TREBALL FI DE GRAU

MODELING AND CONTROL OF AN

ELECTRIC VEHICLE
GRAU EN TECNOLOGIES INDUSTRIALS

Author: Gerard Sabarich Mir

Director: Daniel Montesinos Miracle

Date: June 21, 2020

Location: Barcelona
1 Modeling and control of an electric vehicle

ABSTRACT
Following the rising demand for electric vehicles to fight the increasing factors of pollution
in urban areas, electric mobility is becoming more and more important in the automotive industry.
Therefore, this study is conducted to better understand the dynamics and components of electric
vehicles.

First, a theoretical study on vehicle dynamics is conducted, in which equations describing

the vehicle behavior are proposed. Furthermore, analysis of energy storage, propulsion system,
and less complex subsystems of the vehicle are carried out, from these analyses a model will be
developed. This full vehicle model and the control system designed for it are later used to extract
information on vehicle behavior and performance verification for the given car components.

Second, a more complex model is developed for the road-tire interaction, which
introduces the possibility of wheel spin to appear, requiring a traction controller, based on
maximum transmittable torque estimation, MTTE. The MTTE controller limits the torque provided
to the wheel to prevent instabilities in the vehicle and maximizing its performance. Some issues
where encountered since it works for only given torque profiles.

For the implementation of the full model, the energetic macroscopic representation, EMR,
has been used, since it provides a clear interpretation of actions and reactions in each subsystem.
As for the controller an inversion-based controller, IBC, is proposed to maximize control
performance. The implementation of the model and control system has been carried out in
Simulink.
Modeling and control of an electric vehicle 2

CONTENTS

INTRODUCTION......................................................................................................................................... 8

1 VEHICLE FUNDAMENTALS ............................................................................................................... 10

1.1 VEHICLE DYNAMICS ...................................................................................................................... 10

1.2 TIRE-ROAD INTERACTION ............................................................................................................. 15
1.3 PROPULSION SYSTEM ................................................................................................................... 17
1.4 ENERGY STORAGE ........................................................................................................................ 19

2 VEHICLE MODELLING ...................................................................................................................... 21

2.1 INTRODUCTION ............................................................................................................................ 21

2.2 ELECTRIC VEHICLE MODEL............................................................................................................ 23

3 VEHICLE CONTROL .......................................................................................................................... 29

3.1 INTRODUCTION ............................................................................................................................ 29

3.2 ELECTRIC VEHICLE CONTROLLER .................................................................................................. 31

4 SIMULATION ................................................................................................................................... 39

4.1 FULL VEHICLE SIMULATION .......................................................................................................... 40

4.2 TIRE-ROAD INTERACTION SIMULATION ....................................................................................... 48

5 PLANNING AND BUDGET ................................................................................................................ 52

5.1 PLANNING .................................................................................................................................... 52

5.2 BUDGET ........................................................................................................................................ 53

6 ENVIRONMENTAL IMPACT.............................................................................................................. 55

7 CONCLUSIONS ................................................................................................................................ 56

8 BIBLIOGRAPHY ............................................................................................................................... 58
3 Modeling and control of an electric vehicle

LIST OF FIGURES
FIGURE 1.1.1: FORCE DIAGRAM [1] ...................................................................................................................... 10
FIGURE 1.1.2: PRESSURE DISTRIBUTION [1] ............................................................................................................ 11
FIGURE 1.1.3: GRAVITY EFFECT DECOMPOSITION [1] ................................................................................................ 12
FIGURE 1.1.4: TIRE DEFORMATION AND ROLLING RESISTANCE ON A HARD (A) AND SOFT (B) SURFACE [1] ............................ 12
FIGURE 1.1.5: FREE BODY DIAGRAM OF ONE WHEEL ................................................................................................. 14
FIGURE 1.2.1: CURVE PRODUCED BY THE MAGIC FORMULA [2] .................................................................................. 16
FIGURE 1.3.1: DC MOTOR SCHEMATIC [1] .............................................................................................................. 18
FIGURE 1.3.2: SERIES ARRANGEMENT EQUIVALENT CIRCUIT OF A DC MOTOR [1]............................................................ 18
FIGURE 1.4.1: VOLTAGE CURVE OF A TYPICAL BATTERY [1] ......................................................................................... 19
FIGURE 1.4.2: DISCHARGE CURVE AT DIFFERENT DISCHARGE RATES [1] ......................................................................... 20
FIGURE 2.1.1: GRAPHIC REPRESENTATION OF A REACTION INDUCED BY AN ACTION [3] .................................................... 21
FIGURE 2.1.2: GRAPHIC REPRESENTATION OF BOTH GENERATOR AND RECEPTOR EMR SOURCES. [3] ................................. 22
FIGURE 2.1.3: GRAPHIC REPRESENTATION OF AN EMR ACCUMULATOR [3] ................................................................... 22
FIGURE 2.1.4: GRAPHIC REPRESENTATION OF AN EMR CONVERSION ELEMENT [3] ......................................................... 23
FIGURE 2.2.1: EMR BASED MODEL IMPLEMENTED IN SIMULINK.................................................................................. 28
FIGURE 3.1.1: INVERSION-BASED CONTROL GENERAL SYSTEM REPRESENTATION [3] ........................................................ 29
FIGURE 3.1.2: TIME-INDEPENDENT INVERSION CONTROL, (LEFT) SINGLE INPUT, (RIGHT) MÚLTIPLE INPUT [3]....................... 30
FIGURE 3.1.3: TIME-DEPENDENT INVERSION CONTROL [3] ......................................................................................... 31
FIGURE 3.2.1: SPEED SIGNAL FOR A STEP INPUT, (TOP) TR=0.01 S AND (BOTTOM) TR=0.1 S ............................................. 33
FIGURE 3.2.2: KXY,BRAKE CALCULATOR SYSTEM ............................................................................................................ 35
FIGURE 3.2.3: VOLTAGE SIGNAL FOR A STEP INPUT, (TOP) TR=0.001 S AND (BOTTOM) TR=0.01 S ..................................... 37
FIGURE 3.2.4: IBC MODEL IMPLEMENTED IN SIMULINK ............................................................................................. 38
FIGURE 4.1.1: REFERENCE SPEED AND VEHICLE SPEED FOR A STEP INPUT ....................................................................... 41
FIGURE 4.1.2: FORCE REFERENCE VS.OUTPUTED FORCE ............................................................................................. 41
FIGURE 4.1.3: MAXIMUM TRACTIVE EFFORT VS TRACTIVE EFFORT: (LEFT) FRONT WHEELS, (RIGHT) REAR WHEELS ................. 42
FIGURE 4.1.4: SPEED PROFILE, REFERENCE VS. OUTPUT ............................................................................................. 42
FIGURE 4.1.5: PERCENTAGE OF BRAKING EFFORT PROVIDED BY THE MECHANIC BRAKE ..................................................... 43
FIGURE 4.1.6: MAXIMUM TRACTIVE EFFORT VS TRACTIVE EFFORT: (LEFT) FRONT WHEELS, (RIGHT) REAR WHEELS ................. 43
FIGURE 4.1.7: BATTERY CURRENT AND SOC............................................................................................................ 44
FIGURE 4.1.8: REFERENCE SPEED PROFILE OF 4 NEDC CYCLE CONCATENATED ............................................................... 45
FIGURE 4.1.9: OUTPUT PEED PROFILE OF 4 NEDC CYCLE CONCATENATED ..................................................................... 46
FIGURE 4.1.10: BATTERY EVOLUTION FOR A NS=160 AND NP=100 PACK ..................................................................... 47
FIGURE 4.1.11: BATTERY EVOLUTION FOR A NS=120 AND NP=10 PACK ....................................................................... 47
FIGURE 4.2.1: SLIP (Λ), TOP, AND FRICTION COEFFICIENT (µ), BOTTOM, FOR A CONSTANT SURFACE.................................... 48
FIGURE 4.2.2: REFERENCE TOQUE, MAXIMUM TORQUE AND CAPED REFERENCE TORQUE FOR A CONSTANT SURFACE .............. 49
Modeling and control of an electric vehicle 4

FIGURE 4.2.3: RATIO OF ACCELERATION FOR A CONSTANT SURFACE ............................................................................. 49

FIGURE 4.2.4: SLIP (Λ), TOP, AND FRICTION COEFFICIENT (µ), BOTTOM, FOR A CHANGING SURFACE.................................... 50
FIGURE 4.2.5: REFERENCE TOQUE, MAXIMUM TORQUE AND CAPED REFERENCE TORQUE FOR A CHANGING SURFACE ............. 51
FIGURE 4.2.6: RATIO OF ACCELERATION FOR A CHANGING SURFACE ............................................................................. 51
FIGURE 5.1.1: GANTT DIAGRAM OF THE PROJECT ..................................................................................................... 53
5 Modeling and control of an electric vehicle

LIST OF TABLES
TABLE 1.1: ROLLING RESISTANCE COEFFICIENTS [1] ................................................................................................... 13
TABLE 1.2: BURCKHARDT TIRE MODEL PARÀMETRES.................................................................................................. 16
TABLE 2.1: BATTERY CELL PARAMETERS [4] ............................................................................................................. 24
TABLE 4.1: SIMULATION PARAMETERS.................................................................................................................... 40
TABLE 5.1: ECONOMIC EVALUATION ...................................................................................................................... 54
Modeling and control of an electric vehicle 6

GLOSSARY
Symbols

𝑨𝒇 Frontal area

𝑨𝒆𝒗 Track

𝑪𝑫 Drag coefficient

𝑭𝒘 Aerodynamic resistance

𝑭𝒈 Grading resistance

𝑭𝒓 Rolling resistance

𝑭𝒙𝒚𝒙 Tractive effort

𝒉𝒈 Center of gravity height

𝑱𝒘 Road grading

𝑳𝒂 DC motor inductance

𝑳 Wheelbase

𝑳𝒐𝒏𝒈𝒂 Front-wheel distance to the center of gravity

𝑳𝒐𝒏𝒈𝒃 Rear-wheel distance to the center of gravity

𝑴𝒆𝒗 Vehicle mass

𝑵𝒙𝒚 Normal force

𝑹𝒂 Equivalent resistance of the DC motor

𝒓𝒄𝒖𝒓𝒗 Turning radius

𝒓𝒘 Effective wheel radius

𝒗𝒆𝒗 Linear speed of the vehicle

𝒗𝒙 Linear speed of the wheel center

𝒗𝒘 Equivalent linear speed of the wheel angular speed

Greek symbols

𝝎𝒘 Angular velocity of the wheel

7 Modeling and control of an electric vehicle

𝜶 Road grading

𝜶𝒙𝒚 Duty cycle

𝝆 Air density

Superscript

∗ Reference value

Subscript

𝒃𝒂𝒕 Battery value

𝒆𝒗 Electric vehicle value

𝒇𝒍 Front-left wheel value

𝒇𝒓 Front-right wheel value

𝒎 Motor value

𝒎𝒂𝒙 Maximum value

𝒓𝒆𝒇 Reference value

𝒓𝒍 Rear-left wheel value

𝒓𝒓 Rear-right wheel value

𝒕𝒓𝒂𝒏𝒔 Transmission value

𝒘 Wheel value

𝒙𝒚 General wheel value

Acronyms

𝑺𝑶𝑪 State Of Charge

𝑫𝑪 Direct Current

𝑷𝑰 Proportional Integral

𝑬𝑴𝑹 Energetic Macroscopic Representation

𝑰𝑩𝑪 Inversion Based Control

𝑴𝑻𝑻𝑬 Maximum Transmittable Torque Estimator

Modeling and control of an electric vehicle 8

INTRODUCTION

The following project has the aim of studying the dynamic behavior of a vehicle, its battery
behavior through discharge, and the capabilities of the propulsion system, as well as other internal
parameters of the electric vehicle.

Once the study is done, it is proposed to develop a model of the previous subsystems
studied, using the energetic macroscopic representation to visualize the actions and reactions of
each subsystem more easily. Then, an inversion-based control system will be proposed so the
model behaves as desired.

This controlled model, implemented in Matlab Simulink, will allow for dynamic behavior
analysis as different simulations are carried out, aiming to determine the required vehicle
components, such as a more powerful motor or a higher capacity battery, to fulfill a commanded
speed profile.

Furthermore, a more complex model will be developed, in which the wheel slip is
considered, thus a more complex control system will be required to prevent the wheel from
spinning freely and control the car performance.

Objectives of the project

• General objective: Propose a model for an electric vehicle based on IMR, the required IBC
control system and analyze the dynamic and electric behavior of the electric vehicle.
• Specific objectives:
- Study the dynamic behavior of a vehicle as it interacts with the environment.
- Study the relation between tire slip and friction force.
- Study the propulsion system and battery of the vehicle
- Model vehicle behavior and implement it in Simulink.
- Design an IBC control system for the modeled vehicle and implement it in Simulink.
- Analyze the dynamic and electric behavior of various subsystems of the vehicle.

Scope of the project

The project includes:

• The theoretic study of the dynamic behavior of an electric vehicle.

• The theoretic study of the tire-ground relation and its ideal controller.
9 Modeling and control of an electric vehicle

• The ideal model of the full vehicle.

• The reduced model for the tire-ground interaction.
• The control system for the vehicle.
• The traction control for the reduced model.
Modeling and control of an electric vehicle 10

1 VEHICLE FUNDAMENTALS

1.1 VEHICLE DYNAMICS

As any object moving through a fluid and in contact with a road, a vehicle is subjected to
external forces, such as drag produced by contact with the given fluid, or gravity pushing the
object downwards. Therefore, a tractive effort must overcome these external forces, to propel
the object forward, this is represented by

𝑑𝑣 Σ𝐹𝑡 − Σ𝐹𝑡𝑟
= , (1)
𝑑𝑡 𝑀

where dv/dt is the vehicle acceleration, M is the vehicle mass, ∑Ft is the sum of all tractive efforts,
and last, ∑Ftr is the sum of the resistive forces. These are represented in Figure 1.1.1.

Figure 1.1.1: Force Diagram [1]

1.1.1 Resistances

The sum of the resistive forces can be divided into three, aerodynamic drag, rolling resistance,
and grading resistance, these are introduced below.

• Aerodynamic drag

Drag is created by high-pressure areas in front of the EV and low-pressure areas in the back.
The frontal high-pressure area is created as the vehicle pushes through the air, as it cannot be
displaced immediately, the air in front is compressed creating a force against the vehicle
movement. As for the rear low-pressure area, following the same principle, the air cannot fill the
11 Modeling and control of an electric vehicle

empty space immediately, therefore creating a depression pulling the vehicle backward. This
phenomenon is represented in Figure 1.1.2. These two effects are combined to generate a force
against the vehicle movement,

Figure 1.1.2: Pressure distribution [1]

Drag will be considered a simple force applied at a height hw parallel and opposed to the
vehicle movement, as represented in Figure 1.1.1. As a simplification, on a road car, hw can be
assumed to be equal to hg.

The equivalent force provided by air resistance is described as

1
𝐹𝑤 = 𝜌𝐴𝑓 𝐶𝑑 (𝑣 − 𝑣𝑤 )2 , (2)
2

where ρ is the air density, Af is the frontal area of the vehicle, Cd is the aerodynamic drag
coefficient, v is the vehicle velocity and vw is the wind velocity, this last one will be considered 0
throughout the project.

• Grading Resistance

As gravity pushes downwards on the vehicle, two forces are generated, a force pushing
the vehicle into the ground and, in case there is a slope, a force along the road slope always
pointing downwards, as shown in both Figures 1.1.1 and 1.1.3. This later force is the grading
resistance, it is described as

𝐹𝑔 = 𝑀𝑣 𝑔 sin 𝛼, (3)

where g is the gravitational acceleration and α is the grading slope. It must be stated that grade
slope α is only valid for small road angles since the following simplification is being made:
Modeling and control of an electric vehicle 12

𝐻
𝑖= = tan 𝛼 ≈ sin 𝛼, (4)
𝐿

where H is the height cleared each L distance, as represented in Figure 1.1.3.

Figure 1.1.3: Gravity effect decomposition [1]

• Rolling resistance

As the tires deform while in contact with the ground, an asymmetry is generated in the
pressure distribution underneath the contact patch, due to the material accumulation in front of
the wheel, and lag of it in the rear. This creates a not centered equivalent force, which in turn
generates a torque against the wheel movement, shown in Figure 1.1.4.

Figure 1.1.4: Tire deformation and rolling resistance on a hard (a) and soft (b) surface [1]

The moment generated can be described as

𝑇𝑟 = 𝑃𝑎, (5)

where P is the normal load generated by gravity. If the torque is translated to a force, it can be
stated that
13 Modeling and control of an electric vehicle

𝐹𝑟 = 𝑃𝑓𝑟 cos 𝛼, (6)

where fr is the rolling resistance coefficient. This rolling resistance coefficient is a function of many
road-tire variables which makes it difficult to calculate, therefore a table with the main values in
some situations is proposed in Table 1.1.

Conditions Rolling resistance coefficient

Car tires on concrete or asphalt 0.013
Car tires on rolled gravel 0.02
Tarmac 0.025
Unpaved road 0.05
Field 0.1–0.35
Truck tires on concrete or asphalt 0.006–0.01
Wheels on rail 0.001–0.002
Table 1.1: Rolling resistance coefficients [1]

1.1.2 Dynamic equation

As a reduced version was introduced in (1), and the resistive forces have been presented,
an extended version

𝑑𝑉
𝑀𝑣 = (𝐹𝑓𝑟𝑡 + 𝐹𝑟𝑟𝑡 + 𝐹𝑓𝑙𝑡 + 𝐹𝑟𝑙𝑡 ) − (𝐹𝑟 + 𝐹𝑤 + 𝐹𝑔 ), (7)
𝑑𝑡

is provided, where the tractive effort is divided into the individual contribution of each wheel.
Also, the resistive forces are represented by their contribution.

The maximum tractive effort, overlooking complex tire-road interaction, can be described
as

𝐹𝑥𝑦𝑡,𝑚𝑎𝑥 = 𝜇𝑁𝑥𝑦 , (8)

where µ is the adhesion coefficient and Nxy is the normal load applied to each wheel. The normal
load may be different at each wheel depending on the acceleration profile, the turn steepness and
speed carried through the turn. Therefore, a different expression is provided for each wheel. For
the front wheels and considering only longitudinal dynamics, the normal load is

𝐿𝑏 ℎ𝑔 𝑟𝑤 𝑑𝑣
𝑁𝑓𝑦 = 𝑀𝑣 𝑔𝑐𝑜𝑠(𝛼) − (𝐹𝑤 + 𝐹𝑔 + 𝑀𝑣 𝑔𝑓𝑟 cos(𝛼) + 𝑀𝑣 ), (9)
𝐿 𝐿 ℎ𝑔 𝑑𝑡

and for the rear wheels

Modeling and control of an electric vehicle 14

𝐿𝑎 ℎ𝑔 𝑟𝑤 𝑑𝑣
𝑁𝑓𝑦 = 𝑀𝑣 𝑔𝑐𝑜𝑠(𝛼) + (𝐹𝑤 + 𝐹𝑔 + 𝑀𝑣 𝑔𝑓𝑟 cos(𝛼) + 𝑀𝑣 ), (10)
𝐿 𝐿 ℎ𝑔 𝑑𝑡

For the lateral component, taking the curvature, but overlooking the stirring angle effects
and lateral slip force. The expressions are

𝑀𝑣 𝑣 2 ℎ𝑔
𝑁𝑥𝑙 = 𝑁𝑥𝑦 + , (11)
𝑟𝑐𝑢𝑟𝑣. 𝐴𝑒𝑣

for the left wheels, and

𝑀𝑣 𝑣 2 ℎ𝑔
𝑁𝑥𝑟 = 𝑁𝑥𝑦 − , (12)
𝑟𝑐𝑢𝑟𝑣. 𝐴𝑒𝑣

for the right ones. Where rcurv is the radius of the turn, being positive for left turns and positive for
right turns, and Aev is the track of the vehicle, distance from wheel to wheel.

1.1.3 Wheel dynamics

As the resistive forces and dynamic equations have been discussed, all the parameters
from (1) have been introduced but the tractive force. This one will be dictated by the interaction
between the ground and tire. Before describing the mentioned interaction, an introduction to tire
dynamics is due.

Figure 1.1.5: Free body diagram of one wheel

The dynamic of the wheel is described by the toque balance

𝑑𝜔
𝐽𝑤 = 𝑇𝑚 − 𝐹𝑥 𝑟, (13)
𝑑𝑡
15 Modeling and control of an electric vehicle

where Jw is the wheel inertia, ω is the rotating speed, Tm is the torque provided by the motor, r is
the radius of the wheel and Fx is the friction force acting on the wheel, represented in Figure 1.1.5.
Fx is defined as

𝐹𝑥 = 𝜇(𝜆)𝑁, (14)

where N is the normal load, and µ is the friction coefficient, which is not assumed constant, it will
be defined as a complex function of slip (λ). In the following chapter different approaches and
solutions to the mentioned function are presented.

1.2 TIRE-ROAD INTERACTION

Before defining the relation between the friction coefficient and slip, slip must be defined,
and it is done in [2] as

𝑣𝜔 − 𝑣𝑥
𝜆= , (15)
𝑣𝜔

𝑣𝜔 = 𝑟𝑤 𝜔, (16)

where λ is the slip and vx is the linear speed of the wheel. This description is only valid for an
acceleration profile; therefore, a modification is proposed, described as

𝑣𝜔 − 𝑣𝑥
𝜆= . (17)
max⁡(𝑣𝜔 , 𝑣𝑥 )

As mentioned earlier, the interaction between tire and ground, represented by, µ is non-
trivial, and several approaches have been made to define a reliable model. In this chapter two of
them are discussed: Burckhardt’s model (1993) and Pacejka’s (2002) [2].

1.2.1 Pacejka’s model

Starting by Pacejka, the expression, named the Magic Formula [2]

𝐹𝑥 = 𝐷𝑥 sin(𝐶𝑥 arctan[𝐵𝑥 (1 − 𝐸𝑥 ) · 𝜆𝑥 + 𝐸𝑥 arctan(𝐵𝑥 𝜆𝑥 )]) + 𝑆𝑣𝑥 , (18)

where Fx is the friction force, Svx is the vertical shift, Bx is the stiffness factor, Cx is the shape factor,
Dx is the peak value and Ex is the curvature value. A graphic representation is provided in Figure
1.2.1.
Modeling and control of an electric vehicle 16

Figure 1.2.1: Curve produced by the Magic Formula [2]

The equations for the parameters are complex and may diverge depending on the vehicle,
tires, or road. Therefore, given that the project aims to deliver a generic model with its controller,
a simpler relation between tire and ground will be used. This simpler model is introduced below.

1.2.2 Burckhardt’s model

As the Pacejka model is deemed too complex for the scope of the project since the
necessary values cannot be measured, a simpler model is introduced: the Burckhardt model. The
model is described by

𝜇(𝜆) = 𝑐1 (1 − 𝑒 −𝑐2 𝜆 ) − 𝑐3 𝜆, (19)

where c1 is the maximum value of the friction curve, c2 is the friction curve shape and c3 is the
friction curve difference between the maximum value and the value at λ=1. These coefficients are
represented in Table 1.2.

C1 C2 C3
Asphalt, dry 1.2801 23.99 0.52
Asphalt, wet 0.857 33.822 0.347
Concrete, dry 1.1973 25.168 0.5373
Cobblestone, dry 1.3713 6.4565 0.6691
Cobblestone, wet 0.4004 33.708 0.1204
Snow 0.1946 94.129 0.0646
Ice 0.05 306.39 0

Table 1.2: Burckhardt tire model paràmetres

The virtue of this model is its simplicity, only three parameters are required to define the
relationship between the tire and ground for each surface, although, fidelity to reality will be
inferior. As an example, in Pacejka’s model a slight difference in the camber of the wheel may
affect the curve, this difference would go unnoticed in Burckhardt’s model. Moreover, the
17 Modeling and control of an electric vehicle

differences between the curves are significant even for analog situations. Furthermore, the
Burckhardt model does not follow properly the curvature after the peak value, providing a slight
divergence from reality at slips lager than the optimal.

Given these drawbacks, it would be preferable to use the Pacejka model, but since the
information required to compute its parameters is not available, the generic Burckhardt model
will be enough.

1.3 PROPULSION SYSTEM

As the dynamics of a vehicle have been described, it is time to define how to provide this
motion to the vehicle. The tractive effort will be provided by a motor, but there are multiple
options to choose from.

Since it is an electric vehicle the limitation is clear, it must be an electric machine, but still,
there is a great variety to choose from. The decision can be made considering many factors, such
as power, reliability, cost, availability, or control necessities. The solutions can vary from a simple
DC motor, an induction machine, or a brushless synchronous machine, to name a few, each of
them more complex to control.

As the main objective is not to design a complex motor controller, the simpler motor is
chosen, the DC motor. Its working principle relays on a wire carrying electrical current located
inside a magnetic field, inducing a perpendicular force to both the wire and magnetic field defined
by

𝑇 = 𝐵𝐼𝐿 cos 𝛼, (20)

where T is the torque produced, B is the magnetic field density, I is the current flowing through
the wire, L is the wire length and α is the angle between the coil plane and magnetic field shown
in Figure 1.3.1. The torque is at its maximum when the angle is equal to 0 since the cosine will
equal 1. Therefore, brushes and multiple coils are used to maintain maximum and constant
torque.
Modeling and control of an electric vehicle 18

Figure 1.3.1: DC motor schematic [1]

There are multiple arrangements for DC motors depending on the mutual

interconnections, but in this project, a series excitement arrangement is selected, its equivalent
circuit is shown in Figure 1.3.2.

Figure 1.3.2: Series arrangement equivalent circuit of a DC motor [1]

This arrangement behavior can be described by

𝑑𝑖𝑎
𝑈𝑎 = 𝑅𝑎 𝑖𝑎 + 𝐿𝑎 + 𝐸, (21)
𝑑𝑡

𝐸 = 𝑘𝜙𝑓 𝜔, (22)

𝑇 = 𝑘𝜙𝑓 𝐼𝑎 , (23)

where Ua is the voltage applied to the ends of the motor, Ra is the equivalent resistance of the
circuit and La is the inductance of the coil, E is the back electromotive force, kφf is the emf constant,
and ia is the current flowing through the motor.
19 Modeling and control of an electric vehicle

As the equations are listed, the control problem is presented. For the motor to operate at
a desired torque and speed, the voltage must be controlled, something the battery cannot do
since the voltage supply is constant. Therefore, an element in between the two is required, a
DC/DC converter, which will modulate the voltage fed to the motor.

1.4 ENERGY STORAGE

Once the dynamics and the propulsion have been solved, it is necessary to define the energy
supplier of the vehicle, the battery. As well as the electric machine, the list of solutions for energy
storage is large, therefore a brief introduction to battery theory is presented.

Fist, the first characteristic to describe is the discharge curve. As the battery empties, the
voltage drops, going from the rated open-circuit voltage to the cut-off voltage, represented in
Figure 1.4.1, this curve is found through testing and experimentation and is provided by the
manufacturer. This curve must be monitored at all times to prevent damage to the battery, cutting
the voltage supply when the cell voltage falls below the cut-off voltage.

Figure 1.4.1: Voltage curve of a typical battery [1]

Second, the capacity and discharge rate are described. Capacity represents the amount of
energy the battery can generate at a given discharge rate until the cut-off voltage is reached. The
discharge rate represents the amount of current circulating through the battery and is
represented as a fraction of the rated capacity by the discharging current, the lower it is the higher
the capacity, as shown in Figure 1.4.2.
Modeling and control of an electric vehicle 20

Figure 1.4.2: Discharge curve at different discharge rates [1]

At last, the state of charge is introduced, SOC for short, it is defined as the ratio of remaining
capacity to the fully charged capacity. It is expressed as

𝑖⁡𝑑𝑡
𝑆𝑂𝐶 = 𝑆𝑂𝐶0 − ∫ , (24)
𝑄(𝑖)

where SOC0 is the initial SOC, which will be Q at time 0 if the battery starts fully charged, and Q(i)
is the capacity at a current rate i.
21 Modeling and control of an electric vehicle

2 VEHICLE MODELLING

Once the theory behind vehicle dynamics, propulsion systems and batteries has been
introduced, the next step can be taken, create a model for it.

To create a visual model, the Energetic Macroscopic Representation [3], EMR for short, is
used. Therefore, a brief introduction to the chapter is provided to describe the basics of EMR.

At the end of the chapter the complete model, based on EMR and implemented in Simulink,
is represented in Figure 2.2.1.

2.1 INTRODUCTION

EMR consists of a physic representation of a system with two considerations. First, every
action induces a reaction, as shown in Figure 2.1.1. And second, physical causality is integral and
must be respected, thus the present output of the system can depend on past inputs, but never
on future ones.

Figure 2.1.1: Graphic representation of a reaction induced by an action [3]

Once the considerations have been mentioned, it is due to introduce the elements that
take part in EMR, a small description for each of them is provided.

2.1.1 Energy sources

Their purpose is to define the environment of the system, even if it is an energy generator
or a receptor. An example in Figure 2.1.2.
Modeling and control of an electric vehicle 22

Figure 2.1.2: Graphic representation of both generator and receptor EMR sources. [3]

2.1.2 Energy storage elements

These elements are called accumulators, as the name suggests they accumulate energy,
following the causality principle, and are defined by

𝑦 ∝ ∫ 𝑓( 𝑥1 , 𝑥2 )𝑑𝑡, (25)

where 𝑥1 and 𝑥2 are the inputs in the block and 𝑦 is the output, as shown in Figure 2.1.3.

Figure 2.1.3: Graphic representation of an EMR accumulator [3]

2.1.3 Energy conversion elements

These blocks act as a gain with a slight variation, they can be controlled by a tuning input.
These blocks can be monophysical or multyphysical, depending on the system requirements. The
block follows the expression

𝑦2 = 𝑓(𝑥1 , 𝑧),
(26)
𝑦1 = 𝑓(𝑥2 , 𝑧), (27)
23 Modeling and control of an electric vehicle

where 𝑥1 and 𝑥2 are the inputs of the block, 𝑧 is the tuning vector and 𝑦 is the output, as shown

in Figure 2.1.4.

Figure 2.1.4: Graphic representation of an EMR conversion element [3]

2.1.4 Energy distribution elements

Last, the distribution elements take care of the routing of the signals, being able to divide
it, or distribute it to many blocks, as well as, coupling these signals into a single block.

2.2 ELECTRIC VEHICLE MODEL

As the representation system used has been introduced, the model of the vehicle is
presented.

2.2.1 Battery

As the theory and parameters have been introduced and the complex discharge curve has
been presented, a model is introduced in [4], represented by

𝑉𝑏𝑎𝑡 = 𝐸 − 𝑅𝑖, (28)

𝑄
𝐸 = 𝐸0 − 𝐾 + 𝐴𝑒 −𝐵 ∫ 𝑖𝑑𝑡 , (29)
𝑄 − ∫ 𝑖𝑑𝑡

where R is the internal resistance of a battery cell, E is the no-load voltage, E0 is the battery
constant voltage, K is the polarization voltage, Q is the battery capacity, ∫ 𝑖𝑑𝑡 is the actual charge
of the battery and A and B are experimental values related to the discharge curve.
Modeling and control of an electric vehicle 24

As for the characteristic values of the discharge curve, these are different for each battery
technology, therefore, parameters for a lithium-ion battery cell, used in the vehicle model, with a
rated voltage of 3.6 V, a capacity of 1 Ah and a cut-off voltage of 2.5 V are provided, in Table 2.1.

E0(V) R(Ω) K(V) A(V) B(Ah-1)

3.7348 0.09 0.00876 0.468 3.5294
Table 2.1: Battery cell parameters [4]

These parameters are for an individual cell, therefore, must be scaled up. This is done by
connecting these cells in both series and parallel following the equations

𝑉𝑏𝑎𝑡 = 𝑉𝑐𝑒𝑙𝑙 · 𝑛𝑠𝑒𝑟𝑖𝑒𝑠 , (30)

𝑉𝑏𝑎𝑡⁡𝑟𝑒𝑞.
𝑛𝑠𝑒𝑟𝑖𝑒𝑠 = , (31)
𝑉𝑐𝑒𝑙𝑙

𝑄𝑏𝑎𝑡⁡𝑟𝑒𝑞.
𝑛𝑝𝑎𝑟𝑎𝑙𝑙𝑒𝑙 = , (32)
𝑄𝑐𝑒𝑙𝑙

Since running out of battery in the middle of a simulation is not desired, a greatly oversize
battery is provided, one with 160 cells per row and 100 rows. Through simulation, they could be
trimmed down.

2.2.2 Distributor

As the energy is generated at the battery, it must be distributed to the four DC/DC
converters, one for each wheel. The voltage will be maintained as that outputted by the battery,
but the current flowing towards the battery will be that of the sum of the outputted by the four
DC/DC converters. This is described by

𝑉𝑏𝑎𝑡 = 𝑉𝑓𝑟𝑎 = 𝑉𝑟𝑟𝑎 = 𝑉𝑓𝑙𝑎 = 𝑉𝑟𝑙𝑎 , (33)

𝐼𝑏𝑎𝑡 = 𝐼𝑓𝑟𝑎 + 𝐼𝑟𝑟𝑎 + 𝐼𝑓𝑙𝑎 + 𝐼𝑟𝑙𝑎 , (34)

where Vxy is the voltage fed to the DC/DC converters and Ixy is the current coming from the DC/DC
converters.

2.2.3 DC/DC converter

Once the energy division is done, the next blocks in line are the DC/DC converters, all of
them being equal. The DC/DC converter is modeled as a monophysical conversion element with a
25 Modeling and control of an electric vehicle

control input which converts the fixed DC voltage coming from the battery into the desired DC
voltage required by the controller through the duty cycle. The DC/DC converter is model as

𝑈𝑥𝑦𝑎 = 𝛼𝑥𝑦 𝑉𝑥𝑦𝑎 , (35)

𝐼𝑥𝑦 = 𝛼𝑥𝑦 𝐼𝑥𝑦𝑎 , (36)

where Uxya is the voltage fed to the motor, Ixya is the current coming from the motor, and αxya∈[-
1,1]is the duty cycle and is the control signal used to control the force outputted by the vehicle.

2.2.4 DC motor

At this point, a causal system is encountered, the electric conversion, in which the voltage
differential at the coils of the motors induces a current through it, as described by (21) and (22).
As for the electromechanical conversion, it is a non-causal system described by (23).

The values taken for the simulation are, the rated voltage as 400 V, the rated current as
89.5 A, the back electromagnetic force (kφa) as 1.2396 Vs/rad, the equivalent resistance as 0.35 Ω
and the equivalent inductance as 6.5 mH.

2.2.5 Transmission

The transmission is a simple monophysical conversion block, only in charge of applying

the fixed reduction kgear. It follows the equation

𝑇𝑥𝑦𝑡𝑟𝑎𝑛𝑠 = 𝑘𝑔𝑒𝑎𝑟 𝑇𝑥𝑦𝑚 , (37)

𝜔𝑥𝑦𝑚 = 𝑘𝑔𝑒𝑎𝑟 𝜔𝑥𝑦𝑡𝑟𝑎𝑛𝑠 , (38)

where Txytrans is the torque applied to the wheel and ωxytrans is the angular speed induced by the
wheel. For the transmission modeling, kgear is taken as 5.

2.2.6 Brake distributor

As an electric vehicle can alternate between mechanic braking and regenerative braking
through the electric machine. This will mean two sources of braking torque are available, that of
the ideal brake and that of the motor, where the outputted torque to the wheel can be defined
as

𝑇𝑥𝑦𝑤 = T𝑥𝑦𝑡𝑟𝑎𝑛𝑠 + 𝑇𝑥𝑦𝑏𝑟𝑎𝑘𝑒 , (39)

Modeling and control of an electric vehicle 26

where Txybrake is the torque provided by the brake, which can only be negative.

2.2.7 Wheel

• Ideal wheel

Although the explanation on tire-road dynamics in chapter one was done with the non-
linear and more complex theory, described by the Pacejka and Burckhardt models, a simpler and
easier to control model is proposed to extract initial simulations.

The rigid model of an ideal wheel does not consider the slip of the tire, and understands
de phenomenon as a simple conversion, just like the transmission. It follows the equation

𝑇𝑥𝑦𝑡𝑟𝑎𝑛𝑠
𝐹𝑥𝑦𝑥 = , (40)
𝑟𝑤

𝑣𝑥𝑦𝑥
𝜔𝑥𝑦𝑥 = , (41)
𝑟𝑤

where vxyx is the linear velocity at the center of the wheel and Fxyx is the force transmitted from
the wheel to the chassis. For the model, the effective radius is taken as 0.26 m.

• Non-ideal wheel

As the linear, simpler, model is introduced, the non-linear which models the slip of the
wheel is presented. It takes the same inputs and produces the same outputs but is no longer a
simple converter. As inertia is considered it becomes an accumulator. Considering (13), (14), (16),
(17), and (19) the model can be defined.

For the model, the total motor, transmission, and wheel inertia is taken as 0.5 kg/m2.

2.2.8 Union

The last energy distribution element in the model, required to couple all the tractive
efforts coming from the wheels and transmitting the velocity required at each wheel as a reaction.
The behavior is defined as

𝐹𝑒𝑣 = 𝐹𝑓𝑟𝑥 + 𝐹𝑟𝑟𝑥 + 𝐹𝑓𝑙𝑥 + 𝐹𝑟𝑙𝑥 , (42)

𝑣𝑒𝑣 𝐴𝑒𝑣
𝑣𝑓𝑟𝑥 = 𝑣𝑟𝑟𝑥 = (𝑟𝑐𝑢𝑟𝑣 + ), (43)
𝑟𝑐𝑢𝑟𝑣 2

𝑣𝑒𝑣 𝐴𝑒𝑣
𝑣𝑓𝑙𝑥 = 𝑣𝑟𝑙𝑥 = (𝑟𝑐𝑢𝑟𝑣 − ). (44)
𝑟𝑐𝑢𝑟𝑣 2
27 Modeling and control of an electric vehicle

For the model, the track (Aev) of the car is taken as 1.65 m.

2.2.9 Chassis

Once the total force produced by the vehicle is computed at the union block, it is inputted
into an accumulator and the vehicle speed is outputted, as described by (1) in the first chapter.
Mass for the vehicle is taken as 1000 kg.

2.2.10 Environment

At last, the final block in the model, the environment, takes the vehicle speed as an input and
computes the resistive forces for the given speed, as described in (2), (3), and (6), introduced in
the first chapter.

For environment modeling, the following values are provided. The air density is taken as
1.22521 kg/m3, the drag coefficient as 0.35, the frontal area as 2 m2, and the rolling coefficient as
0.017.
Modeling and control of an electric vehicle 28

Figure 2.2.1: EMR based model implemented in Simulink

29 Modeling and control of an electric vehicle

3 VEHICLE CONTROL

As the vehicle model has been described, the introduction to the control system is provided.
This control system is based, in its majority, in a maximum control structure, which consists of
directly inverting the model. The problem is, being a causal system, some blocks will not allow for
direct inversion, therefore these will need an indirect inversion, in other words, a closed-loop
controller. As it has been done for the modeling, an introduction to control theory and the control
for each subsystem is described.

At the end of the chapter, the complete IBC control system implemented in Simulink is
represented in Figure 3.2.4.

3.1 INTRODUCTION

The control system for the project is based on an inversion-based control, IBC for short. As the
name implies, the objective is to invert the system, as shown in Figure 3.1.1 for a generic system.
It can be seen the desired effect is the input to the controller, which computes the right cause
required by the system which will later compute the actual output of the system. Depending on
the time relationship presented by the system, the inversion will vary.

Figure 3.1.1: Inversion-based control general system representation [3]

3.1.1 Time independent relationship

Also known as non-causal systems, they accept the direct inversion of the system. This
system can be a multiple-input one, although only one will be the reference, the rest must be
measured values from the system itself, as shown in Figure 3.1.2.
Modeling and control of an electric vehicle 30

Figure 3.1.2: Time-independent inversion control, (left) single input, (right) múltiple input [3]

If the system is described by the generic expression

𝑛º⁡𝑖𝑛𝑝𝑢𝑡𝑠

𝑦(𝑡) = ∑ 𝑢𝑖 (𝑡), (45)

𝑖=1

where y(t) is the output value and ui(t) are the inputs. Then the direct inversion is described as

𝑛º⁡𝑖𝑛𝑝𝑢𝑡𝑠⁡∉⁡𝑗

𝑢𝑗 (𝑡) = 𝑦𝑟𝑒𝑓 (𝑡) − ∑ 𝑢𝑖 (𝑡), (46)

𝑖=1

where uj(y) is the chosen variable to act on the output y(t), yref(t) is the reference input, and ui(t)
are the disturbances inputs.

3.1.2 Time-dependent relationship

Also known as causal systems, and contrary to the non-causal ones, they cannot be directly
inverted. Therefore, an indirect inversion mechanism, a controller, is required, as shown in Figure
3.1.3. This controller can be one of many options, but the one used for accumulative systems in
the project will be a PI controller due to its simplicity and easy tuning through simulation.
31 Modeling and control of an electric vehicle

Figure 3.1.3: Time-dependent inversion control [3]

If the system is described by the generic expression

𝑦(𝑡) = ∫ 𝑢(𝑡)𝑑𝑡, (47)

where y(t) is the output value and u(t) is the input. Then the indirect inversion is described as

𝑢(𝑡) = 𝐶(𝑡)[𝑦𝑟𝑒𝑓 (𝑡) − 𝑦𝑚𝑒𝑎𝑠 (𝑡)], (48)

where u(y) is the variable to act on the output y(t), yref(t) is the reference input, ymeas(t) is the
measured output desired to control and C(t) is the controller transfer function chosen for the
system.

3.2 ELECTRIC VEHICLE CONTROLLER

As the control theory has been presented the individual controllers for each presented
model system, in reverse order as they were presented, as the signal flows.

3.2.1 Chassis

The system to invert, a causal system, requires an indirect inversion, through a PI

controller, as mentioned before. The model is described by a first-order transfer function, defined
by

1
𝐺(𝑠) = ⁡, (49)
𝐴𝑠 + 𝐵

where A for the chassis being the vehicle mass M and B equals 0.
Modeling and control of an electric vehicle 32

The PI controller has a transfer function

𝐺(𝑠) = 𝐾𝑃 + 𝐾𝐷 𝑠, (50)

where the KP is the proportional part of the controller and KI is the integral part. These values are
easy to compute for a first-order system from a given rise time tr, through the α parameter defined
as

ln⁡(9)
𝛼= , (51)
𝑡𝑟

and from it, the proportional and integral constants are computed as

𝐾𝑃 = 𝛼𝐴, (52)

𝐾𝐼 = 𝛼𝐵. (53)

Therefore, the PI controller will only be proportional, and

ln⁡(9)
𝐾𝑃 = 𝑀𝑒𝑣 , (54)
𝑡𝑟

where tt can be adjusted as a controller performance is wanted. In Figure 3.2.1 the effect of tr is
shown, the time it takes for the output to go from 10% to 90% of the final value is tr. The expression
for the control system is represented by

∗ ∗
𝐹𝑒𝑣 = 𝐾𝑃 [𝑣𝑒𝑣 − 𝑣𝑒𝑣 ] − 𝐹𝑟𝑒𝑠 . (55)

If a driver were in the vehicle acting on the accelerator pedal, the controller would be
the driver himself and would try to behave as close as the proportional controller does.

For the measurements of vev and Fres estimators are used. These estimators emulate the
job of sensors, if the variable is measurable, or act as estimators. For these variables, speed can
be measured, but force cannot, therefore must be estimated, as it is done in the environment
source block.
33 Modeling and control of an electric vehicle

Figure 3.2.1: Speed signal for a step input, (top) tr=0.01 s and (bottom) tr=0.1 s

3.2.2 Union

The union block becomes a divisor when directly inverted, therefore a selector is
introduced to decide how the energy is distributed. It will be treated as a rigid differential like the
transmission in which the force is equally distributed to each wheel. As an addition, the reference
traction force is limited according to the weight distribution at each wheel. Thus,

∗
𝐹𝑓𝑟𝑥 = 0.52 𝐹𝑒𝑣
∗
, ∈ [−𝐹𝑓𝑟,𝑚𝑎𝑥 , 𝐹𝑓𝑟,𝑚𝑎𝑥 ], (56)

∗ ∗
𝐹𝑟𝑟𝑥 = (1 − 0.5)0.5𝐹𝑒𝑣 , ∈ [−𝐹𝑟𝑟,𝑚𝑎𝑥 , 𝐹𝑓𝑟,𝑚𝑎𝑥 ], (57)

∗ ∗
𝐹𝑓𝑙𝑥 = 0.5(1 − 0.5)𝐹𝑒𝑣 , ∈ [−𝐹𝑓𝑙,𝑚𝑎𝑥 , 𝐹𝑓𝑙,𝑚𝑎𝑥 ], (58)

∗ ∗
𝐹𝑟𝑙𝑥 = (1 − 0.5)(1 − 0.5)𝐹𝑒𝑣 , ∈ [−𝐹𝑟𝑙,𝑚𝑎𝑥 , 𝐹𝑟𝑙,𝑚𝑎𝑥 ]. (59)
Modeling and control of an electric vehicle 34

3.2.3 Wheel

• Ideal wheel

For a rigid wheel, a simple direct inversion is enough, since it is non-causal. This means a
division by the effective radius of the wheel will generate the reference torque required at the
transmission. Thus,
∗ ∗
𝑇𝑥𝑦𝑤 = 𝑟𝑤 𝐹𝑥𝑦𝑥 . (60)

• Non-ideal wheel

A much more complex inversion is required for this system, since not only is causal, many
factors to control cannot be accurately measured in all situations. Therefore, many traction
controllers have been proposed, such as the one used for the model, proposed by [5].

The controller used to limit the wheel slip is called Maximum Transmittable Torque Estimator,
MTTE for short. It consists of computing a maximum torque the wheels can be provided to keep
the slip at an optimal point in which the maximum force can be transmitted.

The working principle is greatly influenced by the ease in which many of the variables can be
measured, for instance, the angular speed of the wheel can be measured easily, such as torque
provided by the motor, but the linear speed of the vehicle is not so simple to accurately measure
since it requires a variety of expensive, and not always reliable, sensors.

Therefore, [5] proposes the following control system to avoid the inversion of the µ-λ relation
and the measurement of vehicle speed. According to (1) and (16), the friction force at the contact
point between tire and ground can be expressed as

𝑇 𝐽𝑤 𝑣̇𝑤
𝐹𝑡 = − 2 . (61)
𝑟 𝑟

where T and 𝑣̇𝑤 will be measured from the system.

Once the friction force is estimated, the next step is to define the control input, this will be
the torque, but it will require saturation in case slip becomes too high, this saturation is proposed
to be

𝐽𝑤 𝐽𝑤
𝑇𝑚𝑎𝑥 = (1 + ) 𝐹𝑡 𝑟 − (𝐹 + 𝐹𝑤 + 𝐹𝑟 ), (62)
𝛼𝑀𝑟 2 𝛼𝑀𝑟 𝑔

where α is the relaxation factor defined by

35 Modeling and control of an electric vehicle

(𝐹𝑡 − 𝐹𝑟𝑒𝑠 )⁄
𝑣𝑥̇ 𝑀 .
𝛼= = (63)
𝑣𝑤̇ (𝑇𝑚𝑎𝑥 − 𝐹𝑡 𝑟)
⁄𝐽
𝑤
and is taken as 0.9 for the controller, to allow for the small slip to maximize the friction force.

3.2.4 Brake distributor

Before the transmission, the controller must decide how much braking torque is provided
by the mechanical brake. Thus, a strategy block calculates how much torque the power train
cannot provide and compensates for the lack of braking power by applying the required force. The
expression is like that of the union, where

∗ ∗
𝑇𝑥𝑦𝑡𝑟𝑎𝑛𝑠 = (1 − 𝑘𝑥𝑦,𝑏𝑟𝑎𝑘𝑒) 𝑇𝑥𝑦𝑤 ⁡, (64)

∗ ∗
𝑇𝑥𝑦𝑡𝑟𝑎𝑛𝑠 = 𝑘𝑥𝑦,𝑏𝑟𝑎𝑘𝑒 𝑇𝑥𝑦𝑤 ⁡, (65)

where kxy.brake is computed by a complex system shown in Figure 3.2.2.

Figure 3.2.2: kxy,brake calculator system

3.2.5 Transmission

As well as the ideal wheel, the transmission is a simple conversion system, which can be
directly inverted. Thus,
∗ ∗
𝑇𝑥𝑦𝑚 = 𝑘𝑔𝑒𝑎𝑟 𝑇𝑥𝑦𝑡𝑟𝑎𝑛𝑠 . (66)

3.2.6 DC motor

Once again, a causal system is encountered, not at the electromechanical coupling of the

motor, which is represented as a conversion system, just as the transmission, by

Modeling and control of an electric vehicle 36

∗ ∗
𝐼𝑥𝑦𝑎 = 𝑘𝜙𝑎 𝑇𝑥𝑦𝑚 . (67)

However, the electric conversion is causal, therefore, it requires a PI controller to properly

invert the system. These PI controller parameters are computed as it was done for the chassis
system, thus

ln⁡(9)
𝐾𝑃 = 𝐿 , (68)
𝑡𝑟,𝑒 𝑎

ln⁡(9)
𝐾𝐼 = 𝑅 . (69)
𝑡𝑟,𝑒 𝑎

The expression for the control system is represented by

∗ ∗
𝑈𝑥𝑦𝑎 = [𝐾𝑃 + 𝐾𝐼 𝑠][𝐼𝑥𝑦𝑎 − 𝐼𝑥𝑦𝑎 ] − 𝐸𝑥𝑦 , ∈ [−400⁡𝑉, 400⁡𝑉]⁡ (70)

Unlike the former proportional controller, this PI controller will have a saturation block at
the exit. This paired to an integral controller means high levels of error if the output is saturated
at any point, thus an anti-windup mechanism is implemented in which the output is compared
and passed through a constant gain and subtracted from the integral part of the controller.

Once again, the behavior of the controller for different tr values is given, see Figure 3.2.3.
37 Modeling and control of an electric vehicle

Figure 3.2.3: Voltage signal for a step input, (top) tr=0.001 s and (bottom) tr=0.01 s

3.2.7 DC/DC converter

At last, the DC/DC converter consists of a conversion system which takes the reference
value from the motor and the current battery voltage output, and routed by the divisor, through
the division and a limitation that more voltage cannot be provided than that of the battery, the
duty cycle is generated. Thus,
∗
∗
𝑈𝑥𝑦𝑎
𝛼𝑥𝑦 = , ∈ [−1,1], (71)
𝑉𝑥𝑦𝑎
Modeling and control of an electric vehicle 38

Figure 3.2.4: IBC model implemented in Simulink

39 Modeling and control of an electric vehicle

4 SIMULATION

Once the theory behind modeling and control has been introduced, and the model and
control systems themselves have been presented, the simulation results are provided for the two
variations of the model developed: the rigid wheel model and the complex tire-road interaction
model.
The simulation parameters are summarized in Table 4.1. bellow.
VEHICLE PARAMETERS VALUE UNITS
Vehicle Mass m 1000 kg
Tire Effective Rolling Radius rw 0.26 m
Reduction Gear kgear 5 -
Air Density ρ 1.22521 kg/m3
Frag Coefficient Cd 0.35 -
Frontal Area Af 2 m2
Wheel Friction Coefficient fr 0.017 -
Gravity g 9.81 m/s2
Wheelbase L 2 m
Center of gravity distance to
Longa 1.2 m
the front axel
Center of gravity distance to
Longb 0.8 m
the front axel
Center of gravity height hg 1 m
Axel inertia Jw 0.5 kg·m2
Track Aev 1.65 m

BATTERY PARAMETERS VALUE UNITS

Battery constant voltage E0 3.7348 V
Equivalent resistance R 0.09 Ω
Polarization voltage K 0.00876 V
Exponential zone amplitude A 0.468 V
Exponential zone time
B 3.5294 As-1
constant inverse
Rated voltage Vnom 3.6 V
Rated cut-off voltage Vtall 2.5 V
Rated capacity Qnom 1 Ah

ENVIRONMENT PARAMETERS VALUE UNITS

Inclination Slope 0 º
Wind speed vw 0 m/s

MOTOR PARAMETERS VALUE UNITS

Rated power Prated 32000 W
Rated voltage Vrated 400 V
Modeling and control of an electric vehicle 40

Rated current Irated 89.5 I

Rated speed rpmrated 2840 rpm
Maximum speed rpmmax 6000 Rpm
EMF constant Kφf 1.2396 Vs/rad
Winding resistance Ra 0,35 Ω
Winding inductance La 6.50E-03 H

CONTROLLER PARAMETERS VALUE UNITS

Anti-windup gain Kaw 10 -
Chassis time constant tr_m 0.1 s
Motor wiring time constant tr_e 0.01 s

BATTERY PACK PARAMETERS VALUE UNITS

Series cells ns 160 Units
Parallel cells np 100 Units
Table 4.1: Simulation parameters

4.1 FULL VEHICLE SIMULATION

As it has been introduced in the model description chapter, this variation is a simplification
of the real road-wheel interaction, as the slip is not considered. This allows for multiple tests and
evaluations, such as vehicle performance, battery consumption, range, or weight distribution
analysis to name a few.

4.1.1 Acceleration and constant velocity

The first test that will be carried out is by giving the reference a step value of 15 m/s, this
will generate an acceleration towards the reference value as shown in Figure 4.1.1. This
acceleration is limited by one of two factors, either the road interaction does not allow for proper
acceleration, or the motors themselves are caping the force to protect from overheating, the
limitation is seen in Figure 4.1.2, as the referenced force is greater than the actual outputted force.

To verify the acceleration limitation Figure 4.1.3 is provided. Due to weight distribution,
the front wheels cannot provide as much tractive effort as the rear wheels, losing on acceleration
potential. Even though the front wheels are not producing the wanted force due to low contact
with the ground, the main reason the acceleration is not higher is the limitation of the motors
themselves as mention before, this can be seen as the reference force is capped at 6000 newtons
due to a saturation block introduced in the controller.
41 Modeling and control of an electric vehicle

Velocity [m/s]

Time [s]

Figure 4.1.1: Reference speed and vehicle speed for a step input
Force [N]

Time [s]

Figure 4.1.2: Force reference vs.outputed force

Modeling and control of an electric vehicle 42

Force [N]

Time [s] Time [s]

Figure 4.1.3: Maximum tractive effort vs tractive effort: (left) front wheels, (right) rear wheels

4.1.2 Deceleration

As acceleration has been tested, braking should be too. This will allow to see the proper
behavior of the brake selector, which chooses the braking source to use. The speed profile is
shown in Figure 4.1.4.
Velocity [m/s]

Time [s]

Figure 4.1.4: Speed profile, reference vs. Output

The steep braking requirement means the motor will not be able to produce the braking
torque required, therefore, the mechanic brake will provide the extra torque, as shown in Figure
4.1.5, where it can be seen that for the front wheel a 60% of the braking is done with the mechanic
brake. Again, braking capabilities are limited by the weight distribution, as the maximum tractive
43 Modeling and control of an electric vehicle

effort limits the braking power, shown in Figure 4.1.6. At last, the energy regeneration can be seen
in Figure 4.1.7, where the negative values of current represent an augment of state of charge of
the battery.

Time [s]

Figure 4.1.5: Percentage of braking effort provided by the mechanic brake

Force [N]

Time [s] Time [s]

Figure 4.1.6: Maximum tractive effort vs tractive effort: (left) front wheels, (right) rear wheels
Modeling and control of an electric vehicle 44

Current [A]
State of charge

Time [s]

Figure 4.1.7: Battery current and SOC

4.1.3 Variable speed input

Finally, for more realistic testing of vehicle performance and behavior, speed cycles are
used, such as the NEDC cycle, which emulates the speed profile a car follows while driving through
a city and entering a highway. In Figure 4.1.8 the profile is displayed.

These profiles allow battery sizing to be more accurate, as the car can be simulated for
long distances at realistic duty cycles. Therefore, many cell arrangements can be tested and
determine optimal distribution. Since acceleration performance has been already tested before,
only vehicle speed and battery parameters are to be analyzed.
45 Modeling and control of an electric vehicle

Velocity [m/s]

Time [s]

Figure 4.1.8: Reference speed profile of 4 NEDC cycle concatenated

Looking at the whole model as a black box, the first conclusion it can be extracted is the
inability of the vehicle to follow the profile at high speeds as seen in Figure 4.1.9, which would
mean that a more powerful motor should be tried, one with a higher rated voltage that does not
cap the speed.
Modeling and control of an electric vehicle 46

Velocity [m/s]

Time [s]

Figure 4.1.9: Output peed profile of 4 NEDC cycle concatenated

Next, the NEDC cycle simulation is useful to find out the required battery pack to
withstand the desired range, as seen in Figure 4.1.10. For the tested cell arrangement, the SOC
charge only drops to 96% and Vbat produces around 600 V, much more than the required by the
motor, 400 V. Therefore, if the pack size is reduced to ns=120 and np=10 the behavior becomes
as shown in Figure 4.1.11, where SOC drops to 60% by the end of the simulation. Furthermore, in
both Figures 4.1.10 and 4.1.11, the effects of regenerative braking can be seen, as the current
becomes negative the SOC rises slightly, extending the vehicle range.
47 Modeling and control of an electric vehicle

Voltage [V]
Current [A]
State of charge

Time [s]

Figure 4.1.10: Battery evolution for a ns=160 and np=100 pack

Voltage [V]
Current [A]
State of charge

Time [s]

Figure 4.1.11: Battery evolution for a ns=120 and np=10 pack

Modeling and control of an electric vehicle 48

4.2 TIRE-ROAD INTERACTION SIMULATION

As a real vehicle has a complex interaction between the ground it travels through and its
tire, it is interesting to evaluate how the model behaves as the tire begins to slip.

4.2.1 Dry road acceleration

For the first simulation, a varying torque profile is taken, simulating the driver's desire of
accelerating the vehicle.

As the vehicle starts in standstill some irregularities are presented at the beginning of the
simulation. These irregularities begin with slip rising towards a value of 400, not displayed in Figure
4.2.1 as the desired zone of research, from 0 to 1, would not be appreciated.

After the irregularities associated with lower speeds, the system stabilizes and behaves
properly, limiting the torque when needed to control the slip, as seen in Figure 4.2.1. It must be
said, although the slip is perfectly controlled at around 20% producing the maximum tractive
effort the vehicle is capable of, the acceleration ratio is kept under the desired value of 0.9, at
around 0.2, as shown in Figure 4.2.3.

As for the input torque, in Figure 4.2.2 it can be seen how the controller limits the steps
of the torque reference to provide the required force to prevent excessive slipping of the tire.
Slip, λ
Friction coefficient

Time [s]

Figure 4.2.1: Slip (λ), top, and friction coefficient (µ), bottom, for a constant surface
49 Modeling and control of an electric vehicle

Torque [Nm]

Time [s]

Figure 4.2.2: Reference toque, maximum torque and caped reference torque for a constant surface
Acceleration ratio, α

Time [s]

Figure 4.2.3: Ratio of acceleration for a constant surface

4.2.2 Dry road to wet road transition

As an extension of the previous simulation, a road transition phenomenon is introduced

at 30 seconds to visualize the effects of increasing slip as the surface friction coefficient drops. In
Modeling and control of an electric vehicle 50

Figure 4.2.4 it is seen how the slip increase as the vehicle enters a less grippy wet area but is later
controlled at a value of 15 % to provide maximum friction as the surface changes.

As it was observed in the previous simulation, the torque reference is caped to prevent
excessive slip when the friction coefficient drops, as represented in Figure 4.2.5.

And again, although the slip is controlled, the acceleration ratio is kept under the desired
0.9, although the slip is controlled at the desired value, as shown in Figure 4.2.6.
Slip, λ
Friction coefficient

Time [s]

Figure 4.2.4: Slip (λ), top, and friction coefficient (µ), bottom, for a changing surface
Torque [Nm]
51 Modeling and control of an electric vehicle

Time [s]

Figure 4.2.5: Reference toque, maximum torque and caped reference torque for a changing Surface
Acceleration ratio, α

Time [s]

Figure 4.2.6: Ratio of acceleration for a changing surface

Modeling and control of an electric vehicle 52

5 PLANNING AND BUDGET

5.1 PLANNING

For the project timeline, a Gantt diagram is provided in Figure 5.1.1, where the main activities
realized in the project are plotted in time.

The project started on the 18th of February, as the first meeting with the project director was
conducted, in it the main objectives where proposed and discussed. Once the objectives were
defined the research for the project started.

First, the bibliography was consulted on vehicle dynamics, motor modeling, DC/DC converter
theory, and battery modeling, creating a theoric basis for the model to be developed upon. As
research was still underway, the model was stared, applying all the acquired knowledge from the
literature consulted in previous weeks. Later, as the model development was concluding,
investigation on control theory started, research on PID tuning, anti-windup mechanisms, and
control performance was conducted, and once again, as the research advanced the control system
model did too. Results analysis and validation were conducted afterward.

Once the ideal model was tested, and the report was started, the research for the complex
tire-ground interaction was conducted. Once again, the research and development of the model
were conducted at the same time. As the deadline closed in, the decision to drop the development
of the control system was made and results were taken with the reduced model and the limited
traction control.

Finally, the report was concluded adding the results of the reduced model and the project
conclusions before the deadline.
53 Modeling and control of an electric vehicle

GANTT DIAGRAM
18-febr 9-març 29-març 18-abr 8-maig 28-maig 17-juny
OBJECTIVE SETTING
IDEAL MODEL
INVESTIGATION
MODEL DEVELOPMENT
CONTROL SYSTEM DEVELOPMENT
TESTING
RESULT ANALYSIS
NON-IDEAL MODEL
INVESTIGATION
MODEL DEVELOPMENT
CONTROL SYSTEM DEVELOPMENT
TESTING
RESULT ANALYSIS
REPORT DEVELOPMENT

Figure 5.1.1: Gantt diagram of the project

5.2 BUDGET

In this chapter, the economic evaluation of the project is conducted. This evaluation is
separated into human resources, software, hardware, and energy consumption.

First, for the estimation of human resources cost, assuming the net salary of a junior
engineer being around 23€/h, and assuming a 3.5 hour a workday, from Monday through Friday,
during 18 weeks, as shown in the Gantt diagram in Figure 5.1.1, the total ascends to 7245 €.

Second, the software is considered free, since the educational licenses for the software
used, MatLab and Microsoft Office, are provided by the university.

Third, the hardware consists of only a personal computer, a Matebook pro 2018, listed at
the time at 1500 €. Given the estimated lifespan of the computer at 6000 hours, and assuming
80% of the project duration requiring a computer, the total hardware cost becomes 62 €.

At last, the energy consumption of the project is computed as the hours of computer usage,
248 hours, times the average power consumption, 50 W, and the average cost of kWh of energy,
0.12 €/kWh. Therefore, the cost is 1.5 €.

A table is provided to summarize the cost breakdown in Table 5.1.

CONCEPT UNITARY COST UNITS TOTAL

HUMAN RESOURCES
Engineering hours 23 €/h 315 h 7245 €
HARDWARE
Huawei Matebook Pro 2018 0.25 €/h 248 h 62 €
ENERGY CONSUMPTION
Modeling and control of an electric vehicle 54

Computer consumption 0.12 €/kWh 248 h 1.48 €

TOTAL 7245.48 €
Table 5.1: Economic evaluation
55 Modeling and control of an electric vehicle

6 ENVIRONMENTAL IMPACT

In this chapter, an evaluation of the environmental impact of the project itself is presented
alongside the environmental repercussions of the use of electric vehicles.

As the project objective was to develop a model and a control system for it without
developing a prototype, the project itself does not pose a real impact other than the energy
consumption of the computer. Thus, if the average consumption of the computer is around 50 W
and the time the computer has been used is around 248 hours, the total energy consumption will
be of 12,4 kWh. Taking the factor mix from 2019 of the company ENDESA ENERGÍA, S.A listed in
[6] as 0.27 kg CO2/kWh, the total carbon dioxide produced is 3.348 kg CO2 for the whole project.

Moving past the emissions of running a computer, a brief analysis is made for the
electrification of the vehicle fleet and its benefits and drawbacks considering environmental
impact.

As the electric vehicle emissions when in movement are zero, unlike an internal combustion
engine vehicle, the air quality of the area they are being used in is improved, an effect greatly
appreciated in crowded urban areas.

That being said, electric vehicles cannot be considered zero-emission vehicles, since the
recharge of the vehicle battery is done through the electric grid of the country it is charged in, and
currently, this grid is composed in its majority by fossil fuel burning electric plants, represented by
the factor mix in [6] not being zero. Therefore, the problem is moved from the location the car is
driven through, to the location of the electric plant generating the energy. Furthermore, the
manufacturing of the vehicle generates carbon dioxide and many more pollutants associated with
the mining of materials required, manufacturing process, and assembly of parts. At last, the
lithium-ion battery packs used for most electric vehicles have a great environmental impact as its
production is highly pollutant.

In conclusion, the electric vehicle has the benefit of not polluting the environment it drives
through, like crowded cities, improving the air quality of the area, but it does not mean the vehicle
is zero emissions, as it has been shown the emissions are generated anyway. Therefore,
development on renewable electric plants must be done to reduce the impact of production and
recharging each vehicle, bringing down the total emissions.
Modeling and control of an electric vehicle 56

7 CONCLUSIONS

First, research on the dynamic behavior of the vehicle has been performed, which allowed
modeling the environment interaction with the vehicle through the chassis and tire. Furthermore,
analysis of the motor, battery, and other subcomponents dynamics and behavior, allowed for a
full vehicle model capable of analyzing the interaction between them and the environment earlier
described.

As it has been mentioned throughout the project, two models had to be developed, one
for the whole vehicle with a simple wheel, and a reduced model with the complex tire-ground
interaction. Although the later one was not possible to implement directly into the whole electric
vehicle model, due to simulation problems induced by the controller implementation, it was still
possible to extract value out of it.

For the control system, the conclusions are more complex. As the controller for the whole
vehicle model worked as expected, the reduced model for the complex tire-road interaction had
some problems. These problems involved the interaction between controllers and poor
performance at slow speeds, non-steady-state situations, and following decelerating profiles. All
these cases would make the simulation to go unstable or spark an error message, therefore
invalidating the simulation. However, some simulations could be performed and provided
promising results, as the vehicle slip was controlled at optimal values.

Overall, the objectives set at the start of the project have been met, except for the full
traction control which is limited to specific situations.

FUTURE WORK

As the project comes to an end, a reflection is done on the future work it could be done to
further develop it.

In terms of research, a more robust definition of road-tire interaction could be found, since
the proposed by Pacejka and Burckhardt presents problems with speeds near 0 and non-steady-
state situations generate great error and oscillation. Therefore, a solution could be found in which
the complex wheel is implemented into the whole vehicle model without problems with its
controller.

As for the model, a few changes could be made to improve the value it has to the industry.
Right now, induction machines and permanent magnets brushless machines are dominating the
57 Modeling and control of an electric vehicle

market, thus, it would be interesting to test the vehicle performance with this kind of motor,
although, it would require a change in the control system, augmenting its complexity. As for the
reduced model, the lateral slip should be considered, adding a layer of complexity to the tire-road
interaction.

Furthermore, the mainline of future work is the actual controller, regarding traction control.
Developing a proper traction control system in which low-speed situations or transitioning from
acceleration to braking do not pose a problem should be of utmost priority, as a control system
cannot be tested in a real vehicle if it will not allow it to brake or alter speed as desired.
Modeling and control of an electric vehicle 58

8 BIBLIOGRAPHY

[1] M. Ehsani, Y. Gao, S. E. Gay and A. Emadi, Modern electric, hybrid electric, and fuel cells
vehicles: fundamentals, theory, and design, 2005.

[2] H. B. Pacejka, Tyre and vehicle dynamics, 2002.

[3] A. Bouscayrol, A. Bruyère and P. Delarue, Teaching drive control using Energetic
Macroscopic Representation, 2007.

[4] O. Tremblay, L. A. Dessaint and A.-I. Dekkiche, A generic battery model for the dynamic
simulation of hybrid electric vehicle, 2007.

[5] D. Yin, S. Oh and Y. Hori, A novel traction control for EV based on maximum transmissible
torque estimation, 2009.

[6] g. d. E. Ministerio para la transición ecológica y el reto demográfico, Factores de emisión:

registro de huella de carbono, compensación y proyectos de absorción de dióxodo de
carbono, 2020.