Department of Physics, Chemistry and Biology

Master’s Thesis

Towards better alternator efficiency

Markus Örn
Master thesis at Scania

Linköping, June 30, 2014

LiTH-IFM-EX--2014/2959--SE

Department of Physics, Chemistry and Biology

Linköping University
SE-581 83 Linköping, Sweden
 Department of Physics, Chemistry and Biology

Towards better alternator efficiency

Markus Örn
Master thesis at Scania

Linköping, June 30, 2014

Supervisor: Weine Olovsson

ifm, Linköpings universitet
Holger Dreher
Scania

Examiner: Peter Münger

ifm, Linköpings universitet
 Avdelning, Institution Datum
Division, Department Date

Organisatorisk avdelning
Department of Physics, Chemistry and Biology 2014-06-30
SE-581 83 Linköping

Språk Rapporttyp ISBN

Language Report category —
 Svenska/Swedish  Licentiatavhandling ISRN
 Engelska/English
  Examensarbete
 LiTH-IFM-EX--2014/2959--SE
 C-uppsats
Serietitel och serienummer ISSN
 D-uppsats Title of series, numbering —
  Övrig rapport


URL för elektronisk version

http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-109098

Titel Optimering av generatorer

Title Towards better alternator efficiency

Författare Markus Örn

Author

Sammanfattning
Abstract

The requirements on vehicle industry are constantly getting stricter, especially when it comes
to emissions. At the same time cars, trucks and buses are needed for our way of living. This
have forced companies to be as efficient as possible in their way of using fossil fuels for
travelling and transport. To increase the efficiency companies investigate all possible fuel
savings to decrease their carbon footprint as much as possible. One area of savings that is
not that obvious to many people is the alternator. Several percent of the total energy used
by a vehicle are needed to operate the alternator. With a typical alternator efficiency of 70%
considerable savings can be achieved.

This thesis that concern alternator efficiency was carried out at Scania in Södertälje, Sweden.
The goal of the thesis is to construct a mathematical model of an alternator. The model is
supposed to consider all losses in the alternator and together with the output power give an
efficiency model of the alternator at different speeds and loads. A great part of the project
has been dealing with the magnetic losses. The magnetic losses have been modeled as an
equivalent circuit with the load angle as a central piece. The equivalent circuit is built up
by the fact that the alternator used in the vehicles is a salient pole alternator. The equivalent
circuit describes a voltage equation where the voltage drop over the magnetic inductance is
described. From that relations between the signals in the alternator and output signals can
be written.

The alternator model is then used together with data recorded from different buses all over
the world, this to be able to investigate how the alternator contributes to the fuel consump-
tion depending on the way that the buses are driven.

The result of this thesis is a mathematical model that describes the losses in the alternator
for different load cases and speeds.

Nyckelord
Keywords salient claw-pole alternator, load angle, synchronous machine, equivalent circuit, alternator
losses
Abstract
The requirements on vehicle industry are constantly getting stricter, especially
when it comes to emissions. At the same time cars, trucks and buses are needed
for our way of living. This have forced companies to be as efficient as possible
in their way of using fossil fuels for travelling and transport. To increase the
efficiency companies investigate all possible fuel savings to decrease their carbon
footprint as much as possible. One area of savings that is not that obvious to many
people is the alternator. Several percent of the total energy used by a vehicle
are needed to operate the alternator. With a typical alternator efficiency of 70%
considerable savings can be achieved.
This thesis that concern alternator efficiency was carried out at Scania in Södertälje,
Sweden. The goal of the thesis is to construct a mathematical model of an al-
ternator. The model is supposed to consider all losses in the alternator and to-
gether with the output power give an efficiency model of the alternator at dif-
ferent speeds and loads. A great part of the project has been dealing with the
magnetic losses. The magnetic losses have been modeled as an equivalent circuit
with the load angle as a central piece. The equivalent circuit is built up by the fact
that the alternator used in the vehicles is a salient pole alternator. The equivalent
circuit describes a voltage equation where the voltage drop over the magnetic in-
ductance is described. From that relations between the signals in the alternator
and output signals can be written.
The alternator model is then used together with data recorded from different
buses all over the world, this to be able to investigate how the alternator con-
tributes to the fuel consumption depending on the way that the buses are driven.
The result of this thesis is a mathematical model that describes the losses in the
alternator for different load cases and speeds.

iii
Acknowledgements
First of all I would like to give a great thank you to my supervisor at Scania,
Holger Dreher, who has helped me very much during this thesis. He has been a
great source of information and helped me investigate the alternator. I would also
like to give a thank you to other Scania employees who have helped me through
my diploma work.
I would also like to thank my family and friends who are always there for me and
have supported me through my education.
Finally I would also like to thank my supervisor at LiU Weine Olovsson for sup-
porting me throughout my report writing and helping me trough the challenging
task of writing a scientific report in an industrial environment, and my examiner
Peter Münger for beeing very open and helpful in my not too obvious choise of
master thesis.
Eskilstuna, June 2014
Markus Örn

v
 Contents

Notation ix

1 Introduction 1
1.1 Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The internal combustion engine . . . . . . . . . . . . . . . . . . . . 2
1.3 Alternators today . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Alternator theory 5
2.1 Rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Stator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Rectifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Voltage regulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.5 Equivalent circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.6 Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.6.1 Mechanical resistance . . . . . . . . . . . . . . . . . . . . . . 19
2.6.2 Magnetic resistance . . . . . . . . . . . . . . . . . . . . . . . 20
2.6.3 Electric resistance . . . . . . . . . . . . . . . . . . . . . . . . 22

3 Method 23
3.1 Model of alternator . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.1 Construction of an initial model . . . . . . . . . . . . . . . . 23
3.1.2 Improvement of the model . . . . . . . . . . . . . . . . . . . 25
3.2 Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2.1 Provided field measurements . . . . . . . . . . . . . . . . . 27
3.2.2 Rotor current . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2.3 Load angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.4 Inductances . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.5 Mechanical losses . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.6 Magnetic losses . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2.7 System measurement . . . . . . . . . . . . . . . . . . . . . . 36
3.3 Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4 Results 39

vii
viii Contents

4.1 Results from measurements . . . . . . . . . . . . . . . . . . . . . . 39

4.1.1 Provided measurements . . . . . . . . . . . . . . . . . . . . 39
4.1.2 Rotor current . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.1.3 Load angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.1.4 Inductances . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.1.5 Mechanical losses . . . . . . . . . . . . . . . . . . . . . . . . 45
4.1.6 Magnetic losses . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.1.7 System measurement . . . . . . . . . . . . . . . . . . . . . . 50
4.2 Model of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2.1 Model of the alternator . . . . . . . . . . . . . . . . . . . . . 52
4.2.2 Model of the engine . . . . . . . . . . . . . . . . . . . . . . . 53
4.3 Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.A Relations used in the alternator block . . . . . . . . . . . . . . . . . 60

5 Conclusion 63

A Details 67
A.1 Circuit diagram for test of mechanical and magnetic losses . . . . 67

Bibliography 69
 Notation

Abbreviations

Abbreviation Meaning
emf Electromotive force
ad-
converter Converts analog signals to digital
q
1 Pn
rms Root mean square, RMS(x) = n i=1 xi 2
dc current Direct current
ac current Alternating current

ix
 Introduction
1
1.1 Requirements
The requirements on vehicle fuel consumption are steadily getting stricter and
this means that all possible savings are investigated to lower the carbon footprint
of vehicles. The requirements comes from several directions where the limits on
emissions are set by governmental institutions and the customers requires a cost
effective product with good performance.
Today several percent of the total energy is used to power the alternator, due to
increased electric loads in the vehicles. These increased electrical loads are due
to improvements of safety, comfort and electric controls. In this master thesis the
efficiency of the alternator and in extension the possible improvements in fuel
consumption by using better alternators are investigated.
The goal in this work is to gain understanding of the alternator losses to be able to
recommend improvements on how to reach a better efficiency. The losses in the al-
ternator are divided between mechanical, magnetic and electric. A mathematical
model considering the alternator efficiency is then constructed from the gathered
theory, measurements of the losses are then used to calibrate the model. Appli-
cation data from several field measurements is provided by Scania and from this
an investigation on the alternator losses and efficiency with respect to the way of
driving will be carried out.

1
2 1 Introduction

1.2 The internal combustion engine

The internal combustion engine is the heart of most motor vehicles. The purpose
of it is to extract chemical energy from fuel through combustion, and then use
the released energy to move the vehicle.
There are a couple of different fuels that can be used; but since diesel engines
have a higher efficiency compared to petrol engines, Scania uses diesel engines
for their buses and trucks. Because of this the focus on this short explanation of
combustion engines will be on diesel engines.
The engine consists of an engine block with cylinders and pistons, the fuel is then
injected to these cylinders. From the start the piston is at its top position in the
cylinder, the piston then moves down, this makes it possible for the cylinder to
fill up with air through inlet valves.
The piston then moves back up with closed valves to compress the air. When the
air gets compressed it heats up due to the adiabatic process where no energy is
added or removed from the cycle, the increased pressure therefore increases the
temperature. Diesel is then injected in the cylinder and an instant combustion
will take place due to the heat of the air. The combustion releases the chemical
energy in form of a pressure and temperature increase. The increase of pressure
presses the piston down in the cylinder.
When the piston is at its lowest point in the cylinder an exhaust valve opens.
Since the piston is attached to the crankshaft the piston will get pushed back up
with the energy of the combustion from the other cylinders as the crankshaft
revolves. This upward movement of the piston will push the exhaust out of
the cylinders exhaust valves and then the exhaust valves will close and the in-
let valves open. This means that the system will be in the same state as in the
beginning, making a new cycle possible.
The rotation energy of the crankshaft is then mainly used to power the wheels
of the vehicle. The rest of the energy is used to power the alternator, oil pumps,
AC compressor and other auxiliaries. To make this cycle work properly electrical
power is also needed. Electricity is for example needed to control the fuel flow
into the cylinders and in the starter motor to crank the engine. With respect to
the whole vehicle, electricity is needed for much more, for example lights, air
conditioning, radio etc.
It would be impractical to power all the electric parts with only a battery since
the battery would have to be taken out and recharged every time it is drained.
This is where the alternator comes in, the alternator enables the battery to charge
while the vehicle is running. The alternator is driven with rotational energy cre-
ated by the engine and the kinetic energy of the moving vehicle. If the vehicle
for example is going downhill potential energy from an earlier drive uphill is re-
leased as kinetic energy making a part of the energy free since no fuel is needed
to generate it.
At the end of the crankshaft there is a pulley, on the pulley a belt is put, the belt
1.3 Alternators today 3

is also put over a pulley on the alternator, this means that when the crankshaft
spins so does the alternator shaft.
This is mainly how the combustion engine powers both the vehicle and the alter-
nator. This way of running the alternator means that the alternator uses a part
of the torque that the engine creates which means that the alternator uses a part
of the consumed fuel. This is why it is important to optimize the alternator to
generate as much energy as possible per unit of fuel. There are also losses in the
engine and in the belt, but these losses will not be investigated in this thesis.
[1], [2].

1.3 Alternators today

The alternator is a rather old invention that has been used for decades. Over
these years the main focus has been an alternator with high output that is cheap
to produce, leading to a rather poor efficiency for alternators.
Since the alternator contributes to the fuel consumption the increasing require-
ments on emissions have started an interest in developing the alternator and sev-
eral changes are investigated to increase the efficiency of the alternator and lower
the fuel consumed by it. A thorough description of how alternators work can be
found in chapter 2.
Much of the research done today concerns the magnetic losses in the alternator.
For example investigations of how well a magnetic equivalent circuit describes
the true system have been carried out [3]. The most focus on recent research is
on the material and the shape of the magnetic active parts used in the alternator.
One example is the investigation of how permanent magnets in the alternator
would impact the path of the magnetic field and further on the magnetic losses
[4]. The theory behind it is that the permanent magnets will force the magnetic
field to take certain paths between the rotor and the stator and in that way lower
the losses that a leakage path over a big air gap between two alternator claws
generates. Another way of lowering the magnetic losses is to change the material
in the alternator. If for example sintered powder is used in the core of the stator
or rotor, instead of a laminated core or a forged part, the magnetic properties can
be improved so that they can handle an alternating field with less magnetic losses
[5], mainly in the form of eddy currents.
The geometry of the alternator is also a part that is frequently researched. This
mainly concerns the rotor and the stator. The shape of the claws in the rotor
are investigated to see the difference between e.g. round and sharp edges. If the
claws are put too close the field will bypass the stator and the largest part of the
magnetic field will go between the poles over the gap between them. Thus not
giving raise to any voltage in the stator [4]. The geometry research also concerns
the winding in the stator, this to make the cooling efficient and in extension get
as much power as possible from the alternator [3], this is a part that in contrast
4 1 Introduction

to the geometry of the claws and the permanent magnets concerns the electric
losses.
 Alternator theory
2
In the section 1.2 it was explained how a combustion engine works and how it
powers both the vehicle and the alternator. As the engine runs it spins an alterna-
tor shaft. The size of the alternator pulley is often smaller compared to the pulley
on the crankshaft, which enables the alternator shaft to spin faster than the en-
gine. The focus in this explanation will be on alternators of claw-pole type since
these are the most common in vehicles like cars, trucks and buses. The alternator
used in different measurements can be seen in figure 2.1.

5
6 2 Alternator theory

Figure 2.1: Mitsubishi alternator, 28V, 100A, used for measurements.

2.1 Rotor
When the alternator shaft spins so does the rotor. The rotor consists of an iron
core that surrounds the alternator shaft. The rotor also consist of pairs of iron
segments. The segments in the rotor are shaped as claws, this is why this type of
alternators are called claw-pole alternators. The claws in every pair are attached
to each side of the iron core. In the gap between the core and the claws a copper
wire is tightly wound. When a DC current, Ir , is run through the copper wire
a magnetic field is induced in the direction of the shaft, this means that a north
and south pole is generated in the core of the rotor, just as expected from an
electromagnet. Since one claw in each pair is attached at each side of the stator,
one of them is a north pole and the other a south pole. Each pair of claws will
correspond to one pole pair.

If only the fundamental frequency (the lowest frequency in the signal) is taken
into account and the rotor is put inside an iron tube, simulating a stator, the
rotor current will generate a magnetic field inside the core with a magnitude
according to eq. (2.1) [6, 7]. In this equation Bmax is the magnetic flux density, Nr
the number of turns in the rotor winding, Ir the rotor current, Rr the reluctance
of the rotor, Ra the reluctance of the air gap (2 for two air gaps in the loop), Rt
the reluctance of the tube and Ar the cross-sectional area of the rotor winding.
Since the rotor consists of iron, which is a ferromagnetic material, the magnetic
2.1 Rotor 7

flow in the middle of the core will be approximately the same as the flow on the
poles, which means that the field strength in the middle is proportional to the
field strength on the poles.
The magnetic field will be enabled to propagate through the tube between the
poles. The geometry of the rotor will give a flux density around the rotor accord-
ing to eq. (2.2) where p is the number of pole pairs and θ describes a position
around the rotor. Since the rotor will spin at a fast pace in the stator the field
from the rotor can be assumed to be constant for constant speeds and rotor cur-
rent, which means that θ will not be used when calculating the stator field, It will
just be used to describe how the rotor field alter around the rotor.

Nr Ir
Bmax = (2.1)
(Rr + 2Ra + Rt )Ar

Br = Bmax sin(pθ) (2.2)

Simple kinds of generators use permanent magnets to produce the magnetic field
in the rotor, for example dynamos in bicycles, but since electromagnets are easily
regulated most cars, trucks and buses uses electromagnets instead of permanent
magnets. The regulation of the alternator means that the output voltage can be
kept at a constant level which enables the battery to operate in a rather small
voltage range which will improve the life length of the battery. A small voltage
range also benefits the life length and performance of the rest of the electrical
system, for example bulbs.
The DC current, Ir , running through the rotor winding is called excitation cur-
rent or rotor current. The excitation current can be fed to the alternator either
from an external battery or from the own circuit, it is said to be external or self
excited. The rotor current is led to the moving rotor by two brushes that are in
constant contact with slip rings on the rotor. These two brushes generate some
friction when the rotor spins which has to be accounted for when investigating
the mechanical losses. If the rotor current is set to zero there will still be a small
polarization in the rotor. This is due to a phenomenon called hysteresis, this phe-
nomenon also give rise to magnetic losses and will be explained more carefully
in section 2.6.2. The field that the small polarization creates is called a remanent
field. If the remanent field is weak the material in the rotor core is said to be soft
and if the remanent field is strong the core is said to be hard [6].
The claw pole geometry requires a certain amount of claws in the rotor. The
alternator used in the measurements have a rotor containing six pairs of claws
giving a rotor with six pole pairs, this kind of rotor is shown in figure 2.2. Each
claw in the figure is either a north or south pole. This set-up with multiple pole
pairs will increase the frequency of changes between poles and further on the
electric frequency of the alternator. The frequency of the electric field follows eq.
(2.3), where fel is the electric frequency and fmech the mechanical frequency of
8 2 Alternator theory

the alternator shaft.

fel = pfmech (2.3)

Figure 2.2: Rotor with six pole pairs and double internal fan. The slip rings
are covered by the fan in the right side of the figure.
2.2 Stator 9

2.2 Stator
The next part of the alternator is the stator. In contrast to the spinning rotor the
stator is stationary. The stator consists of an iron core with tooth-like slots hold-
ing the stator winding extending in from the outer shell of the stator. These teeth
will connect the north and south pole on the rotor and thus close the magnetic
circuit.
In figure 2.3 the stator can be seen. The black plates at the inner diameter of the
stator are the teeth, the copper windings are also easily seen. At the bottom the
connection to the four diode pairs are also visible in form of four pairs of wires
extending in from the stator winding. Between the rotor and the teeth there is a
small air gap. The length of the air gap will alternate due to geometry reasons,
this will be treated in section 2.5.

Figure 2.3: The stationary components of the alternator.

The magnetic field from the rotor through the stator is shown in figure 2.4. When
a magnetic field is applied between the teeth a voltage will be induced in the
winding according to Faraday’s law eq. (2.4), [7], which shows how the induced
EMF depends on a constant, Ns , and the change of the magnetic flux. The mag-
netic flux, Φ B , is the magnetic field flowing through a certain area. The mag-
netic flux can be calculated from the magnetic field density according to Φ B =
Astat Bstat , where Astat is the cross-sectional area of the stator winding.
10 2 Alternator theory

The magnetic flux is proportional to the rotor current and for a fix rotor current
the flux will alternate sinusodialy according to eq. (2.6) and (2.7), where γ de-
scribes an offset between rotor and stator field (further described in section 2.5).
The constant, Ns , is the number of turns on the stator winding. Since the rotor is
rotating with a time invariant magnetic field, the field in the teeth will alternate
and thus generate an AC voltage in the stator. The magnetic flux is proportional
to the rotor current, in a static case the rotor field alternates sinusoidally around
the rotor, the rotor position relative to the stator will generate an additional si-
nusoidal dependence which will lead to a magnetic field that will correspond to
eq. (2.5) in an ideal case. In a real case the magnitude of the stator field will not
be as high as for the rotor field. In eq. (2.5) ωmech is the angular velocity of the
alternator pulley and t the time.

dΦ B
ε = −Ns (2.4)
dt

Bstat = Bmax sin(pθ)sin(pωmech t) (2.5)

To get as much power from the alternator as possible multiple sets of copper
winding between the teeth are used. Typically the number of winding sets are
three per pole pair. This creates a three phase alternator, where each of the sets
of winding corresponds to one phase. The different phases are connected to each
other according to figure 2.5. This configuration is called a star connection and
the middle point where the three phases are connected is called the star point.
If the three phases are connected to each other in series the connection is called
a delta connection, the biggest difference between these two connections is the
possibility to connect an extra output to the star point, this output will make the
influence of overtones that are multiples of three greater, since these overtones
creates a signal from each phase that are in phase with the signal from the other
phases and instead of cancel each other out amplify each other at the star point.

Φ B ∝ Ir (2.6)

dΦ B
∝ sin(pωmech t + γ), (Ir = const.) (2.7)
dt
2.2 Stator 11

Figure 2.4: The path of the magnetic field from the rotor into the stator and
back for one pole pair [8].
12 2 Alternator theory

√
Figure 2.5: Star connection of phases, U = 3Up , [9]. Up and Ip represents
the voltage and current over one phase and U and I represent the voltage
and current between two phases.

2.3 Rectifier
To charge a battery and power the electrical system in a vehicle a DC current is
needed. This is why the AC current from the stator cannot be used directly. The
AC current first has to be rectified to a DC current by a rectifier. The rectifier
uses two diodes connected to each phase. A diode works in the way that it only
lets current through in one direction. If two diodes are connected in parallel,
according to figure 2.6, only current in one direction will be let through creating
a DC-current. When distributing the three phases over one period each of the
phases will be offset to each other by 120◦ . This will enable the DC voltage to
have less ripple.

IU

U
IB

UB

V
W

Figure 2.6: Rectifier bridge and the connection to the phases.

2.3 Rectifier 13

Only the two diodes with the greatest voltage over them will be conductive at any
moment 1 . This means that the voltage induced in each phase will go through
the steps in figure 2.7. At first one sine signal from each phase is illustrated, sep-
arated from each other with a phase of 120◦ . In the second figure the resulting
voltage on the positive and negative side is shown and at last the output volt-
age from the stator [10]. For simplicity only the fundamental frequency is used,
meaning that the phase voltage will look according to eq. (2.8), where, U b is the
amplitude of the voltage, ωel is the angle velocity and γ is the offset. The relation
between ωel and ωmech follows from eq. (2.3) and means that ωel = pωmech .

U =U
b sin(ωel t + γ) (2.8)

Figure 2.7: Voltages through a rectifier.

If the load on the alternator is assumed to be pure resistive the output current
will have the same shape as the output voltage according to Ohm’s law eq. (2.9).
This can be seen in a comparison between figure 2.7 and 2.8.

1 Commutation effects neglected.

14 2 Alternator theory

U = RI (2.9)

The voltage induced in each phase also drives a current according to graphs a-c
in figure 2.8. The reason that the shape of each current wave is almost square is
that only two diodes are conducting at a time only letting a current run through
one phase in each direction. When only taking the fundamental frequency in
consideration the phase current is 78% of the output current [10]. The output
current can be seen in graph d and have the same principal look as the output
voltage since the load is assumed to be pure resistive.

Figure 2.8: Phase and output current over the rectifier.

Two additional diodes can be attached to the star point, these diodes include the
third overtone in the voltage. This allows the output from the rectification to be
increased by as much as 17%. The drawback with the two extra diodes is that
ripple in the output voltage is increased.

2.4 Voltage regulator

At this point a device that generates DC current from mechanical force is com-
plete. But the fact that this device will generate different voltage at different
speeds and that the battery is exposed to minimum wear at a fixed voltage inter-
val and the fact that the rotor current has to be fed to the rotor creates the need
for yet another component, the voltage regulator.
The main task of the voltage regulator is to keep a constant voltage out from
the alternator to the battery. The regulator compares the output voltage to the
set point voltage. If the output voltage is lower than the set point the output
is increased and vice versa. This is done by changing the rotor current, if the
output voltage is too high it can decrease the rotor current and further on the
output voltage and vice versa.
2.5 Equivalent circuit 15

2.5 Equivalent circuit

An alternator consists of two electrical circuits that are connected to each other
through the magnetic interaction. The circuit in the stator can be modeled as one
inductance that describes the voltage drop due to the magnetic interaction and
two resistive loads that describe the magnetic and stator losses.
This ciruit can be seen in figure 2.9 where Up is the voltage induced by the rotor,
Ls the inductance that describes the voltage drop due to the magnetic losses, Rs
the resistance in the stator, Rmag describes the magnetic losses and Rload the re-
sistance that describes the load on the alternator. The main advantage of using
an inductance to model the voltage drop due to magnetic influence is that the
voltage drop can be written as U = jLI where the imaginary voltage drop due to
inductance does not give any power losses in contrast to the resistances.
This circuit can also be described with the voltage equation eq. (2.10) where ϑ
is the load angle and the total inductance Ls is divided into two separate induc-
tances Ld and Lq , these two will be explained further down in this section. The
load angle, ϑ, is a description of the difference between the mechanical angle of
the rotor and the angle of the stator current (or stator field). At no load these
two angles will be in phase, when the load is increased a retardation of the stator
current is created giving a shift between the rotor and stator. The resistance that
describes the magnetic losses will not impact the voltage equation since it is in
parallel to the polrad voltage, therefore it will be of no use during the calculations
and is only used in this section for explanatory purpose.
The load angle is very important for the functionality of the alternator. The load
angle describes the position of a magnet (the rotor) in an external field (from the
stator). If the two are aligned perfectly there will not be any resistance momen-
tum impacting the system and thus no power production. As soon as the load is
increased a phase is created between the rotor and stator, this gives a resistance
momentum which in turn enables a power production when voltages are induced
to resist losses that are created by the movement of the rotor. The higher the dif-
ference gets between the rotor and stator the higher the resistance momentum
gets and this will also enable a higher power production [11, 12].

Up = Us + Is Rs + jωel Is (Ld cos(ϑ) + Lq sin(ϑ)) (2.10)

The inductance that describes the voltage drop due to the magnetic interaction
can be divided into two separate inductances. The two inductances correspond
to the path along different axes that the magnetic field can go from the rotor to
the stator. These two axes are described in figure 2.10, where an alternator with
one salient pole pair is illustrated. The poles are said to be salient since they have
a mushroom shaped top. This are the type of poles used in all alternators in the
thesis.
The alternators used in the measurements have six pole pairs, the d and q axes
16 2 Alternator theory

Is Ls Rs

Up Rmag Rload Us

Figure 2.9: The equivalent circuit of the alternator.

Figure 2.10: The d and q axes represented in a one pole pair set up with a
salient pole [13].

have the same principal look as in figure 2.10 with the difference that the angle
between the axes gets smaller since there are more pole pairs.

Xd describes the inductance when the magnetic field go from the rotor to the
stator along the d axis and Xq when the field goes along the q axis. At no load
the load angle is zero which means that only the d-inductance is acting on the
system. The retardation given by increasing load will then give the q-inductance
a larger impact on the system. The d stands for direct and q for quadrature, the
quadrature axis refers to the axis with an electrical angle that is orthogonal to the
electrical angle of the direct axis. These two axes represent the nearest air gap
2.5 Equivalent circuit 17

between the rotor and stator (along d-axis) and the widest air gap (along q-axis).

The current going through each of the inductances can be calculated with a trans-
formation that projects the current in each phase on the d and q axes. The trans-
formation uses two coordinate systems, one system fixed in the rotor and one
system fixed to the stator voltage. The transformation for a three phase system
results in two currents running through each inductance according to eq. (2.11)
and (2.12) [11, 12]. The current Is is the same in all three phases but with a phase
shift of 120◦ between the phases as mentioned in section 2.3.

Id = Is cos(ϑ) (2.11)

Iq = Is sin(ϑ) (2.12)

According to eq. (2.11) and (2.12) the currents going through each inductance,
Id and Iq , are orthogonal. This is further showed in figure 2.11. By using eq.
(2.11) and (2.12) and the relation between the inductance and the impedance of
the system, eq. (2.13), the voltage equation can be written according to eq. (2.14).

X = ωel L = 2πf L (2.13)

Up = Us + Is Rs + jXq Iq + jXd Id (2.14)

The transformation of the voltages and currents introduces complex numbers in

the equation, this gives vectors with different directions. The right hand side
of eq. (2.14) can with this complex representation be described by a complex
diagram seen in figure 2.11.

In the figure the load angle can be seen, in reality there is also a phase between
the stator voltage and current due to the load on the alternator. The rectifier will
however keep this phase low. In theory cosine of the phase between the current
and voltage in the stator will not be lower than approximately 0.955. Because of
this the angle is estimated to be zero, which is the case in figure 2.11 [10].

As said above the load angle describes an offset between the mechanical angle of
the pulley and the electrical angle of the voltage in each phase. The complex vec-
tor graph also shows how ϑ impacts the different currents running through each
inductance. This angle develops when the alternator is loaded. If the alternator
is loaded harder a retardation between the mechanical and electric field arises. A
heavier load gives a larger load angle. When the load angle is zero, at no load, all
the magnetic field goes along the d axis and when the load angle is 90◦ , at full
load, all the magnetic field goes along the q axis [14–16].
18 2 Alternator theory

−jXd Ibd
bp
U

y = −jXq Ibs
ϑ Rs Ibs

−jXq Ibq

bs
U

Ibq
Ibs
ϑ

Ibd

Figure 2.11: Space phasor diagram of eq. (2.14). Here the approximation
that the stator current, Is , is parallel to the stator voltage, Us . Also the fact
that the current through each inductance is orthogonal can be seen. In this
figure a vector called y can also be seen. This vector is used to calculate the
load angle, ϑ, mentioned in section 3.1.2.

The Polrad voltage, Up , can be calculated from the stator voltage, Us , and the
stator current, Is , from these relations. The Polrad voltage is the true induced
voltage in the stator and is further on used to calculate the rotor current, Ir , this
is described in section 3.2.2. This relations combined with the relation between
the output and stator current are then used to calculate the losses. In the voltage
equation a dependence on the alternator speed can also be seen since the induc-
tances are dependent on the angular speed according to eq. (2.13). The polrad
voltage is proportional to the alternator speed and the rotor current according to
Faraday’s law eq. (2.4) with the magnetic field depending on the rotor current
and alternator speed according to eq. (2.1) and (2.5).
2.6 Losses 19

2.6 Losses
In all alternators there are resistance losses, this resistance is in form of mechan-
ical friction, magnetic resistance and electric resistance. When the alternator is
running, all these resistances generate heat, to keep the alternator from overheat-
ing a fan is used. This fan is driven on the same shaft as the rest of the alternator,
meaning it also causes additional losses in efficiency. At the same time the fan
cools the winding in the alternator causing resistive losses to decrease. To ac-
count for all these losses mathematical models of how they depend on different
parameters are used [8, 11, 17, 18].

2.6.1 Mechanical resistance

The mechanical losses in an alternator depends on the fan and friction in differ-
ent layers. The main task of the fan is to cool the windings and diodes in the
alternator to keep them from overheating. And since the resistive losses increase
with increasing temperature the fan also decreases resistive losses. The mechan-
ical power losses origin from a mechanical resistance momentum that together
with the alternator speed gives the power losses according to eq. (2.15) where
P is the power losses, M the resistance momentum and ωmech the angular speed
of the alternator [7]. The resistance momentum due to the fan depends on the
square of the alternator speed and the momentum due to friction is independent
of the alternator speed [7]. In extension the power losses will depend on the first
and third order of the alternator speed, which can be showed using eq. (2.15)
together with the definition of the resistance momentum of the fan and friction.

P = Mωmech (2.15)

The resistance momentum, M, have a dependence according to eq. (2.16) [7].

This means that the mechanical losses of the alternator can be decided from the
speed and acceleration of the alternator and the moment of inertia.

˙
M = IM ωmech (2.16)

In most alternators there are two fans mounted in the front and the back of the
rotor which means that the air comes in axially at the front and the back and
goes out radially, allowing air to cool the stator winding. However there are
also alternators with only one fan in the front. The two fan set up is used since
it provides better cooling compared to the one fan set up, the two fan set up
also allows smaller fans on the inside of the alternator enabling more compact
alternators.
20 2 Alternator theory

2.6.2 Magnetic resistance

The magnetic losses in the alternator are due to the fact that an alternating field
requires many and sudden changes in the magnetic field in the material. But
since there are limits on how well a material can handle these changes, a retarda-
tion in the changes develops losses.
To understand these losses the relation between the external H-field and the in-
duced B-field is visualized in figure 2.12. The initial state of the core can be seen
in point o where no H- or B-field are present. An H-field can then be applied
in the core by a rotor current. This will lead to an increase of the B-field accord-
ing to the curve oa, the reason for the increase can be seen in eq. (2.1) where an
increase in rotor current will lead to a increasing B-field.
The reason for the behaviour close to a, where the increase of the B-field fades is
that the magnitude of the B-field is limited by the material due to saturation. The
saturation arise from the fact that the magnetic field propagates through the ma-
terial by aligning small magnetic moments to the field and when full alignment
is reached the magnitude of the field that propagates in the material cannot get
stronger and saturation is reached [6].
If the H-field then is decreased to zero along ab it can be seen that a B-field will
still be present at no H-field, this is called a remanent field. This field is due to the
fact that when the H-field is close to zero it is too weak to change the alignment
of the magnetic moments in the material which gives the B-field and therefore
the remanent field will be present. To decrease the B-field to zero an H-field in
opposite direction is needed, the intensity of this field is called the coersive field
intensity. The coersive field intensity can be seen in point c. If the negative H-
field is further increased the core will be saturated in point d, just as in point a.
The core loop will then be closed by the same remanent field and coersive field
intensity for the fields in the opposite directions closing the loop [6].

Figure 2.12: The relation between the H and B-field in the core [19].
2.6 Losses 21

The core losses will then depend on the area of the closed loop in figure 2.12 [11].
This area can be calculated using the integral in eq. (2.17). The core losses that
are present are hysteresis and eddy currents. When modelling the core losses
there are a number of different models that can be used [17]. In this case it corre-
sponds to a number of different expressions with parameters that will be fitted to
measured data. The limited knowledge about material and magnetic properties
leads to a choice of model that has a rather low accuracy but will work well with
the small changes in properties that might show for different alternators. This is
more discussed in [17].

I
H dB (2.17)

The hysteresis losses are due to the fact that energy is needed to alternate the
magnetic field in the core and therefore some losses will be present. The losses
due to hysteresis can be accounted for with the empirical expression eq. (2.18),
where Ph are the hysteresis losses, Kh a constant depending on the core material
and construction, Bm the maximum core flux density, s the Steinmetz exponent
(ranging from 1.5 to 2.5) depending on the core material and f is the frequency
of the field changes [8, 11, 17].

Ph = Kh Bsm f (2.18)

The eddy current losses in the core are due to the alternating magnetic field. The
core is constructed by a ferromagnetic material. The ferromagnetic material is
chosen since it will give a magnetic field strength in the poles close to the strength
in the middle of the winding. The core is therefore also a good electric conductor,
this means that the alternating magnetic field will induce a voltage in the core,
the voltage in its turn will drive a current through it and this will lead to resistive
power losses. The eddy current losses can be accounted for according to eq. (2.19)
that works in the same way as for the hysteresis case [8, 11, 17].
It has also been showed that the constant Ke depends on the thickness of the
core [8, 11, 17], to reduce this thickness as much as possible it is standard to
use laminated cores. This dependence on the thickness is also the reason for
investigating cores with sintered powder mentioned in section 1.3 since a powder
would reduce the thickness even further.

Pe = Ke B2m f 2 (2.19)

The main part of the magnetic losses both due to hysteresis and eddy current
origins in the stator. This is because the field in the stator alternates with the
speed of the alternator and the alternating field giving magnetic losses. In the
static case there will not be any magnetic losses in the rotor since the rotor field
22 2 Alternator theory

only could be changed by changing the rotor current, which is not possible. When
it comes to a dynamic case there will be disturbances that origin from the change
in alternator speed. This change can generate a field from the stator that alternate
the rotor field for a short time and thus gives a small magnetic loss also in the
rotor.
There are also some small secondary losses due to leakage of magnetic flux and
remanent fields. These losses will be accounted for in the other losses models
since they are always present.

2.6.3 Electric resistance

The electric losses in the alternator are resistive losses in the stator and rotor and
also losses due to the voltage drop in the diodes. These losses can be calculated
with Ohm’s law eq. (2.9) and the Power law eq. (2.20) in a rather straightforward
manner.

P = UI (2.20)

The resistance for a metallic conductor is also temperature dependent according

to eq. (2.21), where R is the resistance, R0 the measured resistance in the test,
α the temperature coefficient of the material (3.9 ∗ 10−3 K −1 for copper), T the
temperature and T0 the temperature at the measurement [7]. The temperature
dependence is due to the fact that a higher temperature produces vibrations with
larger amplitudes for the atoms in the material, this increased motion gives more
frequent collisions between the electrons and atoms thus giving more resistance
in the propagation of electrons and further on an energy loss.

R = R0 (1 + α(T − T0 )) (2.21)

[8, 9, 20–22].
 3
Method

In chapter 1 and 2 the theory used in this master thesis was explained. The next
step was to perform a number of measurements on alternators. These measure-
ments are used to validate the theory explained in previous chapters and also
to complement the theory where relations are not exactly explained. Finally the
theory and these measurements are put together to create a mathematical model
of an alternator.

3.1 Model of alternator

When constructing a model of the system, a block structure is used. At first
simple models of each block will be produced. The blocks are then improved to
fit the system. The model of the system is implemented in Matlab [23], this to be
able to use the model on the collected data in a simple way. The data recorded
from different field measurements on the alternator will be handled in Matlab
and the Rotec software [24].

3.1.1 Construction of an initial model

At first a set up of a simple model is carried out. This initial model uses a constant
efficiency for the alternator and also a constant fuel consumption for the engine.
This model is built up in a way so that required properties can be calculated from
measured data. The measured data that is used as input data in the first model is
current and voltage output from the alternator, engine speed and engine torque.
The gear ratio between the engine and the alternator and the fact that the buses
use two alternators in parallel is also accounted for in the model.
The model is built with different blocks, so that one block can be easily exchanged

23
24 3 Method

for another improved model of that part without having to change the rest of the
system. The main block structure is illustrated in figure 3.1. This figure shows
an initial overview of how the voltage and current from the alternator and the
speed and torque from the engine can be used to calculate the fuel flow, electric
power produced and amount of fuel used.

Figure 3.1: Block structure of the model.

At first the data is filtered to get rid of incorrect measurement data, for exam-
ple the current output from the alternator cannot be negative, because of the
rectification where the diodes only let current flow in one direction, and thus
all measuring points with negative current contain a measurement error and are
therefore excluded. The same are for unreasonable values for voltage and engine
speed.
At the end the total energy produced by the alternator, the total amount of fuel
used by the engine and the total amount of fuel used by the alternator during a
cycle of measured data is required.
To get the torque used by the alternator at every given moment the output power
has to be added to the power losses in every given moment and then divided
by the alternator speed since the power, P , relates to the torque used to run the
alternator, Talt , and the speed of the alternator, ωmech according to P = Talt ωmech
[7]. This value can then be compared with the torque produced by the engine.
The ratio between these two torque values then describes the percentage of torque
used by the alternator at every given point of time.
In this case the fact that the alternator cannot use more than 100% of the torque
produced has to be taken into account. A scenario when the alternator uses more
energy than the engine generates is for example when a bus is going downhill
and gravity is adding power to the system. The percentage of torque used by the
alternator can then be multiplied with the fuel flow to get the flow of fuel that
the alternator uses at every measurement point.
The fuel flow used by the engine can be calculated with the help of engine speed
and engine torque, at first a constant weight of fuel per kWh is used. This number
is not actually constant, it depends on different factors but this will work as a first
3.1 Model of alternator 25

simple model.
The fuel used by the alternator can then be calculated by multiplying the fuel
used by the engine with the percentage of power used by the alternator. If the
two fuel flows are integrated over the whole cycle the total amount of fuel used by
the engine respectively the alternator are given. To get the total energy produced
by the alternator the output power has to be integrated over the whole cycle.
The ratio between the fuel used by the alternator and the energy produced by the
alternator then gives the number of units fuel used per unit energy generated.
The initial model of the alternator uses a constant efficiency for the alternator.
This means that it only uses a constant gain on the output power to get the losses
in the alternator at any given moment. This is of course not the way it works in
reality since the losses depends on a number of factors discussed in section 2.6.
The model accounts for the fact that there are two alternators in parallel, this
means that only half the output current comes from one alternator. The output
voltage from one alternator is of course the same as the output voltage for both
alternators since they are parallel to each other [7].

3.1.2 Improvement of the model

When an initial model is implemented and works as it should improvements of
the different blocks are considered. This to be able to improve the model so that
it describes the reality in a better way.
There are a number of improvements to do on the initial model to make it fit
reality. The two blocks that needs an improvement is the engine block with the
fuel consumption and the alternator block with all the alternator losses. Since
this thesis focuses on optimizing the alternator this is the block where the effort
will be put while an improvement of the engine and its consumption will be
taken from known data. Losses from the belt will be included in the losses of the
motor.
First an improved model of the alternator is considered. This model has to ac-
count for mechanical, magnetic and electric losses. These losses are described
more closely in section 2.6.
The mechanical losses that are described in section 2.6.1 only depends on the
alternator speed. The friction losses are proportional to the alternator speed and
the fan losses proportional the alternator speed to the power of three.
There are also magnetic losses. These are the most complex losses and they are
described in section 2.6.2. They will depend on the magnitude of the magnetic
field and the speed of the alternator. The speed of the alternator is directly pro-
portional to the speed of the engine as mentioned before. The magnitude of the
magnetic field is not measured but is directly proportional to the rotor current.
The set up of these measurements is discussed in section 3.2.6.
To simplify the magnetic losses they can be modelled as an electric circuit. This
circuit is called an equivalent circuit and is described in section 2.5. The equiv-
26 3 Method

alent circuit consists of one resistance and one load angle dependent inductance
connected according to figure 2.9.
The equivalent circuit leads to the voltage equation, eq. (2.10), and further on a
complex vector graph shown in figure 2.11 can be drawn. This figure shows the
influence of the load angle on the inductances and in extension the Polrad angle.
By measuring the two inductances and also current, voltage and resistance over
one phase; the load angle can be calculated according to eq. (3.1) which is derived
from the geometry in figure 2.11. The load angle could also be investigated from
a measurement described in section 4.1.3. The measurement is used as validation
of these calculations.

 Xq Is 
ϑ = arctan (3.1)
Us + Rs Is

At last there are electric losses that depends on the current running through the
different parts of the alternator and the resistance or voltage drop in each part.
Since current is running through the stator, the rectifier and the rotor; the stator
current, output current and rotor current will be used together with the resis-
tances or voltage drop to calculate the electric losses in the model with Ohm’s
law and the Power law, eq. (2.9) and (2.20).
All the different categories of losses will be treated separately in the alternator
block. This means that also inside the block improvements of separate losses can
easily be implemented in the whole system.
The improvement of the engine block will focus on fuel consumption and cal-
culating the engine torque. In the data provided by Scania the engine torque is
given as percentage of the maximum torque. The maximum torque is a relation
that is individual for each engine. The maximum torque is dependent on the
engine speed in a relation that is shown in figure 3.2. This curve is needed to
calculate the actual torque that the percentage represent at every point of time.
This means that both the engine speed and the percentage of torque is needed to
calculate the actual torque in every moment. The fuel consumption is then given
in a table provided by Scania that depends on the engine speed and the generated
torque. The fuel consumption is given in grams per minute in the table.
In the end a model of the alternator is constructed. This model can replace ear-
lier fixed values on the alternator load and also giving a better understanding of
which losses that will be impacted by different changes.
3.2 Measurements 27

Figure 3.2: Curve describing the maximum torque that the engine can gen-
erate with respect to the speed of the engine.

3.2 Measurements
To be able to improve the simple initial model and decide the different param-
eters for the losses a number of measurements are carried out. These measure-
ments are used to get information on the parameters in the model and also to
confirm the theory gathered from the literature study.

3.2.1 Provided field measurements

To be able to use the model in a real scenario there are an amount of field data
available. The data is recorded from different buses driving in different climates
around the world to give a complete representation of temperature, humidity etc.
The different variables that are collected from the buses are presented in table
3.1. Further data such as battery current etc. are also collected from the driving
cycles, but since they are not relevant for this model they are excluded from table
3.1. All data is collected at a frequency of 5H z.
The provided data has to be processed since there are measurement errors at
some points and for example at start up the alternator current is negative, which
is not allowed for a functioning alternator. Therefore all measurement points
with data that is not valid has to be excluded. There are also corrupt measure-
ment files. These files have to be excluded in the mathematical model for a
smooth handling of the data.

3.2.2 Rotor current

The first measurement that was carried out was a measurement of the rotor cur-
rent. It is important to have an idea of how it changes with respect to the collected
variables since it is controlled by the voltage regulator and impacts the magnetic
field in the rotor.
To do this test a test bench with an alternator was used, the test bench can be
seen in figure 3.3. The test bench has a belt and motor to be able to power the
28 3 Method

Variable Meaning
ActualEngine_PercTorque Percentage of total torque generated
EngineSpeed Engine RPM
I_alternator Current out from the alternator in A
TCOVehSpeed Vehicle speed in km/h
U_alternator Voltage out from the alternator in V
EstEngParasiticLossesPercTorque Percentage of total torque that is esti-
mated to be parasitic losses

Table 3.1: Variables of collected data from buses.

alternator. The alternator can also be loaded at the test bench, this to be able to
run the alternator under proper conditions. The speed of the alternator and the
alternator load current can then be controlled.
A multimeter is then attached to the rotor circuit to be able to measure the current
going through the rotor. By varying the motor speed and the load current, tables
of how the exciter current changes with alternator speed for different loads could
be extracted.
A polynomial curve of second degree could then be fitted to the measured values
for each load current. One problem with this adjustment is that each load had
maximum four speeds measured and minimum two. This is partly because of
the four fixed motor speeds on the test bench and also because of the fact that to
achieve a high load current high motor speed is required. This gives few measur-
ing points to fit the curve which in extension gives a curve with high uncertainty.
The current in the rotor also depends on the temperature since the resistance is
temperature dependent, which is discussed in section 2.6.3. Since it is hard to
measure the temperature of the copper wire the alternator is run at full load for a
couple of minutes to reach a temperature where it is assumed that the alternator
normally operates.
This measurement is also carried out with more data points and instead of a
multimeter a Rotec system [24]. This measurement generates a graph that better
describes the relation and is described in section 3.2.6.
3.2 Measurements 29

Figure 3.3: Test bench for alternators.

3.2.3 Load angle

The second measurement that was carried out was a measurement of the load
angle. This to be able to find a relation between the losses and the load.
At first the alternator was opened and the rotor was removed. Two wires were
then soldered on each side of a stator phase to be able to measure the voltage
over the phase. This set up is shown in figure 3.4, the soldered wires are the
black and the yellow one on two of the phase wires.
Further on a sensor was set close to the alternator pulley to be able to measure
the speed of the alternator. The sensor and the two soldered wires where then
attached to two measurement cards in a computer. The measuring equipment
was of the brand Dewetron [25], this is also the case for the software used to
analyse the data. The alternator was then accelerated on the test bench and data
from different speeds were collected. A magnetic inductive sensor was used, it
works in the way that when a piece of metal runs in front of it a magnetic circuit
is closed and a voltage is induced. The voltage is then registered by a measuring
card.
The soldered wires were then used to get the stator voltage. Two power pliers
were also used to measure the stator current and ripple in the output current. All
these signals were connected to the measuring equipment.
The same approach, with measurements of pulley speed, stator voltage, ripple
current and output current, was also used with a measurement system from
30 3 Method

Rotec [24]. The measurement system from Dewetron only uses analog signals
while the Rotec system can also handle digital signals that enables a faster update
frequency for the pulley speed, which gives a more robust system. The Rotec sys-
tem also supports a number of functions for pretreatment of the collected signals
[15].

Figure 3.4: Soldering on each side of a stator phase.

The result from the measurement with Dewetron equipment discussed in section
4.1.3, did not follow the expected behaviour. Before the approach with the Rotec
system was carried out another approach to measure the angle between the pul-
ley and the stator field was investigated, as described below.
To do this measurement a lamp with a very distinct voltage step from off to on
was used. This lamp was attached to the wires soldered to the stator. The lamp
was chosen so that it will emit light only at the top voltage of the stator. This
means that the light will turn on and off at the same frequency as the stator field.
A circular plate with marks every ten degrees was glued to the pulley and the
lamp was set to light up a small part of the plate. Figure 3.5 shows the alternator
pulley with the marked plate glued to it.
A mark every ten degrees means that there are 36 marks around the plate. Since
there are six pole pairs on the alternator the frequency of the stator field will be
six times the frequency of the pulley. This means that there are six marks on the
plate for every period of the stator field. Since the light flashes with the same
frequency as the stator field it will light up every sixth mark in the plate making
it appear as if the plate is standing still. This will hold for any constant load but
3.2 Measurements 31

Figure 3.5: Alternator pulley with marked plate.

when the load is changed the plate will appear to move a couple of degrees, this
is due to the fact that a load on the alternator creates a change in the phase angle
between the pulley and the stator field. If the angle is observed for different loads
a relation between the angle and the load can be decided.

Since there are 36 marks on the plate each mark represents 60◦ between the pul-
ley and the stator field. By making a small arc with five marks 12◦ apart a tool
that enables measurements with an accuracy of 15◦ is created. For example if
one line on the pulley is perfectly aligned with one mark on the arc a reference is
gained. If the load on the alternator then is changed the mark on the pulley will
appear to move, if the load is changed further the mark next to the first mark on
the pulley will be aligned with the mark next to the first mark on the arc. This
change represents an offset of 15◦ between the rotor and the stator field. This
extra fine measurement is necessary since a change of the load angle of 60◦ is too
high to get a good estimation.

3.2.4 Inductances
In the equivalent circuit discussed in 2.5 there are two inductances that corre-
sponds to the voltage drop that the magnetic influence give, these two induc-
tances are assumed to be orthogonal to each other and the influence of each of
them is decided by the load angle.

If the whole stator is considered there is also a resistance in series with the two
inductances, which corresponds to the resistive losses in the stator wire. The
load angle then decides the current going through each inductance according to
eq. (3.2) and (3.3), where b
Id and b
Iq are the currents running through the d and q
32 3 Method

inductance, b
Is the stator current and ϑ the load angle. These two equations are
decided from the geometry in the complex vector diagram in figure 2.11.

Iq = b
b Is sin(ϑ) (3.2)

Id = b
b Is cos(ϑ) (3.3)

At no load all current will go through the inductance named Xd and at full load
all the current will go through Xq . The two inductances together with the cur-
rents, voltages and the angles between them are illustrated in figure 2.11.
The resistance of the rectifier can be measured by running a direct current through
the circuit and measuring the voltage at the same time and then using Ohm’s law,
eq. (2.9).
A measurement to gain information about the two inductances between the stator
and rotor was then carried out. An alternating current of 10A was applied over a
phase, the rotor was then rotated at the same time as the current over the phase
was monitored. The current alternated at a sinusoidal manner just as expected.
At minimum current only the rectifier resistance and the small inductance was
impacting the system, and at maximum only the rectifier resistance and the large
inductance was impacting the system. From this behaviour the two inductances
can be calculated and further on the load angle can be calculated. The induc-
tances and the load angle can then be used to validate earlier measurements on
the load angle. Figure 3.6 shows the set up of the measurement.
The current and voltage from the measurements give through Ohm’s law, eq.
(2.9), an impedance that is a combination of the diode resistance and the mag-
netic inductance according to eq. (3.4), where Z is the impedance of the circuit,
X the inductance and R the resistance. Due to the fact that diode resistance and
the impedance are measured the magnetic inductance can be calculated using
complex numbers where the resistance is real and the inductances are purely
imaginary and the impedance is the absolute value of the two.

√
Z= X 2 + R2 (3.4)

These two inductances can also be used to decide the load angle and its depen-
dence of the phase current and voltage. The load angle is decided according to
eq. (3.1).
3.2 Measurements 33

Figure 3.6: Set up for the measurement of the inductances.

3.2.5 Mechanical losses

The next part of the measurements was to construct the models of the different
losses. At first the mechanical losses are investigated. To be able to investigate
the mechanical losses the alternator had to be run without any currents running
through it since the currents also would give magnetic and electric losses. This
meant that the alternator could not be run in the test bench since this requires a
fully functioning alternator.
At first an approach using a drilling machine was investigated. The drilling ma-
chine was used to speed up the alternator.
To be able to speed up the alternator further an approach with two alternators
connected to each other was investigated.
At first the two alternators were disassembled. On alternator I wires are then
soldered on each of the phases and the star point. This to be able to get the
phase and star point voltages as outputs. Alternator I is then reassembled. On
alternator II the rectifier is cut and removed, the rest of the stator is then left in
the alternator. The reason not to remove the whole plate in the bottom, which can
be seen in figure 2.3, is that the connection for the brushes has to be used during
the measurement because the rotor has to be fed with a current, and since this
connection is in the plate it has to remain. On alternator II, wires are soldered to
the cut phase connections. The alternator is then reassembled.
34 3 Method

Alternator I is then put in the alternator test bench. This to be able to run al-
ternator I just as before, with controlled load and speed. The phases are then
connected to each other through a relay to get the stator voltage from alternator
I as input to alternator II with a controlled switch.

Alternator I can then run in the test bench just as before. The connection then
enables alternator II to be run as an engine since the alternating stator voltage
from alternator I makes the rotor rotate. To achieve this a current on the rotor
supplied by a current source is needed. The current will induce a magnetic field
in the rotor enabling the rotor to rotate with the alternating voltage.

If the connection between alternator I and II is activated when the alternator is

out of phase, for example when alternator II is at rest, alternator II is run as
an asynchronous machine. This will give a great loss in the power transmission
between the two alternators, and will lead to a significantly lower speed for alter-
nator two compared to alternator one.

If alternator two instead is accelerated to the same speed as alternator one before
the connection is activated this will give a synchronous machine. This means
that the power conversion will increase significantly and the two alternators will
run at the same speed. This means that alternator II will be able to reach higher
speeds and will therefore give a better description of the losses over the whole
range of speeds.

To synchronize the two alternators before turning on the connection a set up

with three lamps and a drilling machine will be used. At first the drill is used to
accelerate the pulley of alternator II to approximately the speed of the pulley on
alternator I. Then the lamps are used to get the two alternators in phase. These
lamps will be connected between the corresponding phases of the alternators
according to figure A.1 in appendix A. In this connection the lamps will turn
on and off in a cyclic behaviour, one lamp will be lit at the time, and when all
the lamps are dark the voltages between the two alternators are in phase and it is
time to switch on the connection. A complete circuit diagram of the measurement
set-up can be found in appendix A.

By disconnecting the two alternators the speed of alternator II, without any cur-
rent running through it, can be investigated. If the rotor current is turned off
only the mechanical losses will be acting on the alternator since the rotor will not
have any magnetic field. According to the theory explained in section 2.6.1 the
mechanical losses can then be calculated by monitoring the pulley speed of the
alternator.

To monitor the speed of the pulley over time an inductive sensor is used. This
gives a table of how the speed of the pulley changes with time. A derivative can
then be carried out to get the angular acceleration of the pulley with respect to
time. From these two tables a third table of how the acceleration depends on
the speed can be put together. Further on an expression of the acceleration with
respect to the speed can be fitted to the data.
3.2 Measurements 35

By using eq. (2.15) and (2.16) from section 2.6.1 the resistance momentum and
further on the power losses due to friction and fan can be calculated.
Since the mechanical losses depend on the first and third order of the alternator
speed, 2.6.1, the mechanical resistance momentum depend on the zeroth and
second order of velocity. Since P = Mωmech the speed dependence of the friction
moment, M, is reduced by one degree compared to the power losses, P . This also
means that the components in front of orders different to zero and two are close
to zero when fitting an expression to the collected data.
By comparing the acceleration to the square of the velocity a linear curve can be
fitted to the collected data. The constant deciding the friction will be the value
fitted to the zeroth order of the square of the speed. The constant deciding the
fan influence will be the value fitted to the first order of the square of the speed.
To verify the mechanical resistance a test using a rotor where the fan had been
removed was also carried out. Without a fan the mechanical resistance momen-
tum should only depend on friction. In other words the mechanical resistance
momentum should not depend on the pulley speed. The friction constant should
be the same as in the first test since the fan only impacts the dependence to the
second degree of speed.

3.2.6 Magnetic losses

After the investigation of the mechanical losses the magnetic losses were investi-
gated. This was done using the same test as for the mechanical losses with the
exception that the rotor current was left on when the connection between the
two alternators were turned off. The same test was used since the mechanical
and magnetic losses have a similar dependence on the speed of the alternator
which can bee seen in section 2.6.1 and 2.6.2. The presence of rotor current leads
to a magnetic field in the rotor that will slow the pulley down as well.
This means that the alternator speed only depends on the mechanical and mag-
netic losses in this case. The magnetic losses will change with rotor current and
speed while the mechanical losses only depends on the speed. This makes mul-
tiple measurements with different rotor currents necessary. In section 2.6.2 it
is said that the magnetic losses depends on the speed of the alternator and the
magnetic field and since the magnetic field is proportional to the rotor current,
described in 2.1, a change of the rotor current will give a similar change in the
magnetic field.
To investigate the resistance momentum with the magnetic losses eq. (2.15) and
(2.16) are used, just as for the mechanical losses, the power loss and resistance
momentum calculated from this test will depend on both the mechanical and
magnetic resistance. If the mechanical resistance then is subtracted from the
total only the impact of the magnetic losses will be left.
As described in section 2.6.2 the magnetic resistance depends on eddy current
and hysteresis, these resistances depends on the zeroth and first order of velocity
as well as the square of the current. This can be seen in eq. (2.18) and (2.19)
36 3 Method

in section 2.6.2, where the magnetic field, Bm is linearly dependent on the rotor
current Ir according to eq. (2.1). Just as for the mechanical losses curves can be
fitted to measured data. In this case the resistance will be linearly dependent on
the alternator speed for the different rotor currents.
The results gained from measurements with different rotor current can then be
put together to decide the constants for the magnetic losses. In this case the value
fitted to the zeroth order of the alternator speed will correspond to the hysteresis
constant times the square of the rotor current. the value fitted to the first order
of the alternator speed will correspond to the eddy current constant times the
square of the rotor current.
Since the measurement is carried out for a number of different rotor currents
the hysteresis and eddy current constant can be separated from the rotor current
leaving only the constants which then can be put into the model.
To investigate the core losses further a comparison between the theory and the
measured behaviour is carried out. The core losses described in section 2.6.2
describes the relation between the H- and B-field which can be seen in figure
2.12. In reality the H-field relates to the rotor current, and the change in the
B-field relates to the induced voltage in the stator, the Polrad voltage.
Therefore the relation between the H- and B-field (see figure 2.12) will have the
same principal behaviour as the relation between the rotor current and the am-
plitude of the polrad voltage [11]. Since the rotor current only runs in one di-
rection the field from the rotor will always be larger than or equal to zero. This
means that the similarities which can be seen between the H- and B-field rela-
tion and the rotor current and polrad voltage relation is the saturation at large
H-field/rotor current.
The relation between the rotor current and polrad angle are assumed to be lin-
ear. This contradicts the saturation, but in extension a saturated material will
create harmonics that will be included in the output via the fourth diode pair
thus increasing the output power. This diode pair is not accounted for in this
model but a saturated stator will give a overestimated rotor current since the sat-
uration is not accounted for in the model. A higher rotor current will then give
a higher output on the fundamental frequency, compensating for not including
the harmonics from the fourth diode pair.

3.2.7 System measurement

To be able to put the whole system together a final measurement was carried
out. The goal of this measurement was to confirm and complement the relations
between different signals described in chapter 2.
The measured signals in this system measurement are presented in table 3.2.
To get a measurement that follows a real life scenario the alternator was put on
full load and a speed of 3000RP M for 15 minutes before starting the measure-
ment. The measurement was then carried out for a large variation of loads and
3.2 Measurements 37

Measured signal
Rotor current
Rotor voltage
Stator current
Stator voltage
Output current
Output current ripple
Pulley speed
Diode voltage
Diode current

Table 3.2: Signals measured from system measurement.

speeds. Every measurement lasted for 10 seconds and between each measure-
ment the alternator was put at full load and 3000RP M for one minute. The
reason is to maintain the same working temperature throughout the whole mea-
surement.
The recorded signals from this measurement are at the end used together with
the measurement of the different losses to put the mathematical model together.
These measurements are especially important for calculating the relations that
are used in the equivalent circuit, section 2.5, and in the end completing the
model by connecting the different losses to the input signals.
38 3 Method

3.3 Optimization
The finished model of the alternator contains four input variables, nine constants
and relations between output current/voltage and stator current/voltage. The
input variables are engine speed and load and alternator current and voltage
as discussed in section 3.1.1. The nine constants of the model are presented in
table 3.3 where the hysteresis and eddy current are given by expressions instead
of coefficients since they are dependent on both speed and rotor current, these
expressions can be found in 4.A.

Parameter Meaning
Cf rict Friction coefficient
Cf an Fan coefficient
Rs Stator resistance
Rr Rotor resistance
Xq Inductance in q-axis (from section 2.5)
Xd Inductance in d-axis (from section 2.5)
Physt Hysteresis expression
Ped Eddy current expression
Udiode Voltage drop over diodes

Table 3.3: Constants from the alternator used in the model.

A possible change of constants and variables should then be investigated, this to

see if an increase in the efficiency can be achieved. When investigating this the
total amount of energy is important since enough energy to power the electric
system without draining the battery still has to be achieved.
The most simple parameter to alter is the gear ratio between the engine and the
alternator. Because this parameter will only impact the speed of the alternator.
In reality this parameter can be altered just by changing the size of the alterna-
tor pulley, this would not impact any other parameter, just the alternator speed.
Alternating other parameters in the alternator will lead to changes inside the al-
ternator which are harder to predict without simulations.
 4
Results

In this chapter the results from the different measurements and the resulting
model of the alternator are presented.

4.1 Results from measurements

4.1.1 Provided measurements
A script to handle the files with measurement data was implemented. The script
excludes the data that is not necessary for the model, for example battery current.
It also filters the measurement files so that incorrect data, mentioned in 3.2.1, is
excluded before using the model. An issue that emerged was the fact that some
data, for example vehicle speed, does not exists for all files. This made it hard
to calculate travelled distance and mean speed for all data files. This also made
some files useless.

Another problem with the provided data that impacted the model of the system
was lack of information on how the different variables were logged. For example
the engine torque was hard to interpret, this gave some unrealistic events in the
model, for example that all the buses were driving in a steep downhill more than
half the time, which can be considered as highly unlikely and with input from the
employee in charge of the measurements that could be confirmed to be wrong.

This made some further investigation about the data on torque necessary. After
some additional input from Scania employees it was concluded which curves of
the maximum momentum for each engine that were necessary to calculate the
actual torque at every moment.

39
40 4 Results

4.1.2 Rotor current

The result gained from the measurement of rotor current is in form of a table
with different rotor currents depending on the alternator load and speed. The
measurement data was then split into groups having the same load. Curves of
second degree were then fitted to the points. Figure 4.1 shows how the rotor
current depends on the alternator speed for a load of 50A. The shape of the
curve is similar to other loads.

Figure 4.1: Plots of the rotor current with respect to the alternator speed at
a load of 50A with measurement points.

As mentioned in section 3.2.2 the expressions are not very accurate because of
the lack of measuring points. The fact that the rotor current alternate somewhat
with time because of heating also makes this test unreliable.
A measurement containing more data points was carried out later. Since the rotor
current and the stator current have a clearer relation the rotor current is plotted
against the stator current at different speeds in this measurement. The resulting
curves fitted the measured data better. These curves can be seen in figure 4.2.
A problem with the temperature dependence of the wire resistance is that it is
hard to measure the temperature of the copper wire with the equipment avail-
able. The data presented in figure 4.1 represents a temperature around 20◦ Cel-
sius, which is not a temperature where the alternator normally operates. The al-
ternator is therefore heated up before the start of the measurement. This results
in a smaller rotor current since resistance is increasing with increasing temper-
ature. The rotor current illustrated in 4.2 is at working temperature, which is
about 120◦ Celsius.
4.1 Results from measurements 41

Figure 4.2: Rotor current with respect to stator current, the green curve rep-
resent the test made at 2500RP M, the blue 4000RP M and the red 6000RP M
after curve fitting in Matlab.

The rotor current given in figure 4.1 is measured with a multimeter. To get more
accurate values on the rotor current, the experiment in working temperature was
carried out with the Rotec system [24]. Another thing that has to be considered
in this case is the fact that the rotor current also depends on the output current
from the alternator. Since the two different experiments are carried out at differ-
ent temperatures the results from them are hard to compare. But more measure-
ment points, better equipment and being at working temperature makes the later
experiment more reliable.
The measurement of the rotor current carried out with the Rotec system con-
tained an offset in the measurement, this could easily be seen since the measured
rotor current was less than zero in several cases and larger than zero the rest of
the time. This is practically impossible since the rotor circuit contains a diode,
which only let the current flow in one direction.
Since Rotec works with voltage signals it provides more accurate measurements
on the rotor voltage. If the mean current is plotted against the mean voltage a
behaviour according to Ohm´s law can be seen in figure 4.3. This means that
the current increases with increasing voltage linearly with an incline equal to the
resistance, just as expected. If the measured data is plotted the result will be a
plot with a current different from zero for zero volts. This is due to the current
offset, the offset can then easily be read as the current at zero volts in figure 4.3.
The incline which equals the resistance, together with the rotor current cause an
electrical power loss according to Ohm´s law, eq. (2.9), and the power law, (2.20).
42 4 Results

Figure 4.3: The measured rotor current against the measured voltage. The
offset of the curve is also the offset of the current and the incline equals the
resistance.

4.1.3 Load angle

At first a computer with measuring cards from Dewetron [25] are used to mea-
sure the pulley rotations and the stator voltage. The pulley rotation are measured
with a magnetic sensor through an analog signal into the system. This signal ap-
peared to be unreliable at high speeds since it gave an angle that did not alternate
between 0 and 2π as expected of the pulley. The measurements of the stator volt-
age on the other hand were very accurate and a script in Matlab was written to
fit a Fourier series of chosen length to the measured data. This script enables an
analysis of the impact of different overtones of the stator voltage. However since
the signal from the pulley was not correct a comparison between the angle of the
pulley and the stator field could not be carried out. Also the Dewetron software
was hard to combine with Matlab which made it difficult and time consuming to
process the collected data.
Another approach with a lamp that flashes at the same frequency as the stator
field was then investigated. This experiment gave a distinct flashing behaviour
but since the measurement arc was hard to place at a precise manner and the
fact that small changes in the phase angle between the pulley and the stator field
quickly made the margins of error too great for this measurement to give reliable
data.
After this the same approach as before was tried but with a measurement equip-
ment from Rotec and a more precise sensor for measurement of the speed of the
pulley. The speed was measured through a digital input with the Rotec system.
It gave a better signal from the pulley without any interruptions in the middle
of a revolution. This enabled a correct measurement of the pulley angle. The
phase voltage, stator current and output current ripple was measured through
three analog channels. The software was then used to create a plot of how the
angle of the rotor changed with time.
4.1 Results from measurements 43

The plot from the measurement was then combined with the plot of the stator
voltage over time to get a resulting plot of how the stator voltage changes with
the angle of the pulley. A filter on the stator voltage was then used to get the first
order of the stator voltage, this means that the curve is represented by just one
sine signal instead of a combination of multiple sine signals.
To get conclusive data a trigger signal was used, which means that every mea-
surement starts at the same point and all the measured data can be compared
without any more synchronization. This trigger point was detected with another
magnetic sensor and a slit in the pulley, which enabled a trigger signal to start
the measurements.
The data for a number of loads and speeds was then exported to be plotted to-
gether in Matlab. This is illustrated in figure 4.4. In the figure it is easy to see
that the electric signals alternate six times for one lap of the mechanical pulley.
As mentioned in chapter 2, it is due to the fact that the alternator has six pole
pairs. Since the load angle is measured in the electrical signal which means that
an offset of 5◦ in the figure corresponds to a load angle of 30◦ .

Figure 4.4: Plots over how the first order of the stator voltage changes with
angle of the pulley for three different loads.

In figure 4.4 a number of things can be seen, such as the phase shift at different
loads. A change in amplitude can also be seen, which appears because of the
change in load angle which gives different impact of the two inductances. At
higher loads there are also greater influence of harmonics, meaning that filter-
ing out the fundamental frequency removes more of the actual signal for higher
loads.
44 4 Results

4.1.4 Inductances
When applying a direct current of 10A to the phase, a voltage of 401mV was
measured. This means that the resistance of the stator at this specific tempera-
ture is 40.1mΩel . The fact that the resistance changes with temperature is not
a problem in this measurement since the inductances are measured at the same
temperature as the resistance measurement. However the stator resistance has to
be modified with respect to the temperature. The inductances on the other hand
have the same magnitude at every temperature. The resistance dependence of
the temperature follows eq. (2.21) from section 2.6.3.
The resistance in the stator together with the stator current cause an additional
electric power loss according to Ohm´s law, eq. (2.9), and the power law, eq.
(2.20).
When a 10A alternating current was applied to a phase and the rotor was rotated
the maximum voltage gained was 339mV and the minimum voltage 236mV . This
gives the inductances Ld = 294µH and Lq = 177µH. Calculated with eq. (3.4),
where X corresponds to ωel L, just as mentioned in section 2.5.
These values are measured with a multimeter under the assumption that the in-
ductances only depends on the frequency of the signal. When comparing simu-
lations with these values to the collected data there was a great difference. This
difference is due to the fact that the inductances do not only depend on the fre-
quency.
The frequency dependence holds for an ordinary electrical circuit, but since the
inductances are used to replace the magnetic impact on the circuit they will have
the same dependence as the magnetic losses. This means that the inductances
will depend on the frequency as well as on the rotor current.
It is hard to create a test for only the two inductances mainly because of lack
of very specific equipment. The expressions that describe the inductances were
instead set according to eq.(4.1), where it is easily seen that the inductance de-
creases with increasing rotor current and frequency. In eq. (4.1) L0 is the initial
value of the inductance, ki decides the influence of the rotor current on the in-
ductance and kn the influence of the speed. This expression was assumed for
both inductances. A quadratic fit for three parameters on each inductance was
then created in Matlab. The measured inductances gave an indication of the ratio
between them and was used to make an initial assumption of the magnitude of
each parameter. This was done to be able to make the number of values for each
parameter smaller and thus lower the calculation time. The fitted curves and the
measurement points that were used for the fitting can be seen in figure 4.2. The
different curves describe how the inductances impact in the relation between ro-
tor current and stator current. All equations used in this fitting can be found in
section 4.A.

L = L0 (1 − ki Ir 2 − kn ωmech ) (4.1)
4.1 Results from measurements 45

In this test voltage drop over a diode was also measured. This since the voltage
drop cause an electric power loss according to the power law, eq. (2.20). Since
only the diode with highest voltage on each side will be conducting at any mo-
ment the output current can be used to calculate this power loss, together with
two times the voltage drop for the two diodes. Measured values of the voltage
and current gave a voltage drop over two diodes of 1.54V .

4.1.5 Mechanical losses

The mechanical resistance was investigated by collecting data of the pulley speed
depending on time when the alternator first was accelerated with the help of
another alternator as a synchronous machine and then released by disconnecting
them. The two kind of losses that act on the alternator in this case are friction and
fan losses, mentioned in section 2.6.1. The friction and the fan should impact the
resistance momentum with a constant and a parameter depending on the second
degree of the speed respectively. A curve of how the speed of the pulley changes
with time when only the mechanical losses act on the alternator can be seen in
figure 4.5. To reduce the order of the expression the acceleration with respect to
the square of the velocity is calculated. This gives a linear dependence. However
the data that was collected did not give a linear dependence for low speeds. But
a somewhat linear dependence for higher speeds.

Figure 4.5: The change of the pulley speed over time when only mechanical
losses are present.

The non linear dependence for low speeds can be explained with the fact that
the fan does not work properly for low speeds. This together with the fact that
the alternator has an idle speed of 1800RPM was the reason to exclude alternator
speeds below 1800RPM. The mechanical resistance depending on the square of
the velocity with low speeds as well as the error between the estimated value and
the measured value can be seen in figure 4.6. In figure 4.7 the same plots but
46 4 Results

without low speeds are given. In the second figure it can be seen that the fitted
blue curve fits the measured data better.

Figure 4.6: Plots of the mechanical resistance depending on the square of

velocity, low velocity included. The red curve represent the measured data
and the blue is the curve fitted to the data. Also the error between measured
and estimated data with respect to the square of velocity can be seen.

Figure 4.6 and figure 4.7 shows a clear improvement of the curve fitting when
low speeds are excluded. However there is still a non linear dependence for idle
speed. This leads to an error of about 6%. The error of the fitted curve compared
to the measured data can be also be seen in the figures. The error is defined as the
absolute value of the difference between the magnitude of the measured signal
and the fitted curve divided by the magnitude of the measured signal. This error
4.1 Results from measurements 47

Figure 4.7: Plots of the mechanical resistance depending on the square of

velocity, low velocity excluded. The red curve represent the measured data
and the blue is the curve fitted to the data. Also the error between measured
and estimated data with respect to the square of velocity can be seen.

is calculated in the last stage of calculating the losses and the error in this fitting
therefore corresponds to the error of the mechanical losses gained by using the
fitted values.

Since the friction momentum is independent of the speed the friction constant
can be seen in the point where the curve crosses the y-axis. The fan losses that
depend on the square of the speed have a coefficient that is decided from the
slope of the curve.
48 4 Results

To validate the script that was used to calculate the coefficients of the mechanical
resistance some artificial data was created from the theoretical losses. The con-
clusion from this test is that the script was correct since the test result was the
same as the expected result.
One reason for the different behaviour of the mechanical losses are stray losses.
The stray losses are due to leakage of magnetic flux, small remanent magnetic
fields etc. These losses will exist in all scenarios and will therefore be included in
the mechanical loss model as well [8, 11].
A problem that can emerge from this scenario is that running the alternator with
a rotor current in between the measurements can change the magnitude of the
remanent field.
Several tests were carried out to be able to compare the results and eliminate
unexpected errors in the measurement.
To verify the mechanical losses another test with a rotor where the fan had been
removed was carried out. This gave a constant mechanical resistance momen-
tum just as expected. The friction constant varied somewhat from the friction
constant obtained in the first mechanical resistance measurement. This can be
explained with the fact that two different rotors were used for the two tests and
that there can be some differences between them. Also removal of rotors from
the alternator wears out the layers and could also change the friction. Since the
two friction constants have the same order of magnitude a small deviation is not
a problem for the final result. This different friction constant could also depend
on a remanent field that will add some magnetic losses to the system.

4.1.6 Magnetic losses

As mentioned in section 2.6.2 the approach for measuring the magnetic losses is
the same as for the mechanical losses with the exception that the rotor current
was left on during the declutching. Several tests were carried out with different
rotor currents to get a model that covers the whole spectrum of losses including
both the speed and rotor current dependence.
The resistance with respect to the pulley velocity was calculated just as for the
mechanical losses for every different rotor current. The mechanical resistance
for every velocity was then subtracted from this resistance. This gives a number
of graphs representing only the magnetic resistance as a function of the pulley
velocity for different rotor currents. The hysteresis losses depend linearly on the
velocity, which means that the hysteresis coefficient will be of order zero in the
resistance momentum since P = ωmech M. The eddy current power losses depends
on the square of the velocity and the eddy current coefficient will thus be of order
one in the resistance momentum. All this according to section 2.6.2.
Curves were then fitted to these data sets. According to section 2.6.2 the fitted
curves should be linear dependent on the velocity. Figure 4.8 shows the magnetic
resistance depending on the pulley velocity for a rotor current of 0.8A.
4.1 Results from measurements 49

Figure 4.8: Plots of the mechanical resistance depending on the square of

velocity, low velocity excluded. The red curve represent the measured data
and the blue is the curve fitted to the data. Also the error between measured
and estimated data with respect to the square of velocity can be seen.

In figure 4.8 it can be seen that the measured data diverges from a linear de-
pendence for high velocities. This can be explained with the fact that high eddy
currents will lead to saturation in the iron. When the iron gets saturated the mag-
netic resistance will converge towards a constant resistance since the maximum
eddy current is reached. This is also the reason for the same behaviour for high
rotor currents as the B-field reaches its maximum strength.

The second graph in figure 4.8 shows the error between the fitted curve and the
50 4 Results

measured values for the magnetic losses. In this case the stray losses due to the
remanent field in the test of the mechanical losses will not impact these losses
because they were also eliminated when the mechanical losses were subtracted
from the total ones. However there could still be som stray losses due to a rema-
nent field if the remanent field has a different magnitude than for the test of the
mechanical losses.
When the coefficients of the magnetic losses are decided the same approach as
for the mechanical losses is used. In this case the constant will represent the
hysteresis losses and the slope the eddy current losses.
The magnetic losses were measured right after the measurements of the mechan-
ical losses, the rotor current were then slowly increased for each measurement
which gives a remanent field at a constant level. The rotor current was also low-
ered to get more measurement points, this lowering of the rotor current gives
a weaker magnetic field which will change the remanent field in the alternator.
There will also be some stray losses due to leakage. The leakage is depending on
the field strength and will therefore vary with varying rotor current [11].
A measurement of what the ratio of the Polrad voltage and the rpm over a time
period was carried out for several different rotor currents. The result is shown in
figure 4.9, in this figure the same saturation as for the core loss loop can be seen
and thus confirming the saturation in the alternator.
A final remark about the magnetic losses is that the test of them was carried out
on a cold alternator and the fact that eddy current is a resistive loss makes it tem-
perature dependent just as the electric losses. This means that the parameters for
eddy current in the model of the magnetic losses might vary somewhat compared
to a real scenario at working temperature. The problem with measuring the mag-
netic parameters at working temperature is the fact that the alternator does not
run as usual since the wires in the rotor are cut in the connection between the
two alternators, this setup also means that the current running through the al-
ternator where the parameters are measured have a rather low current flowing
through the stator since almost all the current is output current for the alternator
in the test bench. This makes it hard to do measurements on a alternator at work-
ing temperature. However the values measured on the cold alternator should still
give a good representation of the model [26].

4.1.7 System measurement

The system measurement that was carried out at working temperature was a
rather time consuming measurement since the alternator was kept at a quite con-
stant temperature by running the alternator at high load and speed between every
measurement. The results from this measurement did compensate the consumed
time since they described the signals in the alternator at working temperature.
The greatest changes from the case where the alternator is cold when the mea-
surement starts, is that the different currents are lower. The reason for this is that
with increasing temperature the resistance increases (described in section 2.6.3).
4.1 Results from measurements 51

Figure 4.9: Curve of how the ratio between the Polrad voltage and the rpm
relates to the rotor current, showing saturation.

At the same time the voltages in rotor and stator are independent of the temper-
ature. This together with Ohm’s law, eq. (2.9), leads to a smaller current in both
rotor and stator.
These values were then used to construct the relations between signals that were
needed to complete the model.
The model of the losses were also tested against these recorded data. This led to
the realization that the inductances depend on the speed and rotor current. The
inductances were then corrected, which is explained in section 4.1.4.
52 4 Results

4.2 Model of the system

4.2.1 Model of the alternator
The model of the system is divided into blocks as described in chapter 3. The
alternator is divided into three blocks where each block represent one type of loss.
One block represent the mechanical losses, the second represent the magnetic
losses and the third one the electric losses.
The mechanical losses only depend on the speed of the alternator. Since the speed
is one of the input variables in the system this block calculates the mechanical
losses directly from the speed. This block is constructed with the parameters of
the mechanical losses described in section 4.1.5.
The magnetic losses depends on the alternator speed and the rotor current. Since
the rotor current is not one of the input variables this has to be calculated from
the input variables, with the help of the results from 4.1.7. The rotor current is
controlled by the voltage regulator and together with the speed of the alternator
it decides the induced voltage, Polrad voltage, in the stator and further on the
stator current. Since it is hard to get exact measurements of the voltage regulator
another relation is used to decide the rotor current. The rotor current together
with the speed of the alternator induces the Polrad voltage in the stator, this re-
lation is instead used to get the rotor current from the Polrad voltage and the
alternator speed.
The voltage drop due to the magnetic losses together with the resistance of the
stator reduces the Polrad voltage to the measured stator voltage, Us . This means
that the Polrad voltage can be decided from the inductances, stator current, Is and
stator voltage, Us . These signals can in the next step be decided from the output
current. The equations that describes these relations can be found in section 4.A.
All this follows the theory described in section 2.5.
There is also a relation between the final stator voltage and the output voltage,
decided by the rectifier. To gain a better understanding of the relation between
output voltage and stator voltage a simulation in a program called Pspice [27] was
carried out. In Pspice a rectifier with a three phase voltage source in series with
an inductance and a resistance was created. The result from this test was that
the fundamental frequency of the stator voltage had a constant amplitude for
different loads and speeds. This can be explained with the fact that the output
voltage is constant and have a linear relation to the fundamental frequency of the
stator voltage [10].
There is also a relation between the stator current and the output current. This
relation was also simulated in Pspice. In the literature study theory of how a
three phase rectifier works and how the output and stator voltage and current are
related was gained. However this theory is limited to signals only containing the
fundamental frequency. Therefore the relation gathered from the literature study
was verified and complemented with a simulation in Pspice where overtones are
enabled [10].
4.2 Model of the system 53

After testing the rectifier in Pspice it was concluded that an assumption that the
calculations only have to be done with respect to the fundamental frequency is
valid for the model. This means that the RMS value of the rotor current is set
to 78% of the output current just as mentioned in section 2.3. Since the voltage
output is kept at a constant level this should also hold for the stator voltage.
This is seen in the Pspice experiment mentioned above as well as in a series of
measurements, where the amplitude of the fundamental frequency of the stator
voltage is constant at 19.7V . Since RMS values are used this means that the stator
voltage is set to 19.7
√ V.
2

If eq. (4.3) to (4.19), from 4.A, are put together the magnetic losses can be cal-
culated using the parameters for magnetic losses, the speed of the alternator and
the output voltage and current.
The electric losses are divided into three components, one component in the sta-
tor, one component over the diodes and one component in the rotor. The compo-
nent in the stator only depends on the stator current and can be calculated using
Ohm´s law, eq. (2.9), and the power equation, eq. (2.20).
The component that describes the losses over the diodes (in the rectifier) de-
scribes the power losses due to a voltage drop. These losses depends on the
current running through the diodes. But since the rectifier is only conductive
on two diodes at the time these losses can be calculated as the output current
running over two diodes. This means that these electrical losses are calculated by
multiplying two times the voltage drop with the output current according to the
power equation, eq. (2.20). The component of the electrical losses in the rotor
depends on the rotor current and the rotor resistance and is calculated exactly as
the electrical losses in the stator.
These three blocks are then summed up and constitutes the total losses of the
alternator.
The output power together with the alternator power losses and the alternator
speed gives the moment that is used to power the generator at any moment ac-
cording to eq. (4.2).

Pout + Plosses
M= (4.2)
ωmech

4.2.2 Model of the engine

The block that constitutes the engine model calculates the fuel consumption of
the vehicle and the torque generated by the engine. The torque model in the
provided measurements has been a problem since it is hard to know where the
torque is measured. After consulting a number of Scania employees it is clear that
the torque in the measurements is the torque generated by the flywheel. This is
the sum of the torque that the engine generates and the torque that the aggregates
use. Example of aggregates are the alternator and the AC system.
54 4 Results

4.3 Optimization
When the model of the system is finished it is applied on the collected data. This
to be able to see if small changes, for example changes in the gear ratio impacts
the alternator efficiency.
The present set up gives an efficiency that depends on the engine speed and the
load current according to figure 4.12. In this figure it can be seen that the highest
efficiency for the alternator is at an area around an engine speed of 800RP M
and a load current of 30A. This plot is for a gear ratio between the engine and
alternator of 3.6. A gear ratio of 3.6 is chosen since this is the gear ratio of the
alternator used in the measurements. If the model with this gear ratio is applied
to data from a field measurement in Stockholm the alternator uses 3.76% of the
fuel.
The area of where the alternator has its best efficiency in figure 4.12 is at a engine
speed close to idle. This is a profitable area since the buses run most of their time
at engine speeds in this area. The reason that the bus run most of its time at idle
is the many stops that a city bus does at every bus stop.
For comparison the same plots are calculated for gear ratios of 3 and 4, seen
in figure 4.13 and 4.14. To change the gear ratio between the engine and the
alternator is the same as changing the speed of the alternator, a higher gear ratio
gives a higher alternator speed and vice versa. If these gear ratios are applied on
the same data as before from the bus in Stockholm the alternator uses 3.75% of
the fuel for a gear ratio of 3, and 3.77% for a gear ratio of 4. A slight change in
the efficiency can also be seen in the figures.
The efficiency plots that can be seen in figure 4.12, 4.13 and 4.14 are calculated
from the model. The efficiency can easily be calculated as the ratio of the output
power and the input power where the output power follows the power equation,
eq. (2.20), and the input power is the output power and the power losses added.
This indicates that a small change of gear ratio does not impact the fuel consump-
tion of the alternator notable. To be able to see where the losses in the alternator
comes from and how they alter with changing gear ratio the losses of the alterna-
tor for the three different gear ratios is presented in the diagram in figure 4.10.
From this figure it can easily be seen that the biggest source of losses is the sta-
tor. Since the stator losses are resistive they depend only on the current in the
alternator and this is why they do not change with different gear ratios, since it
only changes the speed. The same is for the second biggest source, the diodes. If
the model would be able to calculate the cooling from the fan the resistive losses
should decrease somewhat for higher speeds since this makes the fan more effi-
cient. But to calculate the cooling from the fan more advanced simulations of the
air flow would be needed and this was not included in this diploma work.
The resistive losses in the rotor and the hysteresis losses will decay with increas-
ing speed. The reason for the decay of the rotor losses is that for a greater alterna-
4.3 Optimization 55

Figure 4.10: Diagram of the different losses in the alternator for different
gear ratios.

tor speed less rotor current is required to be able to achieve the right voltage, and
since the losses are resistive less current will give less losses. The decay of the
hysteresis is also due to the decrease of rotor current. An increasing speed of the
alternator will require less magnetic field. From section 2.2 it can be seen that an
increase of the speed will only require a magnetic field approximately inversely
proportional to the change. The reason that the required magnetic field is not
completely inversely proportional to the increase of speed is that an increase in
speed will increase the inductances in the material slightly and thus also increase
the required induced voltage slightly. And since the magnetic field impacts the
losses as a square (from section 2.6.2) and the speed only impacts them linearly
it will result in a decrease of the hysteresis losses, just as expected [11].

The eddy current losses are almost constant through the change of gear ratios. In
the expression in section 2.6.2 it can be seen that constant eddy current losses
can be explained by the same reasoning as for the hysteresis losses since the eddy
current losses are impacted by the square of the speed. The small rise of these
losses can be explained with the slight increase of the inductances mentioned for
the hysteresis.

The friction and fan losses increase with increasing alternator speed. This is also
a natural change since this can be seen directly from the equations presented in
section 2.6.1.

The load current at the best efficiency is rather low, as can be seen in the figures
4.12, 4.13 and 4.14. This current will not make it possible to recharge the battery
and run the electrical system, higher currents are needed for that. This is why
less efficient areas has to be considered. The reason that the alternator efficiency
56 4 Results

decreases at higher currents is due to the rapidly increasing magnetic and elec-
trical losses. The electrical losses are dependent of the square of the load current
and as can be seen in figure 4.10 they have a great impact on the losses and in
extension the efficiency.

The impact of a changed gear ratio is rather low when only changing load current,
this can be seen in figure 4.12, 4.13 and 4.14 where the efficiencies at maximum
load are similar in the different cases. The most obvious difference between the
three cases except the movement of the most efficient area is that for very high
speeds and very high currents a smaller ratio is beneficial. This due to the fact
that the alternator will have a lower speed compared to the cases with higher gear
ratios. This will in extension give less mechanical and hysteresis losses and thus
a better efficiency. If the cooling of the fan would be accounted for this would
impact the alternator in the form of lowered resistive losses at high speeds and
give another behaviour for higher speeds.

When the gear ratio is increased the efficiency curve is moved to the left, this
means that a part of the high efficiency is below engine speeds of 550RP M, which
is idle speed, in extension this means that the efficiency for higher speeds will de-
crease without gaining any efficiency for low speeds, which makes this alternative
a bit less efficient compared to the present gear ratio of 3.6.

If the gear ratio instead is lowered, this will as mentioned before increase the
efficiency for high loads and high speeds. But at the same time it will move the
most efficient area away from idle speed. Since idle speed is the most frequent
engine speed for these buses it is a priority to have a high efficiency at that area.

The conclusion that can be made from this investigation is that the gear ratio does
not impact the final fuel efficiency very much. The three presented gear ratios
changes the consumed fuel with plus or minus 0.01% and with measurement
errors and rounding of parameters this is a very small change. This means that
it would not change the fuel consumption of the alternator by changing the gear
ratio.

To get a comparison on what the alternator fuel consumption looks like for a bus
in a hot climate there are data for a bus running in Dubai. The expected changes
for this bus compared to the one driven in Stockholm is that the hotter environ-
ment would lead to more uses of the AC system, which in turn would lead to more
electricity required. This should increase the fuel consumption of the alternator.
At the same time the bus driven in Stockholm had a rather high average speed of
approximately 30km/h. The bus that have been driven in Dubai may have a lower
average speed since city buses normally have an average speed of about 15km/h,
which would give lower fuel consumption of the engine. An increase of the fuel
used by the alternator and a decrease of the fuel used by the engine would make
the alternator a larger part of the total fuel consumption which in turn would
give small changes in the alternator a bigger impact on the whole system. The re-
sults from running these data in the model is that the alternator uses 5.6% of the
fuel in the Dubai case. Just as expected the average speed for this measurement
4.3 Optimization 57

Figure 4.11: Distribution of losses for bus driven in Dubai.

is 16.7km/h. This contributes to less fuel used per hour. The electricity produced
by the alternator is in contrast to the expectation almost the same as for the bus
in Stockholm. This can be explained by use of heating equipment for the bus
in Stockholm during the winter which also leads to an increase in the electricity
consumption. The total increase in the percentage fuel used by the alternator can
thus be explained by the decrease of the average speed.
The distribution of the losses in this case can be seen in figure 4.11. In the dia-
gram it can be seen that the stator losses are at the same level as for the bus in
Stockholm. This holds also for the diode losses. This is due to a similar fuel con-
sumption. The low gear ratio gives similar values for the other losses as for the
Stockholm case with gear ratio 3, which is expected since both power production
and time at idle speed is similar in the two cases.
58 4 Results

Figure 4.12: Efficiency of the alternator at a gear ratio of 3.6, calculated from
model, the black curve illustrates the maximum load with respect to the
speed.

Figure 4.13: Efficiency of the alternator at a gear ratio of 3, calculated from

model, the black curve illustrates the maximum load with respect to the
speed.
4.3 Optimization 59

Figure 4.14: Efficiency of the alternator at a gear ratio of 4, calculated from

model, the black curve illustrates the maximum load with respect to the
speed.
 Appendix

4.A Relations used in the alternator block

In this appendix part all the equations and relations used to build the mathemat-
ical model are presented.

bP = U
U bs + Rsb
I s + jb
I q X q + jb
Id Xd (4.3)

Iq = b
b Is sin(ϑ) (4.4)

Id = b
b Is cos(ϑ) (4.5)

 Xq Is 
ϑ = arctan (4.6)
Us + Rs Is

Up
Ir = 0.2143 − 0.2264 (4.7)
RP M/1000

19.7
Us = √ (4.8)
2

Is = 0.78Iout (4.9)

Xd = Ld0 (1 − kid Is 2 − knd ωmech )ωmech (4.10)

60
4.A Relations used in the alternator block 61

Variable Meaning
U
bp Polrad voltage
Ubs Voltage over stator
Rs Resistance in stator
Is
b Current trough stator
bIq Current through q inductance
bId Current through d inductance
Xq Inductance along q axis
Xd Inductance along d axis
ϑ Load angle
Ir Rotor current
RP M Speed of the alternator
Iout Output current from the alternator
ωmech Angular speed of the alternator
ωel Angular speed of the voltage in the alternator
f Frequency of alternator
Ld0 Initial value of d inductance
kid Parameter for current influence in d inductance
knd Parameter for speed influence in d inductance
Lq0 Initial value of q inductance
kiq Parameter for current influence in q inductance
knq Parameter for speed influence in q inductance
α Temperature coefficient of material
T Temperature of winding
Pf rict Power losses due to friciton
Pf an Power losses due to the fan
Physt Power losses due to hysteresis
Ped Power losses due to eddy currents
Pr Power losses due to resistance in rotor
Ps Power losses due to resistance in stator
Pdiode Power losses due to voltage drop in diodes

Table 4.1: Variables in the complex vector diagram.

Xq = Lq0 (1 − kiq Is 2 − knq ωmech )ωmech (4.11)

Rs = 40.1 ∗ 10−3 (1 + α(T − 20)) (4.12)

Pf rict = 0.08580ωmech (4.13)

62 4 Results

Pf an = 2.246 ∗ 10−7 ωmech 2 (4.14)

Physt = (1.055Ir 2 + 0.4647)f (4.15)

Ped = (0.006171Ir 2 + 0.001339)f 2 (4.16)

Protor = 0.08Ir 2 (4.17)

Pstator = 3Rs Is 2 (4.18)

Pdiodes = 1.54Iout (4.19)

 Conclusion
5
In this master thesis the losses of alternators are investigated. Chapter 1 con-
tains a background to the thesis and a short introduction in combustion engines
and some examples of what research concerning alternators that are done today.
Chapter 2 contains a thorough theoretical background to alternators, step by step.
Chapter 3 focuses on the method of the thesis, the main parts of this chapter are
the measurements carried out during the thesis to validate and complement the
theory. In chapter 4 the results from the different measurements are presented
together with a mathematical model of an alternator. At the end an optimization
of the gear ratio between the engine and the alternator is also investigated. The
final model of the alternator is gathered from several static measurements and
put together.
For future work in this area a suggestion would be to expand the model to handle
dynamic behaviour, this to investigate how the alternator reacts to changes. An-
other suggestion is to perform a simulation on what impact permanent magnets
in the rotor will have on the total efficiency. This was a part of this thesis at the
beginning, but due to time consuming measurements and complex events there
was not enough time to complete that part.
Further investigations of parameters in the model could also be carried out, one
problem is that most physical changes on the alternator affects many of the pa-
rameters and without further measurements it would be hard to know which
parameters that would change and in which way. A suggestion of a property to
consider would be the thickness of the wires in the windings.
Another thing that would be worth investigating later on is how the fan cools
the windings. To model the temperature changes for different speeds, loads and
surrounding temperatures would be useful since the temperature has a rather

63
64 5 Conclusion

large impact on the resistive losses. To do this a computational fluid dynamics

simulation is probably needed to understand how well the fan cools the windings
for different cases. A more detailed knowledge of all geometries in the alternator
is probably also needed for this since the air flow between windings is relevant
for the cooling.
The final result of this thesis is the mathematical model of an alternator, the
equations behind this model are presented in section 4.A. This model fits well
to measured efficiencies. The benefit from this model compared to information
given to Scania from suppliers is that it handles the process between input and
output, which gives more knowledge of how the alternator works and also give
an indication of what changes could improve the alternator.
Appendix
 A Details

A.1 Circuit diagram for test of mechanical and

magnetic losses
Figure A.1 shows the circuit that was used during the measurement of mechan-
ical and magnetic losses. The numbers on the alternators represent the three
phases and the star represent the star point.
On alternator 2 there are also an input for the rotor current where a direct current
can be applied by a current source. Finally there are two sensors that measure
the alternator pulley. One of the sensors measures the speed of the pulley and
the other one measures a trigger condition.
From the alternator circuit seven parameters are monitored with a measurement
system. The measurement system uses two digital speed channels that record
the pulley speed and the trigger condition. It also uses five analog channels that
records the current in one phase, the voltage in each phase and finally the rotor
current. The analog channels are marked 1-5 and the digital channels are marked
S1 and S2.
The circuit also contains a switch for each phase connection. These switches are
manually closed by running a current over a relay when the two alternators are
in phase. The lamps connected att each side of the switches are used to decide if
the alternators are in phase. This is explained in the method part.

67
68 A Details

Alternator 1
1

3
A
1

V
2

V
3
Measurement System

V
4
5
S1
S2

3
+IR
A

Alternator 2
−IR
T rigger
RP M

Figure A.1: Circuit diagram for measurement of mechanical and magnetic

losses.
 Bibliography

[1] F. Lezsovits A. Penninger et al. Internal combustion engines. Lecture notes,

Technical University of Budapest, 1(1):1–86, 1995. Cited on page 3.
[2] F. Salazar. Internal combustion engines. Lecture notes, University of Notre
Dame, 1(1):1–87, 1998. Cited on page 3.
[3] A. Masmoudi A. Ibala, R. Rebhi. Mec-based modelling of claw pole ma-
chines: Application to automotive and wind generating systems. Interna-
tional Journal of Renewable Energy Reaserch, 1(1):1–8, 2011. Cited on page
3.
[4] S.K. Tidblad Lundmark. Application of 3-D Computation of Magnetic
Fields to the Design of Claw Pole Motors. PhD thesis, CHALMERS UNI-
VERSITY OF TECHNOLOGY, Division of Electric Power Engineering De-
partment of Energy and Environment Chalmers University of Technology
SE-412 96 Göteborg Sweden, 2005. Cited on page 3.
[5] Avo Reinap. Design of Powder Core Motors. PhD thesis, Lund University,
Department of Industrial Electrical Engineering and Automation Lund Uni-
versity Box 118 SE-221 00 Lund Sweden, 2004. Cited on page 3.
[6] David K. Cheng. Field and Wave Electromagnetics. Addison Wesley, 2 edi-
tion, 1989. Cited on pages 6, 7, and 20.
[7] J. Österman C. Nordling. Physics Handbook for science and engineering.
Studentlitteratur AB, 8:4 edition, 2006. Cited on pages 6, 9, 19, 22, 24,
and 25.
[8] Mike Bradfield. Improving alternator efficiency measurably reduces fuel
costs. Delco Remy, 1(1):1–32, 2008. Cited on pages 11, 19, 21, 22, and 48.
[9] Rainer Meyer, Braun and Weinmann. Alternators and starter motors. The
Bosch Yellow Jackets, 1(1):4–46, 2003. Cited on pages 12 and 22.
[10] P. Pejovic. Three-Phase Diode Rectifiers with Low Harmonics. Springer,
third edition, 2007. Cited on pages 13, 14, 17, and 52.

69
70 Bibliography

[11] Charles A. Gross. Electric Machines. CRC Press, 1 edition, 2007. Cited on
pages 15, 17, 19, 21, 36, 48, 50, and 55.

[12] John Chiasson. Modelling and High-Performance Control of Electric Ma-

chines. CRC Press, 1 edition, 2007. Cited on pages 15 and 17.

[13] Thomas Wu. dq theory for synchronous machine with damper winding,
2014. URL http://www.eecs.ucf.edu/~tomwu/course/eel6208/
notes/15dqTheory.pdf. Cited on page 16.

[14] R. Fischer. Elektriche Maschinen. Hanser, 12 edition, 2004. Cited on page

17.

[15] D. Sumina T. Idzotic, G. Erceg. Load angle estimation of a synchronous

generator. IEEE, 1(1):1–4, 2000. Cited on pages 17 and 30.

[16] H. Späth. Wissenschaft technik. G.Braun Karlsruhe, 2 edition, 1986. Cited

on page 17.

[17] J. Soulard A. Krings. Overview and comparison of iron loss models for elec-
trical machines. Journal of Electrical Engineering, ISSN 1582-4594, 10(3):
162–169, 2010. Cited on pages 19 and 21.

[18] G. Henneberger S. Küppers. Numerical procedure for the calculation and

design of automotive alternators. IEEE Transactions on magnetics, 33(2):
2022–2025, 1997. Cited on page 19.

[19] TDK lambda. Clossary of terms, 2014. URL http://se.tdk-lambda.

com/public/kb_glossary_results.aspx?index=H. Cited on page
20.

[20] K. Reif. Automotive Handbook. BentleyPublisher, 8 edition, 2011. Cited on

page 22.

[21] G. Friedrich A. Gimeno. Application of the vda standard for the comparison
and losses reduction of high efficiency car alternators. IEEE, 1(1):1–6, 2004.
Cited on page 22.

[22] J. Cros R. Ivancovic et al. Power Electronic solutions to Improve the Perfor-
mance of Lundell Automotive Alternators. Intech, 1 edition, 2012. Cited on
page 22.

[23] Mathworks. Matlab r2012b, 2014. URL http://www.mathworks.se/.

Cited on page 23.

[24] Vispiron. Rotecfe08, 2014. URL http://www.vispiron.de/en/

business-units/measuring/measurement-systems/. Cited on
pages 23, 28, 30, and 41.

[25] Dewetron. Dewetron measurement system, 2014. URL http://www.

dewetron.com/int/products/. Cited on pages 29 and 42.
Bibliography 71

[26] G. Friedrich A. Gimeno. Experimental and numerical evaluation of iron

losses in a claw pole car generator. XIX International Conference on Electri-
cal Machines, 1(1):1–6, 2010. Cited on page 50.
[27] Electronic Labs. Pspice 9.1, 2014. URL http://www.electronics-lab.
com/downloads/schematic/013/. Cited on page 52.
72 Bibliography
Upphovsrätt
Detta dokument hålls tillgängligt på Internet — eller dess framtida ersättare —
under 25 år från publiceringsdatum under förutsättning att inga extraordinära
omständigheter uppstår.
Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner,
skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för icke-
kommersiell forskning och för undervisning. Överföring av upphovsrätten vid
en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av
dokumentet kräver upphovsmannens medgivande. För att garantera äktheten,
säkerheten och tillgängligheten finns det lösningar av teknisk och administrativ
art.
Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman
i den omfattning som god sed kräver vid användning av dokumentet på ovan
beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan
form eller i sådant sammanhang som är kränkande för upphovsmannens litterära
eller konstnärliga anseende eller egenart.
För ytterligare information om Linköping University Electronic Press se förla-
gets hemsida http://www.ep.liu.se/

Copyright
The publishers will keep this document online on the Internet — or its possi-
ble replacement — for a period of 25 years from the date of publication barring
exceptional circumstances.
The online availability of the document implies a permanent permission for
anyone to read, to download, to print out single copies for his/her own use and
to use it unchanged for any non-commercial research and educational purpose.
Subsequent transfers of copyright cannot revoke this permission. All other uses
of the document are conditional on the consent of the copyright owner. The
publisher has taken technical and administrative measures to assure authenticity,
security and accessibility.
According to intellectual property law the author has the right to be men-
tioned when his/her work is accessed as described above and to be protected
against infringement.
For additional information about the Linköping University Electronic Press
and its procedures for publication and for assurance of document integrity, please
refer to its www home page: http://www.ep.liu.se/

© Markus Örn