 𝑆𝑢 𝑛

where 𝑛 is the factor of safety, 𝑆𝑢 is the tensile strength of the material and 𝑆𝑒 is the endurance limit of
the component. If the maximum applied stress is smaller than 𝑆𝑒 , it is considered that the component
will have an infinite life, so fatigue won’t be the cause of any failure. For the case being presented in this
document, 𝑆𝑒 was considered as the stress value at 𝑁𝑓𝑖 = 107 cycles, in the S-N curve of each material.

When the maximum applied stress is greater than 𝑆𝑒 , fatigue will have an impact, which can be
evaluated with equation (4.4).

𝜎𝑎
, 𝜎𝑎 > 𝑆𝑎
1 𝜎𝑚
𝑆𝑓 = {𝑛 − 𝑆 (4.4)
𝑢
𝑆𝑒 , 𝜎𝑎 ≤ 𝑆𝑎

𝑟𝑆𝑒 𝑆𝑢
𝑆𝑎 = (4.5)
𝑟𝑆𝑢 + 𝑆𝑒

48
𝜎𝑎
𝐴= (4.6)
𝜎𝑚

𝜎𝑚𝑎𝑥 − 𝜎𝑚𝑖𝑛 (4.7)


𝜎𝑎 =
2

𝜎𝑚𝑎𝑥 + 𝜎𝑚𝑖𝑛 (4.8)


𝜎𝑚 =
2

where 𝑆𝑓 is the fatigue strength, 𝑆𝑎 is the limiting value of 𝜎𝑎 , 𝐴 is the ratio between amplitude and

mean stresses, and 𝜎𝑚𝑎𝑥 and 𝜎𝑚𝑖𝑛 are the maximum and minimum applied stresses, respectively [5].
If 𝜎𝑎 > 𝑆𝑎 , there will be reduction in life due to fatigue. That reduction is evaluated through the maximum
number of cycles until failure, 𝑁𝑓𝑖 , calculated with equations (4.9), (4.13) and (4.14) [13].

log 𝑁𝑓𝑖 = 18.21 − 7.73 log(𝑆𝑓∗ (1 − 𝑅)0.62 − 10) (4.9)

log 𝑁𝑓𝑖 = 7.56 − 2.73 log(𝑆𝑓∗ (1 − 𝑅)0.4 − 23.7) (4.10)

log 𝑁𝑓𝑖 = 7.77 − 2.15 log(𝑆𝑓∗ (1 − 𝑅)0.79 − 28.32) (4.11)

𝜎𝑚𝑖𝑛
𝑅= (4.12)
𝜎𝑚𝑎𝑥

where 𝑆𝑓∗ = 𝑆𝑓 /6.895 and 𝑅 is the stress ratio. They represent the S-N curve of the materials

presented in chapter 5, which are aluminium 7075-T6, magnesium ZK60A and steel AISI H13,
respectively.

Based on the past experience of FST Lisboa, the Pugsley safety factor model was used to define
the factor of safety, n, for the structural design. A factor of n=1.5 was obtained after applying this method
[30].

4.4 Stiffness targets

All bodies deform under load, either elastically or plastically [5] If their deflection isn’t controlled,
the system in which they are inserted won’t perform as intended.

The suspension geometry is designed with the assumption that all parts suspension system have
infinite stiffness, and therefore no deformation. Santos [4] has shown that the compliance of the

49
suspension system does in fact affect the overall performance of the car, so a stiffness target must
defined. Reaching for the highest possible stiffness is not necessarily the most correct approach, as for
two parts made of the same material and with similar shape, higher stiffness means more weight, which
is detrimental to the performance of the car. A balance must be found, between stiffness and weight,
that provides the best overall result.

In recent competitions, the performance difference between teams has been relatively small,
which means small gains in the performance may be translated into significant improvements in the final
classification order. The most useful dynamic event to observe for this analysis is the skid-pad event,
as the others are more dependent on the skill of the drivers. Table 4-5 shows the skid-pad best results
from Formula Student Spain 2018.

Table 4-5 – Skid-pad results for Formula Student Spain 2018 - Electric

Skid-pad results in Formula Student Spain 2018 - Electric


Place Car num. Team Best time [s] Score
1 E77 DHBW Stuttgart 4,871 75,0
2 E31 TU München 4,895 73,1
3 E40 Eindhoven University of Technology 5,029 62,7
4 E26 Universität Stuttgart 5,070 59,7
5 E14 DHBW Ravensburg 5,096 57,8
6 E13 UAS Munich 5,101 57,5
7 E33 ETH Zurich 5,142 54,6
8 E81 Karlsruhe Institute of Technology 5,178 52,1
9 E8 Chalmers University of Technology 5,180 52,0
10 E146 Politecnico di Torino 5,181 51,9

It is expected, based on the results for the simulations of section 3.2, to have a skid-pad time of
about 4.9s. On that range, to lose a position the car had to be 0.129s slower, which is equivalent to a
lateral acceleration variation of 0.7m/s2, or 5%, as shown in equations (4.13), (4.14) and (4.15). From
the stiffness point of view, a goal was to assure the entire system (including wheel rim and suspension
links) doesn’t have a toe-angle or camber-angle variation greater those 5%, relative to the angle values
that correspond to the maximum lateral performance (±0.25º or camber-angle and ±0.35º for toe-angle).

𝑣2 (2𝜋𝑅𝑐𝑜𝑟𝑛𝑒𝑟 /𝑡)2 4𝜋 2 𝑅𝑐𝑜𝑟𝑛𝑒𝑟


𝑎𝑦 = = = (4.13)
𝑅𝑐𝑜𝑟𝑛𝑒𝑟 𝑅𝑐𝑜𝑟𝑛𝑒𝑟 𝑡2
1 1
Δ𝑎𝑦 = 𝑎𝑦2 − 𝑎𝑦1 = 4𝜋 2 𝑅𝑐𝑜𝑟𝑛𝑒𝑟 ( 2 − 2) (4.14)
𝑡2 𝑡1
Δ𝑎𝑦
𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑣𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛 = × 100% (4.15)
𝑎𝑦1

50
Analysing the stiffness of a transmission system is a complex task. The shafts and bearings that
hold the gears must have small deflections, otherwise the gears won’t mesh properly. It also must have
low distortion (torsion), or there will be a significant delay in the system’s response. During the structural
design in section 5.1.2, a centre distance tolerance of +/-0.0105mm (JS7) is set, to assure the proper
functioning of the transmission. That tolerance serves two purposes, the first is a machining and
assembly tolerance, and the other is a margin for the deflections of the parts. Both purposes have high
importance on service, so the tolerance for machining and assembly was reduced to JS5 (+-0.0045mm)
and the remaining +/-0.006mm is the maximum acceptable variation due to deflection.

51
5 Final solution

Iterative design is a methodology based on a cyclic process with the several consecutive steps
of the design. Based on the results of the last step of each cycle, changes and refinements are made,
and the process is restarted. This type of process is intended to ultimately improve the quality of the
design. Figure 5-1 shows the iterative process applied to the system being developed [31].

Machinability
CAD Fatigue
& FEA
modelling verification
serviceability

Figure 5-1 – Illustration of the iterative process used in the design

In chapter 2 it was shown that the best solution to the system is to have the motor inside the
upright, followed by a two-stage planetary gear train. Concept 4 is the starting point for the iterative
process and is only a representation of that kind of solution, so further development must be performed,
to obtain a feasible design. This chapter will only show the final result of the iterative process.

There are two main considerations to be taken in mechanical design, while trying to achieve the
performance targets: the systems must be designed to be functional, serviceable and easy to
manufacture and assemble. All the components of the system must have enough resistance to withstand
the applied loads during their expected life.

Different manufacturing processes are considered differently during design process. Only
processes that are applicable to small scale production and accessible to the FST Lisboa team are
considered: CNC machining, conventional machining, hobbing and W-EDM. Laser-sintering, or other
3D printing processes, allow the production of highly complex geometries, however, the anisotropic
properties of the material (after production) and the distortion due to the high temperatures while fusing
layers, means that a deeper analysis must be performed to properly use this process.

For these production processes and the overall temperature and loads applied to the system
(section 4.2), only metallic materials are recommended for the components. Aluminium, titanium,
magnesium or steel alloys are the most common materials used in motorsport for similar applications to
the one being studied [32].

52
Table 5-1 – Mechanical and thermal properties of the considered materials, at 20ºC [13]
𝑬 𝝊 𝑺𝒖 𝑺𝒚 𝝆 𝜶𝒕𝒉𝒆𝒓𝒏𝒂𝒍 𝒄𝒑 𝒌
Material
[GPa] [MPa] [MPa] [MPa] [kg/m3] [µm/mK] [J/kgºC] [W/mK]
Magnesium ZK60A 45 0.35 365 305 1830 26 960 120
Aluminium 7075-T6 70 0.33 572 503 2800 24 960 130
Titanium Ti-6Al-2Sn-4Zr-2Mo 114 0.33 1210 1120 4650 8.6 460 7.7
Steel AISI H13 210 0.3 1500 1300 7800 11.5 460 25

These materials were selected according to their high tensile strengths, which may lead to smaller
thicknesses, resulting in lower weights. They are presented in order of their priority, based on the
material density, so of all the materials that have proved suitable for the part, then the one with the
highest priority is used. Gear material selection is more complex, so its definition is presented with more
detail in section 5.1.1.

Structural design was made with finite element analyses. The geometries analysed with this
method are divided into interconnected elements of finite dimensions (mesh), with a certain shape, to
which algebraic equations are applied. The analyses presented in this document were made in
Hypermesh, a commercial FEA software, using the Optistruct solver.

The first parts to be designed are the ones of the gear train, because the dimensions of all the
other parts of the system surrounding the gear train will be defined by its overall diameter. After that, the
wheel hub, main bearings and oil retention are designed, followed by the upright and brake system.

5.1 Transmission design

Designing the meshing between gears is a complex task, even when aided by the rules set in
international standards. The stress distribution in the contact surfaces of the gears could be estimated
with the Hertzian contact equations, but that is extremely time consuming and complex. Standard ISO-
6336 takes the Hertzian contact into consideration, so it was chosen for the design of all gears. The
amount of interlinked equations and constants presented by the standard is very high, so a commercial
software was used, KISSsoft, that uses this standard in combination with others that have information
which is not included in ISO-6336 [33]. The design of the gears must meet the targets presented in
chapter 4.

5.1.1 Gear material selection

Material selection is a critical factor in the design of any component. In gears, there is a complex
stress distribution in the teeth of the gears, due to their shape and the variable contact between the

53
mating surfaces. For this application, the recommended material for the gears is a wrought alloy steel,
which typically have high fatigue resistance. Nickel and molybdenum should be some of the alloy
elements, as they increase the impact resistance and control the cooling rate while quenching. Surface
hardness must be high to endure the constant sliding between teeth, yet the core must be softer than
the surface, otherwise the teeth would be too brittle and prone to rapid crack propagation, leading to
premature failure. The recommended hardness values for the surface and core are 700HV and 300HV
respectively [34].

Reaching these hardness values is only possible through separate heat treatments for the core
and surface. For the core, a regular quenching and tempering is recommended by the material
manufacturer [35]. For the surface, the most typical surface treatments to achieve those levels of
hardness, are case-hardening or nitriding. Given the small size of the gears in general and he teeth in
particular, great care must be taken during heat treatment, to avoid distortion in the gears that may
interfere with proper gear meshing. This leads to the selection of nitriding, for the surface treatment
process, as the temperature during the nitriding process is below the tempering temperature of the steel,
whereas case-hardening requires a temperature above the one of tempering. Nitriding is a case-
hardening process in which the hardening agents are nitrides formed in the surface layer of the steel,
through the absorption of nitrogen from a nitrogenous medium, usually dissociated ammonia gas. To
assure the nitriding process is made adequately, it’s recommended the alloy has, alongside with nickel
and molybdenum already presented, aluminium, chromium and vanadium [34].

Other materials could be used for the manufacturing of gears, such as cast-iron, which absorbs
vibrations, making the gears more silent under operation, or sintering steel, which is useful for complex
geometries. Those alternative materials have typically less resistance and hardness, making them
suitable for applications with low applied loads, which is not the case being studied [34].

With these material requirements, a specific alloy may be chosen. The actual FST Lisboa sponsor
that manufactures gears frequently uses 34CrNiMo6, which complies with the presented requirements.
Given the availability of this material, it was selected for the gears that are being designed.

5.1.2 Gear design

Standards ISO-6336-2 and ISO-6336-3 present equations to calculate the bending stress, 𝜎𝐹 ,
and contact stress, 𝜎𝐻 , (pitting) on the teeth [36][37].

𝜎𝐹 = 𝜎𝐹𝑂 𝐾𝐴 𝐾𝑉 𝐾𝐹𝛽 𝐾𝐹𝛼 (5.1)

Bending 𝐹𝑡 (5.2)
𝜎𝐹𝑂 = 𝑌 𝑌 𝑌 𝑌 𝑌
𝑏𝐺 𝑚𝑛 𝐹 𝑆 𝛽 𝜖 𝐵

54
𝜎𝐹𝑙𝑖𝑚 𝑌𝑆𝑇 𝑌𝑁𝑇 (5.3)
𝜎𝐹𝑃 = 𝑌𝛿𝑟𝑒𝑙𝑇 𝑌𝑅𝑟𝑒𝑙𝑇 𝑌𝑋
𝑆𝐹𝑚𝑖𝑛

(5.4)
𝜎𝐻 = 𝑍𝐵 𝜎𝐻𝑂 √ 𝐾𝐴 𝐾𝑉 𝐾𝐻𝛽 𝐾𝐻𝛼

Contact (5.5)
𝐹𝑡 𝑚𝐺 + 1
𝜎𝐹𝑂 = 𝑍𝐸 𝑍𝐻 𝑍𝛽 𝑍𝜖 𝑍𝐵 √ 𝐺 𝐺
𝑏 𝑑 𝑚𝐺

𝜎𝐻𝑙𝑖𝑚 𝑍𝑁𝑇 (5.6)


𝜎𝐹𝑃 = 𝑍𝐿 𝑍𝑉 𝑍𝑅 𝑍𝑊 𝑍𝑋
𝑆𝐻𝑚𝑖𝑛

The meaning of each factor is presented in Appendix F. ISO-6336 shows how to calculate most
of these factors. However some of them, referring to characteristics of the geometry of the gear and the
material heat treatments, must be calculated with the help of other standards, also presented in
Appendix F.

Spur gears were selected for the transmission system being presented, because the meshing
between spur gears doesn’t create an axial load on the supporting shaft. With a smaller gear module, a
higher efficiency and lower weight can be achieved, so the smallest standard module, m=1, is used for
the gears of this system. With this module, using equation (5.7), it is possible to see that if the number
of teeth, 𝑧 𝐺 , of a certain gear (with involute teeth and a pressure angle of 20º) is smaller than 17,
undercut will occur, which will have a negative impact on the resistance of the gear [38].

𝑚𝑧 𝐺
𝑚≤ 𝑠𝑖𝑛2 𝜙
2
(5.7)
2 2
𝑧𝐺 ≥ 2
⇒ 𝑧𝐺 ≥ ≈ 17
𝑠𝑖𝑛 𝜙 𝑠𝑖𝑛2 20

To assure the proper meshing between teeth, as well as increase their resistance, even with
gears with low number of teeth, it’s usual to modify the shape of the teeth, using a profile shift coefficient.
The tooth involute is constructed as presented in Figure 5-2, having as reference the base circle of the
gear. A profile shift is obtained by changing the position of the cutting tool relative to the material blank
while manufacturing, which will affect the overall shape of the teeth. Figure 5-3 shows the difference
between a tooth with positive or negative shift, and the “nominal” profile. DIN 3992 [39] shows how to
define the adequate of profile shift values, for a given size of gear and specific application.

55
Involute curve

Figure 5-2 – Illustration of how the involute is constructed [34]

Negative shift Positive shift


(Tooth becomes thinner) (Tooth becomes thicker)
Reference circle
Reference circle

Shifted gear

Standard gear
Shifted gear

Standard gear

Profile shift
Profile shift
Tooth depth Tooth depth

Figure 5-3 - Comparison between “nominal” tooth profile with negative shifted (left), and positive
shifted (right) profiles [38]

The material information presented in section 5.1.1 is inserted into the software, which calculates
the stresses and factors of safety according to the ISO 6336 standard. The first stage has a transmission
ratio of 𝑖1 = 4.313, so the reference torque and speed for the second stage can be calculated with

56
equations (5.8) and (5.9). After this, the iterative process is repeated for the second stage. The results
for both stages are presented in Table 5-2.

T𝑟𝑒𝑓2 = 𝑇𝑟𝑒𝑓1 × 𝑖1 (5.8)

𝑁𝑟𝑒𝑓1
N𝑟𝑒𝑓2 =
𝑖1 (5.9)

Table 5-2 – Results obtained for the second stage of the transmission

First stage Second stage

Sun Planet Ring Sun Planet Ring

Number of teeth 16 18 53 19 23 65

Face width [mm] 15 15 15 30 30 30

Profile shift coefficient


0.577 0.164 -0.325 0.675 -0.025 -0.625
[mm]

Tip diameter [mm] 18.984 20.158 51.65 22.234 24.834 64.25

Number of gears 1 3 1 1 4 1

Root safety 2.46 2.86 4.01 2.57 2.24 3.38

Flank safety 1.17 1.22 2.05 1.19 1.24 2.56

Centre distance [mm] 17.656 - 21.592 -

Module [mm] 1

Pressure angle [º] 20

Helix angle [º] 0

Transmission ratio 4.313 3.421

Efficiency [%] 98.2 97.4

Thickness of the
0.05 0.05 0.05 0.08 0.08 0.08
hardened layer [mm]

57
5.1.3 Lubrication

Lubrication is of great relevance in the resistance of the gears and therefore must be considered
in their design. Work on elastohydrodynamic lubrication (EHD) showed that, in gears and rolling
elements of bearings, a very thin oil film is often developed between the two contact surfaces, that tends
to separate them, so there is minimal or none metallic contact. There are three distinct regimens of
lubrication:

• Regimen I: No appreciable EHD oil film (boundary)


• Regimen II: Partial EHD oil film (mixed)
• Regimen III: Full EHD oil film (full film)

In regimen I, the developed oil film has a thickness comparable to the surface roughness of the
gear. There will be full metal-to-metal contact in the Hertzian contact area. This regimen is typical of
slow-speed, high-load gears. In regimen II, which is where most vehicle gears operate, there is only
partial metal-to-metal contact, in the asperities of the contact surfaces, but there are substantial areas
separated by the film. It’s typical of medium-speed, high-load gears, with relatively thick (thicker than
regimen I, but thinner than regimen III) oil and high surface finish. Finally, in regimen III the film is thick
enough to completely separate the contact surfaces, even the asperities miss each other. Well-
designed, well-built, high-speed gears generally operate in this regimen. The oil film depends on the
operation characteristics and can be calculated using equations (5.10) through (5.16). They are
presented in imperial units, so the result comes in μin, but simply dividing it by 39.37, changes the result
to μm [34].

44.6 𝑟𝑒 (𝑙𝑢𝑏𝑟𝑖𝑐𝑎𝑛𝑡 𝑓𝑎𝑐𝑡𝑜𝑟) ∙ (𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑓𝑎𝑐𝑡𝑜𝑟)


ℎ𝑚𝑖𝑛 = (5.10)
(𝑙𝑜𝑎𝑑𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟)
𝐶𝑠𝑖𝑛𝜙𝑛 𝑚𝐺
𝑟𝑒 = ∙ (5.11)
cos 𝛽 (𝑚𝐺 + 1)2
2 𝐺

𝑙𝑢𝑏𝑟𝑖𝑐𝑎𝑛𝑡 𝑓𝑎𝑐𝑡𝑜𝑟 = (𝜆𝐸′)0.54 (5.12)

𝜋 𝐸
𝐸′ = ∙ (5.13)
2 (1 − 𝜈)2
𝜇0 𝑢 0.7
𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑓𝑎𝑐𝑡𝑜𝑟 = ( ) (5.14)
𝐸𝑟𝑒
𝜋 𝑛𝑝 𝑑 𝐺 𝑠𝑖𝑛𝜙𝑡
𝑢= (5.15)
60
𝑊𝑡 0.13
𝑙𝑜𝑎𝑑𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟 = ( ) (5.16)
𝐹 𝐸 𝑟𝑒

Where 𝑟𝑒 is the effective radius of curvature at pitch diameter (in), 𝐶 is the centre distance (in), 𝜙𝑛
is the normal pressure angle, 𝜓 is the helix angle, 𝑚𝐺 is the ratio between gear and pinion teeth, 𝜆 is
the lubricant pressure-velocity coefficient (in2/lb), 𝐸′ is the effective elastic modulus (psi), 𝜇0 is the

58
lubricant viscosity at operating temperature (cP), 𝑢 is the rolling velocity at pitch line (in/s), 𝑛𝑝 is the
pinion speed (rpm), 𝑑 𝐺 is the pinion pitch diameter (in), 𝜙𝑡 is the transverse pressure angle, 𝑊𝑡 is the
tangential load (lb) and 𝐹 is the facewidth (in).

Regimen I

Regimen II

Regimen III

Figure 5-4 - Typical regimen determination based on surface finish and EHD oil-film
thickness[34]

Figure 5-4 shows the relation between surface finish and EHD film. For the case being studied,
the film thickness will be approximately 0.4μm and the desired surface roughness is Ra=2μm, which
means the transmission system will be operating in regimen II.

There is a substantial difference in the tangential speeds of each gear of the system, with the
input gear having a maximum tangential velocity of 16.75 m/s and the output gear 4.6m/s, and average
tangential speed of 7.8m/s and 2.1m/s. Given these speeds, equation (5.17) from the ASM Handbook
[40] can be used to estimate the ideal oil viscosity, at 40ºC. Because of the difference in tangential
speeds from the input to the output of the system, it’s hard to find the ideal viscosity for the lubricant.
Based on the typical operating speeds, lubrication regimen and commercially available lubricants, an oil
with a kinematic viscosity centre value of 320×10-6 m2 /s, at 40ºC, (ISO VG 320) is used as a lubricant.

7000
𝜈40º𝐶 = 0.5 × 10−6
𝑛𝑝𝜋𝑑𝐺 (5.17)
(0.3048)

where 𝑛𝑝 is the rotation speed of each gear in rpm and 𝑑 𝐺 is the pitch diameter.

59
5.1.4 Shafts

For reasons related to manufacturing and assembly, the gears are mounted on shafts, with needle
roller bearings. These shafts hold the gears in their correct place and must withstand the radial and
tangential loads, originated on the gears contact surfaces. A high surface resistance is necessary due
to the constant rolling of the bearing elements on the shaft, so the material selected for the shafts is the
same as for the gears, as well as the heat and surface treatments. Since the shafts and bearings are
likely to be produced by the same company, having the same material and treatments facilitates in the
manufacturing process.

The mechanical design of the shafts is quite complex. Although they are fixed on both ends and
the main loads applied are the radial and tangential, originated in the gears, they cannot be designed
independently from the rest of the system, as they contribute for the overall stiffness of the system. They
are designed together with the remaining components of the system using FEA.

5.1.5 Connection between stages

Power flows from the carrier of the first stage to the sun of the second stage. Typical shaft
connections, for torque transmission, include either having a key/keyway-like combination, or
connections without interconnecting elements, like splines or polygon curves.

Spline connections are the most common way of transmitting torque while allowing axial
movement of the connected components. Due to its geometry, there are significant stress concentration
areas and the contact between the “teeth” of the mating splines may not be uniform, when multiaxial
loads are applied, leading to high local contact stresses. Polygonal connections have been around for
many years but were unpopular due to their complexity in both their stress analyses and difficulty in
manufacturing by traditional methods. With the appearance of finite element analysis and CNC
machining, as well as the increase in quality of manufacturing technologies, the use of these profiles
started increasing [41].

The polygon shape can be described mathematically by an epitrochoidal curve with a certain
periodicity. German standards DIN 32711 [42] and DIN 32712 [43] are the only existing standards
relative to polygon profiles and describe a 3-lobe (P3G, a curve with periodicity 3) and 4-lobe (P4C, a
curve with periodicity 4) profiles, respectively, as shown in Figure 5-5. Other periodicities are
mathematically possible but not standard.

60
Figure 5-5 – Cross-section of P3G-profile (left) and P4C-profile (right) [44]

As seen in Figure 5-5, the reference dimensions for the profiles are: the mean diameter d1, the
outside diameter d2, the inside diameter d3 and the eccentricity e. P3G-profiles are harmonic, but P4C-
profiles are disharmonic, because of the discontinuities in its circumference. These discontinuities lead
to a lower resistance to fretting fatigue compared to P3G-profile, which makes it less suitable for high
torque applications. The continuous shape of P3G means there is a larger contact area between mating
parts, which leads to lower contact stresses compared to P4C, key/keyway or spline shafts [41][44].

The DIN standards present equations for calculating the resistance of the profile and aid in the
selection of the standard size. Since the components of the transmission that use this connection are
part of the main structure of the system and contribute to its stiffness, the size must selected based on
the results obtained in FEA analysis made to the entire system in sections 5.1.7 and 5.5.

5.1.6 Bearings

As seen in Figure 5-6, there are five different bearings. Some of them are repeated, making a
total of fourteen bearings in the transmission system: twelve K8x12x10_TN needle roller bearings
(marked as B5), one 61815_95x75x10 ball bearing (marked as B1), one 61812_78x60x1 ball bearing
(marked as B2), one W61705_32x25x4 ball bearing (marked as B3) and one 61804_32x20x7 ball
bearing (marked as B4) [10].

61
The needle-roller bearings were selected due to their low thickness and high efficiency, compared
to steel or plastic bushings. The outside diameter of those bearings was defined so there would be no
impact on the resistance of the gear. Naturally, different gear sizes could use bearing with different
sizes, but to simplify the process of acquiring the bearings, all needle roller bearings have the same
size.

Brake disc Brake calliper

Hub
Oil seal

B2

B1
B4

B3
Oil lid

B5
Transmission
1st stage
O-ring

Transmission
Middle support
2nd stage

Figure 5-6 – Transmission system of Concept 4

The length and internal diameter of the roller bearings, as well as the sizes of both ball bearings,
were selected based on the available sizes from the bearing supplier, as well as the results from FEA.

5.1.7 Transmission FEA

The gears were sized based on the equations from international standards, but the same
approach cannot be taken for all other components (shafts and supports), as they have complex shapes
and multiaxial loads applied to them. This makes their design almost impossible with simple stress
calculations and require the aid of computational methods.

62
To reduce the computational time and accelerate the iterative process, the number of elements
of the finite element mesh of the assembly cannot be too high. This can be done by dividing the system
into three parts: the transmission, which is presented in this section; the hub, presented in section 5.2
and the upright, which is presented in sections 5.3. Also, only load cases shown in Table 4-2 and Table
4-3 are considered in this analysis.

All components that make the transmission were inserted into Hypermesh, they were meshed,
the contact between parts was defined, the material properties were set, and the loads were applied.
Appendix G shows this workflow with more detail. There are external loads, originated in the contact
patch, that are also applied to the transmission, so a dummy hub, as well as the connecting bearings,
were included in the model, to accurately represent the load distribution (Figure 5-7). These dummies
have the same stiffness as the parts they represent, yet their shape is simplified to reduce the overall
number of mesh elements and reduce computational time. The meshing between gear teeth shouldn’t
contribute to the overall stiffness and resistance of the system, as that would mean a significant increase
in the wear of the teeth. For that reason, the gears considered in the study do not have any contact
between them.

The mesh used in the parts is made of tetrahedral PSOLID elements, with different element sizes
depending on the component and the respective details. The element shape and size were defined with
a mesh convergence analysis, assuring a good mesh quality.

The geometry of the middle support, which represents the carrier of the second stage, is
constrained by the sizes of surrounding components (as seen in Figure 5-6). For that reason, in the
iterative process, the only modification possible to do in this part is to change its material.

Figure 5-7 – Transmission FEA model

Figure 5-8 shows the deformation and stress results of the analysis. From this analysis it seems
that the aluminium alloy of Table 5-1 has enough stiffness and strength to be used on the middle support.
Despite this, the material selection will only be definitive after the analyses presented in section 5.5.

63
Figure 5-8 – Analysis results: displacement magnitude [m] (left) and Von-Mises stress [Pa] (right),
for a load case in sector O1 of Figure 4-1

The maximum displacement occurs in the shafts of the second stage, which at first glance seems
larger than the defined maximum deformation presented in section 4.4, yet the value shown doesn’t
represent the true variation of centre distance. The centre distance variation, Δ𝑐, can be computed
based on the simulation results with equation (5.18).

Δ𝐶 = √(𝐶𝑜 + 𝑑𝑟 )2 + 𝑑𝑡2 − 𝐶𝑜 (5.18)

where 𝑐𝑜 is the distance from the centre to the sun to the centre of each shaft, in the undeformed state,
𝑑𝑟 is displacement in the radial direction and 𝑑𝑡 is displacement in the tangential direction. For this
analysis, the variation of centre distance is ΔC=0.001275mm, which is well within the defined
displacement threshold.

5.2 Hub (transmission housing)

The wheel hub is a rotating part that is connected to the wheel rim on one side, and on the other
is supported by bearings that constrain all movements, except for the rotation around its axis.

In this system, the hub is also the transmission output and enclosure. Because of this, the internal
dimensions of the hub are all defined by the bearing and gear sizes. On the outside however, the only
constrains are the mounting position of the wheel rim and the brake disk, and the clearance for the brake
calliper and for the assembly/disassembly of the rim.

The operating temperature of this part, as presented in section 4.2, is 80ºC. This low temperature
has minimal effect on the mechanical properties of the materials and low impact on geometrical
distortions due to thermal expansion. All materials shown in Table 5-1 were considered in the analyses.

64
Performing iterations on the shape of the hub is a time-consuming process, so to find the overall
ideal shape of the part, a topology optimization was made using Optistruct. Topology optimization is a
mathematical method that optimizes the material layout within a certain design space, for a given set of
loads and constraints. Optistruct solves the optimization problem using the density method (also known
as the SIMP method), where density of each element should take the value of either 0 or 1, defining the
element as a void or solid, respectively [45].

To perform a topology optimization, a design space must be defined, according to the geometrical
constraints already presented, shown in Figure 5-10. In Hypermesh, the typical FEA workflow is applied,
but before running the analysis, the design objectives and constraints must be defined. Like with the
FEA analysis to the transmission components, a simplified model was used, with a dummy rim,
transmission and upright, with equivalent stiffness’s, to assure the loads distributions are correct (Figure
5-10).

Figure 5-10 – Design space of the hub

Figure 5-10 – FEA model of the hub, used in the topology optimization (Part of the rim is removed
to show the position of the design space relative to the rim)

The hub experiences all the load cases specified in section 4.2, so fatigue must be considered in
the analyses. Using the method presented in section 4.3, together with the topology optimization is
possible, but extremely time-consuming, so that method is only suitable as a verification analysis, when

65
the geometry and material of part have been defined. Like in section 5.1.7, only load cases from Table
4-2 and Table 4-3 are considered. To take fatigue resistance into account in the topology optimization,
the modified Goodman fatigue criteria shown in equation (4.3) was rewritten in function of maximum
stress (together with equations (4.7) and (4.8)), an approximate stress ratio was set based on the
maximum and minimum applied loads, and an approximate endurance limit, 𝑆𝑒 , was set for each
material, calculated with the Marin equation [5].

2𝑆𝑒 𝑆𝑢
𝜎𝑚𝑎𝑥 = (5.19)
𝑛[(1 − 𝑅)𝑆𝑢 + (1 + 𝑅)𝑆𝑒 ]

The hub is a rotating part, but because the most significant loads are originated in the contact
patch of the tire, it’s hard to define an absolute interval of maximum allowable displacements. For that
reason, the optimization objective is to minimize compliance, which is equivalent to maximizing stiffness,
using just a fraction of the initial design space volume, depending on the used material. This volume
fraction, together with the maximum allowable stresses shown in Table 5-3, are the optimization
constraints.

Table 5-3 – Endurance limit and maximum allowable stress for each material
Maximum
Material 𝑹 𝒏 𝑺𝒖 [MPa] 𝑺𝒆 [MPa] 𝝈𝒎𝒂𝒙 [MPa]
volume fraction
Magnesium ZK60A 365 110 75 0.4
Aluminium 7075-T6 572 172 115 0.3
-1 1.5
Titanium Ti-6Al-2Sn-4Zr-2Mo 1210 363 240 0.2
Steel MG50 1500 450 300 0.1

The optimization results for aluminium, titanium and magnesium are shown in Figure 5-11. They
are all very similar in shape, so with the corrections needed to obtain a machinable solution, the final
shape and dimensions of the part are also very similar for all materials. This leads to the selection of
ZK60A magnesium alloy for the part, which is the one with the lowest density resulting in a lightest part,
but this selection is only definitive after checking its suitability in the analysis of section 5.5.

66
Figure 5-11 –Optimization results for aluminium (left), magnesium (middle) and titanium (right)

Figure 5-6 shows an oil seal. This is used to retain oil inside the hub, while allowing its rotation.

Figure 5-12 – Final design of the hub

On the opposite side, there is a lid with an o-ring (72.75x1.78) to prevent oil from leaking on that
side of the hub. To preform simple oil changes, two drain plugs are mounted on the hub, so that when
oil is flowing through one, air can flow through the other.

Under operation, there is always some wear of the gears teeth, and the small debris released are
cleaned off the teeth contact surfaces by the oil. Those debris will move around inside the hub, mixed
with the oil. Since the oil is also lubricating the bearings, the drain plugs have a magnetic tip to attract
those metallic debris and prevent them from disturbing the proper functioning of all components.

67
5.3 Upright

The upright is the part that holds the hub, and consequently the wheel. It only allows the wheel to
rotate, but through the suspension links, it controls the wheel movement in all other directions.

In Concept 4, the upright has the motor mounted inside it, with the motor cooling channels
integrated in it. Connected to the front of the upright, there is the ring of the first stage of the transmission
and the middle support, as well as the bearing B1 and the oil seal, from Figure 5-6. It also holds the
brake calliper. The ideal shape of the cooling channels for this motor in this application was studied by
Fontes [46]. So, the shape he obtained was created inside the upright, as seen in Figure 5-13 (blue
component inside the design spaces).

A similar approach to the hub is taken while designing the upright. A design space was defined
as shown in Figure 5-13, allowing space for the calliper, the motor and clearance for the suspension
links. The FEA model also includes a dummy rim, hub and transmission, with equivalent stiffness’s, to
guarantee the correct load distribution through the model. Only load cases from Table 4-2 and Table
4-3 are considered in this analysis.

Figure 5-13 – Design space for the upright of the front wheels (left, red) and the rear wheels (right,
white)

The position of the wheel relative to the suspension links, is different for each load case, as there
is wheel travel and steering inputs. The FEA model includes the suspension links in their correct position
relative to the wheel, for each load case, to assure the mesh constraints accurately represent what
happens in reality. These links have a stiffness equivalent to the ones currently used by FST Lisboa,
with an equivalent Young Modulus Elink=150GPa.

68
Figure 5-15 – FEA model of the front upright, used in the topology optimization (Part of the rim is
removed to show the position of the design space relative to the rim)

Figure 5-14 – FEA model of the rear upright, used in the topology optimization (Part of the rim is
removed to show the position of the design space relative to the rim)

The allowable displacements presented in section 4.4 and the maximum allowable stresses
shown in Table 5-3 were set as the constraints for the topology optimization, and the objective was set
to minimize volume fraction, which is equivalent to minimize the mass.

The magnesium alloy was not considered for the upright, because its strength is insufficient to
this application. All other materials proved to be suitable. However, because there was a displacement
constraint, all solutions have similar stiffnesses, which means aluminium emerges as the best option,
as it is the one with the lowest weight, although this selection is only definitive after checking it in the
analysis of section 5.5.

The optimal result obtained in the simulation has a very complex shape, so it must be adapted
into a solution that can be machined by any of the processes presented in the beginning of this chapter.

69
Figure 5-17 – Optimization results for the front upright (top) and rear upright (bottom)

Figure 5-16 – Final design of the front upright

70
Figure 5-18 – Final design of the rear upright

5.4 Brake system

FST Lisboa uses a hydraulic brake system in the car that actuates on all four wheels
simultaneously, with two brake lines, one to the front wheels and another to the rear wheels. The
actuation is made through pads on a floating disc, with the pads mounted on a brake calliper and pushed
against the rotating disk, with the pressure from the hydraulic system. The disc rotates together with the
wheel hub.

Ferro [47] showed that ABS (anti-lock braking system) increases the performance of the car, but
since brake-by-wire is not allowed, it is hard to make and effective ABS. To allow the future teams from
FST Lisboa to try and implement this kind of system, all wheels have a two-piston calliper from AP
Racing (CP4227-2S0), each actuating a single brake pad. This way, one pad can be actuated by the
hydraulic system, complying with the competition rules, and the other actuated by the ABS, which would
be considered as a driver aid and not as brake-by-wire.

Figure 5-19 – Model of the brake disc, with the design space for the structural part in blue

71
To design the brake disk, it was separated into two parts (as seen in Figure 5-20): the one where
heat is generated (contact with the pads), and the one that connects to the hub. The first was designed
to have a large surface area of the disc, increasing the amount of energy dissipated through convection,
but without compromising the structural integrity, using holes and slots. The slots are important because,
under braking, the heat generated can be so great that gases are released from the pads into the disk-
pad interface (as a result of the decomposition of the resin in the pads and water-vapor), resulting in a
phenomenon called brake fading, which is translated into loss of braking performance. The slots are an
effective way of removing those gases from the interface.

That part of the disk that connects to the hub was designed in a similar way to the hub and upright.
A design space was created, followed by a topology optimization using FEA, which resulted in the shape
seen in Figure 5-20. To assure each disk can be mounted on any wheel, they all have the same shape
and thickness.

Figure 5-20 – Topology optimization result (left) and final design of the brake disc (right)

The 3.5mm thick disk is connected to the hub through pins, whose diameters are restricted by
possible interferences with the calliper and hub. With that size restriction, only the number of pins and
their material can be iterated. The result is 5 pins made of the same steel as the disc.

5.5 Full system FEA analysis

The system was divided into three parts, as presented in section 5.1.7, to reduce the duration of
the iterative process of selecting the shape and the materials to be used. With those parts defined, an
analysis to the entire system must be performed to confirm that the design constraints of displacement
and stress are met. The FEA model for the front wheel is presented in Figure 5-21. A similar model was
created for the rear wheel. Figure 5-22 shows the displacement plot of the FEA model of the front wheel,
for a load case in sector O12 of Figure 4-1. As in the FEA of the upright, the model includes the
suspension links in their correct position relative to the wheel, for each load case, to assure the mesh
constraints accurately represent what happens in reality.

72
Figure 5-21 – Full-system FEA model (front wheel)

Figure 5-22 – Displacement results for the front wheel full-system FEA model, for the first point
of the load spectrum (pure acceleration)

The maximum displacement experienced by the system was observed in the front wheels. It
occurs in load cases from zone O12 of Figure 4-1 (braking and cornering), with 0.114mm in the
longitudinal direction, and 1.06mm in the lateral direction, measured in the contact patch, which
correspond to 0.266º around the longitudinal axis (camber variation) and 0.021º around the lateral axis
(toe variation). The camber variation is slightly above the limit defined in section 4.4. Despite this, the
presented solution is still valid because is easier to increase the stiffness of the suspension links. Toe
variation is within the limits defined in section 4.4. As described in section 4.3, three load cases from
each sector of Figure 4-1 were applied to the full model of each wheel, with a total of 192 load cases.
For each sector, those three load cases are used to find the constants in equation (4.2), that will then
be used to estimate the stress value of each critical point, of each part, for all load cases in the respective
sector. After using this approach to all sectors, it is possible to estimate all stress values for all critical
points for the entire load spectrum, which allows the usage of the fatigue criterion presented in section
4.3.

73
Table 5-4 – Accumulated damage for the main components of the system

Middle Planet shaft


Upright Hub Brake disc
support (2nd stage)

Front
0 10.314 7.461 1.357 0
wheel
Rear
0 12.498 10.995 1.682 0
wheel

Table 5-4 shows the accumulated damage estimations in the most critical location of each of the
main component. It’s possible to see that neither the middle support, made with aluminium 7075-T6, nor
the hub, constructed of magnesium ZK60A, nor the planets shafts of the second stage, made of steel,
can withstand the entire service life, so one of two approaches should be taken: either produce several
spare parts and replace them periodically (about 85 km for the hub and 50 km for the middle support
and 560 km for the shafts), or simply manufacture the components in different materials. Given the small
weight difference from aluminium to steel for the middle support (0.2kg) and magnesium to aluminium
for the hub (0.17kg), the best solution is to build the middle support in steel AISI H13, and the hub in
aluminium 7075-T6. Titanium also proved unsuitable for the middle support. Even though the shafts
can´t endure the entire life of the car, their simple geometry allows for the easy manufacturing of several
spares. The damage estimations based on the re-analysis with the new materials are presented in Table
5-5.

Table 5-5 – Accumulated damage with updated materials

Middle Planet shaft


Upright Hub Brake disc
support (2nd stage)

Front
0 0.094 0.287 1.357 0
wheel
Rear
0 0.118 0.541 1.682 0
wheel

74
6 Conclusions

6.1 Summary and conclusions

The initial brainstorm process revealed 8 possible concepts for the system. The characteristics of
each concept were inserted into a self-developed transient simulator, to find which of those 8 concepts
is the best. The model was validated by comparing the simulated results with those acquired in testing
of FST08e, the last car of FST Lisboa. The performance difference among concepts was small, 3% on
average. Despite this, Concept 4 showed the best performance in all situations except for aggressive
corner-entry scenarios, which represent a small fraction of the length of a lap.

Mechanical design was divided into three parts: functional design, structural design and
manufacturing design. They are all part of the iterative process of Figure 5-1. Starting with Concept 4,
the system was improved to take into consideration the machinability and assembly of all components.
The loads applied to the system and the expected temperatures, estimated in section 3.2, were used to
assure the resistance and stiffness of the system, within the performance restrictions presented in
chapter 4.

Five different metallic alloys were considered for all components, all suitable for the manufacturing
processes presented in the beginning of chapter 5. Except for gears, FEA analyses were used to find
the best material for each component.

Gears were designed with KISSsoft, a commercial software that uses equations and information
from international standards to design gear trains. The selected steel for the gears was 34CrNiMo6 , as
it has the ideal alloy elements for the application and is available to the team sponsor company that
produces the gears.

Structural analyses were performed using FEA with commercial software Hypermesh. The
system was divided into four parts: transmission, hub, upright and brakes, to simplify the iterative design
process. Topology optimization of the hub, upright and brake disc were performed using Optistruct
(solver of Hypermesh), and its results were adapted into reasonable machine shapes. With all four parts
designed, the entire system was studied with FEA under the loads spectrum, to help estimating the
stress spectrum to evaluate the fatigue resistance.

The results from the full-system FEA were used to estimate the full stress spectrum of each
component. The Rainflow method was applied to the resulting stress spectrum, obtaining the number
of cycles and the respective stresses. Palmgren-Miner method (with equations (4.3) through (4.8)) was
used to evaluate the damage taken by each component and verify the fatigue resistance for their entire
life. This analysis showed that the materials initially selected for the middle support and the hub, were
unsuitable for their specific applications, and the shafts that hold the planets of the second stage must
be replaced before the car makes 560km.

75
6.2 Future work

Except for the brake disc, the temperatures of each component under operation are low and have
small impact on the resistance of the respective materials and on distortions due to thermal expansion.
Despite this, because of the variable temperatures experienced by each component, performing a
transient thermal-structural analysis could return interesting results in terms of stress variation through
time, due to the different thermal expansions of interconnecting components.

The considered manufacturing processes are simple and easily accessible, but they are
restrictive in the amount of shapes that can be obtained. With other alternative processes, like laser
sintering or 3D-printing, high complex shapes can be achieved. The mechanical design associated with
these kinds of processes is different, so in the future, the work presented in chapter 5 should be
repeated, with alternative manufacturing processes taken into account. For the moment FST Lisboa
cannot afford such processes.

Currently FST Lisboa uses commercially available brake callipers and pads, but they could be
changed with ones developed to fit the specific needs a formula student car. Designing a calliper
integrated with the upright can result in a smaller and lighter solution, which would improve packaging
and performance. The brake disc is a rotating component in the un-sprung part of the car, with significant
inertia around its rotating axis. Reducing its weight is improve performance, but the high temperatures
experienced by it, together with the braking loads, restrict the possible materials and shapes of the disc.
Development of discs with alternative materials, such as Cf/C-SiC [48] or other composites, can be a
suitable way of decreasing the weight without compromising the structural integrity of the disc, keeping
the brake power of the vehicle.

The solution presented in this document was developed for the next car produced by FST Lisboa,
FST09e. This means the loads experienced by each component are specific for the characteristics of
that car. Changing the motor, transmission ratio, tires, suspension geometry, mass, CoG height,
aerodynamic coefficients, wheelbase or track mean the presented work should be reanalysed. Despite
this, following the same procedures as presented in this document it will result in an equivalent and
acceptable solution, that takes the made modifications into account.

76
7 References

[1] “Formula E Transmission Evolution | Hewland Transmissions.” [Online]. Available:


https://www.hewland.com/formula-e-transmission-evolution/. [Accessed: 31-May-2019].

[2] L. de Novellis, A. Sorniotti, P. Gruber, L. Shead, V. Ivanov, and K. Hoepping, “Torque vectoring
for electric vehicles with individually controlled motors: State-of-the-art and future
developments,” World Electr. Veh. J., vol. 5, no. 2, pp. 617–628, 2012.

[3] D. Schramm, M. Hiller, and R. Bardini, Vehicle Dynamics: Modelling and Simulation. Springer,
2014.

[4] R. Santos, “Model and Dynamic Simulation Program for Vehicle Analysis Accounting Suspension
Compliance,” Instituto Superior técnico, Tese de Mestrado, 2014.

[5] J. E. Shigley, C. R. Mischke, and R. G. Budynas, Shigley’s Mechanical Engineering Design - 9th
Ed., vol. New York,. 2002.

[6] B. Jacobson, “Vehicle Dynamics Compendium for course MMF062 Preface 2016,” Mats
Jonasson Artem Kusachov, no. 2016, pp. 134–137, 2016.

[7] J. N. Reddy, An introduction to the finite element method. McGraw-Hill, 1993.

[8] U. Krupp, Fatigue crack propagation in metals and alloys, vol. 10, no. 7–8. 2007.

[9] N. E. Dowling, Mechanical behaviour of materials, Fourth Edi. Pearson, 2013.

[10] SKF Group, “SKF Rolling Bearings Catalogue.” SKF Group, 2013.

[11] E. M. Almas, “ORGÃOS DE MÁQUINAS - Volume I - Engrenagens, Trens de Engrenagens.”


Secção de Folhas - IST.

[12] G. Genta and A. Genta, Road Vehicle Dynamics, vol. 88. WORLD SCIENTIFIC, 2017.

[13] D. of D. USA, “Military Hanbook: Handbook Metallic Materials and Elements for Aerospace
Vehicle Structures,” October, no. January, pp. 8–106, 8–109, 2003.

[14] C. Canudas de Wit and P. Tsiotras, “Dynamic tire friction models for vehicle traction control,” in
Proceedings of the 38th IEEE Conference on Decision and Control (Cat. No.99CH36304), vol.
4, pp. 3746–3751.

[15] H. B. Pacejka and I. Besselink, Tire and vehicle dynamics. Butterworth-Heinemann, 2012.

77
[16] W. Hirschberg, G. Rill, and H. Weinfurter, “Tire model TMeasy,” Veh. Syst. Dyn., vol. 45, no.
SUPPL. 1, pp. 101–119, 2007.

[17] L. Nielsen and U. Kiencke, Automotive Control Systems: For Engine, Driveline, and Vehicle,
Second Edition, Second Edition. Springer, 2005.

[18] E. M. Kasprzak and D. Gentz, “The Formula SAE Tire Test Consortium-Tire Testing and Data
Handling,” 2006.

[19] A. M. G. & C. AMK, “AMK RACING KIT 4 wheel drive ‘Formula Student Electric,’” 2015.

[20] R. N. Jazar, Vehicle Dynamics: Theory and Application, Second edi. Springer, 2014.

[21] W. F. Milliken and D. L. Milliken, Race car vehicle dynamics. SAE International, 1995.

[22] M. Guiggiani, The Science of Vehicle Dynamics: Handling, Braking, and Ride of Road and Race
Cars. Pisa, Italy: Springer, 2014.

[23] R. M. Rogers, Applied mathematics in integrated navigation systems. American Institute of


Aeronautics and Astronautics, 2003.

[24] H. Y. Guo, Y. Ji, T. Qu, and H. Chen, “Understanding and Modeling the Human Driver Behavior
Based on MPC,” IFAC Proc. Vol., vol. 46, no. 21, pp. 133–138, Jan. 2013.

[25] “OptimumLap - OptimumG.” [Online]. Available:


http://www.optimumg.com/software/optimumlap/. [Accessed: 31-May-2019].

[26] F. P. Incropera, T. L. Bergman, A. S. Lavine, and D. P. DeWitt, Fundamentals of Heat and Mass
Transfer. Chapter 1: Introduction pp 2-67, Seventh Ed. John Wiley & Sons, Ltd, 2011.

[27] F. A. Talati Salman Jalalifar, F. S. Talati Á Jalalifar, and F. Talati, “Analysis of heat conduction
in a disk brake system,” vol. 45, pp. 1047–1059, 2009.

[28] M. Vidiya and B. Singh, “Experimental and Numerical Thermal Analysis of Formula Student
Racing Car Disc Brake Design,” 2017.

[29] S. D. Downing and D. F. Socie, “Simple rainflow counting algorithms,” 1982.

[30] D. O. Anderson, “Safety Factor,” 2001.

[31] “Three Reasons to Start Designing Iteratively – AndersRamsay.com.” [Online]. Available:


http://www.andersramsay.com/2009/03/01/three-reasons-to-start-designing-iteratively/.
[Accessed: 31-May-2019].

78
[32] G. Savage, “Composite materials technology in formula 1 motor racing,” SPE Automot. Compos.
Div. - 8th Annu. Automot. Compos. Conf. Exhib. ACCE 2008 - Road to Light. Perform., vol. 1,
no. July 2008, pp. 109–139, 2008.

[33] ISO, “ISO 6336-1:2006 - Calculation of load capacity of spur and helical gears -- Part 1: Basic
principles, introduction and general influence factors.” ISO, 2007.

[34] S. P. Radzevich, Dudley ’ s Handbook of Practical Gear Design and Manufacture. 2012.

[35] H. Trea and T. A. B. Le, “Böhler V155: VERGÜTUNGSSTAHL HEAT TREATABLE STEEL.”
BÖHLER EDELSTAHL GMBH & CO KG.

[36] ISO, “ISO 6336-2:2006 - Calculation of load capacity of spur and helical gears -- Part 2:
Calculation of surface durability (pitting).” ISO, 2006.

[37] ISO, “ISO 6336-3:2006 - Calculation of load capacity of spur and helical gears -- Part 3:
Calculation of tooth bending strength.” ISO, 2007.

[38] “Involute Gear Profile | KHK Gears.” [Online]. Available:


https://khkgears.net/new/gear_knowledge/gear_technical_reference/involute_gear_profile.html.
[Accessed: 31-May-2019].

[39] DIN, “DIN 3992: Addendum modification of external spur and helical gears.” DIN, 1964.

[40] ASM International. Handbook Committee., ASM handbook. .

[41] R. Bhatta and W. Reffeor, “Comparison of Load Carrying Capacity of Three and Four Lobed
Polygonal Shaft and Hub Connection for Constant Grinding Diameter,” p. V009T12A010, 2017.

[42] DIN, “DIN 32711-1 Shaft to collar connection - Polygon profil P3G - Part 1: Generalities and
geometry.” DIN, 2009.

[43] DIN, “DIN 32712-1: Shaft to collar connection - Polygon profile P4C - Part 1: Generalities and
geometry.” DIN, 2009.

[44] C. Großmann, “Fretting fatigue of shape optimised polygon-shaft-hub connections,” p. 172, 2006.

[45] Altair Inc., “Practical Aspects of Optimizasyon,” p. 361, 2018.

[46] P. M. de A. Fontes, “Refrigeração do propulsor elétrico de um veículo Formula Student,” Instituto


Superior técnico, Tese de Mestrado, 2016.

[47] J. P. C. Ferro, “Design and Simulation of an ABS Control Scheme for a Formula Student

79
Prototype-master,” Instituto Superior técnico, Tese de Mestrado, 2014.

[48] Z. Li, P. Xiao, B. Zhang, Y. Li, and Y. Lu, “Preparation and tribological properties of C / C – SiC
brake composites modi fi ed by in situ grown carbon nano fi bers,” vol. 41, pp. 11733–11740,
2015.

80
Appendix A

Table A-1 – Characteristics of each concept


Ner of 𝑰𝒎𝒐𝒕𝒐𝒓 𝑰𝑫𝑻 𝑰𝒘𝒉𝒆𝒆𝒍 𝑰𝝍 𝑰𝝋 𝑰𝜽
gears
GR 𝒛𝑮𝟏 𝒛𝑮𝟐 𝒛𝑮𝟑 𝒛𝑮𝟒 𝒛𝑮𝟓 𝒛𝑮𝟔 η [%] 𝒎𝒊 [Kg] 𝒎𝒄𝒉 [Kg] 𝑴 [Kg]
[g.mm2] [g.mm2] [g.mm2]
𝑴𝒆𝒒 [Kg] 𝒉𝑪𝒐𝑮 [m]
(g.mm2) (g.mm2) (g.mm2)

Concept 1 8 14.951 15 43 15 -73 - - 95.28 13.55 230 284.20 2.75E+05 7.04E+05 3.60E+08 319.66 0.275 1.26E+11 2.17E+10 8.34E+10

Concept 2 8 15 15 45 15 -75 - - 93.70 13.42 230 283.68 2.76E+05 6.95E+05 2.18E+08 308.24 0.275 1.28E+11 2.17E+10 8.34E+10

Concept 3 4 15.2 15 19 15 -180 - - 84.58 14.17 230 286.68 2.74E+05 1.91E+03 2.45E+08 310.39 0.275 1.33E+11 2.17E+10 8.34E+10

Concept 4 10 14.96 15 18 -51 - - - 91.88 12.95 230 281.80 2.75E+05 1.29E+05 2.17E+08 303.79 0.275 1.26E+11 2.17E+10 8.34E+10

Concept 5 8 15 45 75 15 -135 - - 83.46 15.63 230 292.52 3.02E+05 2.35E+06 2.38E+08 326.22 0.275 1.37E+11 2.17E+10 8.34E+10

Concept 6 6 14.907 18 27 18 28 18 -115 89.33 13.54 230 284.16 2.76E+05 8.58E+03 2.23E+08 306.10 0.275 1.29E+11 2.17E+10 8.34E+10

Concept 7 8 14.951 15 43 15 -73 - - 93.10 9.81 262 301.72 2.75E+05 7.04E+05 2.29E+08 327.11 0.260 1.34E+11 2.26E+10 1.02E+11

Concept 8 4 15 15 45 15 75 - - 94.04 9.81 260 299.24 2.75E+05 7.04E+05 2.28E+08 324.62 0.270 1.33E+11 2.27E+10 1.02E+11

𝑧𝑖𝐺 are the number of teeth of gear i. A negative value represents an internal gear.
Appendix B

Expanded vehicle model equations

The vertical model presented in section 0 is defined by a state space system ẋ =Ax+Bu, where x
is a vector with the state variables and u is the vector with input variables.

𝑇
x = [𝑍, 𝑍̇, 𝜃, 𝜃̇ , 𝜑, 𝜑̇ , ℎ1 , ℎ̇1 , ℎ2 , ℎ̇2 , ℎ3 , ℎ̇3 , ℎ4 , ℎ̇4 ]

u = [𝐹𝑧 , 𝑀𝜃 , 𝑀𝜑 , 𝐺1 , 𝐺2 , 𝐺3 , 𝐺4 ]

• Expansion of equation (3.3), representing the vertical movement 𝑍 of the sprung mass:

𝑚𝑐ℎ 𝑍̈ = 𝐹𝑧 − [𝐾1 + 𝐾2 + 𝐾3 + 𝐾4 ]𝑍 − [𝐶1 + 𝐶2 + 𝐶3 + 𝐶4 ]𝑍̇ + [(𝐾2 + 𝐾4 )𝑑 −


(𝐾1 + 𝐾3 )𝑐]𝜑 + [(𝐶2 + 𝐶4 )𝑑 − (𝐶1 + 𝐶3 )𝑐]𝜑̇ + [(𝐾1 + 𝐾2 )𝑎 −
(B.1)
(𝐾3 + 𝐾4 )𝑏]𝜃 + [(𝐶1 + 𝐶2 )𝑎 − (𝐶3 + 𝐶4 )𝑏]𝜃̇ + 𝐾1 ℎ1 + 𝐶1 ℎ̇1 + 𝐾2 ℎ2 +
𝐶2 ℎ̇2 + 𝐾3 ℎ3 + 𝐶3 ℎ̇3 + 𝐾4 ℎ4 + 𝐶4 ℎ̇4

• Expansion of equation (3.4), representing the rotation φ:

𝐼𝜑 𝜑̈ = 𝑀𝜑 + [(𝐾2 + 𝐾4 )𝑑 − (𝐾1 + 𝐾3 )𝑐]𝑍 − [(𝐶2 + 𝐶4 )𝑑 − (𝐶1 + 𝐶3 )𝑐]𝑍̇ − [(𝐾1 + 𝐾3 )𝑐 2 +


(𝐾2 + 𝐾4 )𝑑 2 ]𝜑 − [(𝐶1 + 𝐶3 )𝑐 2 + (𝐶2 + 𝐶4 )𝑑 2 ]𝜑̇ + [(𝐾1 𝑎 − 𝐾3 𝑏)𝑐 +
(B.2)
(𝐾4 𝑏 + 𝐾2 𝑑)𝑏]𝜃 + [(𝐶1 𝑎 − 𝐶3 𝑏)𝑐 + (𝐶4 𝑏 − 𝐶2 𝑎)𝑑]𝜃̇ + 𝐾1 ℎ1 𝑐 + 𝐶1 ℎ̇1 𝑐 +
𝐾2 ℎ2 𝑑 + 𝐶2 ℎ̇2 𝑑 + 𝐾3 ℎ3 𝑐 + 𝐶3 ℎ̇3 𝑐 + 𝐾4 ℎ4 𝑑 + 𝐶4 ℎ̇4 𝑑

• Expansion of equation (3.5), representing the rotation φ:

𝐼𝜃 𝜃̈ = 𝑀𝜃 + [(𝐾1 + 𝐾2 )𝑎 − (𝐾3 + 𝐾4 )𝑏]𝑍 + [(𝐶1 + 𝐶2 )𝑎 − (𝐶3 + 𝐶4 )𝑏]𝑍̇ −


[(𝐾1 + 𝐾2 )𝑎2 + (𝐾3 + 𝐾4 )𝑏 2 ]𝜃 − [(𝐶1 + 𝐶2 )𝑎2 + (𝐶2 + 𝐶4 )𝑏2 ]𝜃̇ +
(B.3)
[(𝐾1 𝑐 − 𝐾2 𝑑)𝑎 + (𝐾4 𝑑 − 𝐾3 𝑐)𝑏]𝜑 + [(𝐶1 𝑐 − 𝐶2 𝑑)𝑎 + (𝐶4 𝑑 − 𝐶3 𝑐)𝑏]𝜑̇ −
𝐾1 ℎ1 𝑎 − 𝐶1 ℎ̇1 𝑎 − 𝐾2 ℎ2 𝑎 − 𝐶2 ℎ̇2 𝑎 + 𝐾3 ℎ3 𝑏 + 𝐶3 ℎ̇3 𝑏 + 𝐾4 ℎ4 𝑏 + 𝐶4 ℎ̇4 𝑏

• Expansion of equation (3.6), representing the vertical movement ℎ1 :

𝑚1 ℎ̈1 = 𝐾1 𝑍 + 𝐶1 𝑍̇ + 𝐾1 𝜑𝑐 + 𝐶1 𝜑̇ 𝑐 − 𝐾1 𝜃𝑎 − 𝐶1 𝜃̇𝑎 − [𝐾1 + 𝐾𝑝1 + 𝐾𝑎𝑟𝑏𝐹 ]ℎ1


(B.4)
− [𝐶1 + 𝐶𝑝1 ]ℎ̇1 + 𝐾𝑎𝑟𝑏𝐹 ℎ1 + 𝐾𝑝1 𝐺1

• Expansion of equation (3.7), representing the vertical movement ℎ2 :

2
𝑚2 ℎ̈2 = 𝐾2 𝑍 + 𝐶2 𝑍̇ − 𝐾2 𝜑𝑑 − 𝐶2 𝜑̇ 𝑑 − 𝐾2 𝜃𝑎 − 𝐶2 𝜃̇ 𝑎 − [𝐾2 + 𝐾𝑝2 + 𝐾𝑎𝑟𝑏𝐹 ]ℎ2
(B.5)
− [𝐶2 + 𝐶𝑝2 ]ℎ̇2 + 𝐾𝑎𝑟𝑏𝐹 ℎ2 + 𝐾𝑝2 𝐺2

• Expansion of equation (3.8), representing the vertical movement ℎ3 :

𝑚3 ℎ̈3 = 𝐾3 𝑍 + 𝐶3 𝑍̇ + 𝐾3 𝜑𝑐 + 𝐶3 𝜑̇ 𝑐 + 𝐾3 𝜃𝑏 + 𝐶3 𝜃̇ 𝑏 − [𝐾3 + 𝐾𝑝3 + 𝐾𝑎𝑟𝑏𝑅 ]ℎ3


(B.6)
− [𝐶3 + 𝐶𝑝3 ]ℎ̇3 + 𝐾𝑎𝑟𝑏𝑅 ℎ3 + 𝐾𝑝3 𝐺3

• Expansion of equation (3.9), representing the vertical movement ℎ4 :

𝑚4 ℎ̈4 = 𝐾4 𝑍 + 𝐶4 𝑍̇ − 𝐾4 𝜑𝑑 − 𝐶4 𝜑̇ 𝑑 + 𝐾4 𝜃𝑏 − 𝐶4 𝜃̇𝑏 − [𝐾4 + 𝐾𝑝4 +


(B.7)
𝐾𝑎𝑟𝑏𝑅 ]ℎ4 − [𝐶4 + 𝐶𝑝4 ]ℎ̇4 + 𝐾𝑎𝑟𝑏𝑅 ℎ4 + 𝐾𝑝4 𝐺4

Car characteristics

Each concept presented in section 2.1 has its own characteristics, but the transient model of
section 3.1 requires additional information, which is independent from those concepts.

Table B-1 – Car characteristics used in the transient model.

Spring-rate Front (𝐾𝑠1 and 𝐾𝑠2 ) 35025 Static Ackerman Front 14


[N/m] Rear (𝐾𝑠3 and 𝐾𝑠4 ) 30647 (%) Rear 0
Damping Front (𝐶𝑠1 and 𝐶𝑠2 ) 2000 Front 5
coefficient Steering-ratio
[Ns/m] Rear (𝐶𝑠3 and 𝐶𝑠4 ) 2000 Rear 0

Tire spring-rate Front (𝐾𝑝1 and 𝐾𝑝2 ) 95000 Wheel radius [m] 0.228
[N/m] Rear (𝐾𝑝3 and 𝐾𝑝4 ) 95000 Weight distribution (%front) 0.5
Tire damping Front (𝐶𝑝1 and 𝐶𝑝2 ) 1436 Weight distribution (%left) 0.5
coefficient
[Ns/m] Rear (𝐶𝑝3 and 𝐶𝑝4 ) 1436 Wheel base [m] 1.535

Anti-roll bar Front (𝐾𝑎𝑟𝑏𝐹 ) 227 Track [m] 1.2


stiffness [N/m] Rear (𝐾𝑎𝑟𝑏𝑅 ) 406 Lift coefficient (𝑪𝒍 ) -2.875

Motion-ratio Front 1.27 Drag coefficient (𝑪𝒅 ) 1.375


(MR) 𝒑𝒓𝒐𝒋 𝒑𝒓𝒐𝒋
Rear 1.31 Reference area (𝑨𝒙 and 𝑨𝒛 ) [m2] 1.182

3
Appendix C
Efficiency [%] Motor speed [rpm]
Torque [Nm] 500 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 16000 17000 18000 19000 20000
1.3 66.61 68.99 72.6 74.93 76.3 76.95 77.13 77.03 76.8 76.57 76.45 76.47 76.68 77.06 77.56 78.11 78.6 78.87 78.75 78.02 76.44
2.7 62 67.25 75.23 80.44 83.51 85.03 85.49 85.31 84.84 84.35 84.05 84.06 84.44 85.17 86.14 87.2 88.09 88.5 88.04 86.25 82.58
5.4 49.16 57.41 70.1 78.58 83.81 86.64 87.79 87.91 87.52 87.02 86.72 86.83 87.42 88.49 89.9 91.42 92.7 93.3 92.65 90.09 84.85
7.9 39.49 48.98 63.76 73.86 80.32 84.07 85.91 86.52 86.47 86.23 86.13 86.39 87.13 88.35 89.91 91.57 93 93.7 93.11 90.51 85.1
10.4 32.49 42.42 58.04 68.94 76.14 80.54 82.96 84.05 84.41 84.49 84.63 85.06 85.91 87.18 88.77 90.46 91.91 92.68 92.21 89.84 84.78
12.5 26.93 36.87 52.66 63.9 71.55 76.46 79.38 80.97 81.76 82.2 82.62 83.25 84.21 85.52 87.11 88.77 90.21 91.03 90.73 88.7 84.23
14.4 22.65 32.34 47.9 59.18 67.06 72.33 75.68 77.7 78.91 79.71 80.41 81.24 82.31 83.67 85.23 86.85 88.27 89.14 89.01 87.35 83.53
16 19.08 28.33 43.35 54.44 62.4 67.92 71.62 74.05 75.66 76.82 77.81 78.84 80.04 81.43 82.98 84.55 85.94 86.85 86.89 85.61 82.47
17.4 16.19 24.94 39.29 50.05 57.94 63.6 67.56 70.33 72.29 73.79 75.08 76.34 77.68 79.13 80.65 82.12 83.35 84.07 83.95 82.56 79.41
18.5 13.8 22.08 35.71 46.05 53.76 59.41 63.53 66.54 68.82 70.65 72.26 73.8 75.33 76.85 78.31 79.54 80.34 80.41 79.4 76.85 72.28
19.6 12.6 19.54 31.67 41.66 49.75 56.16 61.1 64.78 67.39 69.1 70.09 70.52 70.52 70.25 69.82 69.35 68.95 68.7 68.69 68.98 69.65
21 9.1 14.27 24.08 33.05 41.01 47.84 53.47 57.85 60.99 62.93 63.76 63.58 62.57 60.93 58.9 56.76 54.84 53.5 53.14 54.22 57.2

Figure C-1 – Expanded efficiency map.


Power (kW) Motor speed [rpm]
Torque [Nm] 500 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 16000 17000 18000 19000 20000
1.3 0.07 0.14 0.27 0.41 0.54 0.68 0.82 0.95 1.09 1.23 1.36 1.50 1.63 1.77 1.91 2.04 2.18 2.31 2.45 2.59 2.72
2.7 0.14 0.28 0.57 0.85 1.13 1.41 1.70 1.98 2.26 2.54 2.83 3.11 3.39 3.68 3.96 4.24 4.52 4.81 5.09 5.37 5.65
5.4 0.28 0.57 1.13 1.70 2.26 2.83 3.39 3.96 4.52 5.09 5.65 6.22 6.79 7.35 7.92 8.48 9.05 9.61 10.18 10.74 11.31
7.9 0.41 0.83 1.65 2.48 3.31 4.14 4.96 5.79 6.62 7.45 8.27 9.10 9.93 10.75 11.58 12.41 13.24 14.06 14.89 15.72 16.55
10.4 0.54 1.09 2.18 3.27 4.36 5.45 6.53 7.62 8.71 9.80 10.89 11.98 13.07 14.16 15.25 16.34 17.43 18.51 19.60 20.69 21.78
12.5 0.65 1.31 2.62 3.93 5.24 6.54 7.85 9.16 10.47 11.78 13.09 14.40 15.71 17.02 18.33 19.63 20.94 22.25 23.56 24.87 26.18
14.4 0.75 1.51 3.02 4.52 6.03 7.54 9.05 10.56 12.06 13.57 15.08 16.59 18.10 19.60 21.11 22.62 24.13 25.64 27.14 28.65 30.16
16 0.84 1.68 3.35 5.03 6.70 8.38 10.05 11.73 13.40 15.08 16.76 18.43 20.11 21.78 23.46 25.13 26.81 28.48 30.16 31.83 33.51
17.4 0.91 1.82 3.64 5.47 7.29 9.11 10.93 12.75 14.58 16.40 18.22 20.04 21.87 23.69 25.51 27.33 29.15 30.98 32.80 34.62 36.44
18.5 0.97 1.94 3.87 5.81 7.75 9.69 11.62 13.56 15.50 17.44 19.37 21.31 23.25 25.19 27.12 29.06 31.00 32.93 34.87 36.81 38.75
19.6 1.03 2.05 4.11 6.16 8.21 10.26 12.32 14.37 16.42 18.47 20.53 22.58 24.63 26.68 28.74 30.79 32.84 34.89 36.95 39.00 41.05
21 1.10 2.20 4.40 6.60 8.80 11.00 13.19 15.39 17.59 19.79 21.99 24.19 26.39 28.59 30.79 32.99 35.19 37.38 39.58 41.78 43.98

Figure C-2 – Map of output power of each motor.

4
Power [kW] Motor speed [rpm]
Torque [Nm] 500 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 16000 17000 18000 19000 20000
1.3 0.07 0.14 0.27 0.41 0.54 0.68 0.82 0.95 1.09 1.23 1.36 1.50 1.63 1.77 1.91 2.04 2.18 2.31 2.45 2.59 2.72
2.7 0.14 0.28 0.57 0.85 1.13 1.41 1.70 1.98 2.26 2.54 2.83 3.11 3.39 3.68 3.96 4.24 4.52 4.81 5.09 5.37 5.65
5.4 0.28 0.57 1.13 1.70 2.26 2.83 3.39 3.96 4.52 5.09 5.65 6.22 6.79 7.35 7.92 8.48 9.05 9.61 10.18 10.74 11.31
7.9 0.41 0.83 1.65 2.48 3.31 4.14 4.96 5.79 6.62 7.45 8.27 9.10 9.93 10.75 11.58 12.41 13.24 14.06 14.89 15.72 16.55
10.4 0.54 1.09 2.18 3.27 4.36 5.45 6.53 7.62 8.71 9.80 10.89 11.98 13.07 14.16 15.25 16.34 17.43 18.51 19.60 20.69 21.78
12.5 0.65 1.31 2.62 3.93 5.24 6.54 7.85 9.16 10.47 11.78 13.09 14.40 15.71 17.02 18.33 19.63 20.94 22.25 23.56 24.87 26.18
14.4 0.75 1.51 3.02 4.52 6.03 7.54 9.05 10.56 12.06 13.57 15.08 16.59 18.10 19.60 21.11 22.62 24.13 25.64 27.14 28.65 30.16
16 0.84 1.68 3.35 5.03 6.70 8.38 10.05 11.73 13.40 15.08 16.76 18.43 20.11 21.78 23.46 25.13 26.81 28.48 30.16 31.83 33.51
17.4 0.91 1.82 3.64 5.47 7.29 9.11 10.93 12.75 14.58 16.40 18.22 20.04 21.87 23.69 25.51 27.33 29.15 30.98 32.80 34.62 36.44
18.5 0.97 1.94 3.87 5.81 7.75 9.69 11.62 13.56 15.50 17.44 19.37 21.31 23.25 25.19 27.12 29.06 31.00 32.93 34.87 36.81 38.75
19.6 1.03 2.05 4.11 6.16 8.21 10.26 12.32 14.37 16.42 18.47 20.53 22.58 24.63 26.68 28.74 30.79 32.84 34.89 36.95 39.00 41.05
21 1.10 2.20 4.40 6.60 8.80 11.00 13.19 15.39 17.59 19.79 21.99 24.19 26.39 28.59 30.79 32.99 35.19 37.38 39.58 41.78 43.98

Figure C-3 – Map of power supplied to each motor.


Power (kW) Motor speed [rpm]
Torque [Nm] 500 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 16000 17000 18000 19000 20000
5.2 0.41 0.79 1.50 2.18 2.85 3.54 4.24 4.95 5.67 6.40 7.12 7.83 8.52 9.19 9.83 10.46 11.08 11.74 12.45 13.26 14.25
10.8 0.91 1.68 3.01 4.22 5.42 6.65 7.94 9.28 10.66 12.07 13.46 14.80 16.07 17.26 18.38 19.45 20.54 21.72 23.12 24.91 27.39
21.6 2.30 3.94 6.45 8.64 10.80 13.05 15.46 18.01 20.68 23.39 26.08 28.66 31.05 33.23 35.22 37.11 39.04 41.21 43.94 47.70 53.32
31.6 4.19 6.76 10.38 13.44 16.48 19.68 23.11 26.77 30.62 34.54 38.42 42.14 45.58 48.69 51.53 54.21 56.93 60.04 63.97 69.47 77.77
41.6 6.70 10.27 15.01 18.96 22.89 27.04 31.51 36.28 41.29 46.40 51.48 56.34 60.85 64.96 68.70 72.24 75.84 79.91 85.04 92.13 102.77
45.8 8.21 12.24 17.45 21.77 26.08 30.64 35.54 40.77 46.26 51.87 57.42 62.76 67.73 72.28 76.43 80.36 84.35 88.84 94.46 102.14 113.55
49.6 10.01 14.46 20.10 24.77 29.43 34.37 39.66 45.31 51.22 57.25 63.24 69.00 74.39 79.34 83.89 88.21 92.59 97.47 103.51 111.67 123.60
52.8 12.13 16.96 22.97 27.94 32.92 38.19 43.83 49.82 56.08 62.46 68.80 74.92 80.67 85.98 90.89 95.57 100.31 105.55 111.94 120.44 132.65
55.6 14.61 19.75 26.06 31.32 36.60 42.17 48.12 54.41 60.97 67.65 74.28 80.68 86.72 92.35 97.61 102.68 107.87 113.64 120.66 129.93 143.17
57.8 17.39 22.68 29.21 34.72 40.27 46.13 52.35 58.90 65.68 72.56 79.36 85.92 92.15 98.02 103.62 109.19 115.08 121.87 130.36 141.86 158.60
60 19.64 26.14 33.43 39.04 44.45 50.07 56.06 62.50 69.38 76.67 84.31 92.20 100.28 108.44 116.66 124.91 133.18 141.53 150.09 159.13 169.26
62.8 27.52 35.96 44.04 49.40 54.34 59.49 65.11 71.36 78.33 86.10 94.72 104.26 114.78 126.32 138.89 152.35 166.24 179.71 191.50 200.19 205.17

Figure C-4 – Map of total power supplied to the all motor.

5
Appendix D
Constants for thermal model

Uprighttop Brake pins

Hubrear
Uprightrear
Hubfront
Calliper Ring

Uprightfront Middle support

Carrier front
Bearingrear
Drive train

Brake disc Bearingfront

Figure D-1 – Identification of each zone considered in the thermal model of Concept 4

Figure D-1 shows Concept 4 with the identification of each zone considered in the thermal model.
The calliper and brake pads are not shown in the figure, as they would be covering other important
zones. Based on this figure, Table D-1 shows the thermal constants used in the model of section 3.2.4.

Table D-1 – Thermal constants used in the thermal model

Thermal constants
Component R [K/W] Component Mass [Kg] Cp [J/KgK]
Brake disc 1/6 Brake disc 0.315 460
Brake pins 2x6 Brake pins 0.05 460
Hubfront 4.42 Hub 0.79 900
Hubrear 1.1 Bearingfront 0.11 460
Bearingfront 9e14 Bearingrear 0.14 460
Bearingrear 1.3e16 Upright 1.05 900
Uprightfront-top 0.38 Calliper 0.35 900
Uprightrear-top 0.4 Brake pad 4x0.018 950
Uprightfront-rear 0.8 Oil 0.2 1800
Uprightcaliper-front 0.38 Drive train 0.8 460
Uprightcaliper-rear 0.38 Middle support 0.108 900
Uprightcaliper-top 0.02 Carrier front 0.075 900
Brake calliper 1 Ring 0.24 460
Brake pad 1 Brake pins 0.105 460

6
Ring 0.66
Middle support 0.76 Component hnatural [W/m2K] hforced [W/m2K] A [m2]
Carrier front 0.06 Upright 3000 - 0.03
Drive train 4x2.068 Hub 5 12 0.056
Oil 9 Brake disc 5 12 0.037

Sensors assembly in FST 08e

Section 3.2.4 shows the data acquired in testing to validate the thermal model. The temperature
sensors used on the car were mounted in the positions shown in Figure D-2.

NTCbrake pad
NTCupright

Infra-red NTCbrake calliper

Figure D-2 – Position of the temperature sensors mounted on FST 08e.

7
Appendix E

As presented in section 3.2.2.2, twelve different track layouts were inserted in Optimum Lap, two
corresponding to the acceleration and skid-pad events and the other ten corresponding to autocross
and endurance events. The acceleration and skid-pad tracks are a simple 75m straight line and the
other is a simple 360º, circular track, with an 8.5m radius. The remaining tracks are much more complex,
and their layouts are presented in Figure E-1. These layouts represent the ones used in the competitions
of 2018.

Test track Formula Student Germany


Autocross and Endurance track

Formula Student Netherlands


Autocross and Endurance track Formula Student Spain
Autocross and Endurance track

Formula Student UK
Formula Student UK
Endurance track
Autocross track

8
Formula Student Czech Republic Formula Student Czech Republic
Autocross track Endurance track

Formula Student Austria Formula Student Austria


Autocross track Endurance track

Figure E-1 – Track layouts used in the steady-state simulations.

9
Appendix F

Meaning of each factor of equations (5.1) through (5.6)

• 𝜎𝐹0 : nominal tooth root • 𝑆𝐹𝑚𝑖𝑛 : safety factor for • 𝑍𝛽 : helix angle factor;
stress; tooth breakage; • 𝑍𝜖 : contact ratio factor;
• 𝐾𝐴 : application factor; • 𝑌𝛿𝑟𝑒𝑙𝑇 : notch sensitivity • 𝑑 𝐺 : reference diameter
• 𝐾𝑉 : dynamic factor; factor; of the pinion;
• 𝐾𝐹𝛽 : face load factor; • 𝑌𝑅𝑟𝑒𝑙𝑇 : relative surface • 𝜎𝐻𝑃 : permissible
• 𝐾𝐹𝛼 : transverse load factor; contact stress;
factor; • 𝑌𝑋 : size factor; • 𝜎𝐻𝑙𝑖𝑚 : allowable stress
• 𝑌𝐹 : tooth form factor; • 𝐹𝑡 : transverse number (contact) ;
• 𝑌𝑆 : nominal stress tangential load; • 𝑍𝑁𝑇 : life factor for
correction factor; • 𝑏 : facewitdh;
𝐺
contact stress;
• 𝑌𝛽 : helix angle factor; • 𝑚𝐺 : Ratio between • 𝑆𝐻𝑚𝑖𝑛 : safety factor for
• 𝑌𝜖 : deep tooth factor; mating gears; pitting;
• 𝑌𝐵 : rim thickness factor; • 𝜎𝐻0 : nominal contact • 𝑍𝐿 : lubricant factor;
• 𝜎𝐹𝑃 : permissible tooth stress; • 𝑍𝑉 : velocity factor;
root stress; • 𝐾𝐻𝛽 : face load factor for • 𝑍𝑅 : roughness factor;
• 𝜎𝐹𝑙𝑖𝑚 : nominal stress contact stress; • 𝑍𝑊 : work hardening
number (bending); • 𝐾𝐻𝛼 : transverse load factor;
• 𝑌𝑆𝑇 : reference stress factor for contact stress; • 𝑍𝑋 : size factor.
correction factor; • 𝑍𝐵 : tooth contact factor;

• 𝑌𝑁𝑇 : life factor for tooth • 𝑍𝐸 : elasticity factor;

root stress; • 𝑍𝐻 : zone factor;

Standards used by KISSsoft to perform the analyses presented in this


document, in association with ISO 6336

• ISO 21771:2007 – Geometry • DNV 41.2:2012 – Safety of the


calculations; hardened layer;
• ISO 286:2010 – Centre distance • ISO/TR 13989:2000 – Scuffing load
tolerance; capacity;
• Q-ISO 1328:1995 – Accuracy grade; • DIN 3967 – Tooth thickness deviation;
• ISO 53:1998 – Reference profile; • ISO/TR 10064-3:1996 – Axis alignment
• ISO/TR 14144-1:2004 – Micro-pitting; tolerance;
• DIN 3992 – Profile shift coefficient.

10
KISSsoft workflow

To perform the gear design in KISSsoft, the following steps must be taken:

1. Select “planetary gear” on the tree on the top left side of the screen, under “modules” tab;
2. In the “Basic Data” tab, insert desired values of gear module, number of planets and
accuracy grade in the respective boxes (Figure F-1). Choose the material and lubricant
from the respective databases;
3. In the “Rating” tab, insert the maximum torque and speed, required service life and upload
the load spectrum presented in section 4.1 (Figure F-2);
4. Perform a “Rough Sizing” analysis, by defining a helix angle and the desired gear ratio
and allowable deviation;
5. From the solutions found after the analysis, select the one closest to the desired (either
by evaluating the total gear ratio, module, or gear size);
6. The information in the several tabs of the software are updated with the information
calculated in the “Rough Sizing”;
7. Perform a “Fine Sizing” analysis, by defining the desired gear ratio, module, pressure
angle, helix angle, centre distance, range for profile shift coefficient and the respective
allowable deviations. Make sure “Permit undercut” option under the “Conditions III” tab is
not selected;
8. From the solutions found, select the one that is closest to the desired (either by evaluating
the total gear ratio, module, gear size or safety factors);
9. The information in the several tabs is once again updated with the calculated information;
10. Further adjustments can be made by changing parameters directly within the several
tabs, like profile shift coefficient, facewitdh, tip relief, tooth crowning,… After making the
modifications, click “Run” to recalculate the safety factors, which will be presented in the
results window (Figure F-1).

11
“Fine Sizing”
“Run”

“Rough Sizing”

Results window

Figure F-1 – KISSsoft “Basic data” interface for planetary gears

Maximum torque and


speed, and servce life

Load spectrum window

Figure F-2 – KISSsoft “Rating” interface for planetary gears

12
Appendix G

Import • Directly from Solidworks


geometry • Figure G-1

• Tetramesh
•2D surface mesh, extruded to a 3D mesh
Mesh
•curvature and proximity based
• Figure G-2

• Create contact surfaces - Figure G-3 - red


Contacts • Create contact properties (PCONT) - Figure G-3 - orange
• Create contact groups - Figure G-3 - yellow

• Specify static properties 𝐸, 𝜈, 𝜌 - Figure G-3 - black


Material
properties • Create property with material properties and element type
(PSOLID) - Figure G-3 - orange

• Create constraints (with 1D RBE3 elements) - Figure G-3 - blue


Loads • Apply static loads (with 1D RBE2 elements) - Figure G-3 - blue
• Create static analysis load step - Figure G-3 - green

13
Figure G-1 – Import geometry in Hypermesh

Figure G-2 – 3D mesh generation in Hypermesh

Figure G-3 – Position of the contact, material and load sub-menus

14

Você também pode gostar