509 2
509 2
509 2
Abstract
In this thesis, various drive cycles, legislative, official real-world and measured, have
been studied and characterized based on their speed and acceleration content. Three ref-
erence vehicles (a City car, a Highway car and a Sport car) were conceptualized after
performance requirements, with data on existing battery electric cars as a frame of ref-
erence. The acceleration performance, energy consumption and efficiency of the power-
train, comprising a traction motor, a power electronic module and a battery, was deter-
mined and analyzed for the various drive cycles. Furthermore, the consequence on ac-
celeration performance, drive cycle fulfilment and energy consumption during re-scaling
of the electric drive system was studied. Moreover, the electromagnetic losses for four
different slot areas were compared, along with the thermal steady state and transient over
load as well as temperature development during drive cycles.
Through comparison between official and measured drive cycles, it was found that
even though the measured cycles reach higher peak acceleration levels for a certain speed
level, on an average they still spend only slightly more time at higher levels of accelera-
tion compared to the official cycles. The resulting cycle average powertrain efficiencies
were fairly similar for both the official and measured cycles, and showed to be slightly
higher for cycles that spend more time at higher speed levels.
During the powertrain sizing regarding torque and power, the acceleration require-
ment turned out to dominate over other requirements such as top speed, and grade levels.
It was found that a down scaling of the electric power train resulted in an energy con-
sumption down to 94% of the original powertrain size.
Finally, the small slot geometry had the highest peak losses during the drive cycles,
however, on a cycle average it had the lowest losses for many cycles. This fact, in combi-
nation with the highest peak torque and lowest material cost, makes it a very interesting
option as an electric vehicle traction motor.
Index Terms: Battery Electric Vehicle, Drive Cycles, Electric Motor, Sizing, Energy
Efficiency, Energy Consumption, Thermal Modelling, Thermal Capacity.
v
vi
Acknowledgements
The financial support through Chalmers Energy Initiative (CEI) is gratefully appreciated.
I would here like to take the opportunity to thank all of those that had a positive
impact during the execution of this work.
Foremost, I want to thank my main supervisor and examiner Torbjörn Thiringer for
his endless commitment to give vital support, feedback, and encouragement, which he did
through all these years. Furthermore, I would like to express my gratitude towards my co-
supervisors: Mikael Alatalo for many fruitful discussions and for sharing his knowledge
and opinions, and Sonja Lundmark for all her help during this project, e.g. by collecting
driving data, and particularly for giving detailed feedback during the finalizing of this
thesis.
Additionally, I am very grateful for all of the friendly colleagues at the division of
Electric Power Engineering, which together contribute to a very pleasant working envi-
ronment. Specifically, I would like to thank my long term room-mates over the years:
Christian Du-bar, Ali Rabiei and Andreas Andersson for their professional insights, as
well as for their kindness and humor.
Finally, I give my warmest thanks to my family whose support after all is the most
essential.
vii
viii
Contents
Abstract v
Acknowledgements vii
Contents ix
1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Previous work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Purpose of the thesis and contributions . . . . . . . . . . . . . . . . . . 2
1.4 List of Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
ix
Contents
3 Road Type Driving Patterns, Road Grade and Daily Driven Distances 25
3.1 Driving patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.1 Legislative cycles . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.1.2 Non-legislative cycles . . . . . . . . . . . . . . . . . . . . . . 30
3.1.3 Driving pattern characterization parameters . . . . . . . . . . . 31
3.2 Road type specification based on speed levels . . . . . . . . . . . . . . 32
3.2.1 Urban . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2.2 Rural . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2.3 Highway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3 Acceleration distributions . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3.1 Urban . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3.2 Rural . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.3.3 Highway driving . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4 Acceleration duration . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.5 Road grade levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.5.1 Measured road grades . . . . . . . . . . . . . . . . . . . . . . 48
3.6 Average daily driving/traveling distance . . . . . . . . . . . . . . . . . 50
x
Contents
xi
Contents
9 Thermal Performance for HighwayII PMSM with Four Different Slot Areas151
9.1 Steady state performance . . . . . . . . . . . . . . . . . . . . . . . . . 151
9.2 Transient over load . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
9.3 Performance during drive cycles . . . . . . . . . . . . . . . . . . . . . 155
References 163
Appendices 187
xii
Chapter 1
Introduction
1.1 Background
One of the major challenges in the global society today is to reduce the negative im-
pacts that road transportation has on the environment due to toxic and green-house-gas
emissions. As a consequence, these type of emissions from vehicles are legally regulated
on national and sometimes regional levels. In order to comply with expected near future
more stringent regulations, vehicle manufacturers are forced to invest in various fuel sav-
ing technologies. This has led to an increased interest in vehicle electrification, foremost
hybrid electric vehicles (HEVs) which can reduce fuel consumption compared to con-
ventional vehicles, but also battery electric vehicles (BEVs). BEVs offer high powertrain
efficiency and no tailpipe emissions, which is why they are so far considered CO2 neu-
tral in the regulations [1]. If charged with electricity that is produced by fossil free and
renewable sources, BEVs have the potential to offer an emission free use phase [2].
Today a large part of the major automotive manufacturers in the world have devel-
oped their own BEV model, and BEV sales have seen increased annual growth rates, as
high as 54%-87% during 2012-2014 [3]. Still, the battery related drawbacks of relatively
short driving range (mainly due to prize constraints) combined with long charging time,
prohibit BEVs from taking up the commercial competition with fuel energized cars on a
large scale just yet [4, 5].
In this light, it becomes important to investigate the effect on energy efficiency as
well as performance that different design choices have, both when it comes to design
of the different components in the powertrain, but also regarding the design of the drive
system as a whole. Another interesting research aspect is to investigate the possibility
to design the drive system according to a specific type of usage, and then to assess the
consequence on energy efficiency and performance. Moreover, due to the often limited
space for drive system components in vehicles, the choice of peak torque versus thermal
capability for a certain electric machine size also becomes highly important.
1
Chapter 1. Introduction
2
1.4. List of Publications
tensive number of existing BEVs. Moreover, a target was to account for the performance
requirements in an adequate way, which brought a need to collect high frequency drive
cycles where also the acceleration was determined using an accelerometer in addition to
just deriving it from a GPS speed signal. An aim was also to investigate the consequence
on acceleration performance, drive cycle fulfillment, and cycle energy consumption due
to powertrain re-scaling. A final goal was to evaluate the effect that various slot sizes
have on the steady state and transient thermal capability of an electric motor when the
external size is unaltered.
The main contributions of this work are:
Some contributions are also made in the following publications, however, their con-
tent is not directly related to the content of this thesis.
3
Chapter 1. Introduction
4
Chapter 2
This chapter deals with basic concepts and what is considered to be necessary informa-
tion for taking part of the rest of the report.
5
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
Battery
VDC M Differential
6
2.2. Vehicle dynamics
keep a constant speed the net resistive force must be exactly matched by the net tractive
force.
7
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
area, air density and vehicle speed. Then, the drag coefficient can be found via (2.2).
In comparison to area, this parameter is often made official and communicated in car
model specifications. Typically the Cd value is in the range 0.25-0.35 in today’s passen-
ger cars [27], yet it may vary between 0.15 for a more streamlined shape up to 0.5 or
higher for open convertibles, off-road vehicles or other rough shaped vehicles. Further-
more, the Cd value will change if the airflow around and through the vehicle is altered
during driving, for instance an open side window may increase the Cd value by about 5
% [27]. During the last few decades the general trend has been decreasing Cd values on
new passenger cars [28], much due to the increased interest in fuel efficiency and emis-
sions. In order not to compromise too much on the design and compartment comfort for
the passengers, most work on aerodynamical drag reduction is likely to be focused on the
Cd value [28] rather than on the area.
8
2.2. Vehicle dynamics
0.025
Ehsani, Cr=0.01(1+vkm/h/160)
0.01
0.005
0 50 100 150 200 250
Vehicle speed (km/h)
Figure 2.2 Rolling resistance coefficient as a function of speed from different literature sources;
Bosch: [29], Wong: [27], Gillespie: [25], Ehsahni: [32] and Guzzella: [31].
9
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
According to [33], based on tire manufacturer data of new tires from 2005, the average
Cr of low speed tires (up to 180 − 190 km/h) is 0.0098, for high speed tires (up to
210 − 240 km/h) it is 0.0101, while for very high speed tires (above 240 km/h) it is
0.0113.
Fg = m g sin(α) (2.4)
where α (rad/s) is the angle between the level road and the horisontal plane as in
rise %grade
α = arctan( ) = arctan( ) (2.5)
run 100
where rise is the vertical rise and run is the horisontal distance. Road slope is often
expressed in terms of % grade, hence this terminology will be used throughout the thesis.
Since the vehicle may be traveling uphill or downhill this force may either be resisting
or contributing to the net tractive force on the vehicle, i.e. it will either be positive or
negative.
From an energy perspective, driving on a non level road will cause buffering and
draining of potential energy in the vehicle. However, since passenger cars are usually
displaced only temporarily over a day or so, from it’s starting position (e.g. at home),
whatever the route traveled the potential energy remains the same when coming back to
the starting point. As with deceleration, a BEV is normally able to recuperate some of
the energy from going downhill.
Table 2.1 Equivalent force for certain acceleration and grade levels.
a (m/s2 ) 1 2 3 4 5 6
Grade (%) 10.3 20.8 32.1 44.7 59.2 77.3
Grade (%) 5 10 15 20 25 30
a (m/s2 ) 0.49 0.98 1.46 1.92 2.38 2.82
10
2.2. Vehicle dynamics
where Facc (N m) is the force required to accelerate the vehicle mass at a certain magni-
tude of acceleration (Facc = m a), see (2.1).
A positive value of Fwheel then strives to accelerate the vehicle, while a negative
value can represent either a regenerative braking force from an electric motor or friction
braking. Finally, if Fwheel (t) = 0 and the friction brake is disengaged, the vehicle is said
to be coasting, that is only Fa , Fr and possibly Fg are acting on the vehicle.
The maximum tractive force on the driving wheels can be limited by either the pow-
ertrain’s maximum force capability or the maximum adhesive capability between tire and
ground that is possible to be applied on the wheel without loosing the grip to the road,
i.e. starting to spin or slide [25] p. 35. The later is limited by the current normal force
on the driving wheels, FN and the coefficient of friction between the tire and the road,
µ [24] as
Fwheel,max = µ FN (2.7)
The normal load on the driving wheels or wheel pair is affected by the weight distri-
bution in the car, hence it varies from car to car, and even from occasion to occasion for
the same car since the loading may vary, and finally by the change in weight distribution
during an acceleration or deceleration, [24, 25].
The friction coefficient depends nonlinearly on the longitudinal tire slip, which is
caused by deformation of the tire during acceleration and decelerations [24]. The slip is
defined as
vcar
slip = (1 − )100 (%) (2.8)
ωr
and it leads to a non unity relation between the car speed, vcar (m/s) and the product of
wheel speed ω (rad/s) and wheel radius r (m), which would otherwise be valid.
Starting from zero slip and friction, the friction coefficient increases with increasing
slip, up to slip values of about 15 to 20 % where the coefficient peaks at values around
0.8 to 1, depending on type of tire and road condition, [24] and [32]. At even higher slip
values, the friction coefficient decreases, but at a lower rate than before. Moreover, high
slip values means that the wheels, hence also the electric machine will spin faster than
calculated directly from the vehicle speed while ignoring the tire slip.
The total consumed energy at the wheel can be found from the time integral of the
power as Z
Ewheel = Pwheel (t) dt (2.10)
During regenerative braking while Fwheel is negative also Pwheel will be negative,
hence the total consumed energy over time will be reduced.
11
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
id Rs Ld ωelLqiq iq Rs Lq ωelLdid
+ +
ωelΨm
ud uq
- -
The dynamic d- and q-axis stator voltage equations as functions of the d- and q-axis
stator currents (id and iq ) are;
did
ud = Ld + Rs id − wel Lq iq (2.11)
dt
diq
uq = Lq + Rs iq + wel Ld id + wel Ψm (2.12)
dt
where Rs is the stator winding resistance, we l is the electrical angular speed (we l =
np wr where wr is the rotor angular speed and np is the number of pole pairs), Ld and
Lq are the d- and q-axis winding inductances, and Ψm is flux linkage related to the
permanent magnet.
When considering electrical steady state, the di/dt-terms may be omitted.
12
2.3. Permanent Magnet Synchronous Machine (PMSM)
where Ψd and Ψq are flux linkage in the d- and q-axis, and K is the scaling constant for
transformation between three phase to two phase space vectors. For amplitude invariant
scaling, K should be set to unity.
The stator inductance relates a change in current with a change in flux linkage (as
Ψ = L i), and for low current levels (i.e. low torque levels) the relation is close to linear,
but at higher current levels the iron becomes magnetically saturated, thus an equally large
increase in current will then only cause a minor increase in the flux linkage (i.e. only a
minor increase of the torque). In order for this effect to be represented in the circuit
diagram, both the d- and the q-axis inductance could be modeled as functions of current.
The saturation also limits the magnet flux linkage, hence this could also be modeled as a
function of current.
The part of the electromagnetic torque production that is caused by the right part of
(2.13) is called reluctance torque. In salient machines, often Ld is smaller than Lq , due
to a higher reluctance of magnetic material compared to iron. Thus, to be able to produce
a positive reluctance torque, the d-axis current must be negative.
Ideally the mechanical output of an electric motor, in terms of torque and power as a
function of speed, can be divided into two main areas of operation; the constant torque
region and the constant power region. In the constant torque region starting from zero
speed, the machine is capable of producing its max torque given that it can be fed by the
same level of max current. As the speed increases, so does the induced voltage, hence
the applied voltage must also increase, until the maximum voltage limit is hit. At this
point the machine is operating at its maximum power limit. The speed level where this
occurs is referred to as base speed. To be able to reach still higher speeds, the effect of
the induced voltage must be decreased. This is done by reducing the flux-linkage in the
d-axis, by utilizing the d-axis current. Therefore, the same level of maximum torque can
no longer be provided. Instead the torque becomes inversely proportional to the speed.
The power, however is ideally kept constant up to the top speed of the motor, hence the
name constant power region.
For a certain machine, the maximum transient low speed torque is often limited by the
maximum converter current, which in turns is set by thermal limitations. The base speed
depends on the maximum available voltage from the voltage source. Naturally both the
current and voltage will affect the maximum available power.
13
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
wire temperature, the resistance increases by about 10 %. This means that, for the same
magnitude of current, the copper losses will increase by the same factor.
Another factor that may increase the resistance during operation is the frequency of
the supply voltage, through the so called skin effect or by the proximity effect. These
effects are however fairly small.
Iron (core) losses:
Iron losses or core losses depend mainly on two phenomena; magnetic hysteresis and
induced eddy currents. The mean losses can be described as
where
kh a hysteresis parameter
f frequency of the flux
Bpk the peak flux density in the B-H hysteresis curve
n depends on Bpk , fr , and steel material (typically 1.6-2.2)
kc an eddy current parameter
The core losses are generally very difficult to estimate correctly. Even with advanced
FEM softwares the error may be quite large. One of the complexities is that, induced
voltages in machines which are fed by switched inverters contain harmonics beside the
base frequency, hence the flux linkage will also contain harmonics that causes excess
core losses. Both the characteristics of the harmonics and their effect in the material are
difficult to predict correctly.
The rotor losses are usually rather small in PMSM machines, and mainly caused by
eddy current losses in the iron core and the magnets, which can be reduced by certain
design choices such as thinner laminations, core material with higher resistivity and by
segmentation of the magnets.
14
2.4. DC-AC Converter loss modeling
Figure 2.4 Steady state model of a PMSM, taking no load losses into account.
15
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
and the average (per switching period) turn-on and turn-off switching losses as
1 Iˆs VDC Kv
Psw. IGBT = fsw E(on+of f ) (2.21)
π Iref Vref
where the following parameters are component dependent (most are extractable from the
semiconductor component data sheet)
VCE0 IGBT threshold voltage of the on-state characteristics, temperature dependent
RCE IGBT on-state resistance, temperature dependent
Eon+of f Energy dissipated during turn-on and turn-off
Iref Reference current, to which switching losses in data sheet correlate
Vref Reference DC voltage, to which switching losses in data sheet correlate
Kv Parameter describing voltage dependency of switching losses, typically 1.3 to 1.4
Due to symmetry in operation, it is enough to model the losses in a single switch, and
to attribute the same power loss in the other switches in order to find the total converter
losses.
According to [38], by utilizing the so called third harmonic injection operation of
the converter, the output amplitude of the AC phase voltage, Ûph , ideally depends on the
present DC voltage, VDC , and the controlled PWM modulation index, ma as
VDC
Ûph = ma √ (2.24)
3
In order to maintain controllability of the current a maximum ma of 0.9 is recom-
mended in [38]. This then sets the limit of the possible output AC voltage relative to the
DC voltage.
16
2.5. Battery modeling
IGBT converter modules are typically designed to withstand specific voltage levels
of around 600 V , 1200 V etc. Then at each voltage level, a number of slightly different
modules are normally available, with various current ratings, such as 200 A, 400 A etc.
Since the losses to a large part depend on the magnitude of current, the current ratings
implies how large temperature rise due to losses that the cooling system is able to handle
without risking overheating of the transistor or diode chips.
Rdis
Ib
+
Rch
VOC Vt
Figure 2.5 Simple battery model, with separate internal resistance for discharge and charge, with
ideal symbolic diodes.
17
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
Normally a lithium ion battery cell has a maximum and minimum allowed terminal
voltage level, and a maximum and minimum allowed current, or C-rate, where a dis-
charge rate of 1C means that the current is such that the battery will be discharged in one
hour. According to [40] the test to determine a battery’s energy content is usually done
for a constant current discharge at a C/3 discharge rate.
With the battery model as in Figure 2.5, the maximum power that can be transferred
2
to the load is Pmax,theoretic = 14 VROC
dis
, however according to [40], for practical reasons
the limit is rather set as
2 VOC 2
Pmax,theoretic = (2.27)
9 Rdis
The output power may also be limited by either a minimum voltage as
VOC − Vmin
Pmax,Vmin = Vmin (2.28)
Rdis
or by a maximum current limit, which may be due to lifetime or thermal issues, as
2.6 Transmission
Automotive gearbox losses generally depend on various operating conditions, where the
main factors are; speed, load level and temperature, resulting in typical vehicle gearbox
efficiencies of 95 − 97 % [31]. According to [41] the losses spring from phenomena that
are both load independent (spring losses; oil churning and air windage) and load depen-
dent (mechanical losses; rolling and sliding), where sliding losses may be the dominat-
ing contributor. The load independent losses cannot easily be modeled accurately with
general analytical expressions. Instead experimental results are required in order to de-
velop empirical loss models whose validity naturally will be rather limited. A number of
these types of models have been suggested by various researchers. For the load depen-
dent losses, physical expressions can be utilized in the loss modeling, however accurate
parameter estimation can still be difficult.
In a BEV the transmission is typically of single-speed spur gear type, which accord-
ing to [42] and [43] can be assumed to have an efficiency of 95%, in energy consumption
assessments.
18
2.7. Auxiliary loads
19
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
external factor such as a fan, pump, or even wind, it is called forced convection, otherwise
it is called natural convection. The convection heat transfer rate qconv (W ) is proportional
to: the surface area A (m2 ) in contact with the fluid medium; the temperature difference
between the surface Tsurf (K) and the fluid Tf l (K); and the convection heat transfer
coefficient h (W/m2 K), as in
For electric machines, natural convection (or free convection) occurs between the
housing/exterior and the surrounding air, where fins can be used to enlarge the surface
area thus enhance the heat transfer rate. In order to improve the heat transfer rate even
further, forced convection can be utilized, often realized by a fan, e.g. attached to the
shaft on one side of the machine. For highly loaded machines, liquid cooling can be used
where the cooling media can be a water or oil mixture, and the cooling ducts can be
placed inside the frame, or sometimes even inside the machine close to the windings.
Convection heat transfer is also present in the machine’s air gap, as well as in the
internal air between the housing endcaps and the lamination core, where the rotor’s rota-
tions cause the air to move [49].
The convection heat transfer coefficient depend on many factors such as: the surface
geometry; the nature of fluid motion; the thermodynamic and transport properties of the
fluid; and on the flow rate. Hence, it has to be experimentally or empirically determined
[22, 45, 48]. As a rough guidance, there are published ranges of typical values of the
heat transfer coefficient depending on the type of convection and media. By combining
the range data in [45, 48, 50, 51], the heat transfer coefficient for natural convection is
typically 2 − 25 W/m2 K for gases, and 10 − 1000 W/m2 K for liquids, whereas for
forced convection it is typically 10 − 300 W/m2 K for gases and 50 − 20, 000 W/m2 K
for liquids.
In practice, the seeking of a convection heat transfer coefficient is often replaced by
the seeking of the dimensionless Nusselt number N u, which is the ratio of the convective
and conductive heat rates. It is thus proportional to the heat transfer coefficient and the
characteristic length (or fluid layer) lc , and inversely proportional to the thermal conduc-
tivity, as in [48]
qconv hA∆T hlc
Nu = = = = f (geometry, Re, P r) (2.32)
qcond λA∆T /lthick λ
The Nusselt number depends on the specific geometry as well as on the two dimen-
sionless parameters: the Prandtl number P r, and the Reynolds number Re [45]. The
Prandtl number is the ratio of a fluids momentum to heat diffusivity, and depend only
on a medium’s material parameters. It is proportional to the fluid specific heat capacity
cp (J/kgK) and to the dynamic viscosity µ (kg/m s) and inversely proportional to the
thermal heat conductivity [45, 48], as in
cp µ
Pr = (2.33)
λ
For P r 1 conductive heat transfer dominates, and for P r 1 convective heat transfer
dominates. According to [48], Pr is typically 0.12-1 for gases, 1.12-13.7 for water, and
20
2.8. Heat transfer modeling in electric machines
50-100,000 for oils. There is also a temperature dependence of all of the three material
parameters.
The Reynolds number is the ratio of the fluid inertia to the viscous forces, and is pro-
portional to the mass density ρ (kg/m3 ), the fluid velocity v (m/s), and the characteristic
length lc (m), and inversely proportional to the dynamic viscosity [45, 48], as in
ρvlc
Re = (2.34)
µ
For high Re the flow easily becomes turbulent, and for low Re the turbulence is more
damped and may even give a laminar flow. Since a high Re usually gives a higher Nu,
a turbulent flow thus gives a higher convection heat transfer coefficient. The transitions
from laminar to turbulent flow may occur at a specific critical Re value. This value depend
on many factors, such as the type of fluid medium, surface geometry and roughness and
temperature of medium and surface, hence it is difficult to predict analytically. However,
historically conducted experiments for a range of often used geometries and fluids can
offer reasonable estimation methods and some times approximated values.
Thermal radiation is heat transfer in the form of electromagnetic waves or photons
that are emitted by a surface to its colder surroundings or to another surface [45, 48]. The
radiated heat transfer rate qrad (W ) is proportional to the surface emissivity ε, surface area
A (m2 ), and difference between surface temperature Tsurf and surrounding temperature
Tsurr (K) both to the power of four, as in (Stefan-Boltzman’s equation)
4 4
qrad = εσA(Tsurf − Tsurr ) (2.35)
where σ = 5.670 10−8 (W/m2 K 4 ) is Stefan-Boltzmann’s constant [48]. The maximum
rate of radiation that can be emitted is that of a black body which has an emissivity equal
to one. For other surfaces the emissivity is a measure of the effectiveness of radiation
compared to that of a black body, and range between 0-1 [45, 48]. When forced convec-
tion cooling is used the radiation is often assumed negligible [48, 52]. It is also possible
to define a heat transfer coefficient hrad for radiation and form an equation similar to
(2.31) [45, 48], as in
21
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
With the lumped-parameter method, execution is fast, and a wide range of load points
e.g. vehicle load cycles, can be analyzed in a relatively short period of time. The primary
effort lies in forming a relevant network where the main heat transfer paths are repre-
sented, and to find suitable estimates of the heat transfer rates. With FEA and CFD soft-
ware, highly detailed and complex machine geometries can be analyzed, however both
model set-up and load point execution can be time consuming. CFD software is espe-
cially competent in analyzing convection problems. Thermal FEA software programs are
usually limited to improving the accuracy of conduction estimation, whereas for convec-
tion and radiation the same approximations must be used as those in a lumped-parameter
network.
In a lumped-parameter thermal model, bulk regions of the electric machine are lumped
together where it is reasonable to assume spatially uniform material properties, ther-
mal energy storage capacity, internal heat generation, and thus temperature. Different
lumped regions are then interconnected using thermal impedances which are primarily
determined by geometrical and material properties. In this way, the thermal network is
analogous to an electrical network, but with thermal resistances and heat energy stor-
ing capacitances instead, where the heat transfer rate has the same role as current, and
temperature the role of voltage.
In thermal steady state, the temperature T difference between the two adjacent nodes
i and j (in an n-node system) can be described by the relation [49]
Ti − Tj = G−1 P (2.38)
where P is a vector that represent internal heat sources i.e. machine losses in selected
nodes, and is defined as
P1
P2
P= . (2.39)
..
Pn
and G is the thermal conductance matrix defined as
X n
1
− R11,2 ... − R1,n1
i=1 R1,i
n
X 1
− 1 1
R2,1 ... − R2,n
G= i=1
R 2,i
(2.40)
.. .. .. ..
. . . .
n
X 1
1 1
− Rn,1 − Rn,2
...
i=1
R n,i
The net thermal resistance Rth (K/W ) of a region due to heat conduction is propor-
tional to the temperature difference over the region and inversely proportional to the heat
transfer rate as in
1 ∆T l
Rth = = = (2.41)
G qcond λA
22
2.8. Heat transfer modeling in electric machines
For a region with a varying area perpendicular to the direction of heat flow, the net ther-
mal resistance instead becomes a line integral as in
Z l
1
Rth = dx (2.42)
0 λA(x)
Due to the typically cylindrical shape of rotating electric machines, a few specific
geometries are often modeled, such as a hollow cylinder and a segment of a hollow
cylinder as in Figure 2.6. For radial heat transfer in a segment of a hollow cylinder, the
thermal resistance, as resulting from (2.42), is
ln( rrout
in
)
Rth = (2.43)
φλlthick
where rout (m) and rin (m) are the outer and inner radii, φ (rad) is the segment angle, and
lthick is the segment thickness. For a whole hollow cylinder, the segment angle is 2π.
r out
lthick
r in
A
Figure 2.6 Segment of a hollow cylinder, for calculation of radially thermal conductive material.
23
Chapter 2. BEV Dynamics, Powertrain Component Modeling, and Heat Transfer
Modelling
For transient thermal analysis, the node increase in temperature from a reference value,
can be found from [49] [53]
dT
= C−1 (P − GT) (2.49)
dt
or, expressed differently as in
dT
C + GT = P (2.50)
dt
24
Chapter 3
For a successful vehicle design, as with the design of any product, knowledge of the use
phase is essential. Vehicles are used in various environments, by different types of drivers
and for numerous purposes. Each of these circumstances put their specific capability
requirements on the vehicle in terms of static and dynamic road load levels.
Through the years much research has been conducted, especially in Europe and the
US, aiming to identify typical driving patterns on different road types. The main reason
has then been to assess in which way, and to what degree, vehicle pollutant emissions
and fuel consumption are effected by different driving patterns and situations.
This chapter attempts to identify typical levels of speed, acceleration and road
grade attributed to different road types, such as urban (or city) driving and high speed
motorway driving, for the purpose of finding suitable BEV powertrain design criteria
regarding torque speed and power. Additionally, typical daily driving distances will be
investigated, since range is an important design factor for a BEV.
25
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
hicles, i.e. passenger cars, as a part of the type approval procedure. Then legally regulated
emissions as well as fuel economy/efficiency are measured while the vehicle is driven ac-
cording to a reference speed over time: a so called drive cycle. In order to make sure that
the legally set emission targets are not exceeded in typical real-world traffic, it is highly
desired that the laboratory test fallout is fairly close to that. Another important outcome of
the test is that they represent a fair estimation of fuel economy/efficiency for customers.
However as recognized by EPA most test drive cycles were developed a few decades
ago, and both legal speed limits and vehicle power specifications thus performance have
increased since, [7], the later which has also been noted in Sweden by [57]. It was to
account for this that EPA updated the fuel economy test procedure in 2008 to also include
two more aggressive cycles.
In recent years, several extensive studies have been conducted all around the world,
targeting investigation and identification of typical driving patterns and their effect on
fuel consumption and emissions. As a fact, apart from the added cycles, EPA has also
considered such studies in order to find other causes of fuel consumption during real-
world driving, such as fuel energy density, wind, tire pressure and road roughness [7].
But instead of expanding the test procedure to include such factors, EPA has developed
an intricate formula where the city and highway fuel economy label data are calculated
based on certain weighting factors on sections of each test cycle.
So far, the test procedures in Europe are the same since the year 2000. Yet, several
large projects have been conducted where driving data has been collected by using instru-
mented vehicles, and corresponding drive cycles have been developed. The most known
studies are the INRETS, HYZEM, ARTEMIS and the latest is the WLTC.
Due to the large interest in gathering information of local driving patterns, a wide
range of real-world cycles have been developed to capture the specific circumstances of
driving on a certain type of road, in a particular city, or region of the world. Cycles of this
type can be found from regions such as; (Pune) India, China, Hong Kong, Tehran Iran,
(Melbourne, Perth) Australia, Manila Filipines, Edingburgh UK, Latvia, Athens Greece
and Vietnam.
In the Chalmers initiated, The Swedish Car Movement Data Project, 714 cars have
been instrumented during the years 2010-2012, in the south west of Sweden and the data
is gathered in a data base [58]. As a researcher, access to the data base can be granted,
however it has not been used within the scope of this project.
Instead, in this project sample in-house-measurements have been conducted, mostly
covering the area in and around Göteborg, but also longer highway sections, e.g. be-
tween Göteborg and Jönköping and Torslanda-Stenungsund. Data regarding speed and
acceleration has been measured with a GPS and an accelerometer, see Appendix B.
26
3.1. Driving patterns
in procedures may be; the used drive cycle/-s, or at which ambient temperature the test
is run. European and Japanese test procedures include only one specified drive cycle
(NEDC) and (JC08) for both test procedures including BEVs, while the US test proce-
dures utilize five dissimilar drive cycles, each aiming to represent various types of driving
on different road types.
Here follows a short review of the history of the development of the legislative drive
cycles in the US, Europe and Japan, many of which are used in this thesis.
27
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
chassis dynamometer tests, however, these modifications were considered to have only
minor influence since they were limited to the fist ten and last twenty seconds of the
cycle [59].
According to a study by CARB in 1990, large accelerations could have dramatic ef-
fects on emission levels, but these events were missing in the FTP cycle [61]. As EPA
found a lack of data on the occurrence frequency of such accelerations during normal
driving, they initiated a new study. Hence in 1992 EPA together with CARB and two
manufacturer organizations1 conducted a vast real-world-driving study based on 100-150
instrumented vehicles in each of the three cities, Baltimore, Spokane and Atlanta [61].
Data was also recorded using the "chase car" method on routes in Baltimore, Spokane
and Los Angeles [61]. The recorded driving data showed that during almost 13 % of the
time, the vehicle was experiencing higher speed and acceleration levels than what was
represented in the FTP cycle [61]. Based on the measured data from Baltimore together
with the collected data from Los Angeles, Sierra Research which was contracted by EPA,
divided the time traces into micro-trips, i.e. cutting the cycle at each stand still. A soft-
ware was then developed which sampled different micro-trips together to form a drive
cycle, according to certain cycle targets. In this way thousands of cycles were created.
Finally those few with the best match to targeted speed-acceleration-distribution were
chosen to be continued with [61]. One of the most famous ones is the REP05 which
represents aggressive highway driving where about 70 % of the time is spend at higher
speed and acceleration levels compared to the FTP cycle [61]. Other known cycles are
the ARB02 and LA92 also called the Unified Cycle (UC).
Later in the mid 1990s the EPA emission test procedure was updated, the so called
Supplemental Federal Test Procedure (SFTP), to also include two new cycles which rep-
resents more aggressive driving and rapid speed fluctuations. One of them was the SC03,
which is a low speed cycle, but contains rapid speed fluctuations. The other one was
the US06 cycle, which is developed as a shortened combination of the REP05 and the
ARB02 cycles, representing high speed and acceleration levels.
Today EPA’s test procedure for emissions comprise of the FTP-75, SC03, UDDS
(FTP-72), US06 and the LA92 cycles, regulated under T ier1 − 3. For fuel economy
and CO2 emission labeling (regulated since 2012 on a fleet average basis), the FTP and
HWFET were the only cycles used until vehicle model year 2008, when also the US06,
SC03 (run at 35 ◦ C) and a cold version of the FTP (−6.7 ◦ C, instead of 23.9 ◦ C as for
HWFET and hot FTP) were added.
When it comes to certifying driving range and fuel economy of BEVs, either the FTP
and HWFET cycles can be used, or the five cycles used for conventional vehicle fuel
economy can be used, until EPA’s rules are updated for BEVs [62].
Another cycle which has been developed by the EPA, is the New York City Cycle
(NYCC), which represents congested traffic in a large city. Even though it is not utilized
in any regulatory test procedure, it can be seen in many published articles regarding
Hybrid Electric Vehicles (HEV) and Plug-In Hybrid Electric Vehicles (PHEV).
1 The American Automobile Manufacturers Association (AAMA) and the Association of International Au-
28
3.1. Driving patterns
29
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
3.1.2.1 Artemis
The Artemis project involved 40 European research laboratories and was founded by the
European Commission, to develop new European methods, tools, models and databases
for accurate estimation of pollutant emissions from transport [66]. Within the project, a
set of real world driving cycles for passenger cars were developed, categorized as urban,
rural and motorway, each with sub-cycles representing various driving conditions such
as traffic density.
The Artemis cycles are based on data from 77 instrumented vehicles in France,
UK, Germany and Greece collected in 1990s in the previous research projects DRIVE-
MODEM and HyZem, each in which a number of cycles were also developed (in the
categories: urban, congested urban, road and motorway) [66].
The Artemis Urban cycle has sub-cycles of congested, dense and free flow traffic,
where the speed level differs between about 10 − 60 km/h as well as the number of and
duration of stops. The Rural cycle has sections of steady and unsteady speeds at rural
secondary to main roads at speed levels around 50 − 100 km/h. Finally the Motorway
cycle also has sub-cycles of steady and unsteady speed at speed levels between about
90 − 150 km/h [66]. There is also a modified version of the Motorway cycle, which has
a top speed of about 130 km/h.
3.1.2.2 WLTC
Since 2008, the United Nations Economic Commission for Europe (UN-ECE) are presently
working on finalizing a proposal for a new Worldwide harmonized Light duty driving
Test Cycle (WLTC), which can be used for type approval fuel consumption and emission
tests. The cycle is aimed to represent typical driving on a global scale. Vehicle data has
been collected from instrumented vehicles in USA, Japan, India, Korea and the European
countries Belgium, Germany, Spain, France, Italy, Poland, Slovenia, Switzerland, United
Kingdom and Sweden. The logged data has then been chopped into micro trips which in
turn have been categorized for further analysis [67].
Three classes of drive cycles have been developed, where Class 1 is to be used by ve-
hicles with power-to-mass ratio of ≤ 22 W/kg, Class 2 by vehicles with power to mass
1 Australian emission regulations are based on European regulations but with some additional selected parts
30
3.1. Driving patterns
ratios > 22 and ≤ 34 W/kg, and finally Class 3 by vehicles with ratios > 34 W/kg
[67]. Each class consist of 2 to 4 cycles called: Low, Middle, High, ExtraHigh, which are
refereing to speed levels. They have different characteristics such as speed and accelera-
tion levels, and cover scenarios like congestion or free flow traffic.
31
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
spent standing still, and in certain speed levels have been used, as well as relative time
spent during positive and negative acceleration, and within certain acceleration level in-
tervals. Finally an oscillatory measure is also used; Relative Positive Acceleration (RPA),
defined as R
speed(t) positive acceleration(t) dt
RP A = R . (3.1)
speed(t) dt = total driven distance
RPA is an indicator of accelerations which demand high power, since a large acceleration
may demand a high torque but not necessarily a high power, unless it is combined with a
high speed [8] [6].
Table 3.1 General categorization of cycles based on time share at speed levels.
60 − 90 km/h > 90 km/h
Urban: < 20 % of the time ≈ 0 % of the time
Rural: > 20 % of the time < 20 % of the time
Highway: - > 20 % of the time
3.2.1 Urban
The time series of speed and acceleration for the Urban Test cycles can be seen in Figure
3.1, and for the Logged Urban cycles in Figure 3.2. Three logged urban cycles were
excluded from the figure, since two of them represent already displayed routes, and one
was similar to the others.
Cycle data regarding time duration, driven distance and speed levels can be seen in
Table 3.2, for both Test cycles and Logged cycles. In order to account for non zero mea-
sured speed from the GPS, the average running speed as well as the time share standing
are based on speed levels at 3 km/h and below, for all cycles, i.e. including the test
cycles.
Regarding the time shares within certain speed intervals, the selected logs match the
test cycles quite well. The main difference is a longer mean time share standing for the
32
3.2. Road type specification based on speed levels
test cycles (27 % compared to 17 %), and thus a shorter time share at low speed, compared
to the logs. Most certainly some of the test cycles represent more dense traffic conditions
than what was experienced in the Logged cycles.
For the test cycles, the maximum speed varies between 45 km/h for the NYCC cycle
to 91 km/h for the FTP cycles, while the difference is smaller for the logged cycles; 49
for the V774 120312 log to 78 km/h for the Prius 120403 log. Still, the logged cycles
have a somewhat higher mean average speed value compared to the test cycles, while
the average running speed is rather similar. The spread in speed levels seen by the speed
standard deviation is a bit larger for the test cycles, especially for those with the highest
top speed.
50 50
0 0
FTP75 ECE
100 100
Speed , Acc.
50 50
0 0
50 50
0 0
0 200 400 600 800 1000 0 100 200 300 400 500
50 50
0 0
50
0 500 1000
Time (s)
Figure 3.1 Acceleration and speed over time for Urban test cycles, [68], [64], [1] and [69].
33
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
50 50
0 0
−50 −50
0 500 1000 1500 2000 2500 0 500 1000 1500
50 50
0 0
−50 −50
0 200 400 600 800 0 200 400 600 800
50 50
0 0
−50 −50
0 500 1000 0 100 200 300
50 50
0 0
−50 −50
0 200 400 600 0 500 1000
Time (s) Time (s)
Figure 3.2 Acceleration and speed over time for Logged Urban cycles.
34
3.2. Road type specification based on speed levels
Table 3.2 Cycle data for Urban Test cycles, and Logged Urban cycles.
!
#$% $%&'% ( '%
" " " "
! " ##$
! "
$
%&
'( )
) * '+, , -$. ,- /$ 00 /% /, $$ , %
"
" " "
! $%& %&'(& ) (&
# # # #
! "#$%&
' # ($ %$ ($ $
' # ($ $ ($ %$
' # ($ $
) ' ) ! " $!*
)
) $$ +
) $$ +
, &
* + , ,,, - .,% %/ 0( ./ ,1 ,1 1% 1 &
3.2.2 Rural
The time series of speed and acceleration for the Rural Test cycles can be seen in Figure
3.3, and for the Logged Rural cycles in Figure 3.4. One logged cycle was excluded from
the figure, since the same route was already displayed in the opposite direction. How-
ever the graphs show the same route at three different occasions, which illustrates the
deviation between different runs on the same route.
Cycle data for both Logged Rural and Rural Test cycles can be seen in Table 3.3.
Also in this case the mean time share standing is larger for the test cycles compared to
the logs (13 % respective 9 %). Then the time share on medium to high speed levels are
somewhat higher for the test cycles compared to the logs. This is also reflected in the
maximum speed levels which are larger for the test cycles. In the same time, the mean
average speed as well as the average running speed are very similar between the test
and logged cycles. The UCLA has the lowest average speed of only 39 km/h, while the
EUDC has the highest at 63 km/h, followed by the Artemis Rural and WLTC High. It is
also EUDC and NEDC that have the highest maximum speed levels. For the logs there
are four cycles with maximum speed levels around 100 to 110 km/h. It can be seen that a
high maximum speed is not necessarily linked with a high average speed, instead cycles
with lower max. speed may have similar level of average speed.
35
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
UC LA92 NEDC
120 120
(km/h)
80 80 2
Speed , Acc.
10 * (m/s )
40 40
0 0
−40 −40
0 500 1000 0 200 400 600 800 1000
80
Speed , Acc.
80
40 40
0 0
−40 −40
0 200 400 600 800 1000 0 100 200 300 400
Time (s)
Figure 3.3 Acceleration and speed over time for Rural test cycles, [68], [64], [1] and [69].
80 80
40 40
0 0
−40 −40
0 200 400 600 800 1000 0 500 1000 1500
Ecar500 120413 CTH−Kllrd C30d2 121219 Stnsnd−Henan
120 120
Speed , Acc.
80 80
40 40
0 0
−40 −40
0 500 1000 1500 2000 0 500 1000 1500
V744 120307 VCC−CTH V744 120316 Klltrp−Cstat
120 (km/h) 120
10 * (m/s2)
Speed , Acc.
80 80
40 40
0 0
−40 −40
0 200 400 600 800 1000 0 200 400 600
V744 120320 Klltrp−Csttn V744 120327 Klltrp−Cstat
120 120
Speed , Acc.
80 80
40 40
0 0
−40 −40
0 100 200 300 400 500 0 100 200 300 400 500
Time (s) Time (s)
Figure 3.4 Acceleration and speed over time for Logged Rural cycles.
36
3.2. Road type specification based on speed levels
Table 3.3 Cycle data for Rural Test cycles, and Logged Rural cycles.
!
( ) &*$ ** #++ *** ,$ ,# -$ *- .# +/ *+
!
! ! !
#$% $%&'% ( '%
" " " "
3.2.3 Highway
The time series of speed and acceleration for the Highway Test cycles can be seen in
Figure 3.5, and for the Logged Highway cycles in Figure 3.6. Cycle data regarding time
duration, driven distance and speed levels can be seen in Table 3.4, for both Logged and
Test cycles.
Here it can be seen that the time share standing is only a few percent for both test
and logged cycles, while the time shares at high speed are in general higher for the
test cycles (58 %) compared to the logged (41 %). The max. speed levels are higher for
the test cycles with one exception; the HWFET cycle. Consequently, also the average
speed levels are higher for the test cycles, with the Artemis cycles at the highest levels.
The highest average speed is logged for the V70 120515 cycle, followed by the C30d2
121219 Vcc-Stnsnd log.
37
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
50 50
0 0
−35 −35
0 200 400 600 0 200 400 600
100 100
Speed , Acc.
50 50
0 0
−35 −35
0 500 1000 0 200 400 600 800 1000
100 100
Speed , Acc.
50 50
0 0
−35 −35
0 200 400 600 800 1000 0 100 200 300
Time (s) Time (s)
Figure 3.5 Acceleration and speed over time for Highway cycles, [68], [64] and [69].
40 40
0 0
−40 −40
0 1000 2000 3000 0 500 1000 1500 2000
80 80
Speed , Acc.
40 40
0 0
−40 −40
0 2000 4000 6000 0 2000 4000 6000
Time (s) Time (s)
Figure 3.6 Acceleration and speed over time for Logged Highway cycles.
38
3.3. Acceleration distributions
Table 3.4 Cycle data for Highway test cycles, and logged Highway cycles.
!
#$% $%&'% ( '%
" " " "
! "# $
) * +,- .- /$0 -.+ ++ '% // / -, .. 1+
"
" " "
!
$%& %&'(& ) (&
# # #
#
3.3.1 Urban
Cycle data regarding acceleration levels can be seen in Table 3.5, for Urban Logged and
Test cycles. It can be seen that the time share of positive and negative acceleration are
similar for both the test and logged cycles, where they are shorter for the test cycles due
to the longer time share standing. Regarding the time shares at certain acceleration levels,
the mean values are very similar for the test and the logged cycles.
The main difference between test cycles and the logged cycles are the much larger
maximum acceleration levels seen in the logs, both positive and negative. Amongst the
test cycles, it is the NYCC and Artemis Urban that have the highest levels of peak accel-
eration, at 2.7 m/s2 and 2.4 m/s2 respectively, while many of the logs have maximum
values over 3 m/s2 . Still the average positive and negative acceleration levels are similar
39
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
for all cycles, while the standard deviations are slightly higher for the logs. Another in-
teresting point is that ECE has the lowest maximum acceleration level between the test
cycles, yet in the same time it has the highest average acceleration, both positive and
negative.
Finally the mean RPA values are similar for the test and logged cycles. Many test cy-
cles have RPA values around 0.17 to 0.20 m/s2 . However two cycles stand out; Artemis
Urban and NYCC, with RPA values of 0.30 and 0.29 respectively. The spread in RPA
values seem to be larger for the logged cycles. The largest values are seen for the V744
120312 (0.37) and XC60 .0.32 m/s2 .
Table 3.5 Acceleration cycle data for Urban test and Logged cycles.
! ! ! ! ! !
#$ $# # # #% &%
" " " " " "
!
" # $$%
" #
%
&'
() *
' ( $) *$ + ,- *, . ,/ ,/ , $ %) %- 0- - , ,
! ! ! ! ! !
" " " " " "
$% %$ $ $ $& '&
# # # # # #
! "#$%& '
( # )$ %$ )$ $ '
( # )$ $ )$ %$ '
( # )$ $ '
* ( * ! " $!+
* ' ' ''
* ' $$ , '
* $$ ,
- & ' ' '
( ) &% *+ % ,- *, - ,- ,. , + +/ -% /& . % ,
The maximum acceleration as well as second based operating points, for the Test
cycles can be seen in Figure 3.7 and for the Logged cycles in Figure 3.8. As can be seen,
the largest acceleration levels are experienced at the lower speed levels.
For the test cycles, at speed levels up to around 40 km/h Artemis Urban, NYCC and
SC03 have the highest levels of positive and negative acceleration at 1.7 − 2.5 m/s2 and
−2.3 to −2.8 m/s2 . At higher speeds it is SC03, FTP75 and WLTC Middle that have the
highest positive acceleration levels, although strongly descending with increasing speed.
At about 50 km/h the peak acceleration levels are around 0.4 − 1.4 m/s2 , while at about
80 km/h they are around 0.2 − 0.5 m/s2 .
As already noted from the cycle parameter data, the peak acceleration levels of the
logged cycles are higher compared to the test cycles, also when considering the different
speed levels. For the logged cycles, at about 50 km/h the peak acceleration levels are
around 1 − 2.5 m/s2 , while at about 80 km/h the cycles have already reached their top
speed.
40
3.3. Acceleration distributions
3
Artemis URBAN
NYCC
2 WLTC Cl.3 Low v5.3
FTP72 UDDS
Acceleration (m/s2) SC03
1 jc08
WLTC Cl.3 Mdl v5.3
ECE
0
−1
−2
−3
0 20 40 60 80 100 120
Speed (km/h)
Figure 3.7 Maximum acceleration over speed along second with operation points per second, for
Urban Test cycles.
4
Ecar500 120413 Linneg−City−CTH
Leaf 130321 CTH−Erksbrg
Leaf 130322 Korsv−Guldh
2
Prius 120329 Mlndl−Mlnlcke
Prius 120329 Mlnlcke−Mlndl
Prius 120403 Mlnlcke−CTH
0
Acceleration (m/s2)
V60PHEV Krsvgn−GalvBron
V744 120312 Cstat−CTH
V744 120318 Klltrp−CTH
−2 V744 120327 CTH−Klltrp
XC60 CTH−Krthsg
−4
−6
−8
0 20 40 60 80 100 120
Speed (km/h)
Figure 3.8 Maximum acceleration over speed along second with operation points per second, for
Logged Urban cycles.
In order to compare the distribution of acceleration over the speed range, between
Urban test cycles and Logged Urban cycles, the relative time spent in bins of 0.5 m/s2
and 5 km/h is depicted in Figure 3.9, along with the highest reached acceleration level
at each speed. When creating the plots, to each speed and acceleration operating point a
gaussian noise was added, which was later compensated for in the final result. The benefit
is smoother contour lines, however a small error is also introduced, it can be regarded as
negligible. Hence the contour lines outside of the maximum acceleration graph should
be disregarded. The method is further illustrated in Appendix C.
As was also indicated by the standard deviations, the logged cycles spend a somewhat
larger share at slightly larger acceleration levels, compared to the test cycles. This can be
seen by comparing the acceleration levels for the same time share contour line.
The figures show that for the test cycles, most of the time is spent around zero speed at
very low levels of accelerations (less than 1 m/s2 ), due to frequent stops. For the logged
41
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
cycles, most time is spent at speed levels around 5 to 10 km/h, but also within 1 m/s2 .
Furthermore, the test cycles spend a somewhat larger time share at higher speed levels
compared to the logged cycles.
3 4 3 4
1 0.1
0. 0.2
0 3.5 3.5
1
0.2 .1
0.
2 2 0.2
2
0.1
0.
0.5
1 0.5 3 1 0 3
0.1 .5
0.5
Acceleration (m/s2)
Acceleration (m/s )
1 0.5
2
1 1 1 0.5
0.
43 2
0.2 2.5 2.5
2
1
2
23
0.2
1
0.1
0 2 0 1 2
4
2
1
0.
1
1.5 1 0.5 1.5
−1 0.5 0.1 −1 0.5
0.2
1
0.5
0.5 0.1
0.2
1 1
0.
−3 −3
0 20 40 60 80 100 0 20 40 60 80 100
Speed (km/h) Speed (km/h)
Figure 3.9 Share of total time (%) of operation in acceleration bins of 0.5 m/s2 and speed bins of
5 km/h, with added noise, for all Urban Test and Logged cycles.
3.3.2 Rural
Cycle data regarding acceleration levels can be seen in Table 3.6, for Rural Logged and
Test cycles. Also for the Rural cycles, the logs show higher peak acceleration levels and
standard deviations compared to the test cycles, while the average values are similar. The
test cycles have a slightly larger time share below 1 m/s2 , while the logs have slightly
a higher time share for acceleration levels between 1 to 2 m/s2 . Furthermore, high RPA
values are more frequent amongst the logs.
For the test cycles it is the UC (LA92) cycle that has the highest peak acceleration
both positive and negative, as well as the highest RPA value. On the contrary, EUDC and
NEDC have the lowest peak acceleration levels and the lowest RPA value.
Amongst the logs, two cycles have similar peak acceleration levels; C30d2 121219
VCC-CTH and V744 120327 Klltrp-Cstat, however the later have a slightly higher aver-
age acceleration, and a much larger RPA value (0.34 compared to 0.19 m/s2 ).
The maximum acceleration as well as second based operating points, for the Test
cycles can be seen in Figure 3.10 and for the Logged cycles in Figure 3.11.
In Figure 3.10 it is clearly seen that the two cycles UC LA92 and Artemis Rural
differs quite a bit from the others since the levels of acceleration are larger over the
whole speed range, and even more so for negative accelerations. Also when comparing
with the Urban test cycles, these two cycles stand out. For the logged cycles it can be
seen in Figure 3.11 that the highest peak accelerations only occur at very low speed
levels (below 20 km/h). Around 40 km/h most cycles have peak accelerations between
1.5 to 2.5 m/s2 , while for higher speed levels the difference between the cycles increase.
42
3.3. Acceleration distributions
Table 3.6 Acceleration cycle data for Rural test and Logged cycles.
!"
& ' #( ) * +* )+ , +$ +- + #* $* ( ./ - + +
! ! ! ! ! !
#$ $# # # #% &%
" " " " " "
' ( ) *% +, *+ , +, +, + % , -) .% / $ +
At about 50 km/h the peak acceleration levels are around 1 − 2.5 m/s2 for the logs
and 0.5 to 1.5 m/s2 for the test cycles, while at about 80 km/h they are 0.5 − 2 m/s2
for the logs and 0.3 to 1 m/s2 for the test cycles.
As for the Urban cycles, the relative time spent in acceleration bins of 0.5 m/s2 and
5 km/h is depicted in Figure 3.12, along with the highest reached acceleration level at
each speed.
Also for the Rural cycles, the logged cycles spend slightly more time at somewhat
higher acceleration levels over the whole speed range. I can also be seen that apart from
the time spent at stand still and very low speed, also much of the time is spent around
40 km/h for both test and logged cycles, 55 km/h for the test cycles, and finally around
70 km/h for both test and logged cycles.
43
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
3
UC LA92
EUDC
2 NEDC
Artemis RURAL
1 WLTC Cl.3 Hgh v5.3
Acceleration (m/s2)
−1
−2
−3
−4
0 20 40 60 80 100 120 140
Speed (km/h)
Figure 3.10 Maximum acceleration over speed along second with operation points per second, for
Rural Test cycles.
4
C30d2 121219 CTH−VCC
C30d2 121219 Henan−Stnsnd
3 C30d2 121219 VCC−CTH
Ecar500 120413 CTH−Kllrd
2 C30d2 121219 Stnsnd−Henan
V744 120307 VCC−CTH
Acceleration (m/s2)
−1
−2
−3
−4
0 20 40 60 80 100 120 140
Speed (km/h)
Figure 3.11 Maximum acceleration over speed along second with operation points per second, for
Logged Rural cycles.
3 2 3 2
0. 0.2
0.5
0.1
0.5
Acceleration (m/s2)
Acceleration (m/s2)
1 1 1
0.5
0.5
0.
0.1
5
1
1.2 1.2
1
1
1
1
1
1
0.5
2
2
0.1
0
1
1 0 2 1
1
0.5
0.2
1
−1 2
0.1
−1 0.5
0.
5
0.1
0.
Figure 3.12 Share of total time (%) of operation in acceleration bins of 0.5 m/s2 and speed bins
of 5 km/h, with added noise, for all Rural Test and Logged cycles.
44
3.3. Acceleration distributions
Table 3.7 Acceleration cycle data for Highway test and Logged cycles.
! #$ $# # # #% &%
" " " " " "
! " #$ %
' ( $ ) * +* )+ * +* +, + $% *- * ./ % $ +
! $% %$ $ $ $& '&
# # # # # #
The maximum acceleration as well as second based operating points, for the Test
cycles can be seen in Figure 3.13 and for the Logged cycles in Figure 3.14.
As could be noted in the cycle parameters, the US06 and REP05 cycles have higher
levels of acceleration over the speed interval than the other test cycles, with many oper-
ating points above 2 m/s2 up to a speed level of 60 km/h. Next to US06 and REP05, the
Artemis Motorway cycles have the highest acceleration levels over the speed range, and
a particulary strong braking acceleration at high speed (about 2 to 3 m/s2 ).
As can be seen in Figure 3.14, the peak acceleration over the speed range of the
logged Highway cycles are close to one another, but also fairly close to those of the
US06 and REP05.
The relative time spent in bins of 0.5 m/s2 and 5 km/h is depicted in Figure 3.15,
along with the highest reached acceleration level at each speed.
When comparing the distribution of logged cycles with the test cycles, the differences
in time spent at various acceleration levels over the speed range is rather small. The main
difference that can be seen is the relatively larger time spent at higher speed levels in the
test cycles compared to the logged cycles. It can be seen that in the test cycles most time
is spent at speed levels close to 80 to 130 km/h (110 km/h in particular), while for the
logged cycles most time is spent at 50 km/h and between 70 to 120 km/h.
45
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
4
HWFET
US06
3
REP05
Artemis MW130
2
Artemis MW150
Acceleration (m/s2)
−1
−2
−3
−4
0 20 40 60 80 100 120 140 160
Speed (km/h)
Figure 3.13 Maximum acceleration over speed along second with operation points per second, for
Highway Test cycles.
4
C30d2 121219 Stnsnd−VCC
C30d2 121219 VCC−Stnsnd
3 V70 120510 CTH−Jnkpng
V70 120515 Jnkpng−CTH
2
Acceleration (m/s2)
−1
−2
−3
−4
0 20 40 60 80 100 120 140
Speed (km/h)
Figure 3.14 Maximum acceleration over speed along second with operation points per second, for
Logged Highway cycles.
3 2 3 5
1.8 4.5
2 0.1 2
4
1
1.6
0.5 0.1
0.1
0.
1
0.
0. 2 0. 0.2
0.10.2
0.
Acceleration (m/s2)
Acceleration (m/s2)
0.2 3.5
0.1 0.2
0.5
5 1.4
2
1 1
1 24
0.1 0.2
0.2
2
0.2
1
0.2
3
1
3
3 2
1.2
1
0.5
54
0.5
0.1
0.5
0.5
1
0.5
0.2
0 0
0.1
0.1
2.5
2
1
0.2
0.2 0.5
2
1
0.5 0.2
0.5
13
1
2
0. 2
2
1 0.8 2 0.5
0.2
0.1
−1 0.5 −1
0.2 1.5
0.
0.1
2
−2 −2 1
1
0.4
0.
0.5
0.2
−3 −3
0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160
Speed (km/h) Speed (km/h)
Figure 3.15 Share of total time (%) of operation in acceleration bins of 0.5 m/s2 and speed bins
of 5 km/h, with added noise, for all Highway Test and Logged cycles.
46
3.4. Acceleration duration
80
Cumulative frequency (%)
80
Figure 3.16 Cumulative frequency distribution of the time duration of each acceleration and decel-
eration, counted between zero-crossings, for the Test and Logged cycles.
47
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
results in three hits, i.e. three different highways in Sweden has sections of grade 6% or
larger. One is on E6 Bohuslän between Munkedal and Tanum, one in Dalarna between
Borlänge and Falun, and one in Skåne south of Skottorp across Hallandsåsen with grades
of 6 − 7% over a distance of 1.8 km. Another steep slope with a grade of about 7 %, is
west of Jönköping on road 40 along a section of about 1 km at a speed limit of 90 km/h.
Further searches leads to about a 170 hits on sections with speed limits of 100 km/h and
grades of at least 6% but maximum 12%, and 1 330 hits on sections with speed limits
of 90 km/h and grades of at least 6 % but maximum 12 %, for 70 km/h the number of
hits is 31 638, and for 50 km/h it is 5905. These numbers are however not to be blindly
trusted, since some of the later measurements are not coherent with the previous ones
and thus indicates much larger grades than previously.
At low speed levels, e.g. in cities or on drive ways the road grade can be even larger,
perhaps up to and above 30 % as indicated by [73] for several streets is San Francisco in
the US.
4 4 5
x 10 x 10 3
x 10
10 10
2.5
8 8
Occ. Freq. (−)
2
Occ. Freq. (−)
6 6
1.5
4 4
1
2 2
0.5
0 0 0
0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10
Road grade (%) Road grade (%) Road grade (%)
Figure 3.17 Road grade histogram of all Logged Urban, Rural and Highway cycles.
Figure 3.18 presents the difference in altitude between the end and start of the Logged
cycles, as well as the difference between maximum and minimum reached altitudes. For a
cycle with a large positive difference between the end and the start altitude, a significant
part of the energy that is consumed during driving is due to the increases in potential
energy. For those cycles that adhere to the same driving route and are driven in opposite
directions, it can be seen that the difference between end and start altitude is similar but
with opposite signs. Thus, the total energy consumption may be different for the two
cycles.
48
3.5. Road grade levels
400
334
334
End − Start
300 Max. − Min.
143
100
97
95
95
94
79
77
72
70
68
68
64
100
60
60
55
55
52
51
50
49
59
56
53
32
45
14
25
13
10
6
2
−7
−7
−7
−8
−14
0
−37
−38
−41
−52
−54
−65
−68
−70
−100
−149
Urban Rural Highway
−200
TH
on
TH
nd
TH
n
ttn
ttn
H
h
l
TH
g
lrd
CC
nd
n
H
CC
g
lnd
ck
ltr
stt
pn
uld
br
na
ths
CT
CT
CT
Br
ns
−C
Cs
Cs
tns
−C
−C
Kl
−C
−V
−V
Kl
ks
lnl
−M
−C
nk
He
C−
g−
ty−
−G
Kr
alv
H−
St
p−
p−
CC
H−
tat
rp
ke
TH
nd
Er
−S
l−M
−J
−
rp
ke
pn
H−
VC
C−
rsv
Ci
CT
G
nd
ltr
ltr
nlc
H−
Cs
CT
l
tns
7V
an
9C
TH
Kl
nlc
nk
nd
g−
CT
VC
Kl
Kl
Kl
tns
Ko
9
gn
CT
en
Ml
13
12
0C
9S
30
18
27
21
5J
21
ne
Ml
Ml
16
20
27
60
rsv
9
04
9H
9S
03
22
20
03
3
03
1
21
21
21
Lin
21
51
51
03
03
03
29
29
2
XC
12
12
03
1
04
12
12
VK
21
03
21
21
21
21
21
20
20
12
12
12
03
03
44
13
13
12
00
13
44
21
21
21
44
44
0d
21
01
01
0d
12
12
HE
44
44
44
V7
04
ar5
V7
21
21
af
ius
V7
V7
C3
af
C3
0d
0d
V7
V7
V7
V7
V7
ius
ius
0P
12
Le
Le
Ec
0d
0d
Pr
C3
C3
V6
Pr
Pr
00
C3
C3
ar5
Ec
Figure 3.18 The net altitude difference between the cycle end and start points, as well as the differ-
ence between maximum and minimum altitude that is experienced during the Logged
cycles.
The time duration of all uphill and downhill cycle segments are calculated for the
Logged cycles. The resulting cumulative frequency distribution is presented in Figure
3.19. It can be noted that the time duration is generally longer for the Highway cycles.
For 80% of the uphill occasions in the Urban and Rural cycles, the time duration is around
25s or shorter, and it is 38s or shorter for the Highway cycles. Furthermore, for 50% of
the uphill cycle segments, the time duration is 10s-14s or shorter. To conclude, the time
duration of grade climbing is generally more than twice the time duration of acceleration.
Grade duration − Typical Logged cycles
100
80
Cumulative frequency (%)
60
40 Urban, Acc
Urban, Dec
Rural, Acc
20 Rural, Dec
Highway, Acc
Highway, Dec
0
0 50 100 150
Time duration of upphill/downhill (s)
Figure 3.19 Cumulative frequency distribution of the time duration of each uphill and downhill,
counted between zero-crossings, for the Test and Logged cycles.
49
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
50
3.6. Average daily driving/traveling distance
100
Nissan − USA
Cumulative share (%) 80 Nissan − China
Nissan − France
60 Nissan − Germany
Nissan − Japan
40 Nissan − England
S. Karlsson − Sweden − RES0506 Survey by Driver
20 TSS − Sweden − 2010−2012 Measurement by Car
Gbg Stad TrfKntr VV − RVU05 Survey by Driver
0
0 50 100 150 200 250 300 350 400
Distance (km)
100
Nissan USA − unknown method
Cumulative share (%)
Figure 3.20 Daily driving distances in various countries, for either drivers or cars, extracted by
different methods.
According to [79], 90 % of the people in European countries drive less than about
120 km per day on an average, while 90 % of the people in USA drive less than 135 km
per day. It is also seen that Japan seem to have similar driving patterns as EU and that
China seem to have similar driving patterns as US.
As can be seen the information regarding Sweden differs a bit between the different
sources, perhaps this is due to different methods used, see [80], [81] and [82]. Based on
results from The Swedish Car Movement Data Project which cover cars in the south west
of Sweden, 90 % of the cars travel less than about 70 km per day on average.
In the US the spread between different published data is relatively large. As a sum-
mery of all presented graphs, 50 % of the average daily driving distances are shorter than
6 to 89 km, and 80 % of them are shorter than 16 km to 148 km. One study indicate that
larger distances are covered during rural driving compared to urban driving, [85].
51
Chapter 3. Road Type Driving Patterns, Road Grade and Daily Driven Distances
52
Chapter 4
Vehicle performance can be summarized in single quantitative measures such as: top
speed, minimum time to accelerate from 0 − 100 km/h, and specifically for BEVs, driv-
ing range. The work in this chapter and the next is concentrated to three concept light
duty BEVs, i.e. passenger cars, each with their own targeted qualitative specification
which can be seen in Table 4.1. Apart from the qualitative targets, all vehicles should be
able to manage highway driving in most countries around the world.
In this section the stated qualitative targets will be further specified into quantitative
performance requirements for each of the concept cars, with collected data on over 30
existing light duty BEV models as a frame of reference.
Table 4.1 Qualitative design targets for the three concept BEVs.
Seats Size Speed and acceleration Range
City 4-5 Small Medium Medium
Highway 5 Medium High Long
Sport 2 Small Very high Very long
53
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
Vehicles
Data on top selling light duty BEV models (including concepts) has been gathered
mainly from vehicle manufacturers. The full list of vehicle models along with the col-
lected data and their associated references is presented in Appendix A.
For some of the data there exists a level of uncertainty on the correctness and reli-
ability, since it is not specified in a standardized way, and many sources do not declare
under which conditions the data is valid.
For example, mass for vehicles are often given as curb weight, which is usually de-
fined as the mass of the specific model with standard equipment and neither passengers
nor extra cargo. However, in manufacturer’s mass specifications it is not always evident
what is included in the presented value. Moreover, the cross sectional area is very seldom
stated, thus it has to be estimated, e.g. by a method such as the ones suggested in Chapter
2, i.e. as a weighted product of vehicle height and width.
Lastly, the vehicle driving range given is usually specified as the certified range
gained by driving according to either American EPA cycles or the European NEDC cy-
cle, however in some sources this is not declared. The same issue is also valid for energy
consumption per distance.
For further ease of benchmarking, the existing vehicle models are sorted into the four
categories: small, medium-large, high performing and sport, where the former two are
based on vehicle external volume calculation while assuming a rectangular block. The
sport category consists of high acceleration capable 2-seat models. The high performing
category also consists of high acceleration capable models, but with five seats.
54
4.1. Performance requirements based on data of existing BEVs
20
0−100 km/h Small models
0−60 mph Medium−Large models
City car Trend line 0−100km/h
Acceleration Time (s)
High Perf. models
15
Sport models
Highway car a=81750 v−1.812
R2=0.8746
10
a: acceleration (m/s2)
v: speed (km/h)
Sport car
0
100 150 200 250 300 350
Top Speed (km/h)
Figure 4.1 Top speed vs. acceleration for the three concept BEVs and for the existing BEVs. Ad-
ditionally, a suggested trend line is provided, which represent 0-100km/h acceleration
data.
It can be noted that those models with the highest top speed levels, are also those with
the shortest time to accelerate. For the sport and the high performing models the time is
around 3s-6s. For the small cars, the spread in acceleration time is relatively large: around
7s-16s, and for the medium-large cars it is 7s-14s.
For the three concept vehicles, only the time to accelerate from 0 − 100 km/h is
specified. For the City car, the time is decided to be maximum 13 s, which corresponds
to a medium performance small car. For the Highway car, it is decided to be 10 s, which
is also somewhat of a medium value for the category. Finally the chosen 0 − 100km/h
acceleration time for the Sport car is set to 4 s, in order to match the sport cars with the
highest performance.
4.1.2 Gradability
A vehicle’s gradability is the maximum grade that a vehicle can climb at a certain speed
while using the maximum power from the powertrain. Data on gradability for the concept
vehicles is not based on typical values from existing BEVs, since this type of data is rare.
There are however a few sources that present general gradability design goals. In [87]
for example, the requirements are that the cars should be able to start in a 25 % uphill
grade, drive at a speed of 88.5 km/h in a 3 % grade, and in 72.4 km/h in a 6 % grade.
When considering typical grade levels on highways there are relatively few passages
with grade levels larger than 6 %, unless perhaps considering mountainous areas. There
are even fewer sections with grade levels of 12 % or higher.
Based on the above, as well as the quantitative requirements in Table 4.1, it is decided
that the Highway car should be able to sustain a speed of 130 km/h at a grade 6%, and
that the Sport car is to be able to sustain a highway speed of 130 km/h even in steep
grades of 12 %. The City car, on the other hand, should have a maximum speed of at
least 90 km/h at the grade of 6 %. Furthermore it is decided that, in coherence with the
minimum goals stated in [87], all three concept vehicles should be able to start in a uphill
gradient of at least 25 %.
55
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
Vehicles
1800
1600
Sport car
City car
1400
Small models
1200
Medium−Large models
1000 High Perf. models
Sport models
800
0 100 200 300 400 500 600
NEDC range (km)
Figure 4.2 NEDC driving range vs. curb weight for the three concept BEVs as well as for the
existing BEVs.
56
4.2. Wheel load analysis of chosen concepts
be 2.05 m2 , which resembles a small car, the Highway car area is chosen to be 2.3 m2
which takes after the large cars, while the Sport car area is chosen to be 2.0 m2 .
With the published Cd values of existing BEV’s as a reference the Cd value of the
City car is chosen to be 0.3, for the Highway car 0.28, and finally the Cd value of the
Sport car is chosen to resemble that of the Roadster, 0.35.
Data on tire radius of existing BEVs is not available, instead it has been estimated
based on tire size, see Figure A.3. Based on this, the tire radius of concept models are
chosen accordingly: the City car 0.31 m, the Highway car 0.32 m, and the Sport car
0.34 m.
inally, the tire rolling resistance coefficient Cr should be set for all of the concept
vehicles. Even though the rolling resistance is known to be speed dependent, there is no
clear consensus on how this should be modeled (as shown in Figure 2.2) unless perhaps
tire specific data is available, which it is not in this case. Another aspect is that the type
of study in this thesis is of comparative nature, hence a possible speed dependence of the
rolling resistance is expected to have a rather small impact on the final result. With these
aspects as a background as well as the information presented on the topic in Chapter
2 and Appendix A, it is assumed that the rolling resistance of the concept BEVs can
be estimated by the mean values: 0.009 for both the City car and the Highway car
representing low rolling resistance tires, while for the Sport car the tires are expected to
be optimized for performance rather than energy efficiency, leading to a value of 0.012.
57
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
Vehicles
time duration of any of the operating points, only to the level of magnitude. The analysis
is also based on finding information that is of interest for a BEV with a single speed
gearbox.
58
4.2. Wheel load analysis of chosen concepts
Wheel Forces at constant speed levels Mass normalized wheel Forces at constant speed levels
50
1800 90 2 Sport − P (kW)
wheel
10
1600 80
50
100
Highway − Pwheel (kW)
10
50
1400 70
20
1.5
Highway − Road load
Wheel Force (N)
60
1200
City − Pwheel (kW)
50
50
1000 100 City − Road load
10
10
40
1
800
50
10
30
600
50
0.5
10
400
10
200 10 10
10
0 0
0 50 100 150 200 0 50 100 150 200
Speed (km/h) Speed (km/h)
Figure 4.3 Road load wheels force as a function of car speed on a level road, and mass normalized
wheel force, for all three concept cars. The contour lines represent wheel power for a
certain combination of speed and force, which is the same for all cars in the left part of
the figure, however not for the right part.
Figure 4.4 shows both absolute wheel forces and mass normalized wheel forces for
each concept car, for road loads at certain grade levels as well as at operating points
corresponding to the specified speed and grade performance requirements.
Starting with the City car, Figure 4.4(a) shows that the top speed and the grading
requirement at high speed, demand about the same levels of wheel power; 23 kW and
26 kW respectively, while the force level at the grading requirement is roughly twice that
at the top speed (about 1 kN and 0.6 kN respectively). The takeoff requirement demands
the largest force, almost 3 kN but since the speed is low, so is the power demand. In
the right part of Figure 4.4(a) it can be seen that the takeoff requirement is equal to an
acceleration of almost 2.5 m/s2 , as was predicted in Table 2.1, and as is also valid for
the other cars. To conclude, if the wheel force is limited to 3 kN at speed levels up to a
power limit of 26 kW , above which the wheel force follows the power line, then all three
specified operating points will be achievable. As mentioned, the difference between the
maximum available wheel force and the road load at a certain grade, equals the vehicle’s
acceleration capacity. With the above limits, the initial acceleration at takeoff would be
about 2.4 m/s2 on a flat road, at 50 km/h it would be about 1.4 m/s2 and at 100 km/h
about 0.47 m/s2 .
For the Highway car, shown in Figure 4.4(b), the top speed demands a force of about
1 kN and a power of 34 kW , while the high speed grading requirement demands a force
of about 1.8 kN and a power of 60 kW , i.e. a larger difference between the requirements
compared to the City car (it can be seen that also for the Highway car the demanded
power at 6 % grade in 90 km/h is almost the same as that at the top speed). For the
Highway car an initial force level of about 4.1 kN is required to manage the takeoff
requirement, up to about 50 km/h where the power thus can be limited to 60 kW .
The Sport car, shown in Figure 4.4(c), evidently demands the highest levels of force
and power compared to the other two cars. As for the City car, both the top speed and
grading requirements demands roughly the same power levels 96 kW and 108 kW re-
spectively. The takeoff requirement demands an initial force of about 4.8 kN . This force
level may be sustained up to about 80 km/h, where the power may be limited to 108 kW .
59
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
Vehicles
5000 4
100
Powerwheel (kW)
100
15
15
3.5
0
26
Road load on flat road
23
0
26
23
4000
50
Wheel force (N)
50
2.5 90 km/h at 6% grade
3000
Top speed on level road
2
2263
2263
2000 1.5
50 50
1
1000
2236 2236
0.5
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Vehicle speed (km/h) Vehicle speed (km/h)
8000 5
Powerwheel (kW)
60
60
4
Wheel force/mass (N/kg)
25
0
6000
0
34
100
100
20
0
0
3
Top speed on level road
4000
60
60
3000 10 2 10
34
0
34
0
2000
60 1 60
1000 34 34
0 0
0 50 100 150 0 50 100 150
Vehicle speed (km/h) Vehicle speed (km/h)
15000 7
400
Powerwheel (kW)
108
96
400
300
50
300
200
10000
200
30
10986
0
30 0
3
10986
0
20
5000 20 0
50
0 2
50
1098 108
6 1 96
50 50
0 0
0 50 100 150 200 0 50 100 150 200
Vehicle speed (km/h) Vehicle speed (km/h)
Figure 4.4 Force on wheels due to road load per speed level, and normalized force per vehicle
mass, for all three concept cars, along with contour lines representing combinations of
wheel force and speed for different levels of wheel power.
4.2.2 Acceleration
Given the above stated demanded levels of force and power versus speed due to the three
mentioned requirement; top speed, grade at high speed and take-off at high grade, it
can be noted that the implied acceleration capacity of each car would be rather limited.
Further investigations must be made to find suitable levels of force and power that will
also fulfill the acceleration requirements.
By assuming a torque and speed characteristic that is typical for an electric machine,
60
4.2. Wheel load analysis of chosen concepts
(i.e. a region of constant torque at low speeds followed by a region of constant power at
higher speeds, where the torque is inversely proportional to the speed), various combi-
nations of initial maximum levels of wheel force and wheel power limits can be found,
such that they will all fulfill the acceleration requirement. Even though the acceleration
time from 0 − 100 km/h is the same for all combinations, the acceleration time from
0 − 50 km/h, hence also 50 − 100 km/h differs. It can thus be concluded that the re-
quirement on acceleration time from 0 − 100 km/h alone, is not enough when seeking to
specify a vehicles acceleration performance over the operational speed range. Few data
regarding acceleration times between different speed levels are available for the existing
BEV models, see Appendix A. Acceleration times for 0 − 50 km/h are 3s-6s and for
0 − 60 km/h 3.7s-7.7s.
6 5
Fwheel, max Acc.
4 norm. Fwheel
Acc. (m/s2)
Fwheel (kN)
4 Road load
3
2
2
45kW 48.5kW 57kW
1 45kW 48.5kW 57kW
5.0 kN 3.8 kN 3.0 kN 4.2 N/kg 3.2 N/kg 2.5 N/kg
0 0
0 50 100 0 50 100
Speed (km/h) Speed (km/h)
300
100
Speed (km/h)
Distance (m)
200
50 29m
5.8s 100
33m 231m
4.6s 13.0s 41m 218m
3.8s 198m
0 0
0 5 10 15 0 5 10 15
Time (s) Time (s)
Figure 4.5 Speed and distance over time, as well as wheel force and acceleration, for base speeds
that are 1/4, 1/3 and 1/2 of the top speed, during 0 − 100 km/h acceleration, for the
concept City car.
The lowest base speed is related to the largest initial force (5 kN), lowest power (45
kW), fastest acceleration of 0 − 50 km/h (just under 4 s) along with the shortest driven
distance during the acceleration, while the acceleration time of 50 − 100 km/h is the
longest. Given that this is the City car, it can be argued that low speed acceleration per-
formance should be preferred over high speed performance. It is thus decided that the
61
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
Vehicles
lowest power level should be the design template. Compared to the power levels of ex-
isting BEVs with similar acceleration times, the chosen power level is smaller. However
the acceleration time 0 − 50 km/h complies rather well when comparing with some of
the existing BEVs.
When comparing the results in Figure 4.5, with the ones found in Figure 4.4(a), the
initial force demanded by the acceleration requirement should be around 2 kN higher
than that for the takeoff requirement (of 3 kN ), while the power should be 19 kW higher
than that of the high speed gradient requirement (of 26 kW). It is thus the acceleration
requirement that will determine the maximum force and power that has to come to the
wheels (via the gear box) from the powertrain.
3
4
2
2 109kW 89kW 82kW 1 109kW 89kW 82kW
5.2 kN 6.3 kN 7.7 kN 3.1 N/kg 3.7 N/kg 4.5 N/kg
0 0
0 50 100 150 0 50 100 150
Speed (km/h) Speed (km/h)
200
100
Speed (km/h)
Distance (m)
150
100 33m
50
28m
3.3s 50 25m 147m
3.9s 10.0s 161m
4.7s 171m
0 0
0 5 10 0 5 10
Time (s) Time (s)
Figure 4.6 Speed and distance over time, as well as wheel force and acceleration, for base speeds
that are 1/4, 1/3 and 1/2 of the top speed, during 0 − 100 km/h acceleration, for the
concept Highway car.
Also here the result is that the combination that has the highest initial force (7.7 kN )
and lowest power (82 kW ) result in the best low speed performance (3.3 s when acceler-
ating 0−50 km/h). However since the 0−50 km/h and 0−60 km/h acceleration times
of the few existing BEVs that was found are around 4 s, this is here chosen as a targeted
value also for the Highway car. Thus targeted initial force is 6.3 kN and power is 89 kW.
That means that the initial force should be about 2.2 kN more than what was demanded
by the takeoff requirement (of 4.1 kN ), and the power should be 29 kW more than that
62
4.2. Wheel load analysis of chosen concepts
of the high speed gradient requirement (of 60 kW ), as seen in Figure 4.4(b). This would
result in an initial acceleration level of about 3.7 m/s2 .
Acc. (m/s2)
Fwheel (kN)
10 5
Distance (m)
50
50 15m
1.7s 14m 58m
1.9s 4.0s 13m 61m
2.0s 65m
0 0
0 1 2 3 4 5 0 1 2 3 4 5
Time (s) Time (s)
Figure 4.7 Speed and distance over time, as well as wheel force and acceleration, for base speeds
that are 1/4, 1/3 and 1/2 of the top speed, during 0 − 100 km/h acceleration, for the
concept Sport car.
63
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
Vehicles
100
90
90
City car
80
80
78
68
Highway car
63
57
57
55
55
54
54
Sport car
49
48
48
43
43
42
41
50
41
41
38
36
36
36
35
35
35
35
35
34
33
33
32
32
31
31
29
28
28
27
26
25
25
22
22
22
22
22
22
max wheel power (kW)
21
18
16
15
15
11
Positie and negative
0
−10
−12
−15
−14
−15
−16
−17
−18
−17
−20
−21
−21
−22
−22
−24
−23
−23
−25
−25
−27
−27
−28
−29
−31
−32
−32
−32
−32
−34
−34
−34
−35
−37
−37
−39
−41
−41
−43
−45
−46
−51
−53
−53
−50
−58
−58
−61
−64
−69
−69
−75
−83
−87
−96
−99
−99
−100
−110
−110
le
gh
h
N
E
06
8
CC
DC
DC
92
S
5
0
0
ral
g
0
jc0
P0
13
15
Lo
FE
dd
EC
DD
BA
Hi
Hi
SC
US
LA
Ru
NY
EU
NE
RE
Mi
HW
is
is
TC
2U
UR
TC
tra
UC
tem
tem
is
TC
WL
Ex
WL
P7
tem
is
Ar
Ar
WL
tem
FT
TC
Ar
WL
Ar
Figure 4.8 Max positive and negative wheel power during Urban, Rural and Highway Test cycles,
for all the three concept cars.
64
4.3. Wheel load analysis for selected drive cycles
The maximum positive and negative wheel power for the logged cycles can be seen
in Figure 4.9. It can be seen that the City car may have problem to reach some operating
points in 2 of the Urban cycles, 3 of the Rural cycles and 3 of the Highway cycles, while
the Highway car may have problem with only one of the Rural cycles, and the Sport car
is likely to not have any problem to fulfill all power levels.
150
City car
106
Highway
94
94
100
85
85
84
82
83
80
max wheel Power (kW)
76
74
76
75
74
Sport car
73
70
71
68
67
67
Positive and negative
67
67
65
62
60
58
60
59
57
59
55
55
54
52
54
53
50
51
51
51
48
49
48
48
45
46
44
42
44
43
43
43
41
42
50
39
36
37
37
36
36
33
35
32
32
31
30
29
29
26
25
22
20
0
−23
−26
−28
−29
−32
−34
−34
−36
−37
−38
−38
−39
−40
−40
−39
−42
−42
−43
−41
−44
−44
−46
−45
−45
−46
−47
−47
−50
−48
−49
−51
−53
−53
−54
−52
−55
−58
−56
−57
−58
−60
−60
−62
−63
−64
−63
−62
−63
−63
−65
−65
−65
−68
−67
−67
−70
−70
−68
−71
−72
−72
−74
−75
−74
−77
−81
−85
−88
−90
−97
−97
−99
−100
−108
Urban Rural Highway
−150
H
n
TH
d
H
TH
ttn
n
ttn
ng
H
h
l
TH
g
lrd
C
nd
an
H
C
g
lnd
ro
ck
n
ltr
stt
uld
CT
br
ths
CT
CT
VC
CT
VC
tns
kp
en
Cs
Cs
tns
−C
−C
Kl
−C
lvB
Kl
ks
lnl
−M
−C
C−
p−
g−
ty−
−G
H−
d−
Kr
H−
Jn
−H
p−
p−
H−
tat
CC
ke
−S
Er
−S
−M
trp
ke
lltr
pn
VC
−
n
T
sv
H−
Ci
CT
−G
nd
ltr
lltr
nlc
H−
Cs
TH
CT
CC
tns
an
9V
9C
l
nlc
nk
8K
lnd
g−
or
Kl
Kl
CT
tns
07
7K
gn
CT
en
C
Ml
13
12
9V
9S
27
21
K
5J
21
ne
Ml
6
20
03
9M
60
rsv
04
9H
9S
0
03
32
22
03
03
03
21
21
21
21
Lin
21
51
51
03
03
12
29
XC
12
12
20
03
32
04
12
12
VK
21
03
1
21
21
21
21
20
20
12
12
03
44
d2
13
1
13
12
00
13
44
21
21
21
44
44
0d
21
01
01
12
12
HE
44
44
44
V7
04
ar5
V7
21
21
af
ius
V7
V7
C3
af
C3
0d
0d
V7
V7
V7
V7
V7
ius
ius
0P
12
Le
Le
Ec
0d
0d
Pr
C3
C3
V6
Pr
Pr
00
C3
C3
ar5
Ec
Figure 4.9 Max positive and negative wheel power during Logged Urban, Rural and Highway cy-
cles, for all the three concept cars.
65
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
Vehicles
energy consumption, while ECE has the lowest. This can be related to Table 3.2 to 3.7,
where it can be seen that NYCC, Artemis Urban and ECE have relatively low maximum
and average speed values, but ECE have much lower maximum acceleration and RPA
value than NYCC and Artemis Urban. For the Rural cycles, UC LA92 has the highest
consumption and NEDC the lowest. Both cycles have similar levels of average speed
but the UC LA92 has a large time share spent at higher speed levels. It is also the UC
LA92 that has the highest maximum acceleration amongst the Urban cycles, as well as
the highest RPA value, while NEDC has the lowest. Finally the Artemis motorway cycles
have the highest consumption between the Highway cycles, and HWFET the lowest. In
this case it is not the cycle with the highest acceleration or RPA values that consume the
most energy, but it is that cycle that has the highest average speed and spend the most
time at high speed levels.
250
214
214
206
City − Aero. drag
201
194
193
187
184
182
182
City − Roll. resist.
175
200
172
165
163
163
161
158
157
157
155
154
153
150
City − Acc.
Positive and negative wheel energy/
149
148
148
147
147
137
137
136
135
135
131
129
127
127
124
124
123
123
120
118
110
110
103
101
101
100
100
91
89
80
−14
−17
−18
−24
−21
−23
−22
−30
−31
−26
−28
−29
−28
−34
−32
−36
−32
−34
−41
−39
−38
−44
−39
−41
−40
−41
−50
−48
−48
−48
−50
−52
−50
−60
−57
−61
−58
−58
−59
−62
−62
−70
−70
−74
−74
−75
−73
−76
−81
CC −117−82
−83
−100
N −119
−124
−127
le
gh
gh
8
03
92
06
E
DC
DC
0
jc0
P0
13
15
Lo
FE
RA
dd
EC
DD
BA
Hi
Hi
SC
LA
US
NY
NE
EU
RE
Mi
HW
is
is
RU
C
2U
UR
tra
TC
UC
tem
tem
T
C
WL
Ex
WL
P7
is
is
T
Ar
Ar
WL
tem
tem
FT
TC
Ar
WL
Ar
Figure 4.10 Positive and negative wheel energy per driven distance during Urban, Rural and High-
way Test cycles, for all of the three concept cars, while acceleration is calculated using
the described forward-backward method.
The positive and "negative" values of wheel energy consumption per driven distance,
are also illustrated as a function of average running speed for each cycle and car, in
Figure 4.11. As a frame of reference, the calculated wheel energy per driven distance
while driving at constant speed levels are also included in the figure, i.e. while only
considering aerodynamic and rolling resistance. As shown, the energy per distance when
driving at constant speed increases with increasing speed, due to the speed dependency
of the aerodynamic drag. The deviations between the positive cycle values and constant
speed values of energy consumption represent the excess energy consumption due to the
acceleration content in each cycle. At the same time, an equally large amount of "negative
energy consumption" can potentially be regenerated with the electric powertrain, which
would reduce the influence of acceleration on the net battery energy consumption.
66
4.3. Wheel load analysis for selected drive cycles
250
City − Ewheel, pos Test cycles
200 City − Ewheel, neg Test cycles
City − Ewheel, pos Const. speed
wheel energy/dist. (Wh/km)
150
Highway − Ewheel, pos Test cycles
Positive and negative
−100
−150
0 20 40 60 80 100
Cycle mean speed (km/h)
Figure 4.11 Wheel energy per driven distance during Urban, Rural and Highway Test cycles, for
all the three concept cars.
67
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
Vehicles
propulsion mode.
The large difference between the acceleration estimation methods is reduced (inverse
linearly) when the time resolution is increased. By re-sampling the speed vector through
linear interpolation, it was noted that, at 6 Hz the difference is 1% or smaller for all cycles
except Artemis URBAN and NYCC, and at 15 Hz the difference is 1% or lower for all
cycles.
Backwards comp. Central method
12.8
Wheel Energy/Dist.
Brake
10
5.8 6.4 5.0 5.7 6.0 5.1 5.9 5.0
4.2 3.9
2.5 2.3 2.4 1.8
1.1 0.8 0.7
0
−0.6 −1.0 0.9 −0.7−0.6
−3.3 −3.4−2.7 3.3 −3.0 3.7
−4.5 4.6
−5.5−4.7 −5.4
−6.4−5.9
−4.7
−10
w
le
UC Cl.3
gh
is High
N
03
U 8
tem S06
HW 92
RE T
UR C
Ar WLT EDC
E L
WL UDC
tra 05
TC E
TC DDS
0
P7 jc0
13
Lo
FE
RA
dd
WL EC
BA
C
Hi
SC
LA
P
NY
U
Mi
TC
is
RU
N
tem C
2
Ex
is
Ar
WL
tem
FT
TC
WL
Ar
Figure 4.12 Change in wheel energy per distance due to Backwards method compared to Central
difference method, during Propulsion and Braking of all Test cycles, for the City car.
68
4.3. Wheel load analysis for selected drive cycles
300
City − Aero. drag
232
City − Roll. resist.
212
201
200
199
199
196
193
energy/dist. (Wh/km)
182
City − Acc. & Grade
180
179
178
175
174
172
200
170
170
170
169
168
168
167
167
167
165
164
164
163
160
155
153
153
151
150
148
147
147
Highway − Aero. drag
143
143
142
141
141
140
140
139
137
135
135
129
125
123
118
117
117
116
113
111
Highway − Roll. resist.
108
108
106
106
103
102
101
98
97
97
97
96
95
88
85
100 Highway − Acc. & Grade
Sport − Aero. drag
Sport − Roll. resist.
0 Sport − Acc. & Grade
−13
−15
−17
−17
−25
−24
−23
−29
−30
−31
−29
−29
−30
−39
−39
−37
−40
−40
−42
−46
−47
−47
−48
−51
−52
−60
−58
−58
−58
−59
−61
−68
−66
−66
−66
−69
−72
−74
−72
−72
−72
−75
−76
−74
−79
−76
−79
−80
−77
−79
−83
−82
−86
−86
−90
−85
−91
TH −154 −93
−100
−101
−105
−106
−108
−109
−115
−100 −114
−117
−121
−120
−128
−164
Urban Rural Highway
−200
H
n
nd
TH
TH
ttn
ttn
g
H
h
ke
l
TH
rg
lrd
CC
n
TH −CTH
CC
g
nd
2 C vBro
stt
pn
uld
CT
n
na
ths
CT
t
b
nlc
tns
s
Cs
ns
t−C
−C
C
Kl
−C
−V
−V
Ml
Kl
rks
−C
−C
nk
e
C−
C−
g−
−G
Kr
H−
−H
St
p−
9 M l−Ml
H−
p
−
rsv lcke
ty
−S
9 H CTH
nd
l
sta
−J
a
rp
p
−E
ltr
pn
VC
H−
VC
−
rsv
Ci
CT
−G
nd
lltr
ltr
ck
CT
CC
tns
an
TH
Kl
nk
d
g−
CT
ln
Kl
Kl
lnl
tns
07
o
19
K
gn
19
n
en
13
9V
0 1 10 C
9S
M
18
27
K
5J
1C
ne
Ml
31
16
20
27
03
12
2 1 1212
60
04
9S
2
03
03
21
Lin
21
51
0
03
03
03
32
12
29
2
32
XC
12
5
12
32
04
12
12
VK
21
21
21
21
21
20
20
12
12
12
0
03
30
44
C3 30d2
3
13
0
12
00
44
21
21
21
44
44
0d
21
01
41
12
12
HE
f1
44
44
44
V7
ar5
V7
21
af
ius
V7
V7
C3
20
0d
0d
V7
V7
a
V7
V7
V7
ius
ius
0P
C
Le
Le
Ec
1
0d
0d
C3
Pr
C3
V6
Pr
Pr
00
C3
ar5
Ec
Figure 4.13 Positive and negative wheel energy per driven distance during Urban, Rural and High-
way Logged cycles, for all of the three concept cars.
250
City − Ewheel, pos Test cycles
200 City − Ewheel, neg Test cycles
100
Highway − Ewheel, neg Test cycles
50 Highway − Ewheel, pos Const. speed
Sport − Ewheel, pos Test cycles
0
Sport − Ewheel, neg Test cycles
−50
Sport − Ewheel, pos Const. speed
−100
−150
−200
0 20 40 60 80 100
Cycle mean speed (km/h)
Figure 4.14 Wheel energy per driven distance during Urban, Rural and Highway Logged cycles,
for all the three concept cars.
4.3.4 Change in potential energy per driven distance for the Logged
cycles
In order to get an indication of the relative energy consumption that road grade causes
for the different cycles, the potential energy increase or decrease between the end and the
starting altitude level is divided by the total driven distance, and is presented in Figure
4.15. Even though there is a large altitude difference (143 and 149 m) for the two cycles
V70 120510 CTH-Jnkpng and V70 120515 Jnkpng-CTH (as seen in3.18), the driven
distance is also quite long, 145 km. Hence it is likely that the relative energy consumption
69
Ec
ar5
00 E inc./dec. (Wh/km)
70
12
04 potential
13
−60
−40
−20
0
20
40
60
80
Lin
Le ne
af g−
13 Ci
03 ty−
14
Vehicles
21 CT
Le
af CT H 20
Urban
13 H− 23
03 Er
Pr
iu 22 ks −9
s1 Ko br
g −13
20 rsv −15
32 −G
Pr 9M 39
ius lnd
uld
h 55
12
03 l−M 61
Pr 29 lnl 19
ius Ml
nlc
ck
e 27
12 k 30
04 e−
V6 03 Ml −12
0P Ml nd
l −17
HE nlc −19
VK ke
rsv −C −2
V7 gn T H −3
44 −G −3
12 alv
03 Br −3
V7 12 on −4
44 Cs −5
12 tat
03 −C 26
V7 18 TH 37
44 Kl
ltr 42
12 p−
03 CT −3
27 H −4
CT −4
XC H−
0d an
21 −S
Ec 21 tns −1
ar5 21 nd −1
00 9V
CC −1
C3 12 − 6
0d 04 CT 9
21 13 H
21 CT 10
21 H −
9S Kl −6
V7 t ns l rd −8
44 nd −9
12 −H
03 en 0
V7 07 an 0
44 VC 0
12 C−
03 1
V7 1 6K C TH 2
44 lltr 2
12
03 p −C
20 −29
V7 K
stt
n −41
44 lltr −46
12 p−
C3 03
27 Cs −30
0d Kl ttn −42
21
21 ltr −47
2 p−
C3 19 Cs −32
0d St ttn −45
21 ns
nd −51
21
21
9
−V
C 3
V7 VC C 4
4
will naturally be larger than these values, due to the looses in the powertrain.
01 C−
20 St
51 n −3
Highway
V7 0C sn
d −4
01 T H− −4
20 Jn
51 kp 3
5J
nk ng 4
pn 5
g− −3
CT −5
H
−5
Chapter 4. Performance Requirements and Wheel Load Analysis of Studied Concept
per driven distance, for all the three concept cars; City-green (left), Highway-blue
Figure 4.15 Change in potential energy between end point and starting point of the Logged cycles
due to road grade is fairly small. In practice the energy consumption due to road grade
Chapter 5
As was found in Chapter 4, the demanded maximum torque and power from the power-
train can be solely determined by the acceleration requirement for all of the three concept
cars. Furthermore, it was found that the acceleration requirements could be satisfied with
more than one combination of maximum wheel force and power. This means that the
output requirement of the powertrain has some degree of freedom to it. Therefore the
final choice will to a large part depend on the characteristics of the chosen components
to be used.
In this chapter each of the concept cars will be assigned a base-line powertrain set-up,
including sizing of, and models for; the electric machine, converter, battery and transmis-
sion.
For simplicity auxiliary loads are excluded from the study, however they are likely to
have an important impact on BEV energy consumption, especially in particulary warm
and cold conditions.
Furthermore it is decided that the acceleration requirement should be achieved at a
10 % SOC level, and that the torque and power at higher SOC levels will not be functions
of the available DC voltage. This means that the powertrain will be slightly oversized at
higher SOC levels.
71
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
5.1.1 Converter
As can be seen in Appendix A, most commercial BEVs have DC voltage levels of around
300 − 400 V . The only exceptions are the sport cars where the voltage levels are higher.
Hence an IGBT module with a voltage rating around 650 V is well suited, since then
there is a margin for increasing battery voltage during braking, but also to handle other
phenomena occurring during operation such as induced voltage peaks during converter
switchings caused by stray inductances. Furthermore, it is assumed that the maximum
converter RMS phase current IRM S,max can be as high as 2/3 of the current rating in the
data sheet. Converter models from Infinieon found at [88] with different current rating
are used for all three cars.
4.5 4
4.2 V Charge − meas. US DOE
3.8 Charge − ave. SOC10−90%
4.08 V
4 Voc, ave.:3.88 V
3.6 Discharge − US DOE
Cell Resistance (mΩ)
2 2.2
0 20 40 60 80 100 0 20 40 60 80 100
SOC (%) SOC (%)
Figure 5.1 Estimated AESC battery cell no load voltage as a function of SOC, and estimated aver-
age values of cell resistances, based on data from [US DOE, AESC].
72
5.1. Components used for modeling
Given the stated peak RMS current (which was a 3 min. value) and winding struc-
ture, the maximum current density can be found to be about 12 A/mm2 (with a slot fill
factor of about 0.45, three turns per slot and two parallel windings the conductor area is
13.47 mm2 ). According to [34], a liquid cooled brushless machine can have a current
density between 10 to 30 mm2 . It is here chosen to increase the machine current density
to 20 A/mm2 , which leads to a max RMS current of about 540 A.
As was done in [36] and [92], the core losses are estimated using the added core
resistance Rc described in Chapter 2. The core loss resistance as a function of speed is
estimated based on measured no load losses over the machine speed range, presented
in [53] (paper E), as
2
3 (wel ψ)
Rc ≈ (5.1)
2 Pf e
For each operating point, the machine core losses are then calculated as ( [92])
3 2 2
Pf e = Rc (id,o − id ) + (iq,o − iq ) ); (5.2)
2
The efficiency and total losses of the originally sized electric machine (i.e. with unity
scaling factors) can be seen in Figure 5.2, as well as maximum torque over speed using
the MTPA control strategy as described in [92].
73
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
400 400 8
ηPMSM, org (%) Ploss,tot,PMSM, org (kW)
95
350 350 8 7
300 90 300 6
6
Torque (Nm)
250
Torque (Nm)
250 5
85 4
200 4
80
200
90
70
150 80 150 3
95
2
100 96 100 2
97 75
50 50 1
0 70 0 0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 5.2 Efficiency and total power losses for the electric machine with the original size, i.e.
unity scaling factors.
where lstk is the active lamination stack length, τp is the winding pole pitch and dext
is the part of the end winding that is estimated to be axially extended outside of the
lamination stack. The stack length is 231 mm, the pole pitch is π2 D
p , where D is the stator
inner diameter (110 mm) and p is the pole pair number (two pole pairs), and finally the
2dext is assumed to be 0.05 m as in [53]. This gives a total average conductor length of
384.7 mm, where the active part is 231 mm or 60 % and the over hang is 153.7 mm or
40 %, ( [53] paper D p.80). The end winding inductance is assumed to be very small,
hence neglected, although as mentioned in [93] it may not be a viable assumption for
motors with a low length/diameter ratio.
The length scaling is implemented using the stack length scaling factor, SFLstk .
74
5.2. Components sizing process
Figure 5.3 BEV powertrain sizing and modeling algorithm for the City car.
75
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
Vt (10% SOC)
ð SFrw
ð IRMS, max
IGBT-module current Rating
≈ 3/2 IRMS, max
Figure 5.4 BEV powertrain sizing and modeling algorithm for the Highway and Sport cars.
76
5.3. Implemented battery models
The theoretical and the voltage limited discharge power limits are rather similar, while
the current limited discharge power limits are substantially lower. Also a constant dis-
charge power of 58 kW is depicted, and shows to be just a bit larger than the 3C limit.
During charging, only the voltage limited charge power is presented. It shows a stronger
dependence on SOC level compared to the discharge power. Similar characteristics are
valid for the Highway and Sport cars as well, although not shown here.
450 350
Pch.max − Vmax
412 V
300 Pdis.max − theoretic
400
401 V Pdis.max − Vmin
250
350 Pdis.max − 3C
Power (kW)
Voltage (V)
200 Pdis.max − 5C
325 V Voc
300 Pdis.max − 58 kW
Vt 3C 150
250 245 V Vt 5C
Vt 58 kW 100
450 450
412 V 412 V
400 400
401 V 401 V
350 350
Voltage (V)
Voltage (V)
Figure 5.5 No load voltage as a function of SOC-level, max and min terminal voltage, terminal
voltage during constant current during 3 and 5C discharge rates, as well as during a
constant EM max power discharge. In the right part of figure a for the City car: max
charge and discharge power related to max voltage see (2.28), theoretical (2.27), current
limited at 3 and 5C see (2.29), and finally for the EM max power.
77
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
Table 5.3 Summary of electric machine and transmission data for all three cars, along with vehicle
power-to-weight ratio.
City Highway Sport
γtm 10.3882 7.62 6.22
Pmax 49 kW 97 kW 153
Tmax 140 Nm 280 Nm 420 Nm
Fwheel,max 4750 N 6600 N 7700 N
nmax 12,000 rpm 9475 rpm 10,191 rpm
URM S,max,L−L 207 V 202 V 180 V
IRM S,ph,max 233 A 468 A 786 A
SFLstk 0.405 0.795 1.1925
Lstk
Douter
0.495 0.972 1.458
SFrw 2.32 1.154 0.687
ηmax at 350 V 96.2 % 96.9 % 97.0 %
Vehicle Power-to-weight ratio 41 W/kg 57 W/kg 153 W/kg
The resulting output torque and power as function of speed for all cars, can be seen
in Figure 5.6, and the resulting wheel forces can be seen in Figure 5.7.
78
5.4. Implemented EM models including transmissions
100
50
0
0 2000 4000 6000 8000 10000 12000
Speed (rpm)
Figure 5.6 Max torque and power as a function of speed for the original electric machine and the
scaled machines at a DC voltage level during a full power discharge, at 10 % SOC.
16 8
Sport − Road load
14 7 Sport − Fwheel,max
Highway − Road load
Mass norn Fwheel (N/kg)
12 6 Highway − Fwheel,max
City − Road load
10 5
Fwheel (kN)
City − Fwheel,max
8 4
6 3
4 2
2 1
0 0
0 50 100 150 200 0 50 100 150 200
Speed (km/h) Speed (km/h)
Figure 5.7 Maximum wheel force and power as a function of car speed from the electric machine
at a DC voltage level during a full power discharge, at 10 % SOC.
79
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
90
50 50 1000
Torque (Nm)
Torque (Nm)
95 96
0 0
95 96
−50 −50 1000
90
2000
80
70
Figure 5.8 Implemented electric machine efficiency and power loss maps, for the City car.
4000
90
200 200
95
2000
96 97
Torque (Nm)
Torque (Nm)
100 100
0 0
−100 97 −100
96
2000
95
−200 −200
90
70
80
4000
−300 −300 6000
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 5.9 Implemented electric machine efficiency and power loss maps, for the Sport car.
80
5.5. Implemented converter model
City
8
80
70
Highway
87
6 Sport
Figure 5.10 Total power train efficiency, including electric machine, converter and transmission,
for the three concept cars, as a function of speed and normalized wheel force.
Table 5.4 IGBT module parameters, for all three cars, where the data for the Sport car represent all
four modules.
City car: Highway car: Sport car:
IGBT Diode IGBT Diode IGBT Diode
Ic,nom 400 A 800 A 4*600 A
Iref : 400 A 550 A 4*400 A
Vref : 300 V 300 V 300 V
On-state threshold voltage 0.709 V 0.803 V 0.673 V 0.761 V 0.66 V 0.775 V
On-state resistance 2.468 mΩ 1.999 mΩ 1.336 mΩ 1.12 mΩ 0.441 mΩ 0.352 mΩ
Turn-on Energy loss 4.2 mJ - 12 mJ - 44 mJ -
Turn-off Energy loss 16.0 mJ 7.25 mJ 25 mJ 9.5 mJ 68 mJ 28 mJ
The total power losses are estimated using (2.20) to (2.23), and can be seen together
with energy efficiency in Figure 5.11 for the City car and in Figure 5.12 for the Sport car,
in the machine torque and speed frame of reference, while the results for the Highway
converter lies in between. The converter losses mainly increase with increasing torque,
81
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
while the efficiency has a strong speed dependency and increases with increasing speed.
Unlike the electric machine, the converter losses vary somewhat depending on the DC
voltage level in the whole operating range. The efficiency is mostly higher at lower volt-
age levels in the same torque and speed operating point, due to lower switching losses.
95
80
98
50 50 500
Torque (Nm)
Torque (Nm)
70
200
98
0 0
−50 −50
−100 −100
−150 −150
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 5.11 Implemented converter efficiency and power loss maps at three DC voltage levels, for
the City car.
V = 412 V V = 412 V
DC DC
400 VDC = 283 V 400 VDC = 283 V
6000
300 300
4000
80
200 200
70
90
95
97
Torque (Nm)
1000
97
0 0
−100 −100
−200 −200
−300 −300
−400 −400
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 5.12 Implemented converter efficiency and power loss maps at three DC voltage levels, for
the Sport car.
82
5.6. Simulator structure
of 0.01 s, which proved to provide reasonable stability. For the logged cycles, also the
estimated road grade levels as a function of time is an input to the simulations.
In order to allow deviation from the speed reference a PI driver model is imple-
mented. Its input is the difference between the reference speed and the simulated speed.
The proportional Kp and integral Ki gains are chosen such that the reference speed
and associated acceleration are followed fairly close by the simulation output (where
Kp = Ki = 3). The output of the driver model is a signal between −1 to 1.
- EM Torque
- EM Speed - EM Power - Inverter Power
- Battery Voltage
- Battery SOC
Transmission EM Inverter Battery - Battery Energy
- Battery Voltage
The positive output from the driver model is regarded as an acceleration pedal sig-
nal and the negative part as a brake pedal signal. These signals are then scaled by the
maximum torque from the electric machine in motoring and generating mode, which are
implemented through look-up tables as functions of machine speed and DC-link voltage.
Thus it gives the electric machine torque reference signals, via linear interpolation. It is
then assumed that the electric machine can fulfill the torque reference. No wheel slip
is assumed, hence the electric machine speed is simply found via the simulated vehicle
speed, the wheel radius and the transmission ratio.
At each time step the sum of the resistive forces on the wheel (the vehicle’s aero-
dynamic drag, rolling resistance and possibly also grading force) is converted to a total
resistive wheel torque. Given the total resistive torque together with the wheel torque
from the electric machine via the transmission, and the vehicles estimated mass inertia,
(as the product of the vehicle mass and the square of the wheel radius, while ignoring
additional rotating inertias), the resulting vehicle acceleration is calculated (as the ro-
tational form of (2.1)). Furthermore, the simulated speed is then calculated as the time
integral of the resulting acceleration.
Both the electric machine and converter power losses are implemented trough look-
up tables that are functions of machine torque and speed, as well as DC-link voltage,
where the output is found through linear interpolation. The power flow between electric
machine and converter is calculated as the added or subtracted losses to the mechanical
machine output, depending on mode of operation. Likewise is the DC-link power flow
calculated via the additional converter losses. The DC-link power flow is then the input
to the battery model, where the current is found from division of the power with the
terminal voltage from the previous time step. The current is then used to estimate the
83
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
SOC level, (as in (2.26)) as well as battery conduction losses together with the discharge
and charge internal resistances. Finally, the terminal voltage is found as the subtracted
resistive voltage drop from the SOC dependent open circuit voltage.
Next follows results from the simulations in terms of requirement and drive cycle
fulfillment as well as powertrain and energy consumption for the different drive cycles.
EM F Calc.
6 wheel, max. 120
46kW calc. Fwheel, max. Sim. 10% SOCinit.
4.5 kN
5 Road load 100 Sim. 90% SOCinit.
Sim. Fwheel SOCinit. 10%
4 Sim. F SOC 90% 80
Speed (km/h)
wheel init.
Fwheel (kN)
3 60
2 40
3.9s 11.3s
4.0s 12.8s
1 20
4.1s 13.0s
0 0
0 20 40 60 80 100 120 0 5 10 15
Speed (km/h) Time (s)
Figure 5.14 Comparison of acceleration test from 0 − 100 km/h, between calculation and simu-
lation, as well as maximum force on wheels from the electric machine, for the City
car.
For the City car, the time to accelerate from 0 − 100 km/h by simulation was 12.8 s,
at a battery SOC level of 10 % see Figure 5.14. This is about 1.5 % faster than the accel-
eration requirement for the city car, which was 13 s. As can be seen in the figure, above
base speed the simulated wheel force is somewhat larger than the ideally calculated force,
hence the acceleration will be faster. The simulated wheel force is even larger at the 90%
SOC level, resulting in an acceleration time of 11.3 s, i.e. 7.8 % shorter then at 10 %
SOC. The simulated acceleration time to 50 km/h is 4 s at 10 % SOC and only slightly
smaller at the 90% SOC level, 3.9 s.
84
5.8. Simulated time to accelerate 0 − 100 km/h with grade
8 120
89 kW Calc.
7 6.3 kN Sim. 10% SOCinit.
52 km/h 100
6 Sim. 90% SOCinit.
80
Speed (km/h)
(kN)
5
EM Fwheel, max.
4 60
wheel
3
Road load 40
2 Sim. Fwheel SOCinit. 10%
3.9s 8.9s
Sim. Fwheel SOCinit. 90% 20
1 3.9s 9.7s
3.9s 10.0s
0 0
0 20 40 60 80 100 120 0 2 4 6 8 10 12
Speed (km/h) Time (s)
Figure 5.15 Comparison of acceleration test from 0 − 100 km/h, between calculation and simula-
tion, as well as maximum force on wheels from the electric machine, for the Highway
car.
Also the Highway car has a slightly shorter acceleration time to 100 km/h compared
to the ideal calculation; 9.7 s, which is 3 % faster. The difference between the two SOC
levels is even larger than for the City car, where the time for the 90% SOC level is 8.9 s
which is 8.2 % faster than at 10 % SOC. The time to 50 km/h is the same for all cases.
15 120
Calc.
290 kW
14.6 kN Sim. 10% SOCinit.
100
Sim. 90% SOCinit.
Speed (km/h)
10 80
Fwheel (kN)
EM Fwheel, max.
calc. Fwheel, max. 60
5 Road load 40
Sim. Fwheel SOCinit. 10% 1.9s
1.9s 3.8s
Sim. Fwheel SOCinit. 90% 20
3.9s
1.9s
4.0s
0 0
0 20 40 60 80 100 120 0 1 2 3 4 5
Speed (km/h) Time (s)
Figure 5.16 Comparison of acceleration test from 0 − 100 km/h, between calculation and simu-
lation, as well as maximum force on wheels from the electric machine, for the Sport
car.
Finally for the Sport car the time to 50 km/h is the same for all cases, while the time
to 100 km/h is 3.9 s and 3.8 s for the 10 and 90 % SOC respectively, which means about
2.5 and 5 % faster then the requirement.
85
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
6 120 Calc.
46kW EM Fwheel, max.
4.5 kN Sim. 10% SOCinit.
5 calc. Fwheel, max. 100 Sim. 90% SOCinit.
Road load
4 80
Speed (km/h)
Fwheel (kN) Sim. Fwheel SOCinit. 10%
2 40
4.8s 16.1s
1 20
5.0s 19.9s
4.9s 20.2s
0 0
0 20 40 60 80 100 120 0 5 10 15 20 25
Speed (km/h) Time (s)
Figure 5.17 Comparison of acceleration test from 0 − 100 km/h, between calculation and simula-
tion, as well as maximum force on wheels from the electric machine, for a road grade
of 6 %, for the City Car.
The City car was to be able to sustain a speed of 90 km/h in a uphill grade of 6 %,
which it does. In this case it takes about 5 s for the City car to reach 50 km/h and about
20 s to reach 100 km/h at 10 % SOC. The higher SOC level gives only minor effects on
the time to 50 km/h, while it needs 3.8 s less to reach 100 km/h compared to the 10 %
SOC.
8 140 Calc.
89 kW EM Fwheel, max.
Sim. 10% SOCinit.
7 6.3 kN calc. Fwheel, max. 120
Sim. 90% SOCinit.
6 Road load
100
Sim. Fwheel SOCinit. 10%
Speed (km/h)
5
Fwheel (kN)
Figure 5.18 Comparison of acceleration test from 0 − 100 km/h, between calculation and simula-
tion, as well as maximum force on wheels from the electric machine, for a road grade
of 6 %, for the Highway Car.
The Highway car is able to sustain the uphill grade of 6 % at the speed of 130 km/h.
In this case the acceleration time difference between the no grade case is much smaller
compared to the City car. Also less difference is noted between the two SOC levels, only
1.3 s for 100 km/h, and none for 50 km/h.
86
5.9. Fulfillment of reference cycle speed in simulation
15 140 Calc.
290 kW Sim. 10% SOCinit.
14.6 kN 120
Sim. 90% SOCinit.
100
10
Speed (km/h)
Fwheel (kN)
EM Fwheel, max. 80
calc. Fwheel, max.
60
Road load
5
Sim. Fwheel SOCinit. 10% 40
Sim. Fwheel SOCinit. 90% 2.4s 4.7s
20
2.4s 4.9s
2.0s 4.4s
0 0
0 50 100 150 0 2 4 6 8
Speed (km/h) Time (s)
Figure 5.19 Comparison of acceleration test from 0 − 100 km/h, between calculation and simula-
tion, as well as maximum force on wheels from the electric machine, for a road grade
of 6 %, for the Sport Car.
Finally, the Sport car can handle the 12 % uphill grade at 130 km/h. The time to
accelerate to 100 km/h only becomes 0.9 s longer compared to the case with no grade,
for both SOC levels.
87
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
150 150
T V Tmax VDC, max
max DC, max
Tmax VDC 350 V Tmax VDC 350 V
100 100
Tmax VDC 325 V Tmax VDC 325 V
Artemis URBAN
EM Torque (Nm)
HWFET
EM Torque (Nm)
50 50
NYCC US06
WLTC Cl.3 Low v5.3 REP05
0 FTP72 UDDS Artemis MW130
0
SC03 WLTC Cl.3 ExtrHgh v5.3
jc08
−50 WLTC Cl.3 Mdl v5.3 −50
ECE
−100 −100
−150 −150
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 5.20 EM operating points during the Urban and Highway cycles for the City car.
For the City car, three points in each of the two cycles US06 and REP05 were outside
of the powertrain’s operational area, causing a maximum deviation of 1.7 km/h from the
reference speed, and 0.4 m/s2 from the reference acceleration. These represent speed and
acceleration levels of 56 km/h at 3 m/s2 , and 80 km/h at 2.2 m/s2 , and finally around
120 km/h at 1.3 m/s2 . The consequence on the average energy consumption is negligi-
ble in these cases. Also one point in the Artemis Motorway cycle during braking at high
speed is outside of the operating area of the electric machine, see Figure 5.20. The oper-
ating points of the City car electric machine can be seen as gray dots in Figure 5.20 for
the Urban and Highway cycles. In the same figure also ideal electric machine operating
points for a lossless system is depicted for comparison, and color marked depending on
the type of cycle.
300 300
Tmax VDC, max Tmax VDC, max
Tmax VDC 350 V Tmax VDC 350 V
200 Tmax VDC 325 V 200 Tmax VDC 325 V
Artemis URBAN HWFET
EM Torque (Nm)
NYCC US06
100 100
EM Torque (Nm)
−200 −200
−300 −300
0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000
Speed (rpm) Speed (rpm)
Figure 5.21 EM operating points during the Urban and Highway cycles for the Highway car.
The US06 and REP05 cycles caused some deviations also for the Highway car of
maximum 1.5 km/h and 0.2 m/s2 , however in this case the operating points that cause
problem are those that demand high acceleration (almost 4 m/s2 ) from stand still, see
Figure 5.21. The problem appears twice in REP05 and once in US06. Also here there is
no effect on the net average wheel energy.
For the Sport car, the operating points from all Test cycles are well within the limits
of the powertrain.
88
5.10. Simulated component efficiency per cycle
Furthermore, all cars are able to follow the Logged Urban cycles, with typical max-
imum speed differences between the reference and simulated cycle as 0.3 − 0.8 km/h.
One Rural cycle for the City car (C30d2 121219 Stnsnd-Henan) and all Highway cycles
have single large speed deviations with a relatively large effect on the average net wheel
energy. The Highway and Sport cars have no major deviations worthy to mention.
82 83 85 84 86 85 79 82 86 87 87 88 86 87 88 89 88 89
100 Conv.
90 Bat.
Efficiency (%)
80 Tm.
PTtot
70
−70
−80
Braking
−90
−100 80 81 83 81 84 83 78 80 83 85 84 86 83 85 84 85 84 85
−110
w
le
gh
gh
AN
8
03
06
E
CC
DC
DC
30
jc0
P0
Lo
FE
RA
dd
EC
DD
Hi
Hi
SC
LA
US
s1
RB
NY
NE
EU
RE
Mi
HW
RU
TC
2U
tra
TC
mi
UC
sU
TC
WL
Ex
WL
P7
te
is
mi
Ar
WL
tem
FT
TC
te
Ar
WL
Ar
Figure 5.22 Total powertrain efficiency per cycle in propulsion vs. braking mode (negative y-axis),
and efficiency broken down per component, from simulation of the City car, with 90 %
initial SOC, for the Test cycles.
89
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
80 Tm.
PTtot
70
−70
−80
Braking
−90
−100 77 79 82 79 83 81 74 77 82 83 84 86 81 84 83 85 83 86 85
−110
w
le
gh
gh
AN
8
03
92
06
E
CC
DC
AL
DC
30
50
jc0
P0
Lo
FE
dd
EC
DD
Hi
Hi
SC
LA
US
1
UR
RB
NY
NE
EU
RE
Mi
HW
is
is
TC
2U
tra
TC
UC
R
tem
tem
U
TC
WL
Ex
WL
P7
is
is
Ar
Ar
WL
tem
tem
FT
TC
Ar
WL
Ar
Figure 5.23 Total powertrain efficiency per cycle in propulsion vs. braking mode (negative y-axis),
and efficiency broken down per component, from simulation of the Highway car, with
90 % initial SOC, for the Test cycles.
76 78 81 80 83 82 73 77 82 83 84 84 82 84 85 86 85 86 86
100 Conv.
90 Bat.
Efficiency (%)
80 Tm.
PTtot
70
−70
Braking
−80
−90
−100 72 74 77 75 80 78 70 73 78 80 80 83 78 81 80 80 81 82 82
−110
w
le
gh
gh
N
8
92
06
E
CC
DC
0
jc0
P0
13
15
Lo
FE
RA
dd
EC
DD
BA
Hi
Hi
SC
LA
US
NY
NE
EU
RE
Mi
HW
is
is
RU
TC
2U
UR
tra
TC
UC
tem
tem
TC
WL
Ex
WL
P7
is
is
Ar
Ar
WL
tem
tem
FT
TC
Ar
WL
Ar
Figure 5.24 Total powertrain efficiency per cycle in propulsion vs. braking mode (negative y-axis),
and efficiency broken down per component, from simulation of the Sport car, with
90 % initial SOC, for the Test cycles.
Similarly, the total powertrain efficiency for the Logged cycles can be seen in Figure
5.25 to 5.27, for the City, Highway and Sport cars respectively.
Generally the same trends can also be seen here. Nevertheless, the spread within the
Urban and Rural road type categories is here a bit larger. The absolute efficiency values
per car, are in line with those for the Test cycles, although a bit lower for the Logged
Highway cycles compared to the Highway Test cycles.
90
Ec
ar5 Efficiency (%) Ec Efficiency (%)
00 ar5
12 Braking 00
04 Propulsion 12
04
Braking
13
L 13 Propulsion
Le inn Le Lin
af eg af ne
−110
−100
−90
−80
−70
70
80
90
100
110
13 −C g−
0
−110
−100
−90
−80
−70
70
80
90
100
110
32 ity 13 Ci
Le 1 −C 03 ty−
af CT Le 21
TH af CT CT
13
03 H− 13 H− H
79
76
Pr Er 03
79
81
ius 22 ks Pr 22 Er
K ius ks
Urban
br
12 g br
Urban
0 or 12 Ko
g
32 s v− 03 rsv
82
80
Pr Pr 2
82
84
ius 9M Gu ius 9M −G
12 lnd ldh 12 uld
03 l − 03
lnd h
80
77
Pr 29 Ml l−M
80
82
85
83
V6 04 ke
84
86
0P 03 −M V6 03 −M
HE Ml lnd 0P lnd
VK nlc l HE Ml
nlc l
84
83
ke VK
84
85
rsv −C rsv ke
V7 gn TH V7 −C
44 −G 44 gn TH
12 a −G
85
83
12
84
86
V7 03 lvB 03 alv
12 ro V7 12 Br
44 Cs n 44 Cs on
12
78
77
tat 12
80
81
V7 03 −C 03 tat
−C
18 TH V7 18
44 Kl 44 Kl TH
12 ltr
78
75
12 ltr
78
80
03 p− 03 p−
27 CT 27 CT
CT H CT H
85
XC
83
H− XC
84
86
6 Kl H−
C3 0C ltr C3 60 Kl
C3
0d TH p 0d C TH
ltr
p
21 − C3
84
21
80
82
86
0d 21 Kr 0d 21 −K
21 21 ths 21 21 rth
21 9C g 21 9C sg
21 T 21 T
76
76
79
79
9H H− 9H H−
C3 en VC C3 en VC
86
21
83
−S
84
87
Ec 21 tns Ec 21
Rural
21 tns
Rural
ar5 nd ar5 21
C3 00 9V
C3 0 01 9 VC
nd
87
12 CC
85
84
87
0d 04 −C 0d 20 C−
85
TH
82
83
86
9S H− 9S −K
Kl
V7
44
tns l rd V7 tns llrd
nd 44 nd
87
83
84
87
12 −H 12 −H
V7 03 en V7 03 en
44 07 an 44 07 an
12 VC 12 VC
87
84
82
86
V7 03 C− 03 C−
16 CT V7 16 CT
44 Kl H 44 Kl H
12 ltr 12 ltr
0
86
82
81
86
32 p− 03 p−
V7 0 Cs V7 20 Cs
44 Kl ttn 44 Kl ttn
120 l trp 85 12 ltr
84
84
85
C3 32 −C C3 03 p−
0d 7K stt 0d 27 Cs
21
21 lltr n 21
21
Kl
ltr
ttn
85
85
85
85
C3 21 p− C3 21 p−
0d 9S Cs 0d 9S Cs
21 tns ttn 21 tns ttn
21 21
87
nd nd
86
86
87
V7 21 −V V7 21 −V
01 9V CC 01 9V CC
20 CC 20 CC
85
−S
83
−S
85
82
V7 51 tns V7 51 tns
0C 0C
01
20 T H−
nd 01
20 T H−
nd
85
Highway
82
85
82
Highway
51 Jn 51 Jn
5J kp 5J kp
nk ng nk ng
pn pn
86
82
86
81
g− g−
CT CT
H H
85
81
85
81
PT
EM
EM
Tm.
Bat.
Tm.
Bat.
91
5.10. Simulated component efficiency per cycle
tot
PTtot
Conv.
Conv.
and efficiency broken down per component, from simulation of the Highway car, with
Figure 5.26 Total powertrain efficiency per cycle in propulsion vs. braking mode (negative y-axis),
and efficiency broken down per component, from simulation of the City car, with 90 %
Figure 5.25 Total powertrain efficiency per cycle in propulsion vs. braking mode (negative y-axis),
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
80 Tm.
PTtot
70
−70
−80
Braking
−90
−100 71 76 73 79 79 79 72 71 80 76 72 79 80 76 80 77 75 79 80 82 74 75 72 72
−110
H
on
nd
H
ttn
n
ttn
ng
H
h
l
TH
g
lrd
CC
nd
an
H
C
g
lnd
ck
ltr
stt
uld
CT
br
ths
T
CT
CT
CT
CT
VC
Br
tns
kp
en
Cs
Cs
ns
−C
Kl
−C
−V
Kl
ks
lnl
−M
−C
C−
p−
C−
g−
ty−
−G
d−
Kr
alv
H−
Jn
−H
t
p−
p−
H−
tat
ke
−S
TH
Er
−S
l−M
p
ke
ltr
pn
VC
H−
VC
sn
rsv
−
Ci
CT
G
nd
ltr
ltr
ltr
nlc
H−
Cs
CT
CC
an
9C
TH
Kl
c
nk
n
nd
g−
CT
Kl
Kl
Kl
lnl
tns
07
o
9
gn
CT
t
en
Ml
13
12
9V
0C
9S
18
27
21
2K
5J
1
ne
Ml
16
20
27
03
9M
12
60
rsv
04
9H
9S
03
03
03
1
21
21
Lin
21
51
3
03
03
32
12
29
2
2
XC
12
5
12
0
2
04
12
12
VK
21
03
21
21
21
21
21
20
20
12
12
2
0
03
03
44
13
1
13
12
00
13
44
21
21
21
44
44
0d
21
01
01
0d
12
12
HE
44
44
44
V7
04
ar5
V7
21
21
af
ius
V7
V7
C3
af
C3
0d
0d
V7
V7
V7
V7
V7
ius
ius
0P
12
Le
Le
Ec
0d
0d
Pr
C3
C3
V6
Pr
Pr
00
C3
C3
ar5
Ec
Figure 5.27 Total powertrain efficiency per cycle in propulsion vs. braking mode (negative y-axis),
and efficiency broken down per component, from simulation of the Sport car, with
90 % initial SOC, for the Logged cycles.
92
5.11. Simulated energy per driven distance, per cycle
or the highest levels of acceleration and RPA value, within the same road type, and in the
same time they often have low maximum and average levels of speed. This is valid for
NYCC, Artemis Urban, UC LA92 and US06.
Battery energy/dist. (Wh/km)
250
Discharge
Urban Rural Highway
200 Charge
171
168
Net
157
155
155
150
148
143
142
136
150
127
121
119
118
117
116
111
110
109
109
105
100
98
97
97
94
90
87
87
86
100
80
77
76
75
71
67
68
65
45
42
41
41
50
35
34
34
32
29
23
23
21
18
17
12
7
0
w
le
igh
gh
N
8
92
06
E
DS
CC
DC
DC
0
jc0
P0
13
Lo
FE
RA
dd
EC
BA
Hi
SC
LA
US
aH
UD
NY
NE
EU
RE
Mi
HW
is
RU
C
UR
TC
UC
tem
T
tr
2
TC
WL
Ex
WL
P7
is
is
Ar
WL
tem
tem
FT
TC
Ar
WL
Ar
Figure 5.28 Average powertrain propulsion, braking and net energy consumption per driven dis-
tance, from simulation of the City car, with 90 % initial SOC, for the Test cycles.
Battery energy/dist. (Wh/km)
250
225
225
Discharge
215
210
201
Urban Rural Highway
Charge
191
186
184
182
200
174
Net
166
165
163
160
157
155
152
150
149
148
148
144
137
133
150
128
127
126
120
116
116
113
110
106
104
101
100
98
96
93
93
100
64
59
59
59
52
49
48
44
44
50
34
33
32
28
27
26
20
12
0
w
le
gh
gh
N
8
03
92
ET
06
E
DC
AL
DC
05
50
jc0
13
Lo
dd
EC
DD
BA
C
Hi
Hi
SC
LA
US
P
s1
F
UR
NY
NE
EU
RE
Mi
HW
is
TC
2U
UR
tra
TC
mi
UC
sR
tem
TC
WL
Ex
WL
P7
te
is
mi
Ar
Ar
WL
tem
FT
TC
te
Ar
WL
Ar
Figure 5.29 Average powertrain propulsion, braking and net energy consumption per driven dis-
tance, from simulation of the Highway car, with 90 % initial SOC.
93
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
275
271
Urban Rural Highway Discharge
258
252
243
Charge
231
226
225
222
215
206
204
203
Net
198
193
192
190
190
186
185
185
182
182
200
177
172
165
163
157
152
152
149
148
144
141
137
137
135
133
99
92
100
66
60
60
56
54
48
48
44
44
34
33
32
29
28
26
19
11
0
w
le
gh
gh
N
8
03
92
06
E
CC
DC
DC
0
jc0
P0
13
15
Lo
FE
RA
dd
EC
DD
BA
Hi
Hi
SC
LA
US
NY
NE
EU
RE
Mi
HW
is
is
RU
TC
2U
UR
tra
C
UC
tem
tem
T
TC
WL
Ex
WL
P7
is
is
Ar
Ar
WL
tem
tem
FT
TC
Ar
WL
Ar
Figure 5.30 Average powertrain propulsion, braking and net energy consumption per driven dis-
tance, from simulation of the Sport car, with 90 % initial SOC.
Similarly, the discharge, charged and net battery energy flow for the Logged cycles
can be seen in Figure 5.31 to 5.33, for all three cars. Perhaps more examples are needed
in order to draw any general conclusions though.
The energy consumption coherence between the Logged road types, is not as high as
for the Test cycles, except when it comes to the Highway cycles which are very coherent.
Not to forget, for the Logged cycles, not only speed and acceleration but also road grade
levels are included in the energy consumption figures shown.
The two cycles with the highest energy consumption for all three cars are the Urban
classified cycles; Leaf 130322 Korsv-Guldh and V744 120312 Cstat-CTH. Also the cy-
cle Ecar500 120413 Linneg-City-CTH has a relatively high energy consumption. From
Figure 4.15, the reason is a rather long uphill climbing which in these short cycles has a
major impact on the energy consumption.
It can also be seen that for the three cycles V744 Klltrp-Csttn at different days, two
of them have similar results while the third (120327) have a higher consumption, even
though the average powertrain efficiency is somewhat higher for the third case compared
to the first two. In Table 3.3 it can be seen that the speed levels are similar, while as seen
in Table 3.5 the maximum and average levels of acceleration as well as the RPA value is
higher in the third case, hence the higher consumption.
94
Ec Ec
a r50 ar5
01 00
20 12
41 Battery energy/dist. (Wh/km) 04 Battery energy/dist. (Wh/km)
3L 13
Le inn Le Lin
af1 eg af ne
13 g−
0
100
200
300
0
50
100
150
200
30 −C 03 Ci
Le 32 ity Le 21 ty−
af 1C −C af CT CT
13 TH TH 219 13 H 152
Pr 03 −E 73 Pr 03 H− 53
ius 22 rks 146 ius 22 Er
ks 99
12 Ko br 173 12 Ko br 124
Urban
Urban
Pr 03 rsv g 69 Pr 03 rsv g 49
ius 29 −G 104 ius 29 −G 75
12 Ml uld 12 Ml uld
237 165
cycles
nd h nd h
03 l−M 69 03 l−M 50
Pr
ius
29
lnl 168 Pr
ius
29
lnl 115
Ml ck Ml ck
12 nlc e 171 12 nlc e 123
V6 04 ke 49 V6 04 ke 34
0P 03 −M 122 0P 03 −M 89
HE Ml lnd 159
HE Ml lnd 114
VK nlc l 72 VK nlc l 49
rsv ke
−C 87 rsv ke
−C 65
V7 gn V7 gn
44 TH 157 44 TH 114
12 −G
a 59 12 −G
a 40
V7 03 lvB 99 V7 03 lvB 73
12 ro 12 ro
44 n 190 44 n 132
12 Cs 75 12 Cs 54
V7 03 tat
−C 115 V7 03 tat
−C 78
18 18
44 Kl TH 266 44 Kl TH 184
12 l 102 12 l 74
03 trp
−C 164 03 trp
−C 110
27 27
CT TH 157 CT TH 114
XC 63 XC 44
C3 60 H−
Kl 94 C3 60 H−
Kl 70
0d
2
CT ltr
p 159 0d
2
CT ltr
p 115
C3 H− 54 C3 H− 38
0d
21
12
12 Kr 105 0d
21
12
12 Kr 76
19 ths 19 ths
21 g 189 21 g 130
21 CT 94 21 CT 68
C3 9H H− 95 C3 9H H− 62
en VC en VC
0d an C 169 0d an C 125
21 61 21 41
Ec 21 −S
tns 108 Ec 21 −S
tns 84
ar5 21 ar5 21
Rural
Rural
00 9V nd 141 00 9V nd 109
C3
12 31 C3
12 20
0d 0
CC
−C 110 0d 0
CC
−C 90
21 41 21 41
21 3C TH 166 21 3C TH 120
21 TH 53 21 TH 35
9S −K 113 9S −K 85
V7 V7
177 134
tns llrd tns llrd
44 nd 60 44 nd 40
V7 1 20 − He 118 V7 1 20 − He 95
44 30 na 44 30 na
7V n 179 7V n 122
12 54 12 26
V7 03
16
CC 125 V7 03
16
CC 95
44 −C 44 −C
Kl
l TH 152 Kl
l TH 113
12 t rp 46 12 t rp 31
V7 03
20 −C 106 V7 03
20 −C 82
44 Kl stt 140
44 Kl stt 105
C3 12 ltr n 72 C3 12 ltr n 49
0d 03
2
p− 67 0d 03
2
p− 55
7 Cs 7 Cs
21
2 Kl
l ttn 131 21
2 Kl
l ttn 100
12 trp 64 12 trp 43
C3
0d 1 − 67 C3
0d 1 − 56
9S Cs 9S Cs
21 tns ttn 181 21 tns ttn 134
21 105 21 71
V7 21 nd
−V 76 V7 21 nd
−V 63
9V 9V
01 C CC 161 01 C CC 116
20
5 C− 23 20
5 C− 14
V7 10 St 139 V7 10 St 102
ns ns
01 CT nd 161 01 CT nd 115
20 H− 21 20 H− 14
51 Jn 140 51 Jn 101
Highway
Highway
5J kp 5J kp
nk ng 163 nk ng 117
pn 20 pn 12
g− 143 g− 105
CT CT
H 153 H 109
17 11
136 98
Net
Net
Charge
Charge
95
5.11. Simulated energy per driven distance, per cycle
tance, from simulation of the City car, with 90 % initial SOC, for the Logged cycles
Discharge
Discharge
tance, from simulation of the Highway car, with 90 % initial SOC, for the Logged
Figure 5.32 Average powertrain propulsion, braking and net energy consumption per driven dis-
Figure 5.31 Average powertrain propulsion, braking and net energy consumption per driven dis-
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
318
Urban Rural Highway Discharge
285
300 Charge
267
Net
232
230
219
219
218
216
214
211
209
207
207
206
204
201
199
195
195
193
193
192
189
189
187
185
183
200
175
174
169
167
161
159
154
151
146
145
144
142
142
138
134
129
122
113
106
103
103
102
93
100
73
72
71
71
69
67
64
63
60
60
58
53
50
50
48
43
29
21
20
18
15
0
H
on
TH
nd
H
ttn
n
ttn
ng
H
h
l
TH
g
llrd
CC
nd
an
TH
CC
g
lnd
ck
ltr
stt
uld
CT
br
ths
CT
CT
CT
Br
ns
kp
en
Cs
Cs
ns
−C
−C
−C
−V
−V
Kl
−K
ks
lnl
−M
−C
C−
p−
C−
g−
−G
Kr
alv
St
Jn
−H
St
p−
p−
H−
tat
e
ity
TH
d
Er
l−M
TH
rp
ke
lltr
pn
VC
−
VC
sn
C−
n−
rsv
−
−G
nd
ltr
ltr
nlc
H−
Cs
C
TH
CT
9C
TH
3C
nlc
nk
K
n
nd
g−
na
VC
Kl
Kl
Kl
tns
07
Ko
9
gn
CT
St
C
Ml
12
0C
18
27
21
5J
21
ne
Ml
Ml
41
He
6
20
27
03
60
sv
19
9S
9
03
1
22
03
03
03
21
21
21
20
Lin
21
51
51
03
03
03
12
29
29
XC
r
2
12
03
04
12
12
VK
21
03
21
21
21
21
01
21
20
20
12
12
2
03
03
44
13
1
13
12
13
44
21
21
21
44
44
0d
21
01
01
d
12
12
0
HE
44
44
44
V7
04
ar5
V7
21
21
af
ius
V7
V7
C3
af
C3
0d
0d
V7
V7
V7
V7
V7
ius
ius
0P
12
Le
Le
Ec
0d
0d
Pr
C3
C3
V6
Pr
Pr
00
C3
C3
ar5
Ec
Figure 5.33 Average powertrain propulsion, braking and net energy consumption per driven dis-
tance, from simulation of the Sport car, with 90 % initial SOC, for the Logged cycles
96
5.12. Simulated driving range
180
ηPT, avg. − Prop. (%)
X: 16
160 Y: 174.6 ηPT, avg. − Brake (%)
Efficiency, Speed, SOC level
140 ηPT, avg. − Net (%)
60
40
20
0
0 2 4 6 8 10 12 14 16
Number of NEDC cycles
Figure 5.34 Average powertrain propulsion, braking and net efficiency from simulation of City car
NEDC driving range.
Based on the simulation results of net battery energy consumption per driven dis-
tance, possible driving range has been calculated for all Test cycles with the City and
Highway cars, see Figure 5.35 and 5.36. In this illustrative example, the battery en-
ergy content is estimated to 80 % of 19.3 kW h, which here results in a NEDC range
of 178 km. The discrepancy with the value in Table 5.5, may be due to the assumed en-
ergy content of the battery which in the simulation varies depending on the load situation.
It can be seen that the lower energy consumption for the Urban cycles generally provides
20 − 50 km more than the NEDC cycle, while the range is shortened by about 50 km for
most Highway cycles.
Similarly, for the Highway car, the range increase for most Urban cycles is 10 −
30 km, and the decrease is about 20 − 90 km for the Highway cycles, see Figure 5.36.
For comparison, the estimated driving range of the Logged cycles for the City car
can be seen in Figure 5.37. The ranges are about 130 − 250 km, 160 − 280 km and
150 − 160 km for the Logged Urban, Rural and Highway cycles respectively, i.e. a larger
variety than for the Test cycles.
97
Ec
ar5
00
98
12
04
13 Battery range. (km)
Le
Lin
ne
Battery range. (km) Battery range. (km)
a
0
100
200
300
f1 g−
30 Ci
ty
0
100
200
300
32 −C
0
100
200
300
Le 1C TH
af TH
156
13 −E EC
03 rks EC
Pr 22 E
Urban
ius Ko br
g
WL WL E
r
206
12 sv TC
03 −G TC
Pr 29 Lo
ius Ml uld
h w Lo
w
257 248
nd
134
12
237 232
03 l−M
Pr 29 lnl FT jc0
Ml ck
ius nlc e P7 8 FT jc0
173
12
04 ke
−M 2U P7 8
V6 03 lnd 2U
0P Ml l
WL DD
HE S
237
nlc DD
k TC WL
VK
rsv e−
CT Mi TC S
V7 gn H dd
−
210
44 Ga le Mi
12 dd
197
44 Cs 03
12 tat SC
−C
225
03
V7 18 TH Ar 03
140
207 200 204 194
44 Kl NY
12 ltr tem CC Ar
03 p− is NY
27 CT tem
H UR CC
221
CT is
XC H− BA
60 K lltr N UR
C3 p
180 189
Urban
CT BA
203
0d H− N
C3 21 Kr NE
21
Urban
0d 21 ths DC
21 9C g WL NE
250
21
21 TH Ar T DC
9H −V WL
C3 en CC tem C Hi
gh T
184
0d an is Ar
21 −S tem C Hi
Rural
21 tns RU gh
Ec 21 nd RA is
ar5 9V
172
00 C L RU
217 207 212
C3 12 C− RA
0d 04 CT
21 13 H EU L
181
177 179 178 165 172
21 CT DC
21 H− EU
9S Kl UC
t ns l r d DC
V7
163
44 nd LA
12 −H 92 UC
en
Rural
V7 03 an
44 07 HW LA
162
12 VC FE 92
03 C−
T
Rural
16 CT
H
199 190 206
V7 Kl
HW
44
187
ltr
WL RE FE
12 p−
C
03 TC P0 T
20 5
155 158 159
stt
V7 K n Ex
159
44
WL
279
12
lltr tra RE
03 p− TC P0
C3 Cs Hi
0d 27
Kl t t n gh Ex 5
273
21 ltr tra
21 p−
C3 21 Cs US Hi
0d 9S ttn Ar gh
246
21 tns
21 nd tem 06
21 −V is US
V7 9V CC
CC 1
152
01 Ar Ar
Figure 5.35 Estimated driving range for the Test Cycles with the City car.
20 −S
121 108 113
tem 06
144 147 137
51 tem 30
Chapter 5. Powertrain Component Sizing, Modeling and Vehicle Simulation
tns
Figure 5.37 Estimated driving range for the Logged Cycles with the City car.
V7 0C nd is is
TH Figure 5.36 Estimated driving range for the Test Cycles with the Highway car.
Highway
153
01 15 13
20 −J 0 0
Highway
51 nk
103
130
5J pn
g
Highway
nk
147
pn
g−
CT
H
157
Chapter 6
99
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
wayII BEVs compared to the City and Highway BEVs in Chapter 4. They all share the
same curb weights, NEDC range, and gradeability requirements. In contrast, both the
CityII and HighwayII BEVs have slightly lower top speed values, lower tire rolling resis-
tance coefficients, and a bit larger aerodynamic drag coefficients. Other parameters are
subject to only small changes.
As in Chapter 4, three different combinations of max force and power are investi-
gated, in order to reach the required acceleration times; i.e. base speed values that are
1/4, 1/3 and 1/2 of the top speed.
In order to prioritise the low speed performance of the CityII BEV, the relation be-
tween base and top speed is chosen to be small: i.e. 1/4. This gives a 0-50km/h acceler-
ation time of 3.9s, and 31m, instead of 4.4s or 5.7s for the other two considered cases.
The max wheel power needed is then about 45 kW and the max force about 5 kN, which
gives a max wheel torque of about 1.5 kNm.
For the HighwayII BEV, the case of base speed as being 1/3 of the top speed is chosen.
This gives a 0-50km/h acceleration time of 3.8s and 34m, instead of 3.3s or 4.7s for the
other two cases. The max wheel power needed is then about 87 kW, the max wheel force
is 6.5 kN, and the max wheel torque is 2 kNm.
Thus, both the CityII and HighwayII concept BEVs have similar low speed acceler-
ation performances, whereas the HighwayII BEV have better high speed performance.
Consequently, the 50-100km/h acceleration time is around 6s for the HighwayII BEV
and about 9s for the CityII.
Table 6.1 Prerequisites for the new concept CityII and HighwayII BEVs used in Chapter 7, 8, and
9.
CityII HighwayII
Seats 4-5 5
Curb weight 1200 1700 kg
Acceleration 0-100 km/h 13 11 s
Top speed 130 145 km/h
NEDC Range 160 200 km
Aearod. drag coeff. Cd 0.31 0.29 −
Area 2.1 2.3 m2
Wheel radius 0.30 0.32 m
Rolling resist. coeff. Cr 0.007 0.007 −
Starting gradability 25 % 25 % %
Gradability (Speed at grade) 90 at 6% 130 at 6% km/h, %
100
6.2. Design, material and mass of machine parts
Energy science and energy laboratory ORNL (Oak Ridge National Laboratory), cover-
ing the electric powertrain in HEVs such as Toyota’s Prius 2004 and 2010 and Camry
2007, Honda Accord 2014, Lexus LS 600h 2008, as well as the BEV Nissan Leaf 2011,
e.g. in [98–100]. Although the motors in these vehicle models are of different power rat-
ings and volumes, they share some common design features. They are all IPM (interior
permanent magnet) SM motors with 48 stator slots, 8 poles, and distributed windings.
The magnets in the rotor are configured in a V-shape in all models, except in Leaf and
LS 600h where also a third magnet is used close to the air gap (forming a ∇-shape). This
machine type is known to fulfill design goals such as high torque and power density, high
efficiency and wide constant power region, which are all highly desired in EVs [101].
The reference IPM SM stator and rotor core lamination designs are similar to that of
the Toyota Prius 2004 motor, with 48 stator slots and 8 poles. However, the stator teeth
and yoke are thinner, and a different angle for the V-shaped magnet arrangement is used.
Furthermore, the three phase winding is distributed, but with two layers, and four parallel
branches. The copper fill factor in the slot area is set to 45%. This is in line with the fill
factor of industrial IMs, which are in the range 35%-45%, according to [102].
The reference machine geometry is scalable, both by core length and diameter, hence
a suitable machine size can easily be found for any desired peak torque performance
and length-to-diameter ratio. One reference machine is designed for the CityII BEV and
one for the HighwayII. Both have length-to-diameter ratios around 0.8, which during
length scaling between 50-200% of the reference length, gives length-to-diameter ratios
between about 0.4-1.6.
The dimensioning is done considering a max dc voltage of 400 V for both motors,
and the max speed is limited to 12,000 rpm. The voltage and speed values are in line with
what has been found for EMs in current BEVs, as shown in Appendix A. The number of
turns per coil is then adjusted in order for field weakening to start at around 3000 rpm
for the CityII EM and 4000 rpm for the HighwayII.
The gear ratios, kgear are simply taken as the relation between the top speed of the
motor and wheel as in (5.4). With an assumed transmission efficiency of 97%, the max
electric motor output torque and power are determined to be about 46 kW and 147.5 Nm
for the CityII BEV, and about 89 kW and 214 Nm for the HighwayII BEV.
The max rms current density Jrms,max in the slot is initially set to 20 A/mm2 (28.3
A peak). The max rms phase current is then proportional to: the max rms current density;
the slot fill factor kf f ; the slot area Aslot ; and the number of parallel branches nnpb , and
inversely proportional to the number of layers nlayer and turns per coil ntpc , as in
Jrms,max kf f Aslot nnpb
Iph,rms,max. = (6.1)
nlayer ntpc
The assumed machine configuration is shown in Figure 6.1, along with an illustration
of the active machine parts, i.e. the parts that contribute to torque production. For sim-
plicity, the machine configuration is limited to the main components: active parts, end
winding, frame, shaft and bearings. Selected machine data are shown in Table 6.2, for
the CityII and HighwayII reference motors.
101
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
Figure 6.1 Axial cross section of machine configuration, and the active machine parts, of V-shaped
reference motors.
Table 6.2 Summary of PMSM data for the CityII and HighwayII BEV motors.
CityII HighwayII
Peak Power 46 kW 89 kW
Max Torque 147.5 Nm 214 Nm
Base speed 3000 rpm 4000 rpm
Max speed 12,000 rpm 12,000 rpm
Gear ratio 10.440 9.984
Max DC voltage 400 V 400 V
Max rms current density 20 A/mm2 20 A/mm2
Max rms phase current 125 A 251 A
Peak Efficiency at 400 Vdc 96.4 % 97.1 %
Core length 136 mm 150 mm
Core stacking factor 0.95 0.95
Lamination thickness 0.3 mm 0.3 mm
Number of slots 48 48
Number of poles 8 8
Stator outer diameter 166 mm 182 mm
Stator inner diameter 112.05 mm 122.85 mm
Stator yoke thickness 9.445 mm 10.365 mm
Tooth width 4.33 mm 4.52 mm
Tooth height 17.53 mm 19.21 mm
Slot opening width 1.66 mm 1.82 mm
Slot area 69.67 mm2 83.74 mm2
Slot fill factor 45% 45%
Turns per coil 10 6
Coils per pole per phase 4 4
Rotor outer diameter 110.81 mm 121.485 mm
Rotor inner diameter 66.4 mm 72.8 mm
Air gap length 0.62 mm 0.6825 mm
Magnet thickness 3.7765 mm 4.1405 mm
Magnet width 14.94 mm 16.38 mm
102
6.2. Design, material and mass of machine parts
6.2.1 Frame
The frame is assumed to be a hollow cylinder with an end cap on each side, all with a
thickness of 11 mm, and to be without any complex textures, for simplicity. The outer
frame diameter is 102 mm and the length 262 mm. Furthermore, it is assumed to be made
of a die cast aluminum alloy (as in Nissan Leaf [94]), for a lightweight yet mechanically
strong construction, with good thermal conductivity. It also includes a spiral cooling
channel for liquid cooling, see Section 8.1.1.
A number of different aluminum alloys are used for this purpose today, and their
net thermal conductivity is less than that of pure aluminum, which is 237 W/mK at
300K [48]. In publications regarding thermal modelling of electric motors, the value of
thermal conductivity of the frame’s aluminum alloy is seen to vary between 150 to 230
W/mK [20, 49, 53, 103]. It is here assumed that the aluminum alloy 195 is used for the
frame, which has a thermal conductivity of 168 W/mK at 300 K, a specific heat of 883
J/kgK, and a density of 2790 kg/m3 , [45, 48].
500
P core, loss (W/kg)
50 Hz
1 400 400 Hz
B (T)
2.5kHz
300 5kHz
10kHz
0.5 200 FEA Approx.
100
0 0
0 500 1000 1500 2000 0 0.5 1 1.5 2
H (A/m) B (T)
Figure 6.2 Data sheet and implemented data for the chosen lamination material, NO30.
6.2.3 Magnets
A permanent magnet is mainly characterized by its remanent flux density, Br (T), and its
ability to withstand demagnetization, or coercivity, Hc (A/m), both of which are temper-
ature dependent. Permanent magnets consisting of NdFeB (neodymium iron boron) are
103
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
commonly used in today’s EVs, so also in Toyotas, Hondas, Lexus’ and Nissans mod-
els [99]. Published measurements indicate that the Br value for the motors in Prius 2010,
LS600h and Camry 2007 are at least 1.3 T at 20◦ C, and at least 1.1 T at 100◦ C, whereas
Hc is between -1220 to -1440 kA/m at 100◦ C [100]. Their temperature capability seem
to be around 180-200◦ C, or maybe even as low as 160◦ C for Prius 2010.
The temperature capability of NdFeB magnets in electric vehicles, is often improved
by substituting some of the Nd for Dy (dysprosium). This additive helps to increase Hc,
and to maintain high values of Hc at higher temperatures, however it also reduces the Br
somewhat. According to [106], typical values of Dy content in EV magnets, during 1985
to 2011, were 8.5-11% of the magnet weight, which represent temperature capabilities
around 200-220◦ C. A similar value is also presented in [107], where a Dy content of
8.7w% is said to be typical for EV drives. However, due to a sudden and substantial
price increase of Dy in 2011, which is not yet fully recovered, magnet manufacturers are
currently seeking to reduce the amount of Dy [106]. As an example, Nissan claims to
have lowered the Dy content by 40% in their 2013 Leaf model compared to the 2011
model [97]. One publication from 2012, indicate that very low Dy contents may be used
in EVs, on an average 3.6-4w%, and single values up to 7.7w% [108], which represents
temperature capabilities around 150◦ C and 180◦ C, according to [106].
For this reason, a magnet with a relatively low Dy content is chosen for the refer-
ence motors; Hitachi NEOMAX NMX-37F, which is expected to have around 3.5-4.5w%
Dy [109]. The material data is shown in Figure 6.3 [110], which indicate a temperature
capability up to 140◦ C. The Br value is around 1.2 T at 20◦ C, and drops to 1.1 T at
140◦ C, whereas the Hc value is about -940 kA/m at 20◦ C, and -790 kA/m at 140◦ C. As
expected, the Hc values are a bit lower compared to Toyotas and Lexus’ magnets, which
thus both can be assumed to have a larger Dy content.
The magnet density is taken to be 7500 kg/m3 , the thermal conductivity as 7.5
W/m K, and the specific heat as 410 J/kg K, as stated by Hitachi [111]. The elec-
tric conductivity is taken to be 1.3 µΩm, and assumed to adhere to 20◦ C [111]. The
temperature coefficient of resistivity is here assumed to be 0.5%/◦ C, which gives an
electrical conductivity of 750 469 S/m at 70◦ C. The considered magnet material proper-
ties at 20◦ C, 70◦ C and 120◦ C are presented in Table 6.6.
1
140 oC
0.8 180 oC
220 oC
0.6
0.4
0.2
0
-1000 -800 -600 -400 -200 0
Coercivity (kA/m)
Figure 6.3 Data sheet data for chosen magnet material [110].
104
6.2. Design, material and mass of machine parts
105
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
Table 6.4 Mass (kg) of selected main motor components for CityII and HighwayII motors.
CityII HighwayII
Frame 4.7 5.5
Stator Yoke 4.5 6.0
Stator Teeth 3.7 4.9
Active winding 1.8 2.4
Slot impregnation 0.3 0.4
End Winding 1.7 2.2
Magnets 0.9 1.2
Rotor laminations 5.0 6.7
Shaft 4.6 5.8
Bearings 0.4 0.4
Total Lamination 13.3 17.6
Total Copper 3.5 4.6
Total Passive Parts 11.6 14.3
Total Active Parts 16.0 21.2
Total 27.6 35.5
(a) Mesh. (b) Flux lines, total losses and current den-
sity.
Figure 6.4 FEA of one pole of the HighwayII motor at 4000 rpm, 251 A rms, showing mesh, vector
potential flux lines, total losses in the core and magnets, as well as current density in the
windings, at a certain time instant.
For each machine geometry, at 13 different speed levels (i.e. at 1 rpm, and at every
1000 rpm up to 12,000 rpm), 11 × 11 operating points in the id-iq plane have been evalu-
ated for two electrical periods, with 101 sample points. Sinusoidal current excitation have
been used in all operating points. The data resolution in the id-iq-plane is then increased
106
6.3. FEA evaluation of performance and losses
through interpolation, as shown in Figure 6.5(a). From this, the MTPA (Max Torque Per
Ampere) operating points as well as the field weakening operation is numerically found
for each integer torque level, within the allowed operating current and voltage limits, as
exemplified in Figure 6.5(b). A final interpolation in the speed range is then conducted.
Selected data mainly consist of losses in the different machine parts, e.g. lamination
losses in the stator yoke, teeth, and rotor, dc resistive losses in the active part of the
winding, and eddy current losses in the magnets.
For the HighwayII motor, these motor simulations have been done for three additional
slot areas apart from the already presented motor geometry (further treated in section
6.7). For all of the four slot geometries the simulations are done at the three different
magnet temperatures; 20◦ C; 70◦ C and 120◦ C, (section 6.6). For the CityII motor, the
simulations are done for a single geometry with magnet and copper temperatures of 70◦ C
(section 6.4). Finally, for both the CityII and the original geometry of the HighwayII
motors, simulations are done at half and double the stack length with a magnet and copper
temperature of 70◦ C (section 6.5).
400
Trq (Nm)
0
30
25
350 |Iph.max|
0
20
1-3000 rpm |Vph.max|
300
4000 rpm
MTPA
250
0
15
I q (A)
0
200 5000 rpm 10
150
100 50
50
12,000 rpm
0
-400 -300 -200 -100 0
I d (A)
(a) FEA operating points and interpolated torque. (b) Result of current control algorithm.
Figure 6.5 FEA simulated operating points for each speed level along with resulting interpolated
torque, and current control id iq reference values for each integer torque level, at 1 rpm
and every 1000 rpm up to max speed.
Both components of the core losses (hysteresis and eddy) increase with increasing
machine speed, as shown in Figure 6.6 for the two speed levels 4000 rpm and 12,000
rpm. The figure also shows the transient behaviour of the hysteresis losses in the FEA
software program over the two simulated electrical periods, as well as the used mean
value over the last period.
Eddy current losses in the magnets arise due to variation of flux density during oper-
ation. In the FEA software the magnets are modeled as zero current carrying coils while
considering eddy effects. As mentioned, the conductivity values used are presented in
Table 6.6.
107
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
4 krpm − Hyst.
4 krpm − Eddy
1500 12 krpm − Hyst.
12 krpm − Eddy
PCore,loss (W)
Average
1000
500
0
0 50 100
Samples
Figure 6.6 Lamination core losses at around 250 A rms, 4000 rpm and 12,000 rpm, for the High-
wayII motor.
The copper losses in the active part of the winding are extracted from the FEA soft-
ware. Skin and proximitys effect is here ignored. Then the total copper losses are found
via the relative length of the active part of the winding, by again using (5.3). The end
winding losses are simply the difference between the total and the active part. For the
HighwayII reference motor, the active part is 52% of the total winding length, thus leav-
ing 48% to the end windings.
108
6.3. FEA evaluation of performance and losses
300
1000
150 1000 150
Torque (Nm)
Torque (Nm)
200
100
500
100 500 100 15
00
50 50
10
20
40
80
00
0
0
0 0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
200 Rotor
Magnets
150
Torque (Nm)
100
10
200
50
5
1
10
10
0
50
0
0 2000 4000 6000 8000 10000 12000
Speed (rpm)
Figure 6.7 Electromagnetic losses as a function of torque and speed, for the HighwayII V-shaped
reference motor.
The percent ratio of total copper losses over total motor losses is presented in Figure
6.8(a). Generally, the copper losses dominate for torque levels above 50 Nm. As copper
and core losses are the two main loss components, the rest of the losses in the figure ad-
here to the core losses. Generally, the core losses are the main loss components for speed
levels above about 2000 rpm and torque levels below 50 Nm. Consequently, core losses
can be expected to dominate in drive cycles with low acceleration levels and relatively
high speed levels, whereas copper losses can be expected to dominate in drive cycles
with higher acceleration levels.
Finally, the percent ratio of core losses in the teeth over the total core losses is pre-
sented in Figure 6.8(b). At speed levels below 6000 rpm around 50-60% of the core losses
are developed in the teeth. At higher speed levels the share increases to about 80% at top
speed. A similar study of the rotor reveals that its core losses are essentially well below
10% of the total core losses in the whole operating area. This means that the rest of the
loss share shown in Figure 6.8(b) principally adhere to the stator yoke.
109
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
200 200
150 90 150
50
Torque (Nm)
Torque (Nm)
75
100 100
55
50
50 50
25
75
70
55
60
65
10
50
0 0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
(a) Share of total copper losses to the total electromag- (b) Share of lamination teeth losses to the total stator
netic losses, in %. and rotor lamination losses, in %.
Figure 6.8 Share of total losses as a function of torque and speed, for the HighwayII V-shaped
reference motor.
110
6.5. CityII and HighwayII motor efficiency at 50%, 100% and 200% of stack length
70
90
70 350 Vdc
90
300 Vdc
100 150
Torque (Nm)
300 Vdc 250Vdc
Torque (Nm)
250 Vdc
250Vdc 250 Vdc
100
50
250Vdc
50
450Vdc 450Vdc
96
95
0 0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 6.9 Electromagnetic efficiency contour lines at different levels of max dc voltage, for the
CityII and HighwayII V-shaped reference motors.
111
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
70 2
CityII CityII
60 HighwayII
40
30 1
20
10 0.5
0.5 1 1.5 2 0.5 1 1.5 2
EM scaling (%) EM scaling (%)
Figure 6.10 Mass of CityII and HighwayII re-scaled motors, and motor masses normalized by the
reference motors.
Simulations are done at all three length scalings: 50%, 100% and 200% of original
stack length. The losses and efficiency maps from the simulations are essentially identical
to the results when simply scaling the active part of the losses with the same factors as
the stack length. For simplicity, it therefore the latter method that is used for intermediate
length scalings of losses and torque levels, and the resulting efficiencies is presented in
Figure 6.11. Additionally, selected data on three scalings for both motors are presented
in Table 6.5.
Table 6.5 Selected data on min and max length scaling of CityII and HighwayII V-shaped reference
motors.
CityII motor Lstk,min Lstk,ref Lstk,max
Core length 68 mm 136 mm 272 mm
Turns per coil 20 10 5
Max rms phase current 63 A 125 A 251 A
Active winding share 34% 51% 67%
Rs,Tot 204 mΩ 68 mΩ 26 mΩ
Rs,ActWi 70 mΩ 35 mΩ 17 mΩ
Rs,EndWi 134 mΩ 34 mΩ 8 mΩ
Motor mass 16.9 kg 27.6 kg 49.1 kg
(61%) (100%) (178%)
112
6.6. HighwayII motor losses at three magnet and copper temperatures
80
90
70
90
Lref
Lref
250 350 Lmin
Lmin
95
96
300
200
Torque (Nm)
Torque (Nm)
250
150 200
80
70
90
90
100 150
95
96
100
70
80
50
90
50
96
90 95
0 0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 6.11 Electromagnetic efficiency contour lines (%) for three different length scalings of the
CityII and HighwayII motors.
Table 6.6 Magnet simulation input data, phase resistance, and resulting max torque and power at
three temperatures of magnets and copper.
20◦ C 70◦ C 120◦ C
Remanent flux 1.236 T 1.1824 T 1.1244 T
(100%) (95.66%) (90.96%)
Relative permeability 1.0432 1.0428 1.0446
Coercivity -942,926 (A/m) -902,285 (A/m) -856,586 (A/m)
Conductivity 769,231 (S/m) 750,469 (S/m) 732,601 (S/m)
Phase resistance 18 mΩ 22 mΩ 26 mΩ
(100%) (119%) (139%)
Max torque 220 Nm 215 Nm 210 Nm
(100%) (97.7%) (95.5%)
Max power 5.7 kW 6.4 kW 6.9 kW
Due to the weakening of the magnet, at a certain torque level outside the field weak-
ening area of operation, a larger current is needed at a higher temperature compared to a
lower. In contrast, within the field weakening area, the weaker magnet leads to a reduced
need for field weakening current, especially at the lower torque levels. Therefore, at a
certain torque level a lower current is needed for higher temperatures compared to lower.
This is illustrated in Figure 6.12(a).
113
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
This reasoning also explains the presented copper losses in Figure 6.12(b), since cop-
per losses are proportional to the square of the rms current. Naturally, also the winding
resistance increase with temperature, which further enhances the differences of the cop-
per losses’ behaviour in the torque-speed diagram between the studied temperatures. Due
to the linear dependency on current for the torque and the quadratic dependency for the
winding losses, the difference in losses between the three temperatures increase as the
torque becomes higher.
150 150
Torque (Nm)
100 100
1000
50 50
100
0 0
0 50 100 150 200 250 0 2000 4000 6000 8000 10000 12000
Iph, RMS (A) Speed (rpm)
(a) Torque per input phase rms current. (b) Winding loss contour lines at 100 W, 1000 W and
3000 W.
Figure 6.12 Produced torque per applied rms phase current, and total winding losses as a function
of torque and speed, for the HighwayII V-shaped reference motor, at winding and
magnet temperature 20◦ C, 70◦ C and 120◦ C.
The lower field weakening current also cause minor deviations in the magnet losses
in the torque-speed diagram shown in Figure 6.13(a).
Finally, the resulting total losses for the three different temperatures are presented in
Figure 6.13(b).
114
6.7. HighwayII motor losses for four slot areas
20 oC 20 oC
200 200
70 oC 70 oC
120 oC 120 oC
150 150
Torque (Nm)
Torque (Nm)
100 100
500
10 1000 0
5 300
0
50 3 50
1
0 0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
(a) Magnet loss contour lines at 1 W, 3 W, 5 W, and (b) Total losses at 1 kW, 3 kW and 5 kW.
10 W.
Figure 6.13 Magnet and total losses as a function of torque and speed, for the HighwayII V-shape
reference motor, at winding and magnet temperature 20◦ C, 70◦ C and 120◦ C.
Figure 6.14 The FEA simulated different stator lamination geometries, as well as one of the wind-
ing layers.
115
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
The numerical values of the different slot areas are presented in Table 6.7, along
with percentage ratios compared to the Original. Furthermore, the copper areas are are
also presented, and it is the smallest in the Small slot and the largest in the Large slot.
Consequently, the current density is the largest in the Small slot and the lowest in the
Large slot. Additionally, the phase resistance, and max torque for all slots at the three
magnet and copper temperatures are also presented in the table.
It should be noted that the axial extent of the end winding is assumed to be different
for the different slot areas. The extent is taken to be linearly proportional to the square
root of the change in copper area from the original slot area to the others. For the small
slot geometry, the extent is shorter (29.7 mm), for the Large and High slot geometries it
is longer (40.2 mm, and 37.2 mm respectively). .
The flux linkage from the magnet as a function of q-axis current for the different slot
areas is shown in Figure 6.15, for the temperature of 70◦ C. In general the flux from the
magnet decreases with increasing q-axis current due to cross saturation. The saturation
effect is the largest in the stator geometry with the thinnest teeth and yoke: the Large slot,
whereas it is the smallest for the geometry with the thickest teeth and yoke: the Small
slot.
116
6.7. HighwayII motor losses for four slot areas
0.085 1.05
Small Small
Orginal 1 Orginal
0.08 Large Large
0.075
0.9
0.07 0.85
0.8
0.065
0.75
0.06
0.7
0.055 0.65
0 100 200 300 400 0 100 200 300 400
Iq (A) Iq (A)
(a) Flux linkage from magnets (Wb). (b) Flux linkage from magnets normalized by the
value at zero q-axis current.
Figure 6.15 Flux linkage from magnets as a function of q-axis current for the different stator slot
areas at 70◦ C copper and magnet temperature.
The four stator geometries also have different dq inductances, where the Small slot
geometry have the largest values and the Large slot have the lowest, in the whole id-iq
plane. Also the absolute difference between the dq inductances (which is relevant for the
reluctance torque production as expressed in (2.13) is the largest for the Small slot and
the smallest for the Large slot. These relations are the same at every speed level.
Consequently, both torque components: magnet and reluctance, are the highest for the
Small slot geometry in the whole dq current operating area, and the lowest for the Large
slot. The net produced torque for the different geometries is presented in the id iq plane
in Figure 6.16(a). As shown, the difference in produced torque between the geometries
decreases with decreasing current, especially with decreasing q-axis current.
The resulting dq current operating points are presented in Figure 6.16(b), for a few
selected speed levels. Outside of the field weakening area the MTPA method result in a
somewhat larger share of the q-axis current for the Small slot geometry compared to the
others. As a consequence, yet bit higher torque levels can be reached for the same current
magnitude. The higher flux from the magnet also result in a stronger speed dependence
of the induced voltage, hence the base speed is lower for the Small slot compared to the
others. In the field weakening area, instead a larger d-axis current is needed in the Small
slot in order to suppress the flux from the magnet, that is even though its d-axis inductance
is larger. This counteracts some of the previous torque producing advantage with the
Small slot. As noted, the difference between the geometries decreases with increasing
speed regarding produced torque and resulting dq current operating points, i.e. when
going further into the field weakening area.
117
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
400
Small
Small 350 Orginal
Orginal I max
Large 4000 rpm Large
400 High 300 High
Torque (Nm)
300 250
I q (A)
200 200
6000 rpm 1 rpm
100 150
0 100
-400 50
400
-200 12,000 rpm 8000 rpm
200 0
-400 -350 -300 -250 -200 -150 -100 -50 0
I (A) 0 0 I (A) I d (A)
d q
(a) Produced torque for different slot areas. (b) Result of current control algorithm for different
slot areas at 1 rpm, 4000 rpm , 6000 rpm, 8000 rpm
and 12,000 rpm.
Figure 6.16 Produced torque and result of current control algorithm as a function of dq current, for
the different stator geometries at 70◦ C copper and magnet temperature.
As a consequence, at low speed levels the largest produced torque per input rms cur-
rent is given by the Small slot geometry and the lowest by the Large slot. At higher speed
levels the difference is quite small. Still, it is then the Large slot that produces somewhat
higher torque per input rms current compared to the other slot sizes. the tendency is a bit
higher produced torque. The results are shown for a few example speed levels in Figure
6.17(a). The resulting max torque as a function of speed is shown as the dashed line in
Figure 6.17(b).
The total lamination core losses are also presented in Figure 6.17(b). As shown, the
increase with speed of the core losses is the smallest for the Small slot geometry and
the highest for the Large and High slot geometries. Likely, the larger cross sectional
lamination area for the Small slot geometry cause somewhat lower flux density levels,
and therefore also lower core losses. At low torque levels the core losses are the largest
in the Large slot geometry, whereas they are the highest in the High slot at higher torque
levels. A similar relation is also noted for both the yoke and teeth losses separately,
whereas the opposite order is noted in the rotor (although not shown here), i.e. largest for
the Small slot and lowest for the Large. Still, the magnitude of the core losses in the rotor
are much lower than in the yoke and teeth, which is why the impact on the total losses is
small.
118
6.7. HighwayII motor losses for four slot areas
PFe,tot (W)
250
4000 rpm
8000 rpm Small Small
12,000 rpm Original Orginal
200 Small High 200 Large
High
Orginal Large
Large
Torque (Nm)
Torque (Nm)
150 High 150
100 100
10
00
500
100
50 50
0 0
0 50 100 150 200 250 300 0 2000 4000 6000 8000 10000 12000
Iph, RMS (A) Speed (rpm)
(a) Torque per input rms phase current. (b) Total lamination core loss contour lines at 500 W
and 1000 W.
Figure 6.17 Produced torque per input rms phase current, as well as total core losses as a function of
torque and speed, for the HighwayII V-shaped reference motor, for the Small, Original,
Large and High slot areas.
In contrast with the core losses, a larger copper cross section leads to a lower winding
resistance. Owing to this, the Small slot has the highest winding resistance, and the Large
slot the lowest. In the field weakening area the phase current is essentially the same in any
speed torque operating point for all slot geometries. Then, the higher winding resistance
in the Small slot causes larger copper losses in this region, as shown in Figure 6.18(a).
Outside of the field weakening area, at the same torque level, the phase current is the
highest in the Large slot and the lowest in the Small. Thus, in this area the resistance and
current counteract each other in the loss creation. The result is still that the copper losses
are the highest for the Small slot geometry also outside of the field weakening area, and
fairly similar between the other geometries.
Nevertheless, the difference in total losses between the four geometries is rather
small, as shown in Figure 6.18(b). These small deviation s were a bit unanticipated, but
indeed interesting to note. Essentially it is only the losses with the Small slot that deviates
notably. In a large part of the operating area the losses with the Small slot are somewhat
higher compared to the other slot geometries.
The main difference between the losses in the motors with different slot geometries
is thus not the total amount of losses, but the relation between copper and core losses (as
shown for the Original slot in Figure 6.8(a)). The motor with the Small slot thus have
increased its share of copper losses and decreased the share of core losses, whereas the
motors with the Large and High slots have instead increased the share of core losses and
decreased the share of copper losses.
119
Chapter 6. V-Shaped Reference PMSMs for CityII and HighwayII Concept BEVs
Small Small
Orginal Orginal
200 Large 200 Large
High High
3000
Torque (Nm)
Torque (Nm)
150 150
2000
100 100
1000
500
50 50
0 0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
(a) Copper loss contour lines at 500 W and 2000 W. (b) Total loss contour lines at 1000 W and 3000 W.
Figure 6.18 Copper and total losses as a function of torque and speed, for the HighwayII V-shaped
reference motor, for the Small, Original, Large and High slot areas.
The ratio of the copper losses in the Small slot machine over the copper losses in
the Original slot is presented in Figure 6.19. As expected, in the field weakening area
where the rms currents are almost the same, the ratio is fairly constant (≈1.36). However,
outside of the field weakening area the ratio decreases with increasing torque (≈1.1-1.3
).
Figure 6.19 The ratio of the Small slot copper losses to the Original slot copper losses.
120
Chapter 7
In this chapter, the electric motor and inverter of the CityII and the HighwayII concept
BEVs (described in Section 6.5) are resized in terms of torque and power rating as well
as losses and motor mass. After this, the consequence on vehicle performance and energy
efficiency for the Test cycles are studied and quantified.
The re-scaling of the motors is simply a linear scaling of the active length, as is
described in Chapter 5. The scaling factors are here in steps of 10%, from 50% to 200%
of the reference motors.
The inverter which is used in the City car in Chapter 5 (current rating 400 A rms), is
here used as a baseline inverter for both the CityII and HighwayII cars, for simplicity. In
order for the converter losses to still be reasonably representative, the on-state resistance
and switching loss parameters are scaled with the same factors as the motor lengths.
The resistances are inversely proportional and the switching loss parameters are linearly
proportional, while the constant voltage drop term in the collector-emitter path is left
unchanged.
The same simulation structure as in Chapter 5 is here used. Also, the configuration
of the gear efficiency and batteries in the CityII and HighwayII cars are identical to the
settings in the City and Highway cars. On the other hand, only the losses in motoring
mode of the electric machine and inverters are here implemented in the models. It is thus
assumed that the loss maps in the generative mode are the same as in motoring mode.
121
Chapter 7. Effects of CityII and HighwayII Motor and Inverter Re-scaling on
Performance and Energy Efficiency
7.1 Performance
As noted in Section 6.5, the initial max torque levels of the re-scaled motors are linearly
proportional to the scaling factor. The max mass normalized wheel force for some of the
re-scaled electric motors can be seen in Figure 7.1, together with the vehicle’s road load,
which remains essentially the same in all cases. Even the smallest motor size will be able
to sustain the vehicle’s top speed requirement.
8
8 Scale 50 % Scale 50 %
Scale 80 % 7 Scale 80 %
7 Scale 100 % Scale 100 %
4
4
3
3
2 2
1 1
0 0
0 50 100 150 0 50 100 150
Vehicle speed (km/h) Vehicle speed (km/h)
Figure 7.1 Max mass normalized wheel force as a function of vehicle speed (at 400 V DC) along
with road load, for the re-scaled CityII and HighwayII powertrains.
The resulting time to accelerate 0-50 km/h and 0-100 km/h, for an initial SOC level
of 90% is presented in Figure 7.2, both as a function of max torque and of max power.
Naturally, the time to accelerate decreases for increasing max torque and power ratings,
and it does so in a nonlinear way. For both cars, the time to accelerate at 50% scale factor
is about 1.8-2.1 times longer compared to the reference length, and at 200% scale factor it
is 0.5-0.6 times the value of the reference length. Furthermore, it can be noted that at any
certain max torque level, the acceleration time is shorter for the CityII car compared to
the HighwayII car. This can be expected since the City car is seen to have slightly higher
mass normalised wheel force values, compared to the HighwayII car. The difference is
even larger when comparing with max power levels, which is due to the different base
speed levels of the two cars’ motors.
The noted nonlinear dependency means that the effect of a change in stack length is
largest for short stack lengths compared to long ones. This effect can be seen since the
base speed is kept the same during the scaling. For the acceleration time 0-100km/h, the
difference between the scale factors 50% and 80% is 12s for the CityII car, and 8s for the
HighwayII. The difference between 120% and 150% is only about 2s for both cars. For
the acceleration time 0-50km/h the differences between the same scale factors are a bit
smaller: around 3s and less than 1s for both cars.
122
7.2. System Efficiency and Cycle Energy Consumption
30 30
CityII CityII
HighwayII HighwayII
25 0−100km/h 25 0−100km/h
0−50mk/h 0−50mk/h
Acceleration Time (s)
15 15
10 10
5 5
0 0
0 100 200 300 400 500 0 50 100 150 200 250
Max EM Torque (Nm) Max Power (kW)
(a) Acceleration time related to max torque. (b) Acceleration time related to max power.
Figure 7.2 Acceleration time 0-50 km/h and 0-100 km/h, as a function of max torque and max
power at 400 V, for the powertrain scale factors: 50%, 80%, 100% (black), 120%, 150%,
and 200%.
123
Chapter 7. Effects of CityII and HighwayII Motor and Inverter Re-scaling on
Performance and Energy Efficiency
4.5 4
50% − ηPT 50% − ηPT
4 3.5
70
50% − max force
80
50% − max force
85
85
80
70
100% − η 100% − ηPT
Norm. Fwheel (N/kg)
88
3
2.5
2.5 88
2
2
70
80
70
85
80
85
1.5
1.5
1
1 88
88
0.5 0.5
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120 140
Speed (km/h) Speed (km/h)
Figure 7.3 Total powertrain efficiency of the 50 % down scaled system and the reference system
(100 %) at 400 V DC.
The vehicle energy consumption for all Test cycles has been simulated for the re-
scaled powertrain systems, and it was done similarly as described in Chapter 5.
For some cycles, the down-scaled systems were no longer able to sustain all of the
operating points in the reference speed’s time trace due to the limited acceleration capa-
bility. In these cases there may be a rather large difference between the reference speed
and the simulated speed. Even though the effect of this deviation on the average cy-
cle energy consumption per driven distance may be relatively small, the max allowed
speed deviation is here set to 2 km/h. The same value is also the max allowed deviation
between a test vehicle’s speed and the reference cycle (within one second) during stan-
dardized consumption tests with NEDC in Europe [1]. In the US the max allowed speed
deviation is 3.2 km/h (2 mph) [113].
The resulting max speed deviations for the different powertrain systems and test
drive cycles are presented in Figure 7.4. As expected, it is the cycles with the highest
levels of peak acceleration that give too large speed errors for larger scale factors, es-
pecially US06, and REP05, which are too large even for the CityII car with the 120%
scale factor. Furthermore, tor the CityII car ArtemisMW130, LA92(UC), WLTC Ex-
trHgh, NYCC, and ArtemisURBAN cause too large speed deviations for the 60% scale
factor and lower. Also for the HighwayII car it is LA92(UC), NYCC, ArtemisURBAN,
SC03, and ArtemisRURAL that cause too large speed deviations for the 60% scale factor
and lower.
124
7.2. System Efficiency and Cycle Energy Consumption
7 6
ECE ECE
WLTC Low WLTC Low
jc08 jc08
6 FTP72 (UDDS) 5 FTP72 (UDDS)
WLTC Mdl WLTC Mdl
SC03 SC03
5 NYCC NYCC
4
Artemis URBAN Artemis URBAN
NEDC NEDC
4 WLTC Hgh WLTC Hgh
Artemis RURAL 3 Artemis RURAL
EUDC EUDC
3 WLTC WLTC
LA92 (UC) LA92 (UC)
2
HWFET HWFET
2 REP05 REP05
WLTC ExtrHgh WLTC ExtrHgh
US06 1 US06
1 ArtemisMW130 ArtemisMW130
0 0
50 100 150 200 50 100 150 200
EM scaling (%) EM scaling (%)
Figure 7.4 Max speed error per single Test drive cycle, as a function of powertrain system scal-
ing factor. The horizontal lines represent the max allowed speed deviations of 2 km/h
(solid), and 3.2 km/h (dashed).
The resulting net battery energy consumption per driven distance for each cycle and
vehicle is presented in Figure 7.5. In the figure, results that adhere to too large max speed
deviations are omitted. As in Chapter 5, the consumption is the highest for Highway cy-
cles, and the lowest for Urban cycles. In general, the energy consumption decreases with
decreasing system scaling. For some cycles, there is a minimum energy consumption for
a certain down scaled value. The reason is that many cycles most often operate relatively
far from the systems max force limit, which thus benefits the systems of lower scaling
factors.
220 200
ECE ECE
WLTC Low WLTC Low
200 jc08 180 jc08
FTP72 (UDDS) FTP72 (UDDS)
WLTC Mdl WLTC Mdl
180 SC03 SC03
160
NYCC NYCC
Ebattery (Wh/km)
Ebattery (Wh/km)
60 60
50 100 150 200 50 100 150 200
EM scaling (%) EM scaling (%)
Figure 7.5 Battery net energy consumption per driven distance per single drive cycle, as a function
of powertrain system scaling factor.
Figure 7.6, presents the results from Figure 7.5 which are all normalized by the con-
sumption with 100% scale factor. It can be noted for both cars that the optimum scale
125
Chapter 7. Effects of CityII and HighwayII Motor and Inverter Re-scaling on
Performance and Energy Efficiency
factor for most cycles is below 100%, with the exceptions of the acceleration intense
NYCC and ArtemisURBAN. For those cycles the lowest consumption is at 120% scale
factor for the CityII car and 140% for the HighwayII car.
For the CityII car, down scaling to 80% gives the lowest consumption for most Urban
cycles, that is 98-99% of the consumption with 100% scale factor. A further down scaling
will instead increase the energy consumption. For Rural cycles, a scale factor of 60%
generally gives the lowest consumption (96%-98% of the reference system), and even
lower with a 50% scale factor for EUDC and WLTC (95%-96%). The largest decrease
in consumption is noted for the Highway cycles HWFET and WLTC ExtrHgh: 94% of
consumption with the reference system.
The noted gains in energy consumption for re-scaling of the HighwayII car are gen-
erally smaller compared to the CityII car, as they are 98%-100%. Also here a lower scale
factor (90%) gives the lowest consumption for the Urban cycles, whereas 70%-80% gives
the lowest for the Rural cycles, and 60% for the Highway cycles. The largest gain in en-
ergy consumption is noted for NYCC.
120 115
ECE ECE
WLTC Low WLTC Low
jc08 jc08
115 FTP72 (UDDS) FTP72 (UDDS)
WLTC Mdl WLTC Mdl
SC03 110 SC03
Norm. Ebattery (%)
90 95
50 100 150 200 50 100 150 200
EM scaling (%) EM scaling (%)
Figure 7.6 Normalized battery net energy consumption per driven distance per single drive cycle,
as a function of powertrain system scaling factor. The normalization is done with respect
to the reference lengths.
The same results are also reflected when studying the average net propulsion power-
train efficiency in Figure 7.7. For the CityII car, the highest efficiency levels are generally
seen for the Urban cycles (up to 81%), whereas Rural and Highway cycles give similar
results (up to 79%). The lowest efficiency, 76%, is seen for HWFET. For the HighwayII
car, several cycles from all categories are amongst the ones with the highest efficiency
(up to 88%). In the same time some of the Urban cycles result in the lowest levels of
average efficiency (down to 85%). To conclude, the average net propulsion efficiencies
are higher for the HighwayII car compared to the CityII car.
126
7.2. System Efficiency and Cycle Energy Consumption
85 89
ECE ECE
WLTC Low WLTC Low
jc08 88 jc08
FTP72 (UDDS) FTP72 (UDDS)
80 WLTC Mdl WLTC Mdl
SC03 87 SC03
NYCC NYCC
Artemis URBAN 86 Artemis URBAN
75 NEDC NEDC
WLTC Hgh WLTC Hgh
Artemis RURAL 85 Artemis RURAL
EUDC EUDC
70 WLTC WLTC
LA92 (UC) 84 LA92 (UC)
HWFET HWFET
REP05 83 REP05
65 WLTC ExtrHgh WLTC ExtrHgh
US06 US06
ArtemisMW130 82 ArtemisMW130
60 81
50 100 150 200 50 100 150 200
EM scaling (%) EM scaling (%)
Figure 7.7 Powertrain cycle average propulsion (battery discharge) efficiency per single drive cy-
cle, as a function of powertrain system scaling factor.
127
Chapter 7. Effects of CityII and HighwayII Motor and Inverter Re-scaling on
Performance and Energy Efficiency
128
Chapter 8
Lumped-Parameter Thermal
Model of HighwayII V-shaped
Reference PMSM
Electric machines in BEVs experience highly dynamic load cycles, which cause similarly
varying internal losses, and therefore strongly time dependent temperature fluctuations.
High temperatures can degrade the winding insulation and shorten the machine lifetime,
or if high enough even cause immediate winding failure, whereas permanent magnets
risk demagnetization [19, 46].
If the machine is sized to be able to continuously operate at the peak power needed
in the application, it will result in an unnecessary large and expensive motor. In addition,
the available space for electric machines in automotive applications, is often limited.
Therefore, a trade-off has to be found between performance, volume and cost, so that the
machine is not unnecessarily thermally over dimensioned.
In practice, often a lower level of torque and power is specified which the motor can
handle continuously, whereas it can handle a higher level for a shorter period of time,
i.e. intermittently. Hence, BEV electric machines utilize temporary over-loading that is
limited by the thermal capability of the machine.
A lumped-parameter transient thermal network is here presented for the HighwayII
motor with the four different slot areas, such that the temperature development in key
machine components can be estimated in various operating points and during load cycles.
129
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
paths, and be able to predict the transient temperature development in crucial machine
parts such as the windings and magnets, while considering losses in bulk regions e.g.:
stator yoke, stator teeth, active winding, end-winding, rotor yoke and magnets.
A suitable 10-node network for TEFC IMs (totally enclosed fan cooled induction ma-
chines) is proposed in [19], where the T-equivalent node configuration is implemented.
It assumes a uniform frame temperature, includes both radial and axial heat transfer in
laminations and winding, as well as a speed dependent air gap convection and fan speed
dependent convection in the end cap region. A similar fan cooled IM 8-node network
is presented in [20], however without the T-equivalent node configuration. Furthermore,
the axial heat transfer in the laminations are considered negligible, whereas bearing heat
transfer is added. This network was later adjusted to a water cooled inset PMSM pre-
sented in [53], by mainly modifying the rotor modelling. Here, the network has been
further adjusted for the used IPM rotor design. It is presented in Figure 8.1. Similar ad-
justments to the rotor parts are also reported in [101, 114, 115]. Another modification
here, is the added consideration of convection heat transfer to the cooling media.
Furthermore, it is assumed that the internal heat generation is uniformly distributed
in each region, as is also assumed in [47, 56, 116]. Axial heat transfer is considered in
the copper windings and the shaft, whereas only radial heat transfer is considered for the
laminations and magnets, as is also done in [47]. Moreover, axial symmetry around the
midpoint is assumed, as well as cylindrical symmetry both geometrically and thermally.
130
8.1. Implemented thermal network
the two main heat transfer modes involved: convection between the coolant medium and
the aluminum frame, and conduction inside the aluminum. Thus both natural convection
and radiation to the ambient air are considered negligible compared to the forced con-
vection [52]. It is also desired to find an approximate value of the convection resistance
for the specific coolant temperatures 20◦ C and 65◦ C, as well as for certain flow rates.
The higher of the two temperatures is here considered as the dimensioning temperature
for the machines’ thermal capability, which is studied in Chapter 9. In order to estimate
these resistances a geometrical representation of the cooling channel is needed, as well
as material data of the coolant.
The cooling medium in current BEV models is found to typically consist of about
50% deionized water and 50% ethylene-glycol, as stated for Ford Focus and Toyota
RAV4 EV in [117, 118] (it was also seen for Fiat 500e, Mercedes B-Class, Mitsubishi
i-MiEV, Nissan Leaf and Smart fortwo). The ethylene-glycol is added to prevent over-
heating, freezing and corrosion. The cooling medium used here is assumed to be a mix-
ture of 48% (by volume) water and 52% mono ethylene glycol [119]. Its material param-
eters varies with temperature such that the density, thermal conductivity and dynamic
viscosity decreases with increasing temperature, whereas the specific heat increases. The
temperature dependencies are shown in Figure D.1 in Appendix D.
Various cooling circuits are possible for this application. However, for simplicity the
cooling channel is assumed to form a spiral of four laps inside the aluminum frame, as
illustrated in Figure 8.2. The cooling ducts have rectangular cross sections with a width
wduct of 30 mm, and a height hduct of 5 mm, and are separated by 10 mm. The radial
distances from the inner and outer frame boundaries lF r,lower to the ducts are 3 mm on
each side. Similar cooling channels with rectangular ducts are also used in BMW’s i3
model with six laps [120], and in Nissan’s 2011 Leaf model with three parallel laps (i.e
no spiral) [94].
Figure 8.2 The four laps spiral cooling channels with rectangular ducts, along with simplified
frame, stator core and end windings.
131
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
between these two boundaries. This part can be considered to be a hollow cylinder with
the inner and outer radii rF r,in and rF r,in + lF r,lower , and the length is here taken to be
that of the active core length lstk , of 150 mm. The conductive thermal resistance can then
be expressed as
r +lF r,lower
ln( F r,inrF r,in )
Rth,F r = (8.1)
2πλAl lstk
7000
65o C
6000 20o C
10o C
5000 5oC
h cool (W/m2 K)
6 L/min
12 L/min h=4023
4000
3000
h=1988
2000 h=1537
1000
h=292 h=291
0
0 5 10 15 20
Flow rate (L/min)
Figure 8.3 The estimated average heat transfer coefficient of the coolant, as a function of flow rate
and temperature.
Furthermore, it is assumed that 50% of the total heat flux between the stator back and
the cooling channel will pass through the inner duct surface. Therefore, a correction fac-
tor kconv,corr of 0.5 is added to the convection thermal resistance of the cooling channel
132
8.1. Implemented thermal network
Rth,cool , which is
kconv,corr
Rth,cool = (8.2)
hcool Acool
where Acool is the surface area of the inner cooling duct boundaries, as in
133
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
1 y1 y3 y2 x1 y2
−lny2 − x1 y2 −
Rth,T e = + + ln
λF e Qs lstk,cond x1 x3 x1 − x2 x1 − x2 x1 − x2
π a a+1
− +√ arctan √ (8.10)
4 a2 − 1 a2 − 1
where a is
x3 + 2y4
a= (8.11)
2y4
134
8.1. Implemented thermal network
as
Rx Ry Rx0 Ry0
Rth,T e−ActW i = 1− (8.15)
Qs lstk,cond (Rx + Ry ) 720(Rx0 + Ry0 )
where Qs is the number of slots, and Rx , Ry , Rx0 and Ry0 represent thermal resistances
per unit length in the width (x) and height (y) directions of the rectangular equivalent
slot. The resistances Rx are Ry are found as
Rx0
Rx = 0.5(Rix + ) (8.16)
6
135
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
Ry0
Ry = 0.5(Riy + ) (8.17)
6
where Ri x and Ri y represent the per unit length thermal resistances of the slot liner and
air film, as in
dslotliner dslotair
Rix = + (8.18)
λslotliner hslot,eq λair hslot,eq
dslotliner dslotair
Riy = + (8.19)
λslotliner wslot,eq λair wslot,eq
The per unit length thermal resistances of the slot winding mix: Rx0 and Ry0 are
found as in
wslot,eq
Rx0 = (8.20)
λW indM ix hslot,eq
hslot,eq
Ry0 = (8.21)
λW indM ix wslot,eq
Since Ry0 is about ten times larger than Rx0 the heat flow from the slot to the yoke
is assumed negligible.
8.1.4.2 Thermal resistance between the active winding and end winding nodes
The thermal resistance between the active winding and end winding is found as in [53,
p.91-93]. The heat flow from the active part of the winding to the end winding is axial,
with the cross sectional area of the copper ACu (m2 )in each slot. Since the heat flow
occurs in parallel in the slots, the thermal resistance between the active and end winding
network nodes is found as in
lav
Rth,AcW i−EnW i = (8.22)
6Qs ACu λCu
where Qs is the number of slots, lav is the average conductor length of half a turn, and
is approximated as the sum of the active and passive winding length’s by (5.3). The end
winding axial extension outside of the lamination stack dext is here assumed to be 35
mm on each side.
136
8.1. Implemented thermal network
or turbulent flows. Different calculation methods are then given for the Nusselt number
depending on the flow type.
The difference between the found models are usually the critical levels for the Taylor
number which dictate the type of flow, and the approximation of the Nusselt number.
Here the method described in [124] is used, which is also used in [20–22, 53, 121],
however with varying critical Taylor values.
The thermal resistance across the air gap is
1
Rth,AirGap = (8.23)
hairgap Aairgap
where hairgap (W/m2 K) is the air gap convection heat transfer coefficient, and Aairgap
(m2 ) the cylinder surface area in the air gap (Aairgap = 2πrairgap lstk ), and rairgap (m)
is the average air gap radius.
The heat transfer coefficient and the Nusselt number are related as in
N u λair
hairgap = (8.24)
2 lairgap
where lairgap is the air gap length.
The Nusselt number is approximated for different ranges of the modified Taylors
number as
Nu = 2 Tam < 1790 laminar flow
0.367
N u = 0.128 ∗ Tam 1790 < Tam < 104 transitional flow
0.241
N u = 0.409 ∗ Tam 104 < Tam < 107 fully turbulent flow
In [124] the critical value of the modified Taylors number between laminar and transi-
tional flow is 1700, however, here the value is adjusted somewhat to avoid discontinuity
at this transition.
Taylor’s number Ta is
ρ2air ωrotor
2 3
rairgap lairgap
Ta = (8.25)
µ2air
where ρair (kg/m3 ) and µair (kg/ms) is the mass density and dynamic viscosity of air,
respectively, and ωrotor (rad/s) is the angular speed of the rotor.
The Modified Taylors number Tam = FTag , is
where the geometrical correction factor Fg− 1 is close to unity, and therefore often ne-
glected.
The material parameters of air are temperature dependent, as shown in Figure D.2
in Appendix D [48]. The air gap Nu number, heat convection coefficient, and resulting
thermal resistance for the two temperatures 20◦ C and 100◦ C are presented in Figure 8.4,
137
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
as a function of rotor speed. Since the air temperature is not calculated by the thermal
network, the values at 100◦ C are used in all cases.
Additionally, the air gap length due to thermal expansion of the machine. At 80 ◦ C,
the air gap length may decrease to 85% of the value at 0 ◦ C. This effect is however
ignored here.
8 140 0.5
o
20 C 20 oC 20 oC
7 o
100 C 120 100 oC 100 oC
Rth,AirGap (W/mK)
h airgap (W/m2K)
0.4
6
Nuairgap (-)
100
5 80 0.3
4 60
0.2
3 40
2 20 0.1
0 5000 10000 15000 0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm) Speed (rpm)
ln( rrout
in
)
Rth,RoS:1,4 = (8.27)
2π λ lstk
The heat flow in segment S2, S3, and S5-S7 as well as the small air gap under the
magnet are modeled as pieces of a hollow cylinder based on (2.43), rendering
ln( rrout
in
)
Rth,RoS:2,3,5−7 = (8.28)
φ npole λ lstk
where npole is the number of poles. Furthermore, the radial heat flow in segment S2,S3,
and S5-S7 is assumed to occur in parallel, and their equivalent resistance is Rth,M agArcEq .
The radii and angles used in the conversion is presented in Table 8.1, as noted in Figure
8.5.
138
8.1. Implemented thermal network
Table 8.1 Parameters in the equivalent rotor geometry used in the thermal network.
rout rin φ
Lamination Rth,RoS1 r1 rRot,in 2π
Lamination Rth,RoS2 r2 r1 2φ1
Lamination Rth,RoS3 r2 r1 φ5
Lamination Rth,RoS4 rRot,out r2 2π
Air Rth,RoS5 r2 r1 2φ2
Air Rth,RoS6 r2 r1 2φ4
Magnet Rth,RoS7 r2 r1 2φ3
Air Rth,M agAirgap r1 + lM agGap r1 2φ3
(a) Electromagnetic rotor geometry. (b) Thermal equivalent rotor geometry outline.
8.1.7 Shaft
The shaft is assumed to be a lossless solid with only axial heat conduction, as also done
in [47, 53]. Furthermore, it is assumed to have a homogenous temperature distribution as
in [53]. The shaft thermal resistance is divided into two sections depending on the shaft
139
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
radius. Within the length of the lamination stack lstk the shaft radius equals the rotor
yoke inner radius rRoY o,in . Outside of the lamination stack the shaft radius rsh,ext is 20
mm, and the distance to the bearings lsh,ext is 45 mm on each side.
Similarly as in [53], the shaft thermal resistance Rth,Sh is then
lstk 2lsh,ext
Rth,Sh = 2 + 2 (8.31)
πrRoY λ
o,in steel πrsh,ext λsteel
where λsteel is the thermal conductivity of the shaft steel material.
8.1.8 Bearings
The thermal resistance of the bearings Rth,Be is modeled as in [53], which is based
on empirical data of IMs in [20], but where the speed dependent term is neglected for
simplicity.
140
8.1. Implemented thermal network
where vr (m/s) is the rotor peripheral speed. It should be noted that the approximation
is a curve fit to data with a relatively large spread [20].
The inner frame surface area consist of the two inner end cap surfaces and the inner
frame shell surface that extends outside both sides of the lamination stack.
The heat transfer coefficient between the rotor end shield surface and the internal
air in [20], is approximated as
hRo = 16.5vr0.65 (8.37)
The rotor end shield area is approximated by the circular surface area with the average
air gap radius.
The heat transfer coefficient between the end winding surface and the internal air is
via measurements empirically approximated in [20], as
where lpassive is the average end winding conductor length as from (5.3), rEnW i,ave is
the average radial distance for the end winding.
A comparison between different found suggested values of convection heat transfer
coefficient is shown in Figure 8.6. The speed range of found data are all up to 20 m/s,
which represent about 3144 rpm of the here used HighwayII motor. The rotor speed de-
pendence in the right part of Figure 8.6 is based on extrapolation of the velocity profiles.
The Kylander [EW] data are similar to the Schubert [EW] data, whereas the Boglietti
data seem to have a larger speed proportionality coefficient. The Kylander [Fr] data are
very similar to the Stokum data for higher speed levels. The Kylander [Ro] data have the
weakest speed dependence between all.
141
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
200 600
Mellor
Mellor eq.
500 Schubert [EW]
Schubert [Brg]
150 Stokum
400 Di Gerlando [ER]
Hamdi [ECap]
h (W/m2K)
h (W/m2K)
Boglietti A [EW]
100 300 Boglietti C [EW]
Kylander [Fr]
Kylander [EW]
200 Kylander [Ro]
50
100
0 0
0 5 10 15 20 0 2000 4000 6000 8000 10000 12000
Air velocity (m/s) Rotor Speed (rpm)
Figure 8.6 Comparison of suggested internal air convection heat transfer coefficients. Graphs in the
left part of the figure are original data from references, whereas graphs in the right part
are extrapolated data for the HW machine speed range. Mellor eq. from [19], Kylander
from [20], the rest from [127].
The three thermal resistances to the internal air, are then converted from a Y to a
∆, since the internal air temperature is not a desired network node. The net thermal
conductance is
1 1 1
GT = + + (8.40)
Rth,InAir−F r Rth,InAir−Ro Rth,InAir−EnW i
142
8.1. Implemented thermal network
of the cylindrical part of the frame. This is to still keep a 10 mm distance from the end
winding to the inner side of the end cap. Therefore, the frame mass is somewhat lower
for the Small slot geometry, and higher for the Large and High slot geometries.
The thermal capacitance of the yoke is modeled as a hollow cylinder.
The thermal capacitance of the teeth is found via the geometrical surface area from
the FEA software, and the stack length.
The thermal capacitance of the active part of the winding is found by taking the sum
of the active winding mass and the slot impregnation mass. The two volumes are found
via the axial copper and slot cross sectional areas, the number of slots and the stack
length.
The thermal capacitance of the end winding is found in a similar way as the active
part of the winding, but the winding length is instead the passive part of half a turn,
lpassive .
The thermal capacitance of the magnets and the rotor laminations is found in the
same way as the stator laminations.
The shaft thermal capacitance is here neglected. This is considered a valid simplifi-
cation as its thermal time constant is about 1 h and thus have little effect on the rotor and
magnet temperatures, and even smaller on the stator laminations and windings.
The thermal capacitance of the bearings is found by using the mass in Section 6.2.5,
and assuming that the specific heat is the same as for the shaft material.
The resulting network thermal capacitances are presented in Table 8.2, and in Figure
8.7. The values adhere to material parameters at 100◦ C. As shown, the thermal capaci-
tance of the end winding is generally lower than it is for the active winding.
Table 8.2 Thermal capacitances in J/K in the thermal network for the four slot sizes.
Small Original Large High
Frame Cth,F r 4403 4568 4729 4636
Stator yoke Cth,Y o 4039 3345 2631 2631
Stator teeth Cth,T e 2686 2725 2695 3136
Active winging and impreg. Cth,AcW i 1247 1716 2251 1934
End winding Cth,EnW i 576 871 1250 1026
Magnets Cth,P M 501 501 501 501
Rotor yoke Cth,RoY o 3717 3717 3717 3717
Bearings Cth,Be 169 169 169 169
5000
Small Original Large High
Thermal capacitance (J/K)
4500
4000
3500
3000
2500
2000
1500
1000
500
0
Cth,Fr Cth,Yo Cth,Te Cth,AcWi Cth,EnWi Cth,PM Cth,RoYo
Cth,Be
Figure 8.7 Thermal network node capacitances for the different slot areas.
143
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
Table 8.3 Thermal resistances (W/m K) in the thermal network for the four slot sizes.
Small Original Large High
Rth,Co−F r 0.0037 0.0037 0.0037 0.0037
Rth,F r−Y o 0.0066 0.0060 0.0055 0.0055
Rth,Y o−T e 0.0098 0.0115 0.0112 0.0115
Rth,T e−AcW i 0.0092 0.0084 0.0079 0.0076
Rth,AcW i−EnW i 0.0889 0.0669 0.0526 0.0603
Rth,EnW i−RoY o 1.3527 1.3002 1.2540 1.2665
Rth,T e−P M 0.2459 0.2482 0.2483 0.2483
Rth,RoY o−P M 0.1446 0.1446 0.1446 0.1446
Rth,EnW i−F r 0.3422 0.3080 0.2796 0.3507
Rth,Be−RoY o 1.2464 1.3301 1.4123 1.3649
Rth,RoY o−F r 0.5153 0.5150 0.5153 0.5169
Rth,Be−F r 0.2450 0.2450 0.2450 0.2450
100%
90% Small
80% Original
70% Large
60% High
50%
40%
30%
20%
10%
0%
Rth,Fr-Yo Rth,Yo-Te Rth,Te-AcWi Rth,AcWi-EnWi Rth,Te-PM Rth,Be-RoYo Rth,EnWi-RoYo Rth,EnWi-Fr
Figure 8.8 Relative thermal resistances for the different slot areas, normalized by the largest value.
144
8.3. Steady state comparison of lumped network and FEA
In the transient solution, an initial temperature guess for the first value of the losses is
done similarly as in the steady state case. However, then the temperature increase during
one time step is calculated. In the next time step the losses are temperature corrected by
using the temperatures from the previous time step.
In both the steady state and the transient solution, some temperature depending pa-
rameters are also updated. These are the copper and laminations’ thermal conductivity
and specific heat, and the thermal conductivity of steel, which is used in the shaft and
bearings. The time dependencies of these parameters are presented in Appendix D.
- Temperature of EM parts
145
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
At the high load, the LPN winding temperatures are up to 3.8◦ C (3.1%) higher com-
pared to the FEA solution. The largest deviations are seen in the same motor segments as
for the medium load. The temperature agreement is quite good, which is positive since
their temperatures are important for the study of the thermal performance of the motor.
Table 8.4 Steady state temperature comparison in ◦ C between the LPN and FEA software.
100 Nm at 4000 rpm 50 Nm at 12,000 rpm
Motor Part LPN FEA (Diff.) LPN FEA (Diff.)
Frame 71.3 71.3 (0) 78.1 78.5 (-0.4)
Stator yoke 80.4 80.2 (0.2) 95.5 96.8 (-1.3)
Stator teeth 95.6 91.9 (3.7) 126.4 122.6 (3.8)
Active winding 103.2 102 (1.2) 138.0 139 (-1)
End winding 127.2 - (-) 164.2 - (-)
Magnets 93.9 91.7 (2.2) 123.7 120 (3.7)
Rotor yoke 92.9 91.4 (1.5) 120.2 118.1 (2.1)
Figure 8.10 Temperature distribution in the motor’s axial cross section, at 4000 rpm and 100 Nm
load, with 65◦ C coolant and 6 L/min.
146
8.4. Steady state sensitivity analysis for selected parameters
Figure 8.11 Temperature distribution in the motor’s axial cross section, at 12,000 rpm and 50 Nm
load, with 65◦ C coolant and 6 L/min.
147
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
Table 8.5 Change in steady state temperature in ◦ C between the modified network and the non-
modified, for the Original slot geometry at 50 Nm and 12000 rpm.
Fr StYo StTe AcWi EnWi PM RoYo Be
Reference temperatures 78.1 95.5 126.4 138.0 164.3 123.7 120.2 91.1
hcool = 1000 12.7 12.7 12.1 12.5 13.9 12.3 12.5 12.7
hcool = 3000 -4.3 -4.2 -3.8 -3.9 -4.4 -3.9 -4.1 -4.3
kconv,korr = 0.4 -2.6 -2.5 -2.3 -2.3 -2.6 -2.3 -2.4 -2.5
kconv,korr = 0.6 2.6 2.5 2.3 2.3 2.6 2.3 2.4 2.5
lcontF r−Y o = 5µm 0.0 -5.2 -4.3 -4.2 -3.3 -3.2 -2.0 -0.3
lcontF r−Y o = 15µm 0.0 5.1 4.3 4.3 3.3 3.2 2.1 0.3
lM agGap = 10µm 0.0 0.0 -0.1 0.0 0.1 -1.3 1.5 0.2
lM agGap = 200µm 0.0 0.0 0.0 0.0 0.0 0.3 -0.4 -0.1
λW indM ix = 0.275m 0.1 0.0 -0.1 5.0 3.8 0.3 0.6 0.1
λW indM ix = 0.825m 0.0 0.0 0.0 -1.7 -1.3 -0.1 -0.2 -0.1
Rth,Be × 0.5 0.0 0.0 0.0 0.0 -0.1 -0.4 -0.7 -5.7
Rth,Be × 1.5 0.0 0.0 0.0 0.0 0.1 0.3 0.6 4.7
hF r,in × 0.5 (64 W/m2 K) 0.0 0.7 2.1 2.9 9.4 6.7 11.3 1.8
hF r,in × 1.5 (192 W/m2 K) 0.0 -0.3 -0.9 -1.3 -4.1 -2.9 -4.9 -0.8
hRo × 0.5 (138 W/m2 K) 0.0 0.0 0.1 0.1 -0.3 3.2 6.3 1.0
hRo × 1.5 (414 W/m2 K) 0.0 0.0 -0.1 0.0 0.2 -1.7 -3.2 -0.5
hEnW i × 0.5 (39 W/m2 K) 0.2 1.3 3.4 5.2 18.9 0.5 -2.4 -0.2
hEnW i × 1.5 (115 W/m2 K) -0.1 -0.7 -1.9 -2.8 -10.3 -0.3 1.3 0.1
Setting the coolant heat transfer coefficient hcool to 1000 W/m2 K gives a tempera-
ture increase in all network nodes of around 12-14◦ C. When setting it to 3000 W/m2 K,
the temperature decrease is 4◦ C in all points. Hence, the temperature change when going
from 1000 W/m2 K to 2000 W/m2 K is almost three times larger, compared to going
from 2000 W/m2 K to 3000 W/m2 K. Thus it seems that it is important to be able to
maintain a flow rate that can give a sufficient coolant heat transfer coefficient.
In the reference case, it is assumed that half (0.5) of the heat flux to the cooling
channels pass through the lower duct surfaces. When instead setting the share to 0.1
more and less, the change in temperature in all network nodes is around 2-3◦ C.
When the contact air gap between the magnet and rotor yoke is set to 10µm the
magnet node temperature becomes about 1.3◦ C colder, and the rotor yoke node 1.5◦ C
warmer. The other nodes are very little affected from this change, and are even less
affected in the case of a 200µm gap length.
The same weak dependence is also noted for changes in the thermal resistance of the
bearing, except in the bearing node itself.
When varying the contact gap length between the frame and stator yoke, on the other
hand, the stator yoke temperature changes about 1◦ C per µm gap length. The other
node temperatures also change, however, the values decrease in the motors inward radial
direction. The end winding and active winding temperatures changes about 0.8-0.9 ◦ C
per µm gap length. In [102], gap lengths of up to 77µm are reported for some machines.
Such a gap length in this operating point would lead to winding temperatures of around
200◦ C.
The temperature changes are notable when varying the internal air heat coefficients
to h × 0.5 and h × 1.5. The changes are the smallest for hRo and the largest for hEnW i .
For hRo the largest temperature changes of around 3%-5% is naturally seen in the
148
8.5. Transient thermal response to load step
rotor yoke, which is the closest node. The second closest node, the magnet, changes 1-
2%. The temperature differences in the windings are very small and about the same in
both cases.
When varying hF r,in the largest temperature changes are noted in the rotor yoke
node, and the second largest in the end winding node. The end winding increases about
6% in the h × 0.5-case, and it decreases around 3% in the h × 1.5-case. The magnet
temperature increase about 5% in the h × 0.5-case, and decreases around 2% in the
h × 1.5-case.
The largest changes in the end wining temperatures are noted for changes of hEnW i .
In the h × 0.5-case the temperature increase is about 12%, and in the h × 1.5-case the
decrease is around 6%. The effect on the active winding temperature is 2%-4%, on the
stator teeth 1%-3%, and on the magnets it is less than half a percent.
When it comes to the studied contact gap lengths, winding mix thermal conductivity,
and internal air heat transfer coefficients, it is difficult to analytically find assumptions
that are more applicable than the empirical data presented in the references. However, if
experimental results are available from a physical machine, it it highly recommended to
calibrate such network parameters towards measurements in order for the lumped param-
eter network to better fit the specific machine.
600 130
EndWi EndWi
ActWi ActWi
Teeth 120 Teeth
500 PM
StYo
Ber RoYo
110 StYo
Rotor
400 PM Be
Temp. (oC)
100 Fr
Ploss (W)
300
90
200
80
100 70
0 60
0 20 40 60 0 20 40 60
Time (min) Time (min)
Figure 8.12 Losses and temperatures as a function of time, during a load step of 100 Nm at 4000
rpm with a coolant temperature of 65◦ C and a flow rate of 6L/min, for the Original
slot geometry.
149
Chapter 8. Lumped-Parameter Thermal Model of HighwayII V-shaped Reference
PMSM
1400 180
EndWi EndWi
ActWi ActWi
1200 Teeth Teeth
160
StYo PM
Ber RoYo
1000 Rotor StYo
140
PM Be
Temp. (oC)
Fr
Ploss (W)
800
120
600
100
400
200 80
0 60
0 20 40 60 0 20 40 60
Time (min) Time (min)
Figure 8.13 Losses and temperatures as a function of time, during a load step of 50 Nm at 12000
rpm with a coolant temperature of 65◦ C and a flow rate of 6L/min, for the Original
slot geometry.
The thermal time constants for the two load cases at two different coolant tempera-
tures (20◦ C and 65◦ C) and related flow rates (12 L/min and 6 L/min), are presented in
Table 8.6. The time constants are a bit longer in the case of the lower motor speed com-
pared to the higher. Furthermore, the time constants at one speed level are only slightly
longer for the case of the cold coolant compared to the warm. It should be noted that the
time constants of the rotor and magnets would be a bit longer if also the shaft thermal
capacitance was included in the lumped network.
150
Chapter 9
A comparison between the the performance when using different slot area geometries,
which were introduced in Section 6.7, is here presented regarding thermal steady state,
over load capability and transient load cycle performance. The purpose is to investigate
the differences between the various slot geometries.
151
Chapter 9. Thermal Performance for HighwayII PMSM with Four Different Slot
Areas
Temp StTe Temp AcWi
500
200 o 200 500 o
Temp. ( C) Temp. ( C)
200 T (Nm) 50
TEM,max (Nm)
180 EM,max
0
150 Road load 200 Road load
180
150 150 150
Torque (Nm)
Torque (Nm)
120
12800
150 120 12800
100 0 0
15
12 0 50
100 0 100 0
100 12
0
10 21080
0 10 2100
80 1500 0 80
80 150
120
50 50 120
80 10 80
70 0 70 10
0
0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
(a) Stator teeth. (b) Active winding.
Temp EndWi Temp PM
500
200 200
1000 Temp. (o C)
o
Temp. ( oC)
TEM,max (Nm) 200 TEM,max (Nm)
500 180
10
200 50 12800
180 0 150
100 0
150 2 10 12
1800 00 0
100 15
0 100
120 0
10
12 500 0 21080
100 0 80 1500
200
180
120
50 80
150 50
10 120 10
0 80 0
80 70
70
0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 9.1 Steady state temperature contour lines in selected motor parts as a function of torque and
speed, for the Original slot geometry, with a coolant temperature of 65◦ C and a flow
rate of 6 L/min. The road load of the HighwayII BEV is also included as a reference.
In order for the lifetime of the motor to be as long as possible it is desired to limit the
max temperature in the windings of an electric machine. The chosen critical temperature
depends both on the winding insulation class and the number of working hours that is
the design target [46]. It is here assumed that the winding temperature should not exceed
150◦ C in transient mode, and be limited to 120◦ C in continuous operation mode.
At the same time the magnet temperature should not exceed 140◦ C, as noted in Sec-
tion 6.2.3.
The torque-speed levels for when the winding reach its critical temperatures 120◦ C
and 150◦ C, and the magnet reach 100◦ C and 140◦ C, are presented for all four slot
geometries in Figure 9.2. As expected, it can be noted that the motor with the Small
slot will reach both the critical winding and magnet temperatures at lower torque levels
compared to the other slot geometries. The highest torque levels for the specified critical
temperatures can be reached with the Large slot geometry. This is due to the lower losses,
especially copper losses, in a large part of the motor operating area with the Large slot
compared to the other geometries (presented in Figure 6.18), and the higher losses for
152
9.2. Transient over load
250 250
o o
Small 120 C Small 100 C
o o
Small 150 C Small 140 C
200 Original 120oC
200 o
Original 100 C
o o
Original 150 C Origianl 140 C
Torque (Nm)
Torque (Nm)
50 50
0 0
0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 9.2 Steady state temperature development in end windings and magnets for all slot geome-
tries, at a coolant temperature of 65◦ C and a flow rate of 6 L/min.
The max torque levels reached at the studied critical temperatures above, are pre-
sented in Table 9.1. The magnet reaches the max temperature at lower torque levels
compared to the torque levels at which the end winding reaches the max temperature.
It seems that the magnet’s temperature capability is not enough for this design and ap-
plication, and perhaps it should be replaced by one that has a higher thermal capability.
When considering the max torque at 120◦ C, the ratio of peak to continuous torque is
around 2.8 for the Small slot, 2.2 for the Original, 1.8 for the Large, and 1.9 for the High
slot. As a frame of reference it is around 1.2-2 times for nine existing BEVs (as presented
in Figure A.4).
Table 9.1 Max torque reached at winding and magnet critical temperatures, for all slot geometries,
at a coolant temperature of 65◦ C and a flow rate of 6 L/min.
Small Original Large High
Max torque at 120◦ C end winding 81.7 95.7 104.2 99.9
(88%) (100%) (104%) (99%)
Max torque at 150◦ C end winding 100.7 116.6 125.2 120.8
(86%) (100%) (107%) (104%)
Max torque at 100◦ C magnets 64.4 76.9 85.3 81.1
(84%) (100%) (111%) (105%)
Max torque at 140◦ C magnets 95.1 110.5 119.3 114.5
(86%) (100%) (108%) (104%)
153
Chapter 9. Thermal Performance for HighwayII PMSM with Four Different Slot
Areas
the torque level where the end winding has already reached 120◦ C in steady state. The
initial steady state temperatures of the other motor parts at this torque level are used as a
starting point. Then load steps in the torque level are taken, whereupon the time until the
end winding reaches 150◦ C is found via the transient thermal response. The results are
presented in Figure 9.3.
As expected from the steady state results, the longest time at high torque levels can be
reached with the Large slot geometry. With a load step up to the peak torque level for the
Large slot geometry, 182 Nm (at 2000 rpm), it takes 33s until the end winding reaches
the critical temperature of 150◦ C. For the High slot it takes 27s, for the Original 25s, and
for the Small slot 13s. Considering that 80% of all accelerations in the Test and Logged
cycles are around 9s-13s (Section 3.4), and that this is a worst case approach by using a
very high coolant temperature, the Small slot geometry may still be viable. It should also
be noted that the comparison is done for a fixed max phase current. In case of a larger
current, even larger torque levels could be thermally possible with the Original, Large
and High slot geometries.
30
40 300600
60
100
100 300 600
50 50
0 0
2000 4000 6000 8000 10000 12000 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
300 600
100 100
50 50
0 0
2000 4000 6000 8000 10000 12000 2000 4000 6000 8000 10000 12000
Speed (rpm) Speed (rpm)
Figure 9.3 Time spent in different operating points, before end winding reach 150 ◦ C, at 20 ◦ C
coolant temperature, for two slot sizes.
154
9.3. Performance during drive cycles
800 Teeth 95
StYo EndWi
ActWi
ActWi
700 EndWi 90 Teeth
PM
Ber
600 RoYo
Rotor
StYo
PM Temp. (oC) 85 Be
P loss (W)
500 Fr
400 80
300
75
200
70
100
0 65
1.08 1.1 1.12 1.14 1.16 1.18 0 2000 4000 6000 8000 10000 12000
Time (s) 4 Time (s)
#10
(a) Losses during one cycle. (b) Temperature during ten consecutive cycles.
Figure 9.4 Losses and temperature development during the NEDC cycle with the original slot, at a
coolant temperature of 65◦ C and a flow rate of 6 L/min.
2500 130
Teeth EndWi
StYo ActWi
ActWi 120 Teeth
2000 EndWi PM
Ber RoYo
Rotor 110 StYo
PM Be
Temp. (oC)
1500
(W)
Fr
100
loss
90
P
1000
80
500
70
0 60
4.18 4.2 4.22 4.24 4.26 4.18 4.2 4.22 4.24 4.26
Time (s) #10 4 Time (s) #104
Figure 9.5 Losses and temperature development during the ArtemisMW130 cycle with the original
slot, at a coolant temperature of 65◦ C and a flow rate of 6 L/min.
The peak and cycle average power losses, normalized to the values of the Original
155
Chapter 9. Thermal Performance for HighwayII PMSM with Four Different Slot
Areas
slot geometry, are presented in Figure 9.6. The highest peak losses can be noted for the
Small slot geometry, and the lowest for the Large slot, except in the ECE case. The cycle
average losses are the lowest for the Small slot during all cycles except the accelera-
tion intense ArtemisURBAN, UCLA92, and US06, as well as in the high speed cycle
ArtemisMW130. In these cycles the losses are instead the highest for the Small slot, ex-
cept in UCLA92 where the High slot cause the largest losses. Otherwise, the losses with
the Large slot are the highest in most cycles. The largest difference between the highest
and lowest cycle average losses are noted for the HWFET cycle, which includes high
speed levels but relatively low acceleration levels.
The fact that the machine with the Small slot area has the lowest average cycle losses
was slightly unexpected. This in combination with the highest peak torque and least
amount of copper (which is more expensive than laminations per unit mass [128]), makes
it a rather attractive design solution.
120% 117%
115% 113%
105% 110%
110% 106% 108%
106% 102%106% 100% 106% 106%
103% 104%
105% 101% 100% 100% 100%
100% 100%
100% 100% 99% 100% 99% 100% 99% 99%99% 100% 100% 100% 100%
100% 97% 97%
95% 92% 92%
91%
90% 86%
85% 82% 83%
80%
115%
Normalized cycle average Ploss
Figure 9.6 Peak and cycle average losses during drive cycles, where all values are normalized by
the results for the Original slot size, at a coolant temperature of 65◦ C and a flow rate of
6 L/min.
156
9.3. Performance during drive cycles
The max magnet and end winding temperatures during the selected drive cycles are
presented in Figure 9.7. The max magnet temperature differ 2◦ C or less between the dif-
ferent slot geometries for the investigated drive cycles. The highest magnet temperatures
are seen for the two cycles with the highest average speed levels, but still they do not
reach more than 99◦ C.
The max winding temperature is the highest for the Small slot geometry for all cycles,
followed by the Original slot, and the lowest for the Large slot geometry. As with the
magnet, the largest peak temperatures are noted for the cycles with the highest speed
levels, and are close to the critical for the Small slot geometry.
100 99 9997
97
95 92 92 91 91
90
85
81 8180 80
787979 79
80 78 7878 78 787979 79 77 7776 77
75 73 7373 73 72 74 7474 74
7271 71
69 6969 69
70
65
155
146 146
145
135 127
122 125
125 116 117
115 112 11 3
109
101 102 103 103
105 97 99
95 92 90 87 92 89 92 89 89
88 8986 87 87 87
8380 81 84 83 85 858484
85 81
74
75 72 71 71
65
Figure 9.7 Max magnet and end winding temperatures during drive cycles, at a coolant temperature
of 65◦ C and a flow rate of 6 L/min.
157
Chapter 9. Thermal Performance for HighwayII PMSM with Four Different Slot
Areas
158
Chapter 10
10.1 Conclusions
In this thesis, various drive cycles, legislative, official real-world and measured within the
frame of the project, have been studied and characterized in terms of speed and accelera-
tion cycle parameters, as well as acceleration and speed distribution. The objective was to
assess typical vehicle usage on different road types, but also to study the implication on
vehicle energy consumption due to the drive cycle’s characteristics. For this evaluation,
three reference vehicles (a City car, a Highway car and a Sport car) were designed after
different set performance requirements, with data on existing BEVs as a frame of ref-
erence. An available traction motor, power electronic module and traction battery were
utilized. Furthermore, the electromagnetic efficiency of two v-shaped PMSM reference
motors (CityII and HighwayII) were evaluated when scaling the active length. Then the
consequence on acceleration performance, drive cycle fulfilment and energy consump-
tion during re-scaling of the electric drive system was studied. Moreover, the electromag-
netic losses for four different slot areas were compared (for the HighwayII motor), along
with the thermal steady state and transient over load as well as temperature development
during drive cycles.
Since most of the legislative drive cycles were developed a few decades ago, when
the performance of passenger cars were generally lower than today’s cars, the drive cy-
cles are not fully representative for today’s typical driving. Through comparison with
measured drive cycles, it was found that the measured cycles report higher peak levels
of acceleration for a certain speed level, and a larger number of speed fluctuations. On
the other hand, when studying the relative time spent at certain levels of acceleration and
speed, the measured cycles spend only slightly more time at higher levels of acceleration
compared to the official cycles, at least on average over a group of similar cycles.
From a literature study, various estimation methods of vehicle frontal area were
found. These were compared with available manufacturer data from nine models. It was
found that a suitable approximate estimation of the area is 86 % of the product of vehicle
track width and height. Furthermore, different, often referred to, estimations regarding
the speed dependency of tire rolling resistance were found to deviate from each other
rather much.
159
Chapter 10. Conclusions and Future Work
Through the mapping of existing BEVs it was found that there is a wide spread re-
garding their top speed and acceleration performance, as well as driving range, which
made a categorization ambiguous.
When comparing the three reference City, Highway, and Sport cars their road load
per mass while driving at a constant speed, it was found that the City car has a stronger
speed dependency than the other two, due to its relatively higher aerodynamic drag. This
causes it to have a higher road load relative to its mass for speed levels greater than
66 km/h, compared to the Highway and Sport cars.
The needed wheel force and power, due to performance requirements regarding top
speed, time to accelerate to 100 km/h, take off at 25 % grade, and driving at a high
grade in a high speed were studied. In a comparison it turned out that the acceleration
requirement was the dominating one, hence the only one considered when specifying the
needed output from the powertrain.
Furthermore, it was found that the time to accelerate can be achieved with various
combinations of initial maximum force and maximum power. For a higher initial max-
imum force, the needed maximum power to reach 100 km/h at the specified time is
lower. This will lead to shorter time to accelerate to 50 km/h, but longer time to accel-
erate from 50 to 100 km/h. The conclusion is that a requirement of time to accelerate
from 0 to 100 km/h is not enough to describe the desired performance of a vehicle, there
must also be an expressed requirement on low and high speed performance, respectively.
Another important finding is that the time resolution of the speed time traces, that
defines the official cycles, is relatively low compared to the rate of change of speed.
Then the estimated correlated acceleration is highly dependent on the method used for
calculation, whether it is a forward and backward looking method or a simpler just back-
ward looking method. The consequence on cycle energy consumption per distance is then
highly sensitive to the method used for estimation of acceleration, where the difference
may be as large as 15 % for some cycles.
The resulting average cycle powertrain efficiency are fairly similar for the City and
Highway cars, and indicate slightly lower values for the Urban cycles, 77%-86%, com-
pared to the Rural 85%-88% and highway cycles 87%-89%. The Sport model give lower
average efficiencies due to the higher power rating of the powertrain. Moreover, the av-
erage efficiency values are very similar for both the test cycles and the logged cycles,
within the same road-type category. For the two road-type categories with the lowest
speed levels, the logged cycles indicate a larger spread compared to the test cycles re-
garding energy consumption per driven distance. Also within the same road-type cate-
gory, those cycles that spend most time at high acceleration and speed levels are those
that consume the most energy per distance during propulsion. At the same time, in these
cycles more braking energy is also available for regenerative braking, which makes the
total energy consumption a bit less sensitive to speed fluctuations and high acceleration
levels, still the tendency is a higher energy consumption per driven distance.
The effect on acceleration due to re-scaling of motor active length, is largest for the
shortest stack lengths and smallest for the longest. For the acceleration time 0-100km/h,
the difference between the scale factors 50% and 80% is 12s for the CityII car, and 8s for
the HighwayII, whereas the difference between the scale factors 120% and 150% is only
about 2s for both cars.
Both the CityII and HighwayII concept BEVs can fulfill the speed reference of most
160
10.2. Future Work
Test cycles down to 70% of the original stack length. Also, for both cars, the energy
consumption per driven distance decreases with decreasing stack length, except for the
two acceleration intense urban cycles NYCC and ArtemsURBAN whose minimum is
reached with a up-scaled stack length. For the CityII car, the lowest values are down to
94% of the consumption with the original stack length, and for the HighwayII car the
lowest values are 98% of the original stack length. With both cars, the largest decrease is
noted for cycles with higher speed levels.
When varying the slot size for the v-shaped PMSM HighwayII reference motor while
keeping the same max phase current and voltage, the amount of copper and core losses
are shifted. For a smaller slot area, the copper losses increase and the core losses de-
crease, whereas the opposite applies to a larger slot and copper area. The resulting total
losses are then largest for the smaller slot geometry in a large part of the motor oper-
ating area. Due to this, during thermal steady state, the small slot geometry reaches the
chosen continuous end winding temperature at lower torque levels compared to the other
slot geometries. The largest continuous torque levels are reached with the large slot area.
During a transient over load step which originates from the steady state torque level, the
motor version with the largest slot area is capable of staying at its peak torque for 33s.
The high, original high and small slot geometries are able to stay at the same operating
point for 27s, 25s, and 13s, respectively.
The small slot geometry had the highest peak losses during the drive cycles, however,
on a cycle average it had the lowest losses for 7 of the 11 studied cycles. This fact,
in combination with the highest peak torque and lowest material cost, makes it a very
interesting option as an electric vehicle traction motor.
161
Chapter 10. Conclusions and Future Work
the lumped parameter network, perhaps the air temperatures could also be evaluated and
it’s parameters temperature adjusted for a higher accuracy.
Moreover, it would be interesting to also study the temperature development during
the Logges real-world drive cycles for the different slot areas, in order to determine the
effect on temperature for a larger number of speed fluctuations and various grade levels.
Finally, the derived drive cycles from the measured GPS and accelerometer signals,
could perhaps be improved using more advanced filtering techniques.
162
References
References
[1] (2012, may) Regulation no 101, official journal of european union. Economic
Commission for Europe of the United Nations. Accessed 2012-10-31.
[Online]. Available: http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:
L:2012:138:0001:0077:en:PDF
163
References
[10] (2012) Advanced powertrain research facility, avta nissan leaf testing and analysis.
Idaho National Laboratory (INL), Advanced Vehicle Testing Activity (AVTA).
Accessed 2013-10-17.
[11] S. Williamson, A. Emadi, and K. Rajashekara, “Comprehensive efficiency model-
ing of electric traction motor drives for hybrid electric vehicle propulsion applica-
tions,” Vehicular Technology, IEEE Transactions on, vol. 56, no. 4, pp. 1561–1572,
July 2007.
[12] A. Boretti, “Analysis of the regenerative braking efficiency of a latest electric ve-
hicle,” in SAE Technical Paper. SAE International, 11 2013.
[13] J. G. Hayes and K. Davis, “Simplified electric vehicle powertrain model for range
and energy consumption based on epa coast-down parameters and test validation
by argonne national lab data on the nissan leaf,” in Transportation Electrification
Conference and Expo (ITEC), 2014 IEEE, June 2014, pp. 1–6.
[14] F. An, M. Barth, and G. Scora, “Impacts of diverse driving cycles on electric and
hybrid electric vehicle performance,” in SAE Technical Paper. SAE International,
08 1997.
[15] M. D. Gennaro, E. Paffumi, G. Martini, U. Manfredi, H. Scholz, H. Lacher,
H. Kuehnelt, and D. Simic, “Experimental investigation of the energy efficiency of
an electric vehicle in different driving conditions,” in SAE Technical Paper. SAE
International, 04 2014.
[16] J.-M. Zaccardi and F. Le Berr, “Analysis and choice of representative drive cy-
cles for light duty vehicles, case study for electric vehicles,” Proceedings of the
Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering,
2012.
[17] H. Neudorfer, A. Binder, and N. Wicker, “Analyse von unterschiedlichen
fahrzyklen für den einsatz von elektrofahrzeugen,” Elektrotechnik und Informa-
tionstechnik, vol. 123, no. 7-8, pp. 352–360, 2006.
[18] X. Wu, D. Freese, A. Cabrera, and W. A. Kitch, “Electric vehicles energy con-
sumption measurement and estimation,” Transportation Research Part D: Trans-
port and Environment, vol. 34, no. 0, pp. 52 – 67, 2015.
[19] P. Mellor, D. Roberts, and D. Turner, “Lumped parameter thermal model for elec-
trical machines of tefc design,” Electric Power Applications, IEE Proceedings B,
vol. 138, no. 5, pp. 205–218, Sept 1991.
[20] G. Kylander, “Thermal modelling of small cage induction motors,” Thesis for
the degree of Doctor of Philosophy in Engineering, Technical Report No. 265,
Chalmers University of Technology, Feb 1995.
[21] J. Nerg, M. Rilla, and J. Pyrhonen, “Thermal analysis of radial-flux electrical ma-
chines with a high power density,” Industrial Electronics, IEEE Transactions on,
vol. 55, no. 10, pp. 3543–3554, Oct 2008.
164
References
165
References
166
References
[49] Y. Chin and D. Staton, “Transient thermal analysis using both lumped-circuit
approach and finite element method of a permanent magnet traction motor,” in
AFRICON, 2004. 7th AFRICON Conference in Africa, vol. 2, Sept 2004, pp.
1027–1035 Vol.2.
[50] A. F. Mills, Heat Transfer. Prentice-Hall Inc., 1999.
[51] D. Staton, S. PICKERING, and D. LAMPARD, “Recent advancement in the ther-
mal design of electric motors,” in SMMA 2001 Fall Technical Conference "Emerg-
ing Technologies for the Electric Motion Industry", Raleigh-Durham, North Car-
olina, USA, Oct 2001.
[52] A. Boglietti, A. Cavagnino, and D. Staton, “Determination of critical parameters
in electrical machine thermal models,” Industry Applications, IEEE Transactions
on, vol. 44, no. 4, pp. 1150–1159, July 2008.
[53] J. Lindström, “Development of an experimental permanent-magnet motor drive,”
Licentiate of Engineering, Technical Report No. 312L, Chalmers University of
Technology, apr 1999.
[54] I. J. PEREZ and J. G. KASSAKIAN, “A stationary thermal model for
smooth air-gap rotating electric machines,” Electric Machines & Power
Systems, vol. 3, no. 3-4, pp. 285–303, 1979. [Online]. Available: http:
//dx.doi.org/10.1080/03616967908955346
[55] D. Kuehbacher, A. Kelleter, and D. Gerling, “An improved approach for transient
thermal modeling using lumped parameter networks,” in Electric Machines Drives
Conference (IEMDC), 2013 IEEE International, May 2013, pp. 824–831.
[56] G. Dajaku and D. Gerling, “An improved lumped parameter thermal model for
electrical machines,” in 17th International conference on electrical machines
(ICEM2006), 2006.
[57] F. Sprei, S. Karlsson, and J. Holmberg, “Better performance or lower fuel con-
sumption. technological development in the swedish new-car fleet...” Transporta-
tion Research Part D: Transport and Environment, vol. 13:2, s. 75-85, 2008.
[58] S. Karlsson. (2013) The swedish car movement data project final report.
Division of Physical resource Theory, Chalmers University of Technology.
[Online]. Available: http://publications.lib.chalmers.se/records/fulltext/187380/
local_187380.pdf
[59] H. Watson. (1978) Vehicle driving patterns and measurement methods for energy
and emission assessment. Bureau of Transport Economics, Canberra. Accessed
2014-01-15. [Online]. Available: http://www.bitre.gov.au/publications/1978/files/
op_030.pdf
[60] I. M. Berry, “The effects of driving style and vehicle performance on the
real-world fuel consumption of u.s. light-duty vehicles,” Master of Science
in Mechanical Engineering and Master of Science in Technology and Policy,
167
References
168
References
[69] (2013, jan) Finalized test cycle, wltc. UNECE, United Nations Economic
Commission for Europe, Worldwide harmonized Light vehicles Test Procedure
(WLTP). Accessed 2013-11-14. [Online]. Available: https://www2.unece.org/
wiki/download/attachments/5801079/WLTP-DHC-16-06e_rev.xlsx?api=v2
[70] (2012) Krav för vägars och gators utformning. Trafikverket. Accessed 2013-10-
01. [Online]. Available: http://publikationswebbutik.vv.se/shopping/ShowItem.
aspx?id=5848
[71] (2005) A policy on design standards interstate system. Ameri-
can Association of State Highway and Transport Officials. Accessed
2013-10-01. [Online]. Available: http://www.dot.state.fl.us/rddesign/qa/Data/
AASHTO-InterstateDesignStandards.pdf
[72] (2013) Pmsv3 database. Trafikverket. Accessed 2013-10-01. [Online]. Available:
https://pmsv3.trafikverket.se/Pages/EnkelSok/EnkelSokView.aspx
[73] (2009, nov) San fransisco, ca, usa, streets colored by slope. Data
Pointed. Accessed 2013-10-01. [Online]. Available: http://www.datapointed.
net/visualizations/maps/san-francisco/streets-slope/
[74] N. S. Pearre, W. Kempton, R. L. Guensler, and V. V. Elango, “Electric vehicles:
How much range is required for a day’s driving?” Transportation Research Part C:
Emerging Technologies, vol. 19, no. 6, pp. 1171 – 1184, 2011. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0968090X1100012X
[75] (2011) Summary of travel trends: 2009 national household travel survey.
US Department of Transportation, Federal Highway Administration. Accessed
2013-10-09. [Online]. Available: http://nhts.ornl.gov/2009/pub/stt.pdf
[76] (2007) Res 2005-2006, den nationella resvaneundersökningen. SIKA Statistik,
Statens institut för kommunikationsanalys. Accessed 2010-12-21. [Online].
Available: http://trafa.se/PageDocuments/ss_2007_19_1.pdf
[77] (2007) Körsträckor øor 2006. SIKA Statistik, Statens institut för
kommunikationsanalys. Accessed 2010-12-21. [Online]. Available: http:
//www.scb.se/Statistik/TK/_dokument/Korstr_2006.pdf
[78] (2012) Driving and parking patterns of european car drivers, a mobility
survey. European Commission. Accessed 2013-10-10. [Online]. Available:
http://publications.jrc.ec.europa.eu/repository/handle/111111111/26994/1/
[79] (2010) Ample driving range for everyday deeds. Nissan Motor Corporation.
Accessed 2013-04-02. [Online]. Available: http://www.newsroom.nissan-europe.
com/download/media/specialfile/41486_3_6.aspx
[80] S. Karlsson and E. Jonson, “The importance of car movement data for determining
design, viability and potential of phevs,” Proceedings of International Advanced
Mobility Forum (IAMF) 2011, Geneva Switzerland, 2011.
169
References
[81] L.-H. Kullingsjö and S. Karlsson. (2012) The swedish car movement
data project. Division of Physical resource Theory, Chalmers University
of Technology. [Online]. Available: http://publications.lib.chalmers.se/records/
fulltext/local_165497.pdf
[82] (2007) Resvanor i göteborgsregionen 2005. Göteborgs Stad, Vägverket and
Västtrafik. Accessed 2011-10-11. [Online]. Available: http://www2.trafikkontoret.
goteborg.se/resourcelibrary/Resvanor.pdf
[83] J. Krumm, “How people use their vehicles: Statistics from the 2009 national
household travel survey,” SAE 2012 World Congress & Exhibition, 04 2012.
[Online]. Available: http://dx.doi.org/10.4271/2012-01-0489
[84] (2007) Axp course design - baseline driving statistics. Automotive Xprice.
Accessed 2011-10-08. [Online]. Available: http://www.progressiveautoxprize.
org/files/downloads/auto/AXP_FHWA_driving_stats.pdf
[85] R. van Haaren. (2012) Assessment of electric car’s range requirements and usage
pattern based on driving behavior recorded in the national household travel
survey of 2009. Automotive Xprice. Accessed 2011-10-07. [Online]. Available:
http://www.solarjourneyusa.com/HowFarWeDrive_v1.3.pdf
[86] E. A. Grunditz, “Bev powertrain component sizing with respect to performance,
energy consumption and driving patterns,” Thesis for the degree of Licentiate
of Engineering, Chalmers University of Technology, 2014. [Online]. Available:
http://publications.lib.chalmers.se/records/fulltext/198921/198921.pdf
[87] (1999, oct) 1999 ev america technical specifications. Electric Transportation
Applications. Accessed 2014-02-04. [Online]. Available: http://avt.inl.gov/fsev.
shtml
[88] (2014) Igbt modules up to 600v, 650v. Infineon. Accessed 2014-03-
20. [Online]. Available: http://www.infineon.com/cms/en/product/power/igbt/
igbt-module/igbt-module-600v-650v/
[89] (2014, jan) Aesc home page. AESC, Automotive Energy Supply Corporation.
Accessed 2014-03-14. [Online]. Available: http://www.eco-aesc-lb.com/en/
aboutus/company/
[90] (2011) 2011 nissan leaf - vin 0356. Idaho National Laboratory (INL), Advanced
Vehicle Testing Activity (AVTA). Accessed 2014-02-04. [Online]. Available:
http://avt.inel.gov/pdf/fsev/batteryleaf0356.pdf
[91] J. Hellsing, “Design and otimization of a permanent magnet motor for a hybrid
electric vehicle,” Licentiate of Engineering, Technical Report No. 282L, Chalmers
University of Technology, mar 1998.
[92] A. Rabiei, T. Thiringer, and J. Lindberg, “Maximizing the energy efficiency of a
pmsm for vehicular applications using an iron loss accounting optimization based
on nonlinear programming,” in Electrical Machines (ICEM), 2012 XXth Interna-
tional Conference on, Sept 2012, pp. 1001–1007.
170
References
[93] M.-F. Hsieh, Y.-C. Hsu, D. Dorrell, and K.-H. Hu, “Investigation on end winding
inductance in motor stator windings,” Magnetics, IEEE Transactions on, vol. 43,
no. 6, pp. 2513–2515, June 2007.
[94] Y. Sato, S. Ishikawa, T. Okubo, M. Abe, and K. Tamai, “Development of
high response motor and inverter system for the nissan leaf electric vehicle,”
in SAE Technical Paper. SAE International, 04 2011. [Online]. Available:
http://dx.doi.org/10.4271/2011-01-0350
[95] S. Oki, S. Ishikawa, and T. Ikemi, “Development of high-power and
high-efficiency motor for a newly developed electric vehicle,” SAE Int.
J. Alt. Power., vol. 1, pp. 104–111, 04 2012. [Online]. Available: http:
//dx.doi.org/10.4271/2012-01-0342
[96] H. Shimizu, T. Okubo, I. Hirano, S. Ishikawa, and M. Abe, “Development
of an integrated electrified powertrain for a newly developed electric vehicle,”
in SAE Technical Paper. SAE International, 04 2013. [Online]. Available:
http://dx.doi.org/10.4271/2013-01-1759
[97] T. Nakada, S. Ishikawa, and S. Oki, “Development of an electric motor for a
newly developed electric vehicle,” in SAE Technical Paper. SAE International,
04 2014. [Online]. Available: http://dx.doi.org/10.4271/2014-01-1879
[98] T. Burress and S. Campbell, “Benchmarking ev and hev power electronics and
electric machines,” in Transportation Electrification Conference and Expo (ITEC),
2013 IEEE, June 2013, pp. 1–6.
[99] B. Ozpineci. (2014) Annual progress report for the power electronics and electric
motors program. Oak Ridge National Laboratory. Accessed 2015-04-23. [Online].
Available: http://info.ornl.gov/sites/publications/Files/Pub52422.pdf
[100] T. A. B. et. al. (2011) Evaluation of the 2010 toyota prius hybrid synergy
drive system. Oak Ridge National Laboratory. Accessed 2013-11-27. [Online].
Available: http://info.ornl.gov/sites/publications/files/Pub26762.pdf
171
References
[104] (2015) Hi-lite no30 data sheet. Cogent. Accessed 2015-05-19. [Online]. Available:
http://cogent-power.com/cms-data/downloads/Hi-Lite_NO30.pdf
[105] M. D. Magnus Lindenmo and T. C. Support, “No30 hi-lite material data,” Cogent
Power / Surahammars Bruks AB, private Communication 2016-03-01.
[106] S. Constantinides. (2015) The important role of dysprosium in modern
permanent magnets. Arnold Magnetic Technologies Corp. Accessed 2016-03-18.
[Online]. Available: http://www.arnoldmagnetics.com/en-us/Technical-Library/
Technical-Publications
[107] (2011) Critical materials strategy. U.S. Department Of Energy. Accessed 2016-
03-18. [Online]. Available: http://energy.gov/sites/prod/files/DOE_CMS2011_
FINAL_Full.pdf
[108] S. Hoenderdaal, L. T. Espinoza, F. Marscheider-Weidemann, and W. Graus, “Can
a dysprosium shortage threaten green energy technologies?” Energy, vol. 49,
pp. 344 – 355, 2013. [Online]. Available: http://www.sciencedirect.com/science/
article/pii/S0360544212008055
[109] (2014) Hitachi review vol. 63, no. 3, p.113. Hitachi Metals, Ltd. Accessed
2015-05-28. [Online]. Available: http://www.hitachi.com/rev/specialissue/2014/
ebook/index.html#page=1
[110] (2014) Neomax series, data sheets. Hitachi Metals, Ltd. Accessed 2015-05-29.
[Online]. Available: http://www.hitachi-metals.co.jp/e/products/auto/el/pdf/nmx_
a.pdf
[111] (2015) Neodymium-iron-boron magnets neomax, home page. Hitachi Metals,
Ltd. Accessed 2015-05-27. [Online]. Available: http://www.hitachi-metals.co.jp/
e/products/auto/el/p03_21.html
[112] (2016) Ball bearings. SKF. Accessed 2016-03-30. [Online]. Available: http:
//www.skf.com/group/products/bearings-units-housings/ball-bearings/index.html
[113] (2012, oct) Federal report, vol. 77, no. 199, final rule. US Environmental
Protection Agency (EPA), Department of Transportation, National Highway
Traffic Safety Administration (NHTSA). Accessed 2013-12-08. [Online].
Available: http://www.epa.gov/otaq/climate/regs-light-duty.htm#new1
[114] P. Juris, A. Brune, and B. Ponick, “A coupled thermal-electromagnetic energy
consumption calculation for an electric vehicle with wheel hub drive considering
different driving cycles,” in Vehicle Power and Propulsion Conference (VPPC),
2012 IEEE, Oct 2012, pp. 28–31.
[115] A. Abdelli, “Optimal design of an interior permanent magnet synchronous
motor for wide constant-power region operation: Considering thermal and
electromagnetic aspects,” SAE Int. J. Alt. Power., vol. 3, pp. 129–138, 04 2014.
[Online]. Available: http://dx.doi.org/10.4271/2014-01-1889
172
References
[116] J. Fan, C. Zhang, Z. Wang, and E. Strangas, “Thermal analysis of water cooled
surface mount permanent magnet electric motor for electric vehicle,” in Electrical
Machines and Systems (ICEMS), 2010 International Conference on, Oct 2010, pp.
1024–1028.
[117] “2015 focus electric owner’s manual,” Ford Service Con-
tent, 2014, accessed 2015-06-26. [Online]. Available: http:
//www.fordservicecontent.com/Ford_Content/Catalog/owner_information/
2015-Focus-Electric-Owners-Manual-version-1_om_EN-US_11_2014.pdf
[118] “Rav4 ev 2012, quick reference guide,” Toyota, 2012, accessed 2015-05-17.
[Online]. Available: http://www.myrav4ev.com/docs/RAV4EV_Owners_Manual.
pdf
[119] (1991) Table 8-3-1 - properties of mixture water/glycol. Detector Cooling.
Accessed 2016-02-02. [Online]. Available: https://detector-cooling.web.cern.ch/
detector-cooling/data/Table%208-3-1.htm
[120] “Technical training. product information. bmw service i01 high-voltage
components,” Bayerische Motorenwerke Aktiengesellschaft, 2014, ac-
cessed 2015-06-25. [Online]. Available: http://www.kneb.net/bmw/I01/06_
I01High-voltageComponents.pdf
[121] G. Li, J. Ojeda, E. Hoang, M. Gabsi, and M. Lecrivain, “Thermal-electromagnetic
analysis for driving cycles of embedded flux-switching permanent-magnet mo-
tors,” Vehicular Technology, IEEE Transactions on, vol. 61, no. 1, pp. 140–151,
Jan 2012.
[122] G. I. Taylor, “Distribution of velocity and temperature between concentric
rotating cylinders,” Proceedings of the Royal Society of London A: Mathematical,
Physical and Engineering Sciences, vol. 151, no. 874, pp. 494–512, 1935.
[Online]. Available: http://rspa.royalsocietypublishing.org/content/151/874/494
[123] C. Gazley, “Heat transfer characteristics of rotational and axial flow between con-
centric cylinders,” Transactions of the ASME, vol. 80, no. 1, pp. 79–90, 1958.
[124] K. M. Becker and J. Kaye, “Measurements of diabatic flow in an annulus with an
inner rotating cylinder,” Journal of Heat Transfer, vol. 84, no. 2, pp. 97–104, may
1962. [Online]. Available: http://dx.doi.org/10.1115/1.3684335
[125] B. Zhang, R. Qu, J. Wang, W. Xu, X. Fan, and Y. Chen, “Thermal model of to-
tally enclosed water-cooled permanent-magnet synchronous machines for electric
vehicle application,” Industry Applications, IEEE Transactions on, vol. 51, no. 4,
pp. 3020–3029, July 2015.
[126] (2013) Ball bearings catalogue. SKF. Accessed 2016-04-30. [Online]. Available:
http://www.skf.com/binary/138-121486/SKF-rolling-bearings-catalogue.pdf
[127] A. Boglietti and A. Cavagnino, “Analysis of the endwinding cooling effects in tefc
induction motors,” Industry Applications, IEEE Transactions on, vol. 43, no. 5, pp.
1214–1222, Sept 2007.
173
References
174
References
[141] “Paris 2012,” Toyota Motor Europe, 2012, accessed 2015-03-23. [Online]. Avail-
able: http://toyotamedia.iconicweb.com/mediasite/specs/Paris_MS_2012_EN.pdf
[142] “Scion iq ev electric 2013 model emergency response guide,” 2012 Toyota
Motor Corporation, 2012, accessed 2015-03-23. [Online]. Available: http:
//afvsafetytraining.com/erg/Scion-IQEV-2013.pdf
[143] “2014 scion iq specifications,” Scion USA, 2014, accessed 2015-03-23. [Online].
Available: http://www.scion.com/cars/iQ2014/specs/
[144] “2015 fiat 500e specifications,” Fiat Chrysler Automobiles, 2014, accessed
2015-03-20. [Online]. Available: http://www.media.chrysler.com/
[145] “Fiat 500e,” Fiat USA, 2013, accessed 2015-03-26. [Online]. Available:
http://www.fiatusa.com/assets/pdf/brochures/fiat_500e.pdf
[146] “Citroen c-zero caracterisyiques techniques et equipements,” Citroen France,
2014, accessed 2015-03-20. [Online]. Available: http://www.citroen.fr/vehicules/
citroen/citroen-c-zero.html
[147] “Citroen c-zero,” Citroen France, 2015, accessed 2015-03-27. [Online]. Available:
http://www.citroen.fr/vehicules/citroen/citroen-c-zero.html#sticky
[148] “Peugeot ion, preise, ausstattungen und technische daten,” Peugeot Ger-
many, 2015, accessed 2015-03-27. [Online]. Available: http://www.peugeot.de/
showroom/ion/5-turer/
[149] “2014 i-miev specs,” Mitsubishi Canada, 2014, accessed 2015-03-27.
[Online]. Available: http://www.mitsubishi-motors.com/en/showroom/i-miev/
specifications/
[150] “2014 fuel economy datafile,” Office of Transportation & Air Quality, U.S.
Environmental Protection Agency, U.S. Department of Energy, 2014, accessed
2015-03-20. [Online]. Available: http://www.fueleconomy.gov/feg/download.
shtml
[151] “2014 i-miev specs,” Mitsubishi Canada, 2014, accessed 2015-03-27. [Online].
Available: http://www.mitsubishi-motors.ca/en/vehicles/i-miev/2014/specs/
[152] “Tekniska data nya e-up!” Volkswagen Sweden, 2013, accessed 2015-03-18.
[Online]. Available: http://personbilar.volkswagen.se/sv/models/up/broschyrer.
html
[153] “e-up specifications,” Volkswagen Japan, 2015, accessed 2015-03-13. [Online].
Available: http://web.volkswagen.co.jp/cms4/trimlevel/e-up_specs.html#url=
[154] “10 fragen zum e-up!” Volkswagen Das Auto Magazin, 2015,
accessed 2015-07-02. [Online]. Available: http://magazin.volkswagen.de/
e-up-elektromobilitaet-technik.html
175
References
176
References
177
References
[180] “2013 model year ford hybrid car and electric vehicle warranty guide,” Ford,
2013, accessed 2015-07-02. [Online]. Available: http://www.ford.com/resources/
ford/general/pdf/2013HybridWarranty.pdf
[181] “Kia soul ev faktablad,” Kia Motors Sweden AB, 2015, accessed 2015-04-
02. [Online]. Available: https://www.kia.com/se/-/media/se/forms/brochure/pdf/
faktablad/2015/20150216-soul-ev-faktablad.pdf
[182] “Kia soul ev technical specification,” Kia Motors UK, 2015, accessed 2015-04-02.
[Online]. Available: http://www.kia.co.uk/new-cars/range/compact-cars/soul-ev/
specification/technical%20specification.aspx
[183] “All-new 2015 soul ev,” Kia Motors America, 2014, accessed 2015-06-30. [On-
line]. Available: http://cdn.dealereprocess.com/cdn/brochures/kia/2015-soulev.
pdf
[184] “Mercedes-benz b-klasse electric drive: Technische daten,” Mercedes-
Benz Germany, 2015, accessed 2015-04-02. [Online]. Available:
http://www.mercedes-benz.de/content/germany/mpc/mpc_germany_website/
de/home_mpc/passengercars/home/new_cars/models/b-class/w242/facts/
technicaldata/model.html
[185] “The mercedes-benz b-class electric drive: Electric driving with-
out compromises,” Daimler Global Media Site, 2014, accessed
2015-04-02. [Online]. Available: http://media.daimler.com/dcmedia/
0-921-1740090-1-1751040-1-0-0-1751123-0-1-12759-614216-0-0-0-0-0-0-0.
html
[186] “2014 bmw i3 vs. 2014 mercedes benz b-class electric drive, final scoring,
performance data, and complete specs,” Car And Driver, 2014, accessed
2015-04-02. [Online]. Available: http://www.caranddriver.com
[187] “Byd e6 specs,” BYD Latin America, 2015, accessed 2015-04-10. [Online].
Available: http://www.byd.com/la/auto/e6.html
[188] “Byd e6,” BYD North America, 2015, accessed 2015-04-10. [Online]. Available:
http://www.byd.com/na/auto/e6.html
[189] “Byd e6 specifications,” BYD AUTO, 2015, accessed 2015-04-4. [Online].
Available: http://www.byd.com/na/auto/e6.html#p2
[190] “Nissan e-nv200 combi, evalia,” Nissan Sweden, 2015, accessed 2015-
05-15. [Online]. Available: http://www.nissan.se/etc/medialib/nissaneu/_SE_
sv/_Other_pdf/pricelists.Par.67618.File.dat/e-NV200_Kundprislista_Combi_
Evalia_03-03-15_0522.pdf
[191] “Nissan e-nv200,” Nissan UK, 2015, accessed 2015-05-18. [Online]. Available:
http://www.nissan.co.uk/GB/en/tool/brochure/e-nv200.html
[192] “2014 rav4 ev,” Toyota, 2014, accessed 2015-05-17. [Online]. Available:
http://www.toyota.com/content/ebrochure/2014/rav4ev_ebrochure.pdf
178
References
[193] “Rav4 ev,electric vehicle dismantling manual,” Toyota, 2012, accessed 2015-05-
17. [Online]. Available: https://techinfo.toyota.com/techInfoPortal/staticcontent/
en/techinfo/html/prelogin/docs/rav4evdisman2nd.pdf
[194] “Tesla model s,” Tesla Motors USA, 2015, accessed 2015-10-20. [Online].
Available: http://www.teslamotors.com/models
[195] “Tesla model x,” Tesla Motors USA, 2015, accessed 2015-10-20. [Online].
Available: http://www.teslamotors.com/modelx
[196] “Roadster features & specifications,” Tesla Motors, 2012, accessed 2012-04-17.
[Online]. Available: http://www.teslamotors.com/roadster/specs
[197] “Tesla roadster, emergency responder guide,” Tesla Motors, 2009, accessed
2015-03-31. [Online]. Available: https://www.teslamotors.com/sites/default/files/
downloads/Tesla_Roadster_Emergency_Responder_Guide.pdf
[198] “Roadster, technology, battery,” Tesla Motors, 2015, accessed 2015-03-31.
[Online]. Available: http://my.teslamotors.com/roadster/technology/battery
[199] “Concept one, specifications,” Rimac Automobili, 2015, accessed 2015-
07-01. [Online]. Available: http://www.rimac-automobili.com/concept_one/
specifications-10
[200] “Rimac automobili and vredestein at top marques monaco,” Rimac Automobili,
2012, accessed 2014-02-12. [Online]. Available: http://www.rimac-automobili.
com/press/rimac-automobili-and-vredestein-at-top-marques-monaco-p8
[201] “Sls amg electric drive, accumulated data,” Mercedes-Benz, 2015, accessed
2015-03-20. [Online]. Available: http://www.mercedes-amg.com/webspecial/sls_
e-drive/eng.php
[202] “Mercedes-benz sls amg coupé electric drive, electrifying the world’s
most powerful electric super sports car,” Daimler Media, 2012, ac-
cessed 2015-03-20. [Online]. Available: http://media.daimler.com/dcmedia/
0-921-614307-1-1535313-1-0-0-0-0-0-11702-0-0-1-0-0-0-0-0.html
[203] “Mercedes-benz sls amg coupé electric drive,” Daimler Press Information,
2013, accessed 2015-03-13. [Online]. Available: www.daimler.com/Projects/c2c/
channel/documents/2292390_PM_SLS_AMG_en.doc
[204] “Lightning gt,” Lightning Car Company, 2015, accessed 2015-04-10. [Online].
Available: http://www.lightningcarcompany.co.uk/Lightning/home.html
[205] “Lightning gt,” Car body design, 2008, accessed 2015-07-02. [Online]. Available:
http://www.carbodydesign.com/archive/2008/07/25-lightning-gt/
[206] “Sp:01,” Detroit Electric, 2015, accessed 2015-07-01. [Online]. Available:
http://detroit-electric-group.com/sp01.html
179
References
180
References
[220] “2014 chevrolet spark ev vs. 2013 fiat 500e, 2014 ford focus electric, 2013
honda fit ev, 2013 nissan leaf sl, 2013 smart fortwo ed cabriolet, final
scoring, performance data, and complete specs,” Car and Driver, 2013, accessed
2015-03-30. [Online]. Available: http://www.caranddriver.com
[221] “2015 chevrolet spark product information,” Chevrolet News, GM Media,
2015, accessed 2015-03-25. [Online]. Available: http://media.gm.com/media/us/
en/chevrolet/vehicles/spark-ev/2015.html
[222] “Bmw i3 models,” BMW USA, 2015, accessed 2015-07-08. [Online]. Available:
http://www.bmwusa.com/bmw/bmwi/i3
[223] “Bmw i3 and i3 with range extender (i01),” BMW UK, 2013, accessed 2015-
03-26. [Online]. Available: https://www.press.bmwgroup.com/united-kingdom/
download.html?textId=190985&textAttachmentId=234889
[224] “Overview: Renault at the paris motor show,” Under the Skin, Automotive
Technology Online, 2010, accessed 2014-02-13. [Online]. Available: http:
//under-the-skin.org.uk/news_renault_1010.html
[225] “Tekniska data 2012 volvo utg 3,” Volvo Car Sverige AB, 2011, accessed
2015-03-20. [Online]. Available: https://www.media.volvocars.com/se/sv-se
[226] “Volvo c30 electric, siemens fleet in cooperation with volvo car corporation,”
Siemens AG, Digital Factory, eCar Powertrain Systems, Feb 2015. [Online].
Available: www.siemens.com/electriccar
[227] “World premieres of the e-up! and e-golf, international motor show
(iaa) frankfurt, september 2013,” Volkswagen, 2013, accessed 2015-
10-24. [Online]. Available: http://www.smmt.co.uk/wp-content/uploads/sites/
2/Innovation-Stream-public-3.pdf
[228] “Nissan releases specs for the 2011 leaf ev,” Nissan in the News, 2010,
accessed 2015-03-27. [Online]. Available: http://www.nissaninthenews.com/
nissan-releases-specs-for-the-2011-leaf-ev/
[229] “2013 honda fit ev (model since july 2012 for north america u.s.) specifications
& performance data review,” Automobile catalog, 2014, accessed 2015-03-
30. [Online]. Available: http://www.automobile-catalog.com/car/2013/1761020/
honda_fit_ev.html
[230] “Renault fluence z.e.” Renault Denmark, 2014, accessed 2014-02-07. [Online].
Available: http://www.renault.co.uk/cars/electric-vehicles/fluence-ze/fluence-ze/
specifications/index.jsp
[231] “Renault fluence z.e. drive the change,” Renault UK, 2013, accessed 2015-03-30.
[Online]. Available: http://motorline.co.uk/renault/pdfs/fluence-ze.pdf
181
References
182
References
183
References
[259] “2015 fuel economy datafile,” Office of Transportation & Air Quality, U.S.
Environmental Protection Agency, U.S. Department of Energy, 2015, accessed
2015-07-16. [Online]. Available: http://www.fueleconomy.gov/feg/download.
shtml
[260] “Introducing the byd e6 electric car,” Tree Hugger, 2008, ac-
cessed 2015-04-10. [Online]. Available: http://www.treehugger.com/cars/
introducing-the-byd-e6-electric-car.html
[261] “Byd e6a auto owner’s manual,” BYD AUTO, 2009, accessed 2015-04-
10. [Online]. Available: http://www.produktinfo.conrad.com/datenblaetter/
250000-274999/251704-in-01-en-BYD_LITHIUM_FE_BLOCK_12V_10_AH_
B_BMS.pdf
[262] “Model s specifications,” Tesla Motors USA, 2015, accessed 2015-03-31.
[Online]. Available: http://www.teslamotors.com/support/model-s-specifications
[263] “Tesla model s 60kwh,” Miljofordon.se, 2015, accessed 2015-03-31. [Online].
Available: http://www.miljofordon.se/fordon
[264] “Model s,” Tesla Motors Sweden, 2015, accessed 2015-03-20. [Online].
Available: http://www.teslamotors.com/sv_SE/models
[265] “2013 tesla model s overview,” Hybrid Cars, 2013, accessed 2015-03-31.
[Online]. Available: http://www.hybridcars.com/2013-tesla-model-s-overview/
[266] “2013 fuel economy datafile,” Office of Transportation & Air Quality, U.S.
Environmental Protection Agency, U.S. Department of Energy, 2013, accessed
2013-09-05. [Online]. Available: http://www.fueleconomy.gov/feg/download.
shtml
[267] “Operator’s manual, smart fortwo coupé and smart fortwo cabriolet electric
drive,” Smart USA, 2013, accessed 2015-05-25. [Online]. Available: http://
owners.smartusa.com/userfiles/file/2014-smart-electric-drive-owners-manual.pdf
[270] “2016 fuel economy datafile,” Office of Transportation & Air Quality, U.S.
Environmental Protection Agency, U.S. Department of Energy, 2015, accessed
2015-10-21. [Online]. Available: http://www.fueleconomy.gov/feg/download.
shtml
[271] E. N. Zongguang Wang, Tetsuji Ukita and Y. Masuda, “Prediction of air cooling
system for ev/hev battery pack,” in SAE Technical Paper. SAE International, 05
2011.
184
References
[282] “2014 mercedes b-class ed battery much bigger than previously stated,” AutoBlog,
2014, accessed 2015-04-08. [Online]. Available: http://www.autoblog.com/2014/
06/23/2015-mercedes-b-class-ed-battery-much-bigger-than-we-knew/
[283] “Byd e6a auto owner’s manual,” BYD AUTO, 2015, accessed 2015-
04-10. [Online]. Available: http://www.byd-auto.net/medias/Owner’sManual/
BYDE6Owner’smanual2011.6.22-usethis_en-update.pdf
[284] “Charging solutions,” BYD Europe, 2015, accessed 2015-07-07. [Online].
Available: http://www.bydeurope.com/innovations/technology/index.php
185
References
[297] “Tesla model s,” Tesla Motors USA, 2015, accessed 2015-07-14. [Online].
Available: http://www.teslamotors.com/models
186
References
[298] “Tesla model x,” Tesla Motors Sweden, 2015, accessed 2015-10-20. [Online].
Available: http://www.teslamotors.com/modelx
187
References
188
Appendix A
189
Appendix A. Sales and Specification BEV data
9 000
8 000 Japan 2015 Japan 2014
120 000 7 000 Europe 2015 Europe 2014
Japan 2015 Japan 2014 6 000
5 000 USA 2015 USA 2014
100 000 Europe 2015 Europe 2014 4 000
80 000 USA 2015 USA 2014 3 000
2 000
60 000 1 000
0
40 000
20 000
0
Nissan LEAF Tesla Model S BMW i3 Renault Zoe Volkswagen
e‐Golf
Figure A.1 Top 22 of highway capable BEV sales in USA, Europe and Japan during 2014 and
2015, [132–135].
gray color in the tables, which indicates a somewhat lower level of confidence. SI units
are used throughout the tables, hence unit conversions have been made for some of the
data.
Through a deal between Mitsubishi Motors (Japan) and PSA Peugeot Citroën (France)
Mitsubishi is manufacturing both Peugeot iOn and Citroen C-Zero similarly as Mit-
subishi i-MiEV. Therefore, the three BEVs can be expected to have similar specifica-
tions. It was however found that advertised specifications differ between countries on the
manufactures’ web sites even for the same model. The deviations regard: the curb weight
(1065-1185 kg), the motor maximum torque (180 Nm or 196 Nm), the speed at which
maximum torque and power is reached (4 and 5 different data respectively), and the bat-
tery energy capacity (14.5kWh or 16kWh). In contrast, data of the top speed and time
to accelerate 0-100km/h does not differ. The here presented data is assumed to be fairly
representative for the latest models.
190
Table A.1 Brand, model, drive, tire, warranty and battery type data.
Standard Powertrain Battery
Conv. Tire size Warranty Warranty
Model from on driving Distance/ Distance/ Battery
Brand Model year CEV? Seats Drive wheel pair Years Years Type
Smart fortwo 2014 [136] Yes 2 [136] Rear [136] 175/55R15 [136] 50kmi/4yrs [137] 50kmi/4yrs [137] Li-ion [136]
el.dr. Coupe 2015 [138] Pol. [139]
Smart fortwo BRABUS 2014 [140] Yes 2 [140] Rear [140] 225/35R17 [140] Li-ion [140]
El.Dr. Coupe
Toyta Scion iQ EV 2012 [141] Yes 4 [142] Front [141] 175/60R16 [143] Li-ion [141]
Fiat 500e 2015 [144] Yes 4 [144] Front [144] 185/55R15 [144] 50kmi/4yrs [145] 100kmi/8yrs [145] Li-ion [144] Pri. [139]
Citroën C-ZERO 2014 [146] Yes 4 [146] Rear [146] 175/55R15 [146] 100kkm/8yrs [147] Li-ion [146]
Peugeot iOn 2014 [148] Yes 4 [148] Rear [148] 175/55R15 [148] 50kkm/5yrs [148] 50kkm/5yrs [148] Li-ion [148]
Mitsubishi i-MiEV 2014 [149] Yes 4 [149] Rear [150] 175/60R15 [151] 100kkm/5yrs [151] 160kkm/8yrs [151] Li-ion [151] Pri. [139]
Volkswagen e-up! 2013 [152] Yes 4 [153] Front [152] 165/65R15 [152] 160kkm/8yrs [154] Li-ion [152]
Chevrolet Spark EV 2015 [155] Yes 4 [155] Front [155] 185/55R15 [155] 100kmi/5yrs [155] 100kmi/8yrs [155] Li-ion [155]
Bollore Bluecar 2015 [156] 4 [156] Li-ion Pol. [156]
Mitsubishi MiniCab MiEV 2014 [157] Yes 4 [157] Rear [158] 145R12 [157] Li-ion [158]
BMW i3 2014 [159] No 4 [159] Rear [159] 155/70R19 [159] 50kmi/4yrs [160] 100kmi/8yrs [160] Li-ion [159] Pri. [139]
62kmi/8yrs [161]
Renault Zoe 2015 [162] No 5 [162] Front [163] 185/65R15 [162] 100kmi/5yrs [162] 60kmi/5yrs [162] Li-ion [162] Pol. [139]
Volvo C30 Electric 2013 [164] Yes 4 [164] 205/55R16 [165] Li-ion [164]
Volkswagen e-Golf 2015 [166] Yes 5 [166] Front [166] 205/55R16 [166] 60kmi/5yrs [167] 100kmi/8yrs [167] Li-ion Pri. [167]
160kkm/8yrs [168]
Nissan Leaf (Visia) 2015 [169] No 5 [169] Front [169] 205/55R16 [169] 100kkm/5yrs [170] 100kkm/5yrs [170] Li-ion Lam. [171]
Nissan Leaf (Acenta,SV) 2016 [172] No 5 [172] Front [172] 205/55R16 [172] 160kkm/8yrs [172] Li-ion Lam. [173]
Honda FIT EV 2014 [174] Yes 5 [174] Front [174] 185/65R15 [174] 60kmi/5yrs [175] 60kmi/5yrs [175] Li-ion [174] Pri. [139]
Renault Fluence Z.E. 2012 [176] Yes 5 [176] Front [177] 205/55R16 [176] 100kkm/5yrs [176] Li-ion [176]
Ford Focus EV 2015 [178] Yes 5 [178] Front [178] 225/50R17 [179] 100kmi/8yrs [180] 100kmi/8yrs [180] Li-ion [178]
Kia Soul Electric 2015 [181] Yes 5 [182] Front [181] 205/60R16 [181] 100kmi/10yr [183] 100kmi/10yrs [183] Li-ion Pol. [181]
150kkm/7yrs [181]
Mercedes-Benz B-Class El. Dr. 2015 [184] Yes 5 [184] Front [184] 205/60R16 [184] 100kkm/8yrs [185] Li-ion [184] Cyl. [186]
BYD e6 2015 [187] No 5 [187] Front [150] 235/65R17 [187] 10yrs [188] LiFePO4 [189]
Nissan e-NV200 (Evalia) 2015 [190] Yes 5 [190] Front [190] 185/65R15 [190] 100kkm/5yrs [190] 100kkm/5yrs [190] Li-ion [191]
Toyota RAV4 EV 2014 [192] Yes 5 [192] Front [192] 225/65R17 [192] 60kmi/5yrs [192] 100kmi/8yrs [192] Li-ion [192] Cyl. [193]
Tesla Model S 2015 [194] No 5 [194] Rear/AWD [194]245/45R19 [194] unlim./8yrs [194] unlim./8yrs [194] Li-ion [194] Cyl. [139]
Tesla Model X 2015 [195] No 6/7 [195] AWD [195] unlim./8yrs [195] unlim./8yrs [195]
Tesla Roadster 2012 [196] No 2 [196] Rear [197] 225/45R17 [196] 100kmi/7yrs [196] Li-ion [198] Cyl. [139]
Rimac Concept One 2015 [199] No 2 [199] AWD [199] F:245/35R20 [200] LiFePO4 [199]
R:295/30R20
Mercedes-Benz SLS AMG El.Dr. 2015 [201] Yes 2 [201] AWD [201] F:265/35R19 [202] Li-ion [203]
R:295/30R20
Lightning Lightning GT 2015 [204] No 2 [204] Rear [204] 245/45R20 [205] Li-titan [204]
Detroit Electric SP:01 Pure 2015 [206] No 2 [206] Rear [206] 225/45R17 [206]
Detroit Electric SP:01 Perf. 2015 [206] No 2 [206] Rear [206] 225/45R17 [206]
191
A.2. Specification BEV data
C30 El. 1725 [164] 1546 [225] 1782 [226] 1447 [226] 4266 [226] 2.18 [225] 0.28 [225] 0.61 cal.
e-Golf 1510 [166] 422 [166] 1960 [166] 1528 [166] 1799 [166] 1450 [166] 4270 [166] 2.19 cal. [227] 0.27 [166] 0.59 cal. [227]
Leaf’15 1474 [169] 471 cal. 1945 [169] 1538 [169] 1770 [169] 1545 [169] 4445 [169] 2.27 [228] 0.291 [169] 0.66 cal.
Leaf’16 1516 [172] 454 cal. 1970 [172] 58/42 [173] 1537 [173] 1770 [172] 1550 [172] 4445 [172] 0.29 [172]
FIT EV 1475 [174] 55/45 [174] 1474 [174] 1721 [174] 1579 [174] 4114 [174] 0.34 [229]
Fluence Z.E. 1535 [230] 388 cal. 2023 [176] 46/54 [176] 1546 [176] 1809 [231] 1458 [176] 4748 [176] 0.67 [231]
Focus EV 1642 [178] 365 [232] 2007 cal. 1525 [178] 1823 [178] 1478 [178] 4391 [178] 0.295 [233]
Soul Electric 1490 [234] 470 cal. 1960 [181] 1582 [235] 1800 [181] 1593 [181] 4140 [181] 0.35 [236]
B-Class El. Dr. 1650 [184] 445 [184] 2170 [184] 1548 [184] 1812 [184] 1599 [184] 4358 [184] 0.28 [184]
e6 2380 [187] 375 [187] 2755 cal. 1573 [187] 1822 [187] 1645 [187] 4560 [187]
e-NV200 1571 [191] 483 [190] 2220 [190] 48/52 [190] 1530 [190] 1755 [190] 1858 [190] 4560 [190]
RAV4 EV 1828 [192] 1559 [192] 1816 [192] 1684 [192] 4574 [192] 0.30 [192]
ModelS 2108 [194] 482 cal. 2590 [237] 48/52 [194] 1680 [194] 1963 [194] 1445 [238] 4978 [194] 2.34 [239] 0.24 [240] 0.56 cal.
ModelX 0.24 [195]
Roadster 1235 [196] 249 cal. 1484 [241] 35/65 [242] 1482 [196] 1851 [196] 1126 [196] 3939 [196] 0.36 [215]
Concept One 1950 [243] 42/58 [199] 1997 [243] 1198 [243] 4548 [243]
SLS AMG El.Dr.2110 [203] 250 [203] 2435 [203] 47/53 [244] 1667 [203] 1939 [203] 1262 [203] 4638 [203]
Lightning GT 1850 [245] 1668 [246] 1940 [245] 1200 [245] 4445 [245]
SP:01 Pure 1155 [206] 1751 [206] 1117 [206] 3880 [206]
SP:01 Perf. 1175 [206] 1751 [206] 1117 [206] 3880 [206]
Fetish 1200 [207]
America 1525 [209] 2060 [209] 1450 [209] 4125 [209]
192
Coupe 1474 [210]
R8 e-tron 1841 [213] 40/60 [213] 1617 [247] 1929 [247] 1252 [212] 4431 [212] 0.28 [211]
LAMPO 3 1700 [214] 1998 [214] 1307 [214] 4687 [214]
A.3. Comments on selected data
193
Table A.3 Electric Motor Data
EM EM EM EM Speed EM Speed EM
EM Max. Max. Cont. at max. at max. Max.
Motor Torque Power Power Torque power Speed EM Transm.
Model type (Nm) (kW) (kW) (rpm) (rpm) (rpm) Cooling Ratio
fortwo PM [136] 130 [136] 55 [136] 35 [136] Water [138] 9.922 [250]
BRABUS PM [140] 135 [140] 60 [140] 35 [140]
iQ EV PMSM [142] 163 [141] 47 [141] Air [141]
500e PMSM [251] 200 [144] 83 [144] 12’800 [252] Water [252]] 9.59 [144]
C-ZERO PMSM [146] 196 [146] 49 [146] 300 [146] 4’000-8’800 [146]
iOn PMSM [148] 196 [148] 49 [148] 35 [148] 300 [148] 4’000-8’800 [148] 9’500 [148]
i-MiEV PMSM [151] 196 [151] 49 [151] 300 [151] 4’000-8’800 [151] 9’900 [151] Water [151] 7.065 [151]
e-Up! PMSM [152] 210 [152] 60 [152] 2’800 [152] 2’800 [152] 12’000 [152] 8.162 [153]
Spark EV PMSM [155] 444 [155] 105 [155] Liquid [155] 3.87 [155]
Bluecar 50 [156] 35 [156]
MiniCab PMSM [158] 196 [157] 30 [157] 25 [157] 3’000 [158] 2’500-6’000 [158] 7.065 [157]
i3 HPMSM [161] 250 [159] 125 [159] 75 [159] 5’000 [120] 11’400 [161] Liquid [120] 9.7 [159]
Zoe SM, w.r. [162] 220 [162] 65 [162] 2’500 [162] 3’000-11’300 [162] 11’300 [162] 9.32 [253]
C30 El. 250 [164] 89 [164]
e-Golf PMSM [166] 270 [166] 85 [166] 50 [254] 3’000 [255] 3’000 [255] 12’000 [166] 9.747 [166]
Leaf’15 IPMSM [256] 254 [169] 80 [169] 3’008 [169] 3’008-10’000 [169] 10’500 [169] Water [256] 8.1938 [169]
Leaf’16 SM [172] 254 [172] 80 [172] 3’008 [172] 3’008-10’000 [172] 10’500 [172] 8.1938 [172]
Appendix A. Sales and Specification BEV data
FIT EV PMSM [174] 256 [174] 92 [174] 3’056 [174] 3’695-10’320 [174] 10’320 [174] 8.058 [174]
Fluence. SM, w.r. [176] 226 [176] 70 [176] 35 [257] 2’500 [231] 3’000-8’900 [231] 11’000 [258] 9.34 [176]
Focus EV PMSM [259] 250 [178] 107 [178] 5’500 [220] Liquid [179] 7.82 [220]
Soul El. IPMSM [182] 285 [181] 81.4 [235] 2’730 [181] 2’730-8’000 [181] Liquid [235] 8.206 [181]
B-Class El.Dr. IM [184] 340 [184] 132 [184] 65 [184] 9’900-12’500 [184] Liquid [185] 9.73 [186]
e6 PMSM [187] 450 [187] 90 [187] 7’000 [260] Liquid [261] 6.844 [260]
e-NV200 IPMSM [191] 254 [190] 80 [190] 10’500 [191] 7.938 [190]
RAV4 EV IM [192] 370 [192] 115 [192] Liquid [118] 9.73 [192]
ModelS 60 IM Cu. [262] 430 [263] 283 [264] 5’000 [263] 5’000-8’000 [265] 16’000 [237] Liquid [237] 9.73 [262]
ModelS 70 2 IM Cu. [194] 441 [194] 285 [194] Liquid [194]
ModelS 70D 2 IM Cu. [194] 525 [194] F,R:193 [194] Liquid [194]
ModelS 85 2 IM Cu. [194] 441 [194] 285 [194] 5’800 [263] 6’000-9’500 [265] 16’000 [237] Liquid [194] 9.73 [262]
ModelS 85D 2 IM Cu. [194] 658 [194] F,R:193 [194] 5’800 [263] 16’000 [237] Liquid [194] 9.73 [262]
ModelS P85D 3 IM Cu. [194] 967 [194] F:193 [194] 5’100 [263] 5’000-8’600 [265] 16’000 [237] Liquid [194] 9.73 [262]
R:375
ModelX 90D F,R: 193 [195]
ModelX P90D F:193 [195]
R:375 [195]
Roadster IM [196] 370 [196] 225 [196] 5’400 [196] 5’000-6’000 [196] 14’000 [196] Air [196] 8.28 [196]
Concept One 4 PMSM [199] 1600 [199] 800 [199] 6’500 [199] 12’000 [199] Oil [199] 3.5 [243]
SLS AMG El.Dr. 4 PMSM [201] 1000 [201] 552 [201] 13’000 [201] Liquid [203]
Lightning GT 2 EM [204] 4000 [204] 300 [204] 5.5 [245]
194
SP:01 Pure 280 [206] 150 [206]
SP:01 Perf. 280 [206] 210 [206]
Fetish PMSM [207] 380 [207] 220 [207]
America 2 PMSM [208] 480 [209] 300 [209] 6’000 [208] 10’500 [208]
Coupe 2 Ax.Flux [210] 1355 [210] 368 [210]
R8 e-tron 2 EM [213] 920 [211] 340 [211]
LAMPO 3 3 HSM [214] 900 [214] 420 [214] 4’500 [214] Water [214] 6.4 [214]
Table A.4 BEV Battery and Charging Data
Battery Standard DC Fast
Battery Battery (Cont.) Battery (Optional) Charging
Energy Charge Battery Max. Battery Specific AC Charge Method/
Capacity Capacity voltage Power Number weight Energy Battery Power Peak Power
Model (kWh) (Ah) (V) (kW) of cells (kg) (Wh/kg) Cooling (kW) (kW)
fortwo el.dr. 17.6 [136] 52 [266] 339 [266] 93 [136] 178 cal. 98,9 [266] Liquid [267] 3.3 [138]
(22) [268]
fortwo BRABUS 17.6 [140] 93 [140] 3.3 (22) [140]
iQ EV 12 [141] 43 [266] 277.5 [141] 150 [141] 220 [142] 54.7 [266] Air [141] 3 [141] CHA [269]
500e 24 [144] 63 [259] 364 [144] 97 [144] 291 [251] 88 [259] Liquid [144] 6.6 [144]
C-ZERO 14.5 [146] 300 [146] 80 [146] 3.2 [146] 50 [146]
iOn 14.5 [148] 80 [148] 221 [148] 66 cal. 3.2 [148] 50/CHA [148]
i-MiEV 16 [151] 50 [150] 330 [151] 145 cal. 110 [270] Air [271] 3.7 [149] CHA [149]
e-Up! 18.7 [152] 374n [152] 75 [272] 204 [152] 230 [152] 81 cal. Passive [272] 3.6 [152] 40/CCS [152]
Spark EV 18.4 [155] 54 [155] 400t [155] 120 [155] 192 [155] 215 [155] 85 [270] Liquid [155] 3.3 [273] CCS [155]
Bluecar 30 [156] 410n [156] 45 [156] 300 [156] 100 cal.
MiniCab 16 [157] 330 [157] 3 [158] CHA [274]
i3 22.0 [159] 60 [259] 360n [120] (40) [120] 96 [275] 233 [276] 93 [259] Refrig. [120] 7.4 [277] 50/CCS [277]
147 [276]
Zoe 22 [162] 400n [162] 192 [162] 290 [162] 76 cal. 43 [278]
C30 El. 24 [164] 330 [164] 73 cal. 22 [279]
e-Golf 24.2 [166] 75 [270] 323n [166] 264 [166] 318 [166] 76.1 [270] Passive [167] 7.2 [218] 50/CCS [218]
Leaf’15 24 [169] 66 [259] 360 [169] 90 [280] 192 [169] 273 cal. 87.9 [259] Passive [171] 3.6 (6.6) [169] 50/CHA [169]
Leaf’16 30 [172] 83 [270] 360 [270] 192 [172] 294 cal. 102 [259] 3.6 (6.6) [172] 50/CHA [172]
FIT EV 20 [174] 20 [150] 330 [150] 100 [174] 224 cal. 89.3 [150] 6.6 [174] CHA [248]
Fluence 22 [176] 65 [176] 398 [176] 192 [231] 280 [231] 79 cal. 3.5 (43) [177]
Focus EV 23 [178] 75 [259] 350 [259] 287 cal. 80.2 [259] Liquid [178] 6.6 [178]
Soul El. 27 [181] 75 [259] 360 [181] 90 [235] 192 [182] 274.5 [182] 98.4 [281] Air [183] 6.6 [181] 50/CHA [181]
B-Class El.Dr. 36 [282] 120 [259] 367 [259] 222 cal. 125.9 [259] Liquid [185] 10 [186]
e6 61.4 [187] 200 [150] 307 [150] (162) [283] 624 cal. 98.4 [150] Air [283] 40 [284] 100 [188]
270 [283]
e-NV200 24 [190] 360 [191] 192 [191] 3.6 (6.6) [190] CHA [190]
RAV4 EV 41.8 [192] 130 [150] 386m [192] 129 [192] 4500 [285] 383.5 [192] 110.3 [150] Liquid [286] 10 [192]
ModelS 60 60 [264] 245 [259] 400 [259] 353 cal. 170 [259]
ModelS 85 85 [264] 245 [264] 400 [259] 500 cal. 170 [259] Liquid [237] 10 (20) [262] 120/Spr [287]
ModelX 90 [195] 250 [270] 350 [270] 600 cal. 150 [270]
Roadster 56 [198] 366n [197] 215 [198] 6 831 [196] 450 [198] 121 [198] Liquid [198] 16.8 [241]
Concept One 82 [199] 650n [199] (1000) [199] 1 400 [243] Liquid [199] 22 [199] 50/CCS [199]
SLS AMG El.Dr. 60 [201] 400 [201] 600 [202] 864 [203] 548 [202] 109 cal. Liquid [203] 22 [202]
Lightning GT 44 [204] 450 [204] 9 [204] 50 [204]
SP:01 37 [206] 7.7 [288]
Fetish 54 [207]
195
A.3. Comments on selected data
Leaf’16 144 [172] 11.5 [172] 150 [169] 170 [270] 207 [270] 186 [270] 250 [172] 172 [270]
FIT EV 148 [292] 8.5 [220] 3.1 (48) [220] 159 [150] 200 [150] 177 [150] 132 [150]
Fluence Z.E. 135 [176] 13.7 [176] 140 [293] 185 [176]
Focus EV 137 [294] 11.4 [232] 154 [232] 190 [259] 211 [259] 200 [259] 162 [294] 122 [259]
Soul Electric 145 [181] 11.2 [181] 10.8 [182] 147 [181] 175 [181] 228 [259] 199 [259] 212 [181] 150 [259]
B-Class El. Dr. 160 [184] 7.9 [184] 3.9 (60) [184] 166 [184] 245 [259] 256 [259] 250 [259] 200 [184] 140 [259]
e6 140 [187] 9.2 [295] 7.7 (60) [188] 341 [150] 322 [150] 333 [150] 204 [150]
e-NV200 / Evalia 122 [191] 14 [190] 165 [191] 170 [191]
RAV4 EV 161 [192] 7.0 [285] 267 [150] 285 [150] 275 [150] 166 [150]
ModelS 60 190 [264] 6.2 [264] 5.9 [262] 188 [263] 222 [259] 217 [259] 219 [259] 390 [264] 335 [259]
ModelS 70 225 [194] 5.8 [296] 5.5 [194] 442 [296] 386 [194]
ModelS 70D 225 [194] 5.4 [296] 5.2 [297] 420 [296] 370 [194]
ModelS 85 225 [194] 5.6 [296] 5.4 [194] 170 [263] 238 [259] 232 [259] 235 [259] 502 [296] 426 [259]
ModelS 85D 249 [194] 4.4 [296] 4.2 [194] 170 [263] 220 [259] 198 [259] 210 [259] 528 [296] 435 [259]
ModelS P85D 249 [194] 3.3 [296] 3.1 [194] 188 [263] 234 [259] 215 [259] 225 [259] 491 [296] 407 [259]
ModelX 90D 249 [195] 5.0 [298] 4.8 [195] 232 [270] 222 [270] 227 [270] 414 [270]
ModelX P90D 249 [195] 4.0 [298] 3.8 [195] 236 [270] 233 [270] 235 [270] 402 [270]
Roadster 201 [196] 3.9 [196] 160 [299] 340 [299] 393 [300]
Concept One 305 [200] 2.8 [199] 600 [243]
SLS AMG El.Dr. 250 [201] 3.9 [201] 268 [301] 250 [201]
Lightning GT 209 [204] 5 [204] 241 [204]
SP:01 Pure 170 [206] 5.6 [206] 5.3 [206] 288 [206]
196
SP:01 Perf. 250 [206] 3.9 [206] 3.7 [206] 288 [206]
Fetish 200 [207] 4 [207]
America 220 [209] 4.5 [209] 14 (200) [209]
Coupe 193 [210] 3.4 [210]
R8 e-tron 250 [213] 3.9 [211] 450 [211]
LAMPO 3 220 [214] 4.5 [214] 200 [214]
A.3. Comments on selected data
Table A.7 Front area vs. height times width or track width.
Model iQ iOn e-Up! i3 Zoe C30 e-Golf Leaf ModelS
A
wh 85% 90% 85% 85% 96% 85% 84% 83% 82%
A
tw h 97% 103% 98% 96% 110% 97% 99% 96% 96%
3
2.5
2
1.5
1000 1200 1400 1600 1800 2000 2200
Curb weight (kg)
0.35
Cd
0.3
0.25
1
Cd A (m2)
0.8
0.6
1000 1200 1400 1600 1800 2000 2200
Curb weight (kg)
Figure A.2 Data on area, Cd and Cd A, as well as estimates of area and Cd A, using the found
Cd values and area estimates of 80-100% of height and width, all as functions of curb
weight.
classes are labeled A-G. Class A represent tires with the lowest coefficients ≤0.0065,
class B 0.0066-0.0077, C 0.0078-0.009, as so on up to G ≥0.0121 [302].
Still, neither original tire brand and model, nor it’s rolling resistance coefficient
197
Appendix A. Sales and Specification BEV data
0.34
0.32
0.3
198
A.3. Comments on selected data
2.2
1.8
Pmax./Pcont.
1.6
1.4
1.2 Small
Medium−Large
1
0 50 100 150
EM Pmax. (kW)
199
Appendix A. Sales and Specification BEV data
200
Appendix B
GPS-accelerometer
Measurement System
201
Appendix B. GPS-accelerometer Measurement System
(a) One of the measurement units. (b) The measurement unit strapped in a car.
3 Meas.
LP Meas.
2 GPS acc.
LP Grade−corr. Meas.
Acceleration (m/s2)
−1
−2
202
B.3. Ambiguity of measurements
As can be seen in Figure B.3, the exact same route was driven.
V744 CTH−Klltrp
57.72 1
2
57.715 3
4
57.71 5
Latitude (DD)
57.705
57.7
57.695
57.69
57.685
As can be seen in the upper right subplot in Figure B.4, the speed varies from log to
log. The difference may be about 20 km/h.
As can be seen in the lower right subplot in Figure B.4, the estimated road grade may
also vary from log to log. But there is a fair level of consistency.
203
Appendix B. GPS-accelerometer Measurement System
Speed (km/h)
Speed (km/h)
4 4
40 5 40 5
20 20
0 0
0 100 200 300 400 500 600 700 800 900 0 1 2 3 4 5 6 7 8 9
timeacc (s) Distance (km)
Grade (rad)
4 4
5 5
0 0
−0.05 −0.05
−0.1 −0.1
0 100 200 300 400 500 600 700 800 900 0 1 2 3 4 5 6 7 8 9
timeacc (s) Distance (km)
V744 CTH−Klltrp
4
1
2
Acceleration (m/s2)
2 3
4
0 5
−2
−4
0 100 200 300 400 500 600 700 800 900
Timeacc (s)
V744 CTH−Klltrp
4
1
2
Acceleration (m/s2)
2 3
4
0 5
−2
−4
0 1 2 3 4 5 6 7 8 9
Distance (km)
204
B.4. Logged vehicles
205
Appendix B. GPS-accelerometer Measurement System
206
Appendix C
In order to get an idea of the total and relative time spent at different speed and accel-
eration operating points for the Urban, Rural and Highway cycles, each cycle value is
attributed to a certain acceleration and speed bin, of 0.5 m/s2 and 5 km/h, see Figure
C.1(a) for the Urban Test cycles. The result can also be visualized in a contour plot, see
Figure C.2(a) for total time in each bin and Figure C.2(b) for the share of total time in
each bin.
In order to get a more smooth looking contour plot, a random gaussian noise (with
zero mean, and 0.5 m/s2 respectively 1 m/s standard deviation) was added to each op-
erating point, so called dithering. Furthermore, each drive cycle with the added noise is
looped 100 times, after which the total and relative time spent in each bin is divided by
the number of loops. The results can be seen in Figure C.1(b) and C.3(b), for the Urban
cycles, and in Figure C.4 and C.5 for the Rural and Highway cycles respectively.
4
x 10
2000 5
Total time in bin (s)
Total time in bin (s)
4
1500
3
1000
2
500 1
0 0
4
2 2
80
0 80
0 60 60
40 −2 40
−2 20 20
0 −4 0
Acceleration (m/s2) Speed (km/h) Acceleration (m/s2) Speed (km/h)
(a) No added noise. (b) With added noise during 100 loops.
Figure C.1 Share of total time (%) of operation in acceleration bins of 0.5 m/s2 and speed bins of
5 km/h, for all Urban test cycles.
207
Appendix C. Speed and Acceleration Dither
4 300 3 20
1
3 00. .2 18
250 2 0.1
5
10 3 16
0.2
2 1
150 0.5 14
Acceleration (m/s2)
3
Acceleration (m/s2)
30 200 1 0.1
0.5
1 60 60 30 5101 1 0.2 12
1
180 12 30 0.5
2
2
120
0.5
.5
0 3
601280
60 240 240 0
1
10
60
890
0 60
1
3
2
150
1 2345671
30
20
60 10 1 0.2 1
0.
0.
5 8
0.5
−1
30 −1
0.1
5
31 100
0.
30 0 6
−2 510 51 0.5
3 0.2
0.
0
1
4
1
3 −2 .1
5
0.
2
1 50
−3
2
−4 −3
0 20 40 60 80 100 0 20 40 60 80 100
Speed (km/h) Speed (km/h)
Figure C.2 Total time (s) vs. Share of total time (%) of operation in acceleration bins of 0.5 m/s2
and speed bins of 5 km/h, for all Urban test cycles, no added noise.
3 20 3 4
1
1
00. .2 18 0.
0 3.5
2 2 0.2 .1
2
0.1
0.
16
0.2
0.5 14
1 0.5 3
Acceleration (m/s )
0.1
0.5
Acceleration (m/s2)
0.5
2
1 0.1 1 1
0.5
0.5
23
2
2
0.2
0.5
0.5
3
1
0
1
10 0
890
1
3
2
1 2345671
2
1 0.2
1
2
1
0.
0.
0.
8
0.5
1.5
−1
0.1
−1 0.5 0.1
5
0.5
0.
6
0.5
0.2
1
0.
0
1
4
−2 .1
0.
0.2 0.1
0.
2
−2
2
2 0.1 0.5
−3 −3
0 20 40 60 80 100 0 20 40 60 80 100
Speed (km/h) Speed (km/h)
(a) No added noise. (b) With added noise during 100 loops.
Figure C.3 Share of total time (%) of operation in acceleration bins of 0.5 m/s2 and speed bins of
5 km/h, for all Urban test cycles.
208
Appendix C. Speed and Acceleration Dither
600 15000
Total time in bin (s)
200 5000
0 0
2 4
2
0 100 0 100
−2 50 −2 50
−4
−4 0 0
Acceleration (m/s2) Speed (km/h) Acceleration (m/s2) Speed (km/h)
(a) No added noise. (b) With added noise during 100 loops.
Figure C.4 Share of total time (%) of operation in acceleration bins of 0.5 m/s2 and speed bins of
5 km/h, for all Rural test cycles.
3 6 3 2
0.1
0.1 1.8
2 5 2
0.1
0.1 1.6
0.1
0.
0.1
0.2
1
0.2
Acceleration (m/s2)
0.1
0.1
Acceleration (m/s2)
1 0. 2 4 0.2 1.4
0.1
1
0.5 0.5
0.1
0.5
0.2
1 0.21 1.2
0.2
1
46 3 0. 1
0.5
54
2
3
4
2
0
5
3 0
0.5
0.5
0.1
2
1
0.21
2
3
0.5
1
2
2
1 0.1 0.1
1
0.2
0.1
0.2
0.10.2
2 1
0.5 0. 0.5 0.8
0.1 0.5
−1
0.
0.2 2 −1
0.1
2
0.
2
0.1 0.1
0.1
0.2 0.6
0.1
0.1
0.
−2
0.1
1 −2 0.4
0.1
0.1
1
0.1
0.2
0.1
−3 −3
0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160
Speed (km/h) Speed (km/h)
(a) No added noise. (b) With added noise during 100 loops.
Figure C.5 Share of total time (%) of operation in acceleration bins of 0.5 m/s2 and speed bins of
5 km/h, for all Highway test cycles.
209
Appendix C. Speed and Acceleration Dither
210
Appendix D
Temperature Dependence of
some Material Parameters
Water Water
1050 4000
1000 3500
950 3000
0 50 100 0 50 100
Thermal Conductivity (W/m K)
o o
Temperature ( C) Temperature ( C)
Dynamic Viscosity (kg/m s)
Glycol/Water Mix
0.03 0.7 Water
Glycol/Water Mix
Water
0.6
0.02
0.5
0.01
0.4
0 0.3
0 50 100 0 50 100
o o
Temperature ( C) Temperature ( C)
Figure D.1 Glycol/water mixed coolant material properties as a function of temperature [119]. The
properties of water are also included as a reference [48]
211
Appendix D. Temperature Dependence of some Material Parameters
cp,air (J/kg K)
1020 1.01
1 1
(-)
1010 1
0.8 0.8
6 air (W/m K)
7 air (kg/m s)
Figure D.2 Air material properties as a function of temperature, both absolute values and normal-
ized to the value at 20◦ C. [48]
The temperature dependence of thermal conductivity and specific heat, for the used
lamination core material is presented in Figure D.3.
36 1.8
norm 6 lam by 22 oC (-)
cp,lam (kg/m s)
o
32 norm 1.6 700 norm 1.6
30 1.5
28 1.4 600 1.4
26 1.3
24 1.2 500 1.2
22 1.1
20 1 400 1
0 50 100 150 200 0 50 100 150 200
o o
Temp. ( C) Temp. ( C)
Figure D.3 Thermal conductivity and specific heat of lamination material (NO30) as a function of
temperature [105], both absolute values and normalized to the value at 22◦ C.
The temperature dependence of thermal conductivity and specific heat, for copper
(the used winding material) is presented in Figure D.4.
212
Appendix D. Temperature Dependence of some Material Parameters
norm 6 Cu by 300K
c p,Cu (J/kg K )
(W/m K)
norm norm
400 0.98
400 1.05
380 0.96
Cu
6
Figure D.4 Thermal conductivity and specific heat of winding copper as a function of temperature
[48], both absolute values and normalized to the value at 300 K.
Finally, the temperature dependence of thermal conductivity for the used shaft mate-
rial (carbon-silicon steel) is presented in Figure D.5.
52 1
50 norm 0.95
48 0.9
46 0.85
44 0.8
0 100 200 300 400
Temp. ( o C)
Figure D.5 Shaft material thermal conductivity as a function of temperature, both absolute values
and normalized to the value at 300 K. [48]
213
Appendix D. Temperature Dependence of some Material Parameters
214
Appendix E
The Prandtl number of the coolant is much larger than one in the whole temperature
range, as is shown in Figure E.1. This means that convection heat transfer is likely to
dominate over conductive heat transfer in the coolant medium.
800 60
Glycol/Water Mix Glycol/Water Mix
Water 50 Water
600
Prandtl (-)
Prandtl (-)
40
400 30
20
200
10
0
-40 0 40 80 10 20 30 40 50 60 70
o
Temperature ( C) Temperature (oC)
As mentioned in section 2.8.1, the heat transfer coefficient is often found via empir-
ical approximations of the Nusselt number (Nu). There exist a number of different Nu
approximations for circular tubes in literature, for the three flow types: laminar, transi-
tional and turbulent, which is determined by the Reynolds number (Re). The flow can be
assumed to be laminar for Re ≤ 2300, and fully turbulent for Re ≥ 10, 000 [45, 48, 50].
According to [45], these critical values can also be assumed for non-circular tubes. In
practice, the critical Re values additionally depend on factors such as surface roughness
and vibrations [48], which makes them difficult to predict.
For these non-circular tubes, the hydraulic diameter Dh is used when calculating Nu
and Re [48, 305]. It is four times the cross sectional area divided by the perimeter. For a
rectangular tube, with side a and b, Dh is
2ab
Dh = (E.1)
a+b
215
Appendix E. Cooling Channel Modelling
The Reynolds number ReD is here depending on the hydraulic diameter Dh , and
the average flow velocity vave (m/s), which in turn depends on the flow rate f lL/min
(liters/min), as
ρ Dh vave 2ρ f lL/min
ReD = = (E.2)
µ µ(a + b) 60 × 103
The Re number as a function of flow rate is presented in Figure E.2, for four different
coolant temperatures. The flow rate needed to reach the Re value for critically turbulent
flow is lower when the coolant is warmer, and higher when the coolant is colder.
15000
65o C
20o C
10o C
5oC
10000 6 L/min
Re D,cool (-)
12 L/min
Re Crit. Turb
Re Fully Turb.
5000
0
0 5 10 15 20
Flow rate (L/min)
Figure E.2 The Reynolds number of the coolant as a function of flow rate and temperature.
For fully developed laminar flow the Nusselt number can be approximated as a con-
stant that only depends on the cross sectional geometry [45, 48, 50, 305]. For rectangular
tubes with width-to-height ratio of 6 (30/5 = 6), and under the assumption of a constant
heat flux, Nu is 6.05 [48]. There are also expressions for how the Nu decays to that of the
fully developed laminar flow value as a function of tube length, which also include en-
trance effects [50, 306]. For simplicity, the entrance effect is here neglected, even though
it was noted that it could result in up to three times larger Nu values.
For turbulent flow with ReD 3000 − 5 × 106 and Prandtl (Pr) 0.5-2000, the Nusselt
number can be approximated to a reasonable accuracy [45, 48, 50, 306], as
(f /8)(ReD − 1000)P r
Nu = (E.3)
1 + 12.7(f /8)1/2 (P r2/3 − 1)
where the friction factor f , for smooth tubes can be approximated as
216
Appendix E. Cooling Channel Modelling
The resulting Nu number as a function of flow rate is shown in Figure E.3, for dif-
ferent coolant temperatures. Similarly to the ReD , the Nu number increases with both
temperature and flow rate for turbulent flows.
140
65o C
120 20o C
10o C
100 5oC
6 L/min
12 L/min
Nucool (-)
80
60
40
20
0
0 5 10 15 20
Flow rate (L/min)
Figure E.3 The Nusselt number of the cooling circuit as a function of flow rate and temperature.
The temperature increase of the coolant is proportional to the amount of losses Ploss
absorbed and inversely proportional to the flow rate [116], as in
Ploss
∆Tcool = f lL/min
(E.5)
ρcool cp,cool 60×103
The temperature increase of the coolant for different levels of total losses and flow
rates is presented in Figure E.4, for the initial coolant temperature of 65◦ C. Only minor
differences were noted for different initial coolant temperatures. For the highest losses
in the Original slot HighwayII motor: 6 kW, with a flow rate of 6 L/min, the coolant
temperature increase is at most 17◦ C, and for moderate losses of 3 kW the temperature
increase is 8◦ C at most. At 12 L/min the temperature increase is 8◦ C with 6 kW losses,
and 4◦ C with 3 kW.
35
Ploss: ½ kW
30 P : 1 kW
loss
Ploss: 3 kW
25 Ploss: 6 kW
"T cool (oC)
20
15
10
0
0 5 10 15 20
Flow rate (L/min)
Figure E.4 The expected increase in coolant temperature for different heat absorptions.
217