You are on page 1of 88

On Inter-bar Currents in Induction Motors with Cast

Aluminium and Cast Copper Rotors

ALEXANDER STENING

Licentiate Thesis
Stockholm, Sweden 2010
Electrical Machines and Power Electronics
TRITA-EE 2010:027 School of Electrical Engineering, KTH
ISSN 1653-5146 SE-100 44 Stockholm
ISBN 978-91-7415-682-9 SWEDEN

Akademisk avhandling som med tillstånd av Kungl Tekniska högskolan framlägges


till offentlig granskning för avläggande av teknologie licentiatexamen tisdagen den
15 Juni 2010 klockan 10.00 i E2, Kungl Tekniska högskolan, Lindstedtsvägen 3,
Stockholm.

© Alexander Stening, May 2010

Tryck: Universitetsservice US AB
iii

Abstract

This thesis presents a study of the effects of inter-bar currents on induction


motor starting performance and stray-load losses. The work is focused on the
performance differences between aluminium and copper casted rotors.
A method to predict the stator current when starting direct-on-line is
developed. This includes modelling of skin-effect, saturation of the leakage
flux paths and additional iron losses. The results are verified by measure-
ments. An analytical model accounting for inter-bar currents is derived, and
the dependency of the harmonic rotor currents on the inter-bar resistivity is
investigated. It is found that the inter-bar currents can have considerable
effect on motor starting performance and stray-load losses, the amount being
strongly dependent on the harmonic content of the primary MMF.
Based on measurements of inter-bar resistivity, the starting performance
of an aluminium and a copper casted rotor is simulated. The results indicate
a higher pull-out torque of the aluminium rotor than for the equivalent copper
rotor. This is rather due to an increase of the fundamental starting torque of
the aluminium rotor, than due to braking torques from the space harmonics
in the copper rotor. The results are verified by measurements. It is found
that the difference between the pull-out torques is even larger than calculated
from the model. Thereby, it can be concluded that the inter-bar currents have
a considerable effect on motor starting performance.
At rated speed the braking torques are larger in the aluminium rotor than
in the copper rotor. This is seen as increased harmonic joule losses in the ro-
tor cage. Simulations have shown, that these losses can be as large as 1% of
the output power for the studied machine.

Keywords: Induction motors, Inter-bar currents, Copper rotors, Aluminium


rotors, Starting torque, Asynchronous torques, Starting current, Stray losses.
v

Sammanfattning

Denna licentiatavhandling presenterar en studie av tvärströmmars påver-


kan på startegenskaper och tillsatsförluster för asynkronmaskiner med gjutna
aluminium- och kopparrotorer.
En metod för estimering av startströmmen i asynkronmaskiner vid direk-
tstart mot nätet utvecklas. Metoden inkluderar strömförträngning, järnmät-
tning av läckflödesvägar samt järnförluster på grund av läckflöden. Resultaten
verifieras med mätningar. En analytisk modell för beräkning av tvärström-
mar härleds, med vilken beroendet av rotorns övertonsströmmar på kontak-
tresistivitet mellan rotorledare och rotorplåt utreds. Simuleringar visar att
tvärströmmar kan ha stor inverkan på asynkronmaskinens startmoment och
dess tillsatsförluster. Effekten av tvärströmmars inverkan är direkt kopplad
till övervågsinnehållet i den av statorlindningen skapade MMK:n.
Baserat på mätningar av kontaktresistivitet mellan rotorledare och rotor-
plåt, beräknas startprestanda för en gjuten aluminium- respektive koppar-
rotor. Resultaten indikerar att aluminiumrotorn har ett högre kippmoment
än motsvarande kopparrotor. Enligt simuleringar beror detta mer på ett
ökat grundtonsmoment i aluminiumrotorn än på ett reducerat totalmoment
i kopparrotorn. Mätningar visar att denna skillnad existerar och att den
dessutom är större än beräknat från modellen. Det kan således konstateras
att tvärströmmar har en betydande effekt på asynkronmaskinens startegen-
skaper.
Moment av högre ordning än grundtonsmomentet skapar vid nominell
drift ett resulterande bromsande moment, vilket visar sig vara större i alu-
miniumrotorn än i kopparrotorn. Detta leder till en ökning av högfrekventa
resistiva förluster i rotorkretsen. Simuleringar av den studerade asynkron-
maskinen visar att dessa förluster kan vara så stora som 1% av märkeffekten.

Sökord: Asynkronmotorer, Tvärströmmar, Kopparrotorer, Aluminiumro-


torer, Startmoment, Asynkrona moment, Startström, Tillsatsförluster.
Acknowledgements

This work has been carried out within the High Performance Drives program of the
Center of Excellence in Electric Power Engineering at the department of Electrical
Machines and Power Electronics. Since the start of this project, several people have
been involved and have contributed to this thesis in different ways; I am grateful
to them all.
First of all, I would like to thank my supervisor Prof. Chandur Sadarangani for
his help throughout this project and for sharing his knowledge in our conversations.
I am also grateful to Assoc. Prof. Juliette Soulard, for being available and inspiring
me whenever I needed it. I would like to thank the personnel at ABB LV Motors
and ITT Flygt for giving me a rewarding stay outside KTH. A special thanks goes
to Bo Malmros and Jörgen Engström, for their useful inputs and for helping me
with supplies of prototype motors.
I would like to thank Jan Timmerman and Olle Brännvall, for always helping
me to find the best solutions to my problems in the laboratory, and also for the
nice moments we have had discussing hunting and boating. Further, I would like
to thank Dr Stephan Meier for his help during the time he worked in the labora-
tory. Thanks to Prof. Hans-Peter Nee for reading the thesis and for his valuable
comments.
I am very grateful to all employees at EME, for contributing to a pleasant
atmosphere. In the same way I would like to thank the former employees at EME,
for keeping up the good spirit with different kinds of Roebel-activities. A special
thanks to my colleges and friends, Henrik Grop and Dmitry Svechkarenko, for the
nice times we have had and for the times to come. To my former college Rathna
Chitroju, I would like to say that I am glad you are still in Sweden, thanks for
the joy you are bringing. Further, a thank goes to my office room mate Alija
Cosic, with whom I have shared many laughs. Thanks to Eva Pettersson and Peter
Lönn for helping me with administration and computers, things I sometimes do not
understand.
Finally, I would like to express my deepest gratitude to my family, for their
support and understanding. This truly means a lot to me. To my beloved cohabitee
Ida Axelsson; thanks for your patience during the late hours when I have been
working with this thesis, you mean everything to me.
Alexander Stening
Stockholm, May 2010

vii
Contents

Contents ix

1 Introduction 1
1.1 The need of accurate induction motor models . . . . . . . . . . . . . 1
1.1.1 Improved efficiency . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Starting performance . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Die cast aluminium and copper rotors . . . . . . . . . . . . . . . . . 3
1.3 Rotor skewing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Scientific contribution . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.6 Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.7 Studied motors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.8 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Measurements of inter-bar resistance 7


2.1 Test setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Modelling of the rotor . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Results from measurements . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3 Model for the analysis of inter-bar currents 17


3.1 Rotor circuit taking inter-bar currents into account . . . . . . . . . . 17
3.2 Stator flux linkage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2.1 Airgap flux density due to stator current . . . . . . . . . . . . 20
3.2.2 Stator flux linked by the rotor circuit . . . . . . . . . . . . . 24
3.3 Rotor flux linkage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3.1 Airgap flux density due to rotor current . . . . . . . . . . . . 26
3.3.2 Flux caused by the phase belt harmonics . . . . . . . . . . . 28
3.3.3 Flux caused by slot harmonics . . . . . . . . . . . . . . . . . 29
3.4 General set of equations . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.5 Effects of a finite inter-bar resistance on rotor current distribution . 31
3.5.1 Rotor without skew . . . . . . . . . . . . . . . . . . . . . . . 32
3.5.2 Rotor with skew . . . . . . . . . . . . . . . . . . . . . . . . . 36

ix
x CONTENTS

3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4 Effects during a direct-on-line start 41


4.1 Skin effect in the rotor bars . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.1 Numerical method used to account for skin effect . . . . . . . 42
4.1.2 Verification with FEM . . . . . . . . . . . . . . . . . . . . . . 45
4.2 Saturation of the leakage paths . . . . . . . . . . . . . . . . . . . . . 48
4.2.1 Model used to account for saturation . . . . . . . . . . . . . . 49
4.2.2 Iron losses due to leakage flux . . . . . . . . . . . . . . . . . . 52
4.3 Rotor losses and starting torque . . . . . . . . . . . . . . . . . . . . . 52
4.3.1 Rotor losses during a start . . . . . . . . . . . . . . . . . . . . 52
4.3.2 Effects of a finite inter-bar resistance on starting torque . . . 53
4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5 Simulation results and measurements 59


5.1 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.1.1 Starting torque . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.1.2 Rotor losses at rated speed . . . . . . . . . . . . . . . . . . . 63
5.2 Measurements of starting torque . . . . . . . . . . . . . . . . . . . . 64
5.2.1 Measurement setup . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2.2 Results from measurements . . . . . . . . . . . . . . . . . . . 66
5.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

6 Conclusions and Future work 71


6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.2 Future work guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . 72

Bibliography 73

List of Figures 76
Chapter 1

Introduction

Electrical machines are widely used for high efficient conversion between electrical
and mechanical energy. Ever since the induction motor was invented in 1886, it
has been serving the ever growing industry. Due to the simplicity and the robust
design of the induction motor, it has become the most commonly used electrical
machine.

1.1 The need of accurate induction motor models


Despite the simple working principle, the designer requires extensive knowledge of
the induction motor, in order to construct an efficient motor. During the years, the
induction motor designs have been refined, which increases the need for accurate
motor models.

1.1.1 Improved efficiency


In order to reduce the consumption of electricity, the manufacturing of high efficient
motors has become a topic of current interest. New efficiency standards make the
induction motor design even more challenging. This requires not only the mini-
mization of the well known stator and rotor copper losses, iron losses and friction
losses, but also the reduction of the additional losses. These losses are defined as
the additional losses that occur in the machine over the normal losses that are con-
sidered in usual induction motor performance calculations.

At rated load the additional losses are referred to as stray-load losses. For small
to medium sized induction motors these losses vary typically within the range 0,5%
- 3% of the motor input power [1]. Measurements have, however, shown that these
losses can be even larger [2]. Measuring these losses with a reasonable accuracy
is a difficult task. As there are different efficiency standards defining this measur-
ing procedure, the amount of stray-load losses depend on the standard used [3, 4].

1
2 CHAPTER 1. INTRODUCTION

Investigations of the different stray-load loss components have been performed,


among others by Nishizawa in [2]. Figure 1.1 shows the obtained stray-load loss
components for a set of small to medium-size induction motors.

3%
10%

40%
17%

Leakage flux losses


High-frequency losses

Pulsation losses
Inter-bar current losses
30% Surface losses

Figure 1.1: Stray-load loss components (0,2-37 kW induction motors) [2].

According to this study, the largest portion of the stray-load losses is composed
of surface losses caused by high frequency flux. However, these losses can be suffi-
ciently suppressed by the use of non-machined rotors [2], reducing the eddy currents
at the rotor surface. The second largest portion of the stray-load losses, according
to [2], are the losses caused by inter-bar currents. These currents, flowing between
the rotor bars through the iron lamination, can cause considerable losses in the in-
duction motor unless the rotor bars are insulated [5]. These losses, primary caused
by the stator slot harmonics, result in increased torque dips during a direct-on-line
start, which can reduce the pull-out torque [6]. Therefore, the torque-speed curve
contains information regarding these losses.

1.1.2 Starting performance


The starting characteristic of the induction motor is an important design factor.
For the motor to start, the starting torque must be larger than the load torque.
Furthermore, to ensure a reasonable margin of overload capability, it is usual to
require that an induction motor is able to deliver momentarily at least twice its
rated torque at rated voltage [7]. In traction applications, the pull-out torque is
1.2. DIE CAST ALUMINIUM AND COPPER ROTORS 3

an important design criteria since the speed range of the motor is limited by its
pull-out torque [8]. This reinforces the need of inter-bar current models.

1.2 Die cast aluminium and copper rotors


Large motors are manufactured with fabricated aluminium or copper bar rotors.
The prefabricated bars are inserted into rotor slots that are punched around the
periphery of the rotor lamination and the short circuit rings are welded or brazed to
the bars. However, this choice is unattractive for small to medium sized motors due
to cost reasons. These machines are generally equipped with die cast aluminium
rotors. The casting process results in a low resistive path between the rotor bars
and the iron core. This is referred to as inter-bar resistance. Resistance, however,
is not an appropriate unit to use for this contact region as it depends on the stack
length. Therefore, this resistance is usually multiplied with the stack length, defin-
ing the inter-bar resistivity.

Due to advancements in casting technology, it is possible to manufacture die cast


copper rotors. Thanks to the higher conductivity of copper, the motor efficiency
can be increased. Measurements have shown, however that the inter-bar resistivity
in copper rotors can be as much as 10 times lower than in casted aluminium rotors
[9], this promotes the flow of inter-bar currents.

1.3 Rotor skewing


The stator, creating the primary MMF, is usually equipped with semi-closed slots.
These openings create a non-uniform distribution of the air-gap permeance, dis-
torting the fundamental MMF. This give rise to airgap space harmonics referred
to as slot harmonics, the order depending on the number of stator slots. In an
unskewed rotor, these harmonics induce high frequency currents in the rotor cage,
the resulting cage losses can form a considerable part of the stray losses [10].
If the rotor is skewed by one stator slot pitch and the rotor bars are insulated,
these currents are efficiently suppressed, improving the motor efficiency. However,
by the introduction of casted rotors, the inter-bar resistivity being low, inter-bar
currents start to flow. The magnitude of these currents are highly dependent on
rotor skew and inter-bar resistivity.

1.4 Objectives
The objective of this thesis is to study the effects of inter-bar currents on aluminium
and copper casted rotors. This is achieved by developing analytical models to
simulate the starting performance and to calculate additional rotor losses. The
models should be verified by measurements. The objectives can be summarized as
follows:
4 CHAPTER 1. INTRODUCTION

• Measure the inter-bar resistance on a set of aluminium and copper rotors.

• Develop a computer program for the calculation of the effects of inter-bar


currents on motor performance.

• Develop an analytical model that that can account for saturation of the leak-
age paths during a direct-on-line start.

• Verify the analytical models by measuring the starting performance of the


induction motor equipped with either an aluminium or a copper rotor.

1.5 Scientific contribution


This work has resulted in the following contributions:

• Measurements have shown that the inter-bar resistivity in the studied casted
rotors is lower in the copper rotors than in the aluminum rotors.

• Measurements have shown results indicating an uneven distribution of the


inter-bar resistivity in the studied aluminium rotors, while in the copper ro-
tors, the inter-bar resistivity is evenly distributed.

• A numerical method to account for skin-effect has been verified by finite


element simulations.

• A method to estimate the starting current of induction motors has been de-
veloped and verified by measurements.

• An analytical model to predict the effects of inter-bar currents on starting


performance has been verified by measurements.

• The starting torques of one aluminium- and one copper rotor skewed by one
stator slot pitch have been measured. The results show that the pull-out
torque is lower for the copper rotor than for the equivalent aluminum rotor.
This is verified by simulations.

1.6 Publications
The work presented in this thesis has resulted in two international conference papers
listed below:

• A. Stening and C. Sadarangani, The effects of inter-bar currents in cast alu-


minium and cast copper rotors, In Proc. International Conference on Elec-
trical Machines, Vilamoura, Portugal, September 2008.
1.7. STUDIED MOTORS 5

• A. Stening and C. Sadarangani, Starting performance of induction motors


with cast aluminium and copper rotors including the effects of saturation and
inter-bar currents, In Proc. International Conference on Electrical Machines
and Systems, Tokyo, Japan, November 2009.

1.7 Studied motors


In this thesis the models developed are used to simulate the performance of two
different motors, referred to as Motor A and Motor B.

Motor A 11 kw, 4-pole, 36 stator slots and 44 rotor slots with aluminum casted
bars.

Motor B 11 kW, 4-pole, 36 stator slots and 28 rotor slots with both aluminium
and copper casted bars.

Measurements have been performed on Motor B, this motor is therefore used to


verify the analytical models. The aluminum and the copper rotors have the same
geometry, except for a small difference in the short-circuit ring design. The same
stator is used when measuring the performance of the two rotor concepts.
Motor A is used for analytical studies to demonstrate the dependency of the
inter-bar resistivity on motor performance.

1.8 Outline of the thesis


Chapter 1: This chapter introduces the thesis, gives a brief introduction to the
topic and presents the objectives.

Chapter 2: A method for measurements of inter-bar resistance is presented. An


equivalent circuit of the rotor is used to calculate the resistivity from measured
voltages. Results are presented from the measurements on five different rotors.

Chapter 3: In this chapter the main model used to account for inter-bar currents
is derived. Motor A is simulated, results are presented showing the distribution of
the inter-bar currents in a skewed and an unskewed rotor at different inter-bar re-
sistivities.

Chapter 4: Models used to account for inter-bar effects during a direct-on-line


start are presented. A numerical method to account for skin effect is verified with
FEM-simulations. A method to include saturation of the leakage flux paths is de-
rived. The rotor losses and the produced torque is derived from the rotor currents.
Simulation results are presented showing the dependency of the starting torque on
the inter-bar resistivity for Motor A.
6 CHAPTER 1. INTRODUCTION

Chapter 5: This chapter highlights the differences in starting performance be-


tween the aluminum and the copper rotor used in Motor B. The additional losses
created in the rotor cage at rated speed are simulated as a function of inter-bar
resistivity. The simulated starting performance is verified by measurements.

Chapter 6: This chapter concludes the thesis, the results are summarized and
suggested future work is presented.
Chapter 2

Measurements of inter-bar resistance

The casting process results in a distributed low resistive path between the rotor
cage and core. To determine this resistance accurately is a difficult task. It is not
possible to measure the inter-bar resistivity directly; it has to be calculated from
measurements. The methods known can be categorized as non-destructive and de-
structive methods.

In 1958, Odok presented a method to measure inter-bar resistance on casted


rotors [6]. A direct current is fed into one short-circuit ring and taken out through
the shaft on the opposite side. The voltage drop between the ring and iron core is
measured along the axial direction. Based on the average value of this voltage, the
inter-bar resistivity is calculated. Odok also came to the important conclusion that
the inter-bar impedance can be assumed to be purely resistive. Odok’s method is
simple but not so accurate since it does not take the distribution of the bar currents
into account. Odok’s method was further developed, among others by Dabala in
[11]. Assuming an equally distributed inter-bar resistivity, this method takes the
distribution of the bar currents into account.

When casted copper rotors were introduced the measurements became even
more challenging. Dabala suggested an improved method for measurements on
casted copper rotors [9]. The improved method is not only taking the distribution
of the bar currents into account, it also considers the resistivity of the iron sheets.
The method in [9] is used to determine the inter-bar resistivity for a set of
aluminium and copper rotors. All rotors are made for the same 4-pole stator, rated
at 11 kW. The geometries of the rotors are the same for both concepts, except for
a minor difference in the short-circuit ring design.

7
8 CHAPTER 2. MEASUREMENTS OF INTER-BAR RESISTANCE

2.1 Test setup


A test-rig was built with the intention to avoid unnecessarily destruction of the
rotor. This setup, shown in Figure 2.1, makes it possible to measure the inter-bar
resistance with a negligible impact on the rotor construction. The top of the rig,
on which the rotor is standing, consist of a smooth aluminium plate. One end of
the rotor shaft is insulated with a thin plastic film and inserted into a hole in the
center of this plate. Due to the rotor weight a conducting path is created between
the plate and the short-circuit ring. In order to create a uniform distribution of the
current in this contact path, the aluminium plate is machined as well as the surface
of the short-circuit ring. To complete the current path a copper ring is mounted
on the other side of the shaft. With a potential difference between this copper ring
and the aluminium plate, a current will flow from one short-circuit ring to the shaft
on the opposite side via the bar to core region.
By the use of an equivalent circuit of the rotor assuming that the current is
evenly distributed between the rotor bars, it is possible to determine the ring to
ring voltage UAB and the ring to shaft voltages UAD and UBC , as a function of the
inter-bar resistivity.

It is, however, appropriate to note some important issues regarding these mea-
surements. As this setup basically is a short-circuit, it requires a relatively high
current in order to obtain voltage levels that are possible to measure. And it is of
great importance to exclude the connection points of the rotor to the test-rig from
the voltage measuring circuit. It turned out, during the development of the test-rig,
that the currents where not evenly distributed between the rotor bars. Especially
for the copper rotors which where incidently manufactured without any fins on the
short-circuit rings. One reason for this is of course that the inter-bar resistivity
might be unevenly distributed. But an improvement was obtained by placing a
conducting washer between the aluminium plate and the rotor short-circuit ring,
according to Figure 2.2. This washer, being quite soft, distributes the force more
equally around the short-circuit ring, resulting in a smoother distribution of the
current in this contact region.

2.2 Modelling of the rotor


The rotor is modelled as proposed by Dabala [9], with a parameter network dis-
tributed in the axial direction x. As indicated by Odok in [6], the inter-bar
impedance is assumed to be purely resistive. However, the equivalent circuit in
Figure 2.3 has been further developed taking full consideration of the voltage drop
along the shaft. This is obtained by the parallel connection of all the rotor bars
instead of representing one bar. With the total current flowing through the shaft,
the corresponding voltage drop is then modelled correctly.
In the equivalent circuit the following notations are used:
2.2. MODELLING OF THE ROTOR 9

A B

D C

(a) Circuit.

(b) Setup in the lab.

Figure 2.1: Rotor test setup for measurements of inter-bar resistance.


10 CHAPTER 2. MEASUREMENTS OF INTER-BAR RESISTANCE

Figure 2.2: Conducting washer between test-plate and rotor short-circuit ring.

U: Bar to shaft voltage [V].


Ib , Is : Sum of all bar currents, and shaft current, respectively [A].
Ω
Rb , Rs : Bar and shaft resistance per unit length, respectively m .
ℓ: Length of the rotor bars [m].
Qr : Number of rotor bars.
1
 
gtn : Inter-bar conductivity Ωm .
 1 
gF e : Iron core conductivity Ωm .

From Kirchhoff’s voltage law the first differential equation is obtained as:
dU (x) Rb
= Is (x)Rs − Ib (x) (2.1)
dx Qr
And the second equation is obtained from Kirchhoff’s current law, resulting in:
dIb (x) gtn gF e
= −U (x) (2.2)
dx gtn + gF e
By realizing that the change in the shaft current is caused by the current flowing
through the iron core, the third and the last equation within the system becomes:
dIs (x) gtn gF e
= U (x) (2.3)
dx gtn + gF e
The boundary conditions are expressed with the total current flowing through the
rotor I, as;
Ib (0) = Is (ℓ) = I (2.4)
2.3. RESULTS FROM MEASUREMENTS 11

Rb
Ib (x) Qr ∆x Ib (x + ∆x)
A B
+
+
gtn ∆x

U (x) U (x + ∆x)

gF e ∆x

Is (x) Rs ∆x Is (x + ∆x)
D C

0 x x + ∆x ℓ

Figure 2.3: Equivalent circuit of the rotor used for the calculation of the inter-bar
resistivity.

and
Ib (ℓ) = Is (0) = 0. (2.5)
Based on the solution of the equations presented above, the ring to ring voltage
UAB and the ring to shaft voltages UAD and UBC are determined as a function of
the inter-bar resistivity. Figure 2.4 shows these voltages for the studied aluminium
rotor at a total current of 200 A. The corresponding results for the equivalent copper
rotor is shown in Figure 2.5.

These results show that the inter-bar resistivity can be determined, based on
the assumptions made in the model, through measurements of the corresponding
voltages on the considered rotor.

2.3 Results from measurements


A direct current power supply, Delta Elektronika SM15-400, was used to supply the
rotors with a total rotor current of 200 A. The voltages UAB , UAD and UBC were
then measured using sharp probes connected to the multimeter Agilent 34410A.
Based on these measured voltages, the inter-bar resistivities were calculated from
the model described in the previous section. It could be noted from the measure-
ments, that the accuracy for calculating the inter-bar resistivity was poor when the
derivative of voltage to inter-bar resistivity was low, measurements within these
areas were therefore avoided in the analysis. The results obtained for the studied
12 CHAPTER 2. MEASUREMENTS OF INTER-BAR RESISTANCE

260 120
UAD
240
100 UBC

220
80

Voltage [mV]
Voltage [µV]
200
60
180
40
160

140 20

120 −8 0
10 10−7 10−6 10−5 10−4 10−8 10−7 10−6 10−5 10−4
Inter-bar resistivity [Ωm] Inter-bar resistivity [Ωm]
(a) Ring to ring voltage UAB . (b) Ring to shaft voltages UAD and UBC .

Figure 2.4: Calculated voltages for the aluminium rotor at a total current of 200
A.

140 120
UAD
130
100
UBC
120
80
Voltage [mV]
Voltage [µV]

110
60
100
40
90

80 20

70 −8 0
10 10−7 10−6 10−5 10−4 10−8 10−7 10−6 10−5 10−4
Inter-bar resistivity [Ωm] Inter-bar resistivity [Ωm]
(a) Ring to ring voltage UAB . (b) Ring to shaft voltages UAD and UBC .

Figure 2.5: Calculated voltages for the copper rotor at a total current of 200 A.

rotors are shown in Table 2.1.

From these results it can be concluded that the inter-bar resistivity is higher
in cast copper rotors than in cast aluminium rotors. For the studied rotors, the
difference is at least a factor of ten. These results are however consistent with the
findings in [9]. Measurements of inter-bar resistivity on aluminium rotors with the
same slot shape as the rotors studied in this thesis have shown very similar results
[12]. Even though, in that work, the author removed the short-circuit rings and
measured directly between the rotor bars.

It shall be noted that, depending on the casted rotor material, some of the
2.3. RESULTS FROM MEASUREMENTS 13

Figure 2.6: Two of the studied aluminium and copper rotors.

Rotor Al 1 Al 2 Cu 1 Cu 2 Cu 3

UAB [µV] 134 130 125 124 123


Rtn [µΩm] - - 0,3 0,4 0,4
UBC [mV] 15,15 12,22 2,87 2,60 3,49
Rtn [µΩm] 9,0 7,0 0,4 0,3 0,6
UAD [mV] 5,35 3,04 20·10−3 22·10−3 15·10−3
Rtn [µΩm] 7,0 4,0 - - -
Mean value
Rtn [µΩm] 8,0 5,5 0,35 0,35 0,5
Table 2.1: Measured voltages at a total rotor current of 200 A and the resulting
inter-bar resistivities Rtn .
14 CHAPTER 2. MEASUREMENTS OF INTER-BAR RESISTANCE

voltages in Table 2.1 are unsuitable for the determination of the inter-bar resistivity.
Regarding the aluminium rotors, the measured ring to ring voltage UAB results in
an infinitely high inter-bar resistivity. This could be due to the fact that the model
assumes an equally distributed inter-bar resistivity along the rotor bars, and the
ring to ring voltage is strongly dependent on this distribution. One can conclude
that the inter-bar resistivity is unevenly distributed in the aluminium rotors. This
effect is studied further by measuring the voltage drop along the rotor bars with
reference to one short-circuit ring, referred to as UAX .
Z ℓ
Rb
UAX = Ib (x, Rtn )dx (2.6)
0 Qr

The measurements were performed along one fourth of the total number of rotor
bars. The results are presented in Figure 2.7 together with the calculated values
using the inter-bar resistivity from Table 2.1. According to the measured voltage
profile, it can be concluded that the bar currents have decreased to zero in the last
third of the rotor. This implies, for the studied aluminum rotor, that the inter-
bar resistivity might be unevenly distributed. Probably due to the existence of
aluminium oxide along the rotor bar surface, naturally created through the reaction
with oxygen. This process is enhanced by the high casting temperature [13].

The corresponding results for an equivalent copper rotor is shown in Figure 2.8.
In this case, the measured and the calculated voltage shows good correlation. For

160

140

120
Analytical using
Rtn = 5, 5 · 10−6 [Ωm]
100
Voltage [µV]

Measured bar 1
80 Measured bar 2
Measured bar 3
60
Measured bar 4

40 Measured bar 5
Measured bar 6
20 Measured bar 7

0
0 0.05 0.1 0.15
Axial position x [m]

Figure 2.7: Measured and calculated voltage UAX for rotor Al 2 at a total current
of 200 A
2.4. SUMMARY 15

70

60

Analytical using
50 Rtn = 0, 35 · 10−6 [Ωm]
Voltage [µV]

Measured bar 1
40
Measured bar 2

30 Measured bar 3
Measured bar 4
20 Measured bar 5
Measured bar 6
10 Measured bar 7

0
0 0.05 0.1 0.15
Axial position x [m]

Figure 2.8: Measured and calculated voltage UAX for rotor Cu 2 at a total current
of 100 A

the studied copper rotor, the theory of an evenly distributed inter-bar resistivity
seems to hold, indicating an important difference between aluminium and copper
casted rotors.

2.4 Summary
A test-rig has been built for the measurement of rotor voltages, from which the
inter-bar resistivity can be calculated through an equivalent circuit of the rotor.
Measurements have shown that the inter-bar resistivity is as much as ten times
higher in cast aluminum- than in cast copper rotors. An important difference was
noted between the two rotor concepts. Aluminium rotors show results indicating
an unequal distribution of the inter-bar resistivity, while the copper rotors seem to
have a more equal distribution of this resistivity.
Chapter 3

Model for the analysis of inter-bar currents

The analytical model used to include the effects of inter-bar currents is derived from
Behdashti´s work in [14]. The inter-bar currents are taken into account by intro-
ducing a transverse bar to bar resistivity distributed along the rotor bars. Based
on Behdashti´s proposed equivalent circuit of the rotor the inter-bar current dis-
tribution along the rotor core can be obtained. Apart from the machine geometry,
this model requires the fundamental stator current as an input parameter.

3.1 Rotor circuit taking inter-bar currents into account


The equations describing the distribution of the inter-bar currents along the iron
core is derived in the rotor reference frame by studying a small element of the rotor
circuit. In Figure 3.1, the bar current of order n in bar number k, at time t and
axial position x is denoted ibn,k (t, x). The inter-bar current distribution at the
corresponding time and position is referred to as Jtn,k (t, x), given in [A/m]. At
point A in the rotor circuit Kirchhoff’s current law gives:

ibn,k (t, x) − ibn,k (t, x − dx) + (Jtn,k−1 (t, x) − Jtn,k (t, x)) dx = 0 (3.1)

In the rotor reference frame the fundamental component of these currents are vary-
ing with slip frequency. Assuming sinusoidal rotor bar currents and using peak-
value scaling, Equation 3.1 is written with complex notion as:


Ibn,k (t, x) = Jtn,k (t, x) − Jtn,k−1 (t, x) (3.2)
∂x
In the following text, bold symbols represent complex quantities. The phasor of
the inter-bar currents at a certain time can be illustrated as in Figure 3.2, where
the phase displacement of the inter-bar currents of order n between two adjacent
slots, is determined from the number of poles p and the number of rotor bars Qr ,
i.e.

17
18 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

α k k+1

Jtn,k (t, x + dx)


x + dx
D C

ibn,k (t, x) ibn,k+1 (t, x)

Jtn,k (t, x)
x
A B

Figure 3.1: Definition of bar- and inter-bar currents in a small element of the rotor
circuit.

npπ
Jtn,k−1 (t, x) = Jtn,k (t, x)ej Qr (3.3)
Combining Equation 3.2 and Equation 3.3 gives the relation between the inter-bar
current distribution and the bar current as;
npπ
e−j 2Qr ∂
Jtn,k (t, x) = −   Ibn,k (t, x). (3.4)
npπ ∂x
2j sin 2Qr

Jtn,k−1 (x)

npπ
Qr
Jtn,k (x)

Figure 3.2: Phase displacement between inter-bar currents.


3.1. ROTOR CIRCUIT TAKING INTER-BAR CURRENTS INTO ACCOUNT 19

The currents in the rotor element ABCD in Figure 3.1 are linked by both the
flux from the stator current φsn , and the flux produced by the rotor current φrn .
The voltage equation for this current loop is therefore given by;

0 = Zbn Ibn,k (t, x)dx + Rn Jtn,k (t, x) − Zbn Ibn,k+1 (t, x)dx
∂  (3.5)
− Rn Jtn,k (t, x + dx) + φsn (t, x) + φrn (t, x) ,
∂t

where Zbn is the bar impedance per unit length and Rn is the inter-bar resistivity.

In Figure 3.3(a), Rn is defined as the resistance between two adjacent bars


multiplied with the stack length. The inter-bar current path can also be defined
via the rotor shaft, according to Figure 3.3(b). The induced voltages in the rotor
bars are the same for these two cases. If the inter-bar current losses should be the
same for these two models, the following equation must be satisfied [14]:
 
2 npπ
Rn = 4Rtn sin (3.6)
2Qr

As the measurements of the inter-bar resistance resulted in the bar to shaft resistiv-
ity Rtn , Equation 3.6 is used for the conversion into bar to bar resistivity. In order
solve the voltage equation of the rotor circuit, the phase angle of the bar currents
are expressed in a similar way as for the inter-bar currents.
npπ
Ibn,k+1 (t, x) = Ibn,k (t, x)e−j Qr . (3.7)

Combining the Equations; 3.4, 3.5, 3.6, and 3.7 gives the following differential

Rn Rn

Rtn Rtn Rtn

(a) Resistance between rotor bars. (b) Resistance between rotor bars and shaft.

Figure 3.3: Definitions of inter-bar resistances.


20 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

equation for the rotor bar currents:

∂2
0 =Zbn Ibn,k (t, x) − Rtn Ibn,k (t, x)
∂x2
npπ (3.8)
1 ej 2Qr ∂ 
+ φsn (t, x) + φ rn (t, x)
dx 2j sin( npπ
2Qr ) ∂t

The solution gives the axial variation of the bar currents. The inter-bar currents
for the solution of Equation 3.8 can then easily be obtained through Equation 3.4.
Required are the stator and rotor fluxes linked by the rotor element ABCD.

3.2 Stator flux linkage


In this section the stator flux linked by the rotor element is calculated in the
rotor reference frame. The stator flux of order n is expressed as a function of the
fundamental stator current.

3.2.1 Airgap flux density due to stator current


The time varying current in the distributed three phase stator winding results in
in a rotating MMF-wave in the airgap. The magnitude of the resulting flux density
is depending on the airgap permeance. Due to the stator and rotor slotting the
rotating MMF-wave sees a permeance that is varying in both time and space.
The airgap flux density is conventionally determined as the product of the MMF-
wave and the permeance function. This method is described further among others
in [7, 15]. However, in this thesis the airgap flux density caused by the stator cur-
rent is calculated by the use of a different method. The results are verified with
the finite element method (FEM).

The method used is based on the fourier analysis of a simplified flux density
distribution along the airgap, resulting from the current in one phase. The total
airgap flux density is then obtained by superposition of the other two phases. Two
different shapes of possible flux density distributions in the air gap are studied,
referred to as Model A and Model B. It turns out that the two models give very
similar results.
First the calculation is simplified by introducing the following assumptions:
• The stator and rotor iron is assumed to have infinite permeability.
• The rotor slotting do not contribute to the permeance variation along the
airgap circumference.
• The current in phase a is assumed to have the following variation in time:

ia = î cos(ω t). (3.9)


3.2. STATOR FLUX LINKAGE 21

Model A
The distribution of the airgap flux density in space due to the current in phase a
is approximated according to Figure 3.4. By applying Amperes law along the line
C at time t = 0 the following is obtained;
→ −
− → qNs1
I
H · dl = î (3.10)
C Cs
which gives:
µ0 qNs1
B̂δ = î (3.11)
2δ Cs
Where q is the number of slots per pole per phase, Ns1 is the number of conductors
per slot and Cs is the connection factor. It is assumed that the airgap flux density
created by the current in one phase is zero beneath the stator slot openings defined
by the distance d in Figure 3.4, given in [16] as:
 
2π 1
d= 1− . (3.12)
Qs Cf s
Where Qs is the number of stator slots and Cf s is the Carter factor due to the stator
slotting. Based on this distribution of the flux density, the harmonic components
are calculated by means of Fourier analysis. The amplitude of these harmonic
components of order n are obtained as:
 
npπ
2 µ0 qNs1 sin 2Cf s Qs
B̂nδ =   î (3.13)
nπ δ Cs sin npπ
2Qs

stator

rotor
C

B̂δ
d

θ1

Figure 3.4: Airgap flux density due to current in phase a, Model A.


22 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

With the contribution from the other two phases, the peak value of the total airgap
flux density becomes;
3
B̂nm = B̂nδ Krn (3.14)
2
which is equal to:
 
npπ
3 µ0 qNs1 sin 2Cf s Qs
B̂nm = Krn   î (3.15)
nπ δ Cs npπ
sin 2Q s

Where Krn is the winding factor for the wave of order n. With known Fourier
coefficients the resulting airgap flux density can be expressed in both time and
space as:

X  p 
Bnm (t, θ1 ) = B̂nm cos ωt − n θ1 (3.16)
n=1
2

Where θ1 is the mechanical angular position given in stator coordinates.

Model B
The distribution of the airgap flux density in space due to the current in phase a
is approximated according to Figure 3.5. The resulting airgap flux density of order
n is then determined by Fourier analysis to [17];

3 µ0 qNs1
B̂nm = Krn Kf n î (3.17)
nπ δ Cs

stator

rotor

B̂δ
2β σbsys

θ1

Figure 3.5: Airgap flux density due to current in phase a, Model B.


3.2. STATOR FLUX LINKAGE 23

where the form factor is given by


   
ks nπ p nπ p
sin 2Qs 2 sin Qs 21 − k2s
Kf n = 1 − 2 β + 2β ks nπ p
  . (3.18)
nπ p
2Qs 2 sin Qs 2

Where ks is defined as:


σbsys
ks = . (3.19)
τs
The distance bsys is the slot opening and τs is the slot pitch. The coefficients σ and
β are geometry dependent coefficients which can be found in [15, 17].

Verification with FEM


The two models are slightly different when it comes to the modeling of the perme-
ance change beneath a stator slot. This will affect the harmonic components of the
airgap flux density. It is believed that Model B is closer to the actual flux density
distribution in the machine.

In order to verify the analytical models, a comparison is made with the results
from a FEM simulation. The FEM results are obtained from a no-load test per-
formed with the simulation software Flux2d. As the electrical steel is assumed to
have infinite permeability in the analytical models, the same assumption is adapted
to the FEM model, i.e. saturation is not taken into account. The stator current
obtained from this simulation is used to calculate the corresponding flux densities
with the analytical models.

1 0.9
0.8 0.8 Model A
0.6 Model B
Airgap flux density [T]

0.7
Airgap flux density [T]

Flux2d
0.4
0.6
0.2
0.5
0
0.4
-0.2
-0.4 0.3
-0.6 0.2
-0.8 0.1
-1 0
0 20 40 60 80 100 120 140 160 180 1 5 7 11 13 17 19
θ1 [mek o] Harmonic order
(a) FEM simulated airgap flux density at no- (b) Harmonic spectrum of the airgap flux den-
load. sity at no load.

Figure 3.6: Comparison between analytical and FEM-simulated airgap flux density
at no-load.
24 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

The FEM simulated airgap flux density distribution at no load is shown in Fig-
ure 3.6(a). From this simulation it is obvious that the stator slotting has a large
impact on the flux density distribution, and hence, also on the harmonic compo-
nents describing this distribution. It can be seen that the rotor slotting affects the
airgap flux density, but to a lesser extent than the stator slotting. In Figure 3.6(b)
the harmonic spectrum is compared with the corresponding analytical values. In
the analysis, harmonic components up to the order of the first pair of slot harmonics
are considered.

The fundamental component calculated with Model A is 3.4 % higher than


the FEM simulated value, the corresponding value for Model B is 2.4 %. Apart
from the leakage flux, the main difference between the models is believed to be
descended from the modeling of the airgap permeance. The overall harmonic spec-
trum obtained from the two analytical models, for the harmonics considered, shows
acceptable correlation with the FEM results. Based on the values of the fundamen-
tal components, Model B will be used for the calculation of the airgap flux density
created by the stator current.

3.2.2 Stator flux linked by the rotor circuit


In the previous section the airgap flux density produced by the stator current was
derived in stator coordinates. Based on these results this section presents the cal-
culation of the stator flux linked by the rotor circuit.

As the calculations are performed in the rotor circuit, the airgap flux density
produced by the stator is expressed in rotor coordinates. This requires information
regarding the position of the rotor relative to the stator flux waves. Therefore, it
is assumed that a wave of order n at the time t = 0, has a position relative to the
rotor as shown in Figure 3.7. Based on these conditions, the relation between the
stator and the rotor positions is given by:
2
θ1 = θ2 + (1 − s1 )ωt (3.20)
p
The airgap flux density of order n seen by the rotor can then be expressed with
rotor coordinates in time and space as:
 p 
Bnr = B̂nm cos sn ωt − n θ2 (3.21)
2
Where sn is the slip of a wave of order n given by;

sn = 1 − n(1 − s1 ), (3.22)

n should be positive or negative according to the direction of rotation of the wave.


The flux linked by the element ABCD in Figure 3.7 is equal to;
3.2. STATOR FLUX LINKAGE 25

x + dx
D C

A B
x

θ2
2π αx
Qr (k − 1) ℓ

2π αx
Qr k ℓ

Figure 3.7: Position of a wave of order n at time t = 0 in the rotor reference frame.


Z k+ αx ǫ
ℓ −2
Qr  p 
φsn (x, t) = rdxB̂nm cos sn ωt − n θ2 dθ2 (3.23)

Qr (k−1)+ αx ǫ
ℓ +2
2

where ǫ is given by [16].


 
2π 1
ǫ= 1− (3.24)
Qr Cf r

The rotor skewing is taken into account by introducing the mechanical skewing
angle α, given by:
26 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS


α= αs (3.25)
Qs
Where αs is the skewing in number of stator slots. This gives the linked stator flux
expressed with complex notation as:
 
2rdx npπ π(2k−1)

j sn ωt− np + αx
φsn (x, t) = B̂nm sin · e 2 Qr ℓ
(3.26)
n p2 2Cf r Qr

3.3 Rotor flux linkage


The stator flux linked by the rotor circuit induce currents in the rotor bars, the
magnitude and frequency of these currents are depending on the rotor speed. As
a result, a rotor flux is created which counteracts the stator flux. In this section,
the rotor flux linked by the small rotor element is derived as a function of the bar
current of order n.

3.3.1 Airgap flux density due to rotor current


The calculation is simplified by introducing the following assumptions:

• The stator and rotor iron is assumed to have infinite permeability.

• The rotor slotting do not contribute to the permeance variation in the airgap.

• The rotor bar currents are assumed to vary sinusoidally in time.

The current of order n in rotor bar k, ibn,k , and the corresponding MMF, Mn,k , is
shown in Figure 3.8. Applying amperes law along the line C gives:

Mn,k (x, t) − Mn,k−1 (x, t) = ibn,k (x, t) (3.27)

The phase displacement between the MMF over two adjacent rotor teeth is given
by:
npπ
Mn,k−1 (x, t) = Mn,k (x, t)ej Qr (3.28)
The rotor bar current ibn,k is replaced with the complex bar current Ibn,k introduced
to define the rotor circuit equation in Section 3.1. This gives the following expression
for the complex airgap MMF created by the rotor current.
npπ
e−j 2Qr
Mn,k (x, t) = −   Ibn,k (x)ejsn ωt (3.29)
npπ
2j sin 2Qr
3.3. ROTOR FLUX LINKAGE 27

Mn,k−1 Mn,k

D C

ibn,k−1 ibn,k ibn,k+1


x
θ2
C
A B

Figure 3.8: Current in bar number k and the corresponding MMF in the airgap.

The corresponding airgap flux density at the point θ1 , is obtained by multiplying


the MMF with the airgap magnetic permeance per unit area.

Bn,k (x, θ1 , t) = Mn,k (x, t)Λ(θ1 ) (3.30)

According to the previously stated assumption the airgap permeance variation


along the circumference is only created by the stator slotting, i.e. the rotor is as-
sumed to have closed slots. This simplifies the forthcoming analysis considerably.
As the effect of the rotor is neglected, the airgap permeance function can be ex-
pressed independent of time in stator coordinates. In analogy with the calculation
of the airgap flux density due to the stator current in Section 3.2.1, the airgap
permeance function is approximated, Model A is used for simplicity reasons. By
the use of Fourier analysis, the distributed permeance defined by Figure 3.9, can
be expressed in stator coordinates as;
 
∞ sin γπ
µ0 2µ0 X Cf s
Λ(θ1 ) = + cos (γQs θ1 ) (3.31)
δCf s πδ γ=1 γ

where γ is the order of the permeance harmonic considered. In this thesis, only
the permeance harmonics of the first order are considered. With complex notation,
this simplifies Equation 3.31 to;

Λ(θ1 ) = Λ0 + Λs1 ejγQs θ1 (3.32)

where
µ0
Λ0 = (3.33)
δCf s
28 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

Λ
τs d
µ0
δ

θ1
Figure 3.9: Permeance variation along the airgap circumference as defined by Model
A.

and  
2µ0 γπ
Λs1 = sin . (3.34)
γπδ Cf s
The airgap flux density is now calculated as the product of the MMF and the
permeance function. Note that the permeance function has to be expressed in
rotor coordinates by the use of Equation 3.20.

X
Bn,k (x, θ2 , t) = Λ(θ2 , t) Mn,k (x, t) (3.35)
n=1

From this definition a series of harmonic components is obtained, they are treated
separately in the following.

 2

Bn,k (x, θ2 , t) = (M1,k + M5,k + M7,k ...) Λ0 + Λs1 ejQs γ(θ2 + p (1−s1 )ωt) (3.36)

3.3.2 Flux caused by the phase belt harmonics


Phase belt harmonics are due to the concentration of the MMF in slots, in a three
phase machine they fulfil the condition:
• n = 1 ± 6k
The dominating terms of the airgap flux densities caused by rotor currents in-
duced by the phase belt harmonics are those acting through the average permeance
Λ0 . The interaction with the harmonic permeance Λs is neglected in this study.
Thus, the airgap flux densities due to these harmonics are simply calculated as:

Bn,k (x, t) = Mn,k (x, t)Λ0 (3.37)


3.3. ROTOR FLUX LINKAGE 29

The corresponding flux is obtained by integrating the flux density over the area
ABCD in Figure 3.7.
Z Q2π k+ αx ǫ
ℓ −2
r
φrn (x, t) = rdx Bn,k (x, t)dθ2 (3.38)

Qr (k−1)+ αx ǫ
ℓ +2

Which gives the same result as in [17]:


npπ
rπdx µ0 e−j 2Qr jsn ωt
φrn (x, t) = j Ibn,k (x)  e (3.39)
Cf r Cf s Qr δ sin npπ 2Qr

3.3.3 Flux caused by slot harmonics


The slot harmonics are partly made up of the MMF slot harmonics of order:
• n = 1 + γ 2Q
p
s

These rotor fields are caused by the harmonic rotor current of order n = 1 + γ 2Q
p ,
s

acting through the average permeance. The corresponding flux densities are deter-
mined by Equation 3.37.

The slot harmonics are also generated through the interaction with the perme-
ance harmonics of order γ, having the same number of poles. These harmonics
arise from the interaction between the fundamental rotor field and the harmonic
permeance. In the rotor reference frame this is seen as a wave, that can be sepa-
rated into a wave rotating in the positive direction and one in the negative direction
respectively, i.e. γ = ±1. Each of the waves produces a flux density given by:

Λs1 jγQs (θ2 + p2 (1−s1 )ωt)


Bγ,k (x, θ2 , t) = M1,k (x, t)
e (3.40)
2
The backward rotating permeance field corresponding to γ = −1, produces to-
gether with the fundamental MMF, the same rotor frequency as the MMF harmonic
of order n = 1 + 2Q p , rotating in the forward direction. Therefore, the resulting
s

rotor flux of order n, linked by the element ABCD in Figure 3.7 is calculated as:


Z Qr k+ αx ǫ
ℓ −2
φrn (x, t) = rdx (Bn,k (x, t) + B−γ,k (x, θ2 , t))dθ2 (3.41)

Qr (k−1)+ αx ǫ
ℓ +2

Which gives similar results as in [17]:


npπ
rdxµ0 jsn ωt  π In,k (x)e−j 2Qr
φrn =j e npπ
 +
δ Qr Cf s Cf r sin 2Q r
pπ (3.42)
1 γπ  I1,k (x)e−j 2Qr Qs π  −jγ( Q
Qs
2π(k− 12 )+Qs αx
ℓ )

sin pπ
 sin e r
γπQs Cf s sin 2Q Qr Cf r
r
30 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

3.4 General set of equations


For a given stator current the rotor Equation 3.8 can be solved by inserting Equa-
tions; 3.39, 3.42 and 3.26. The solution gives the corresponding rotor bar current
as a function of the axial position x. In the following, the rotor current of order n
in bar number k = 1, is denoted Ibn . The solution is depending on the order of the
considered current.

In case of a phase-belt harmonic the differential equation for the rotor circuit
becomes:

• If n 6= 1 + γ 2Q
p , γ = ±1
s

∂2 npαx
Z2n Ibn (x) − ρtn 2
Ibn (x) + En0 e−j 2ℓ = 0 (3.43)
∂x
Where:  
npπ
2rℓB̂nm ω sin 2Cf r Qr
En0 =   (3.44)
np sin npπ
2Qr


Z2n = Zbn + jXn (3.45)
sn
2πrℓ µ0 ω
Xn =   (3.46)
Qr Cf s Cf r δ 4 sin2 npπ
2Qr


ρtn = Rtn (3.47)
sn

If the considered harmonic is a slot harmonic of the first order, the corresponding
equation becomes:

• If n = 1 + γ 2Q
p , γ = ±1
s

∂2 npαx x
Z2n Ibn (x) − ρtn 2
Ibn (x) + En0 e−j 2ℓ + jX1 Ib1 (x)e−jγQs α ℓ = 0 (3.48)
∂x
Where:    
γπ πQs
rℓω µ0 sin Cf s sin Cf r Qr
X1 =     (3.49)
2γπQs δ sin npπ sin pπ
2Qr 2Qr

Due to the interaction with the permeance harmonics, the rotor bar current of or-
der n is now also a function of the fundamental bar current Ib1 , obtained from the
3.5. EFFECTS OF A FINITE INTER-BAR RESISTANCE ON ROTOR CURRENT
DISTRIBUTION 31

solution of Equation 3.43.

The boundary conditions for these two differential equations are obtained by
studying the rotor circuit at the point where it connects to the adjoining short-
circuit ring, as in Figure 3.10(a). At this point the ring voltage and the bar to bar
voltage are equal, i.e.
Ian,k Zan = Jtn,k Rn (3.50)
Where Zan is the impedance of the short-circuit ring segment with respect to the
frequency of the nth harmonic. Using the phasor relation between the bar- and
the short-circuit ring currents given in Figure 3.10(b), together with Equation 3.4,
gives the boundary conditions as:
(
∂ ℓ ℓ
 Z 
∂x Ibn −2 − Rn Ibn − 2 = 0
an

∂ ℓ Zan ℓ (3.51)
∂x Ibn 2 + Rn Ibn 2 = 0

From this equation it is obvious that the inter-bar current density at the rotor
boundary is determined by the impedance of the short-circuit ring, regardless the
rotor skew. Thereby, it can be concluded that inter-bar currents are present also
in unskewed machines, unless the impedance of the short-circuit ring is zero.

Jtn,k−1 Jtn,k Rn
− 2ℓ
Ian,k−1
Ibn,k Ibn,k+1
Ibn,k
Ian,k−1 Ian,k Zan npπ
Qr Ian,k
(a) Rotor circuit. (b) Ring- and bar current phasors.

Figure 3.10: Rotor currents at the boundary x = − 2ℓ .

3.5 Effects of a finite inter-bar resistance on rotor current


distribution
According to the previous section, a finite inter-bar resistance introduces the inter-
bar current density Jtn,k . With the contribution of the inter-bar current density in
the adjacent rotor tooth Jtn,k−1 , a resulting bar current is obtained dIbn,k .
dIbn,k (x) = (Jtn,k (x) − Jtn,k−1 (x)) dx (3.52)
32 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

This current defines the change in the bar current when moving a small distance
dx in the positive x-direction, according to Figure 3.11. In this figure, the airgap
flux density caused by the stator current Bnm is used as reference phasor.

From this figure one can see that the inter-bar currents causes a non uniform
distribution of the rotor bar currents. This effect is enhanced by the rotor skew, but
the dependency of the size of the inter-bar resistance, or the required bar insulation
to cancel out this effect, is not that obvious.
With the intention to get a better understanding of this effect, a case study is
performed on an 11 kW 4-pole machine. The machine having 36 stator slots and
a full pitch single layer winding is equipped with a cast aluminium rotor with 44
rotor slots, referred to as Motor A. The machine is simulated both with and without
rotor skew. The rotor bar- and inter-bar currents are then studied along the core
at different values of inter-bar resistance.

Bnm

γn

Ibn (x)
Brn (x)

βn

dIbn Ibn (x + dx)

Figure 3.11: Change in rotor bar current due to the interaction with inter-bar
currents.

3.5.1 Rotor without skew


The inter-bar current density in a rotor without skewing is focused towards the
rotor ends, and its magnitude is directly determined by the ratio of short-circuit
ring impedance to inter-bar impedance. This analysis is somewhat simplified by,
as stated in Chapter 2, assuming that the inter-bar impedance is purely resistive.
3.5. EFFECTS OF A FINITE INTER-BAR RESISTANCE ON ROTOR CURRENT
DISTRIBUTION 33

The magnitude of the fundamental locked-rotor inter-bar current density along


the rotor core is shown in Figure 3.12(a). Due to the low impedance of the short-
circuit ring, it requires a quite low inter-bar resistivity to introduce fundamental
inter-bar currents. These currents, having their maximum values at the rotor ends
decreases towards zero in the middle of the rotor.

Figure 3.12(b) shows the angle β1 defined in Figure 3.11, giving the phase angle
of the inter-bar currents flowing along the rotor bar, contributing to the bar current.
At the rotor ends, this angle is set by the voltage over the corresponding short-
circuit ring segment. In the middle of the rotor, where the inter-bar currents are
zero, a phase shift of 180 degrees occurs. In other words; the voltage over the
ring segments at each end of the rotor introduces two circumferential current paths
through the rotor teeth, opposing each other. In a symmetrical rotor, assuming an
equally distributed inter-bar resistivity, these currents cancel each other out in the
middle.
As a result the bar current is reduced towards the ends of the rotor. This can
be seen in Figure 3.13(a) which shows the fundamental locked rotor bar current.
But the effect is of minor importance, as well as for the phase of the corresponding
current phasor shown in Figure 3.13(b).

In the following, the variation of the inter-bar currents with rotor speed, and
the influence on the rotor currents of higher order is studied. The analysis is per-
formed with the inter-bar resistivity Rtn = 5 ·10−5 Ωm. Figures 3.14(a) and 3.14(b)
show the fundamental bar and inter-bar current as a function of both slip and axial
position in the bar. It can be seen that the inter-bar currents have their maximum

50 80
45 60
Rtn = 5 · 10−4 Ωm Rtn = 5 · 10−4 Ωm
40
i
mm

40
A

20 Rtn = 5 · 10−5 Ωm
Rtn = 5 · 10−5 Ωm
h
Inter-bar current density

35 0
30 Rtn = 5 · 10−6 Ωm -20
-40
β1 [◦]

25
-60
20 -80
15 -100
-120
10
-140
5 -160
0 -180
-10 -8 -6 -4 -2 0 2 4 6 8 10 -10 -8 -6 -4 -2 0 2 4 6 8 10
Axial position x [cm] Axial position x [cm]
(a) Inter-bar current density. (b) Resulting fundamental inter-bar current an-
gle β1 .

Figure 3.12: Magnitude of the fundamental locked rotor inter-bar current density
and the resulting angle β1 , for Motor A with unskewed rotor.
34 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

3.56 91.2

91
3.54
90.8

Rotor bar current [kA]


3.52 90.6

Phase [◦]
90.4
3.50
90.2
Rtn = 5 · 10−4 Ωm
3.48 90
Rtn = 5 · 10−4 Ωm Rtn = 5 · 10−5 Ωm
89.8
3.46 Rtn = 5 · 10−5 Ωm
Rtn = 5 · 10−6 Ωm
Rtn = 5 · 10−6 Ωm 89.6

3.44 89.4
-10 -8 -6 -4 -2 0 2 4 6 8 10 -10 -8 -6 -4 -2 0 2 4 6 8 10
Axial position x [cm] Axial position x [cm]
(a) Fundamental bar current. (b) Fundamental bar current angle γ1 .

Figure 3.13: Magnitude of fundamental locked rotor bar current and the corre-
sponding angle γ1 , for Motor A with unskewed rotor.

value at the speed corresponding to the largest slip. However, their influence on
the fundamental rotor current is of minor importance.

The cases when the currents are caused by higher order space harmonics are
different. As these fields have low synchronous speeds, the corresponding rotor
currents create torques that counteract the fundamental torque at nominal speed.
The rotor currents of order higher than the fundamental are therefore sources of
stray losses. Figure 3.14(c) and Figure 3.14(e) shows the rotor currents caused by
the first pair of slot harmonics, having the order n = 1 − 2Q p and n = 1 + p ,
s 2Qs

respectively.
These currents, caused by the slot harmonics, are large in the unskewed rotor.
Especially at low speeds when the high fundamental current creates large slot MMF
harmonics. Thus, large asynchronous torques are expected during a start.
Furthermore, at nominal speed the magnitude of these harmonic currents are
still quite large. This is also true at no-load, when the magnitude of the slot
space harmonics are determined by the fundamental no-load current. The resulting
cage losses can form a considerable part of the no-load losses [10]. This effect is
enhanced by main flux saturation increasing the fundamental current, which has
been neglected in this thesis.

Rotor skewing by one stator slot pitch is a common practice to reduce the rotor
currents caused by slot harmonics. But this promotes the flow of inter-bar currents.
3.5. EFFECTS OF A FINITE INTER-BAR RESISTANCE ON ROTOR CURRENT
DISTRIBUTION 35

Inter-bar current density [ mm


A
]
4.0 8
Rotor bar current [kA]

3.0 6

2.0 4

1.0 2

0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
(a) Rotor bar current, n = 1. (b) Inter-bar current density, n = 1.
Inter-bar current density [ mm
A
]

500 0.08
Rotor bar current [A]

400 0.06
300
0.04
200
0.02
100
0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
2Qs 2Qs
(c) Rotor bar current, n = 1 − p
. (d) Inter-bar current density, n = 1 − p
.
Inter-bar current density [ mm
A
]

300 0.2
Rotor bar current [A]

250
0.15
200
150 0.1
100
0.05
50
0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
2Qs 2Qs
(e) Rotor bar current, n = 1 + p
. (f) Inter-bar current density, n = 1 + p
.

Figure 3.14: Magnitude of the currents in the unskewed rotor caused by the fun-
damental and the first pair of slot space harmonics when Rtn = 5 · 10−5 Ωm.
36 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

3.5.2 Rotor with skew


The analysis continues with the same motor as in the previous section but with
a rotor skew of one stator slot pitch, i.e. αs = 1. When the rotor is skewed
the voltages induced in the skewed conductors are reduced in comparison to the
unskewed conductors. The ratio of these voltages defines the skewing factor for the
nth rotor voltage as:  
npπαs
sin 2Qs
ksk = npπαs ≤1 (3.53)
2Qs

This expression becomes zero if the skewing in number of stator slots is equal to:

Qs
αs0 = np (3.54)
2

For the first pair of slot harmonics ksk is close to zero when αs0 is close to one. As
a result, when skewing insulated rotor bars by one stator slot pitch, a large reduc-
tion of the corresponding rotor currents can be expected, improving the machine
performance, both in terms of reduced asynchronous torques and rated efficiency.
However, in casted rotors the rotor skewing might have an opposite effect.

Figure 3.15 shows the fundamental locked- rotor inter-bar current density and
the resulting angle β1 for different values of inter-bar resistivity. The inter-bar
current densities at the rotor ends are approximately the same as for the unskewed
rotor. The important difference is that the inter-bar currents now are focused
towards the middle of the rotor, somewhat depending on the inter-bar resistivity.
By studying the angle β1 , one can expect a continuous increasing bar current in

80
45
Rtn = 5 · 10−2 Ωm 60 Rtn = 5 · 10−4 Ωm
40 Rtn = 5 · 10−3 Ωm
i
mm

40
A

Rtn = 5 · 10−4 Ωm Rtn = 5 · 10−5 Ωm


35
h

= 5 · 10−5 Ωm
Inter-bar current density

Rtn 20
30 Rtn = 5 · 10−6 Ωm
Rtn = 5 · 10−6 Ωm
0
Phase [◦]

25
-20
20 -40
15 -60
10 -80
5 -100
0 -120
-10 -8 -6 -4 -2 0 2 4 6 8 10 -10 -8 -6 -4 -2 0 2 4 6 8 10
Axial position x [cm] Axial position x [cm]
(a) Inter-bar current density. (b) Angle β1 .

Figure 3.15: Magnitude of the fundamental locked-rotor inter-bar current density


and the resulting angle β1 for Motor A with skewed rotor.
3.5. EFFECTS OF A FINITE INTER-BAR RESISTANCE ON ROTOR CURRENT
DISTRIBUTION 37

3.50 100
Rtn = 5 · 10−2 Ωm
3.45 −3
98
Rtn = 5 · 10 Ωm
3.40 Rtn = 5 · 10−4 Ωm 96
Rotor bar current [kA]

0.8
Rtn = 5 · 10−5 Ωm 94
3.35
Rtn = 5 · 10−6 Ωm 92

Phase [◦]
3.30
90
3.25
88
3.20 Rtn = 5 · 10−4 Ωm
86
3.15 Rtn = 5 · 10−5 Ωm
84
3.10 82 Rtn = 5 · 10−6 Ωm

3.05 80
-10 -8 -6 -4 -2 0 2 4 6 8 10 -10 -8 -6 -4 -2 0 2 4 6 8 10
Axial position x [cm] Axial position x [cm]
(a) Rotor bar current. (b) Bar current angle γ1 .

Figure 3.16: Magnitude of fundamental locked-rotor bar current and the corre-
sponding angle γ1 , for Motor A with skewed rotor.

the positive x-direction, except for very low values of inter-bar resistivity. This can
be seen in Figure 3.16(a), showing the fundamental rotor bar currents.
An interesting result can be found in Figure 3.16(b), showing the angle of the
fundamental bar current phasor with respect to the stator flux density. For low
values of inter-bar resistivity the phase change between the rotor ends is equal to
the electrical skewing angle, indicating that the skewing is ineffective. These results
are consistent with what was found in [18].

Figure 3.17 shows the rotor currents caused by the fundamental and the first
pair of slot harmonics as a fuction of speed, when the inter-bar resistivity Rtn =
5 · 10−2 Ωm. The overall inter-bar current density is low, the value of the inter-
bar resistivity is thereby to be considered as high. In comparison to the unskewed
case the currents caused by the slot harmonics are drastically reduced. The corre-
sponding asynchronous torques will vanish and an improved motor efficiency can be
expected. With high values of inter-bar resistivity the theory, assuming negligible
inter-bar current flow, holds. i.e. the skewing is effective.
Measurements have shown that the inter-bar resistivity in cast rotors is much
lower than the value referred to as high in the previous case. In Figure 3.18 the
corresponding results are shown with an inter-bar resistivity of Rtn = 5 · 10−5 Ωm.
Inter-bar currents now appears for all the considered harmonics. High fundamental
inter-bar currents at start indicate an increased locked rotor torque. But most
notable is the large increase of inter-bar currents caused by the slot harmonics,
resulting in a huge increase and a distortion of the corresponding bar currents. In
this case the skewing is not effective. As the inter-bar resistivity is much higher
than the bar resistivity, large additional losses are created in the bar to core region.
Large asynchronous torques and increased stray losses are expected.
38 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

Inter-bar current density [ mm


A
]
4.0 0.2

Rotor bar current [kA]


3.0 0.15

2.0 0.1

1.0 0.05

0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
(a) Rotor bar current, n = 1. (b) Inter-bar current density, n = 1.

Inter-bar current density [ mm


A
]
40 0.02
Rotor bar current [A]

30 0.015

20 0.01

10 0.005

0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
2Qs 2Qs
(c) Rotor bar current, n = 1 − p
. (d) Inter-bar current density, n = 1 − p
.
Inter-bar current density [ mm
A
]

8 0.03
Rotor bar current [A]

0.025
6
0.02
4 0.015
0.01
2
0.005
0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
2Qs 2Qs
(e) Rotor bar current, n = 1 + p
. (f) Inter-bar current density, n = 1 + p
.

Figure 3.17: Magnitude of the currents in the rotor skewed by one stator slot
pitch, caused by the fundamental and the first pair of slot space harmonics when
Rtn = 5 · 10−2 Ωm.
3.5. EFFECTS OF A FINITE INTER-BAR RESISTANCE ON ROTOR CURRENT
DISTRIBUTION 39

Inter-bar current density [ mm


A
]
4.0 30
Rotor bar current [kA]

25
3.0
20
2.0 15
10
1.0
5
0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
(a) Rotor bar current, n = 1. (b) Inter-bar current density, n = 1.
Inter-bar current density [ mm
A
]

500 8
Rotor bar current [A]

400 6
300
4
200
2
100
0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
2Qs 2Qs
(c) Rotor bar current, n = 1 − p
. (d) Inter-bar current density, n = 1 − p
.
Inter-bar current density [ mm
A
]

400 10
Rotor bar current [A]

300 8
6
200
4
100 2

0 0
10 10
5 1.5 5 1.5
0 1 0 1
-5 0.5 -5 0.5
-10 0 -10 0
Axial position x [cm] Slip Axial position x [cm] Slip
2Qs 2Qs
(e) Rotor bar current, n = 1 + p
. (f) Inter-bar current density, n = 1 + p
.

Figure 3.18: Magnitude of the currents in the rotor skewed by one stator slot
pitch, caused by the fundamental and the first pair of slot space harmonics when
Rtn = 5 · 10−5 Ωm.
40 CHAPTER 3. MODEL FOR THE ANALYSIS OF INTER-BAR CURRENTS

3.6 Summary
A model to calculate the inter-bar currents in cage induction motors has been de-
rived. This model requires the stator current and the inter-bar resistivity as input
parameters. Simulations have shown that the inter-bar current density is increasing
rapidly with decreasing inter-bar resistivity. This affects the bar current distribu-
tion and phase angle.

In unskewed rotors the inter-bar currents are focused towards the rotor ends, and
their magnitude is directly determined by the ratio of short-circuit ring impedance
to inter-bar impedance. Unless this ratio is very high, inter-bar currents are small
in unskewed rotors and can therefore be neglected.

When the rotor is skewed by one stator slot pitch inter-bar currents increase.
This effect is most significant for rotor currents that are caused by the slot space
harmonics. For the studied machine, the inter-bar currents are reduced to a negli-
gible level when the bar to shaft resistivity is larger than 5 · 10−2 Ωm. In this case
rotor skewing becomes effective.
Chapter 4

Effects during a direct-on-line start

When a cage induction motor is started directly against the grid, the presence of
skin effect and leakage path saturation has large impact on the machine perfor-
mance. The high frequency of the fundamental slot leakage flux gives rise to high
current density in the upper parts of the rotor bars. As a result, the effective bar
resistance is increased and the effective bar inductance is decreased. The start-
ing currents will generate large differential and slot leakage fluxes. This saturates
the differential and slot leakage flux paths, which will reduce the corresponding
leakage inductances. It also creates additional iron losses in the stator and rotor
teeth. These factors have to be taken into to account when calculating the starting
performance of the machine.

4.1 Skin effect in the rotor bars


A good starting performance usually implies high starting torque and low starting
current, requiring a high rotor resistance and a high rotor leakage inductance. On
the other hand, to obtain a high full load efficiency and to ensure a reasonable
margin of overload capability, low rotor resistance and leakage inductance is prefer-
able. This trade-off between starting performance and rated efficiency is a classical
problem for the induction motor designer [19].

However, during an induction motor start when the rotor frequency equals the
mains frequency, advantage is taken of the skin effect. In the deep bar rotor shown
in Figure 4.1(a), the self inductance is highest at the bottom of the bar. During
a start when the rotor frequency is high, the rotor current is focused towards the
upper parts of the bar. As a result, the rotor bar losses are increasing and a larger
torque is produced. This is normally modeled by an increased equivalent bar re-
sistance and a decreased bar inductance. When the rotor has accelerated up to
nominal speed the rotor frequency is very low, resulting in an evenly distributed
bar current. Utilizing the whole rotor bar area ensures a low slip and therefore

41
42 CHAPTER 4. EFFECTS DURING A DIRECT-ON-LINE START

(a) Deep bar. (b) Double cage. (c) Bar in the studied
machine.

Figure 4.1: Different types of rotor bars.

reduced losses. Analytical expressions for the frequency dependent impedances of


different deep bar rotors have been derived by Liwschitz-Garik in [20].

With the introduction of the double cage rotor shown in Figure 4.1(b), the
AC-resistance of the bar can be increased even further. The idea is to create two
parallel current paths in the rotor bar, one with low resistance and high inductance
to conduct most of the current at rated speed, and one with high resistance and
low inductance which will conduct the largest part of the current during start. An
analytical method for performance calculations of multiple squirrel cage rotors has
been presented by Alger in [19]. Finite element modeling of a double cage rotor has
shown good correlation with analytical methods [21]. In this work Williamson and
Gersh highlight the effects of magnetic saturation, showing that the phenomena has
a considerable effect on the leakage inductance, while the effects on the frequency
dependent resistance is of minor importance.

The drawback with the analytical models is that they are restricted to a cer-
tain slot geometry, which is often quite simple. In the case of more complicated
geometries numerical methods are preferable.

4.1.1 Numerical method used to account for skin effect


In this thesis the skin effect is taken into account by the use of a one-dimensional
numerical model described in [22]. Besides the accuracy, the main advantage of
using this method is that it can easily be adapted for different slot geometries. The
drawback might be that it requires a software with a numerical solver. However,
the flexibility of the method definitely makes it worth the effort. In the following
a brief description of the model is presented, giving the assumptions used and the
main equations describing the theory.
4.1. SKIN EFFECT IN THE ROTOR BARS 43

The following assumptions are made:

• The magnetic field lines are crossing the rotor bar perpendicular to the slot
sides.

• The rotor iron reluctance is negligible.

• The current density is axial-symmetrical in the slot.

• The rotor bar current density J varies sinusoidally in time with angular fre-
quency ω.

The bar is modeled with a large number of rectangular segments of equal height.
Figure 4.2 shows the dimensions of the v th segment. Within this section Maxwell’s
induction law defines the induced electric field E ~ v , produced by the flux density
~
field Bv .
~
∇×E ~ v = − ∂ Bv (4.1)
∂t
With rectangular slot sections defined in a cartesian coordinate system according to
Figure 4.2, and with a bar current defined in the positive y-direction, the induction
law can be rewritten as:
∂Ey (z)v ∂Bx (z)v
−êx = −êx (4.2)
∂z ∂t

With section resistivity ρv and a current density J~v varying sinusoidally in time,
the first equation within the system can be defined as:

dJy (z)v
ρv = jωµ0 Hx (z)v (4.3)
dz

Neglecting the displacement current, Ampere’s circuit law states that the mag-
~ v is generated by an electrical current according to:
netic field H
~ v = J~v
∇×H (4.4)

Assuming that the bar width is equal to the slot width, which is reasonable in a
casted rotor, the following is obtained:

∂Hx (z)v
êy = êy Jy (4.5)
∂z
Which defines the second and the last equation within the system as:

dHx (z)v
= Jy (4.6)
dz
44 CHAPTER 4. EFFECTS DURING A DIRECT-ON-LINE START

hv y

x
bv

Figure 4.2: The v th section of the rotor slot.

If section i carries the current Ii , the first boundary condition can be obtained from
Ampere’s law.
v
X
bv Hx (z = hv )v = bv+1 Hx (z = 0)v+1 = Ii (4.7)
i=0

As the sections are short-circuited at the bar ends the voltages over all sections are
equal, giving the final boundary condition as:
ρv+1 Jy (z = 0)v+1 = ρv Jy (z = hv )v (4.8)

These equations are part of the iterative process described in Figure 4.3. For
a known bar current Ibar with the angular frequency ω, this method calculates
the bar current distribution and phase. The corresponding impedance correction
factors are then obtained from:
PAC
kr = (4.9)
PDC
WAC
kx = (4.10)
WDC
Where the active power P and the stored magnetic energy W are simply calculated
as:
n sec
X 2 ℓ
P = |Iv | ρv (4.11)
v=0
b v hv

nsec
v 2
1X hv X
W = µ0 ℓ Ii (4.12)
2 v=0
bv
i=0

4.1. SKIN EFFECT IN THE ROTOR BARS 45

Ibar
Iv=0 = nsec

For: nsec Calculate Jv+1

| ni=0
P sec
Ii |−Ibar
Error = Ibar

NO YES
OK?

Iv=0 = Iv=0 (1 − Error) Calculate kr and kx

Figure 4.3: Procedure for the calculation of the impedance correction factors kr
and kx that accounts for skin- effect.

4.1.2 Verification with FEM


In order to verify the model, the radial distribution of the locked rotor bar current
was calculated using harmonic analysis in Flux2D. The studied 4-pole machine is
equipped with a rotor having 28 rotor slots, casted with either aluminium or copper.
Figure 4.4(a) shows the magnitude of the locked rotor bar current density for a
number of rotor bars corresponding to one pole. The result is somewhat depending
on the relative position between the bar and the stator core. The corresponding
current distribution obtained by the proposed model, for same current, is shown
46 CHAPTER 4. EFFECTS DURING A DIRECT-ON-LINE START

140 140
bar 1 One-dimensional numerical method

i
i

mm2
mm2
120 120 FEM - average

A
bar 2

h
h
bar 3

Magnitude of current density


Magnitude of current density
100 bar 4 100
bar 5
80 bar 6 80
bar 7
60 60

40 40

20 20

0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Radial position in rotor bar [mm] Radial position in rotor bar [mm]
(a) FEM-simulation. (b) Proposed numerical method.

Figure 4.4: Locked rotor bar current density for the aluminium rotor.

150 150
bar 1 One-dimensional numerical method
bar 2 FEM - average
bar 3
100 bar 4 100
bar 5
Phase [◦]
Phase [◦]

bar 6
bar 7
50 50

0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Radial position in rotor bar [mm] Radial position in rotor bar [mm]
(a) FEM-simulation. (b) Proposed numerical method.

Figure 4.5: Locked rotor bar current phase angle relative to the bottom of the bar
for the aluminium rotor.

in Figure 4.4(b). The method used shows good correlation with the FEM results,
except in the upper parts of the bar. This is due to saturation of the rotor tooth
tips. As a result, the current density in the top of the bar is overestimated by 22
%.

Figure 4.5(a) shows the phase angle of the bar current density with reference
to the current flowing at the bottom of the slot. The result obtained from the
proposed model is shown in Figure 4.5(b), which indicates a slightly overestimated
inductance in the top of the bar.
As part of the goal of this thesis is to study the differences between aluminium
and copper casted rotors, the corresponding simulation is performed with a copper
4.1. SKIN EFFECT IN THE ROTOR BARS 47

rotor. The results are shown in Figures 4.6 and 4.7. Due to the lower resistivity of
copper the skin depth is smaller, resulting in a more pronounced skin-effect. This
is seen in the results as a slightly increased saturation of the tooth tips. As a con-
sequence, the current density at the top of the bar is overestimated by 28 %.

0 250 250
bar 1 Analytical
i

i
mm2

mm2
FEM - average
A

A
bar 2
200 200
h

h
bar 3
Magnitude of current density

Magnitude of current density


bar 4
bar 5
150 150
bar 6
bar 7
100 100

50 50

0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Radial position in rotor bar [mm] Radial position in rotor bar [mm]
(a) FEM-simulation. (b) Proposed numerical method.

Figure 4.6: Locked rotor bar current density for the copper rotor.

200 200 Analytical


bar 1
180 180 FEM - average
bar 2
160 bar 3 160
140 bar 4 140
Phase [◦]

Phase [◦]

120 bar 5 120


100 bar 6 100
bar 7
80 80
60 60
40 40
20 20
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Radial position in rotor bar [mm] Radial position in rotor bar [mm]
(a) FEM-simulation. (b) Proposed numerical method.

Figure 4.7: Locked rotor bar current phase relative to the bottom of the bar for
the copper rotor.

When simulating the machine performance the skin effect correction factors are
multiplied with the corresponding DC-values, giving the effective AC-values. Based
on the presented theory these factors vary with the rotor frequency according to
Figure 4.8. The more pronounced skin-effect in the copper rotor will somewhat
compensate for the lower resistivity. If the locked rotor torque should be main-
48 CHAPTER 4. EFFECTS DURING A DIRECT-ON-LINE START

tained when changing from aluminium rotor to copper rotor, the rotor slot must be
redesigned. In this case, it could be obtained by the use of a double cage conductor,
increasing the skin-effect even further, but at the expense of reduced power factor
at rated speed.

2.6
2.4 kr - Al

Skin effect correction factors


2.2 kx - Al
2 kr - Cu
1.8
kx - Cu
1.6
1.4
1.2
1
0.8

0 5 10 15 20 25 30 35 40 45 50
Frequency [Hz]

Figure 4.8: Skin-effect correction factors for the studied rotor slot.

4.2 Saturation of the leakage paths

At low rotor speeds when the current is high, the large differential and slot leakage
fluxes saturate the stator and rotor tooth tips. Regarding the rotor slots, this was
already indicated in the previous section. The slot design has a large impact on
the saturation level during a start, especially the design of the tooth tips. In [23],
Agarwal and Alger show that if the tooth tips are designed correctly, a reduction
of the starting current can be obtained with a negligible change of the power fac-
tor at rated speed. There are different techniques to account for these saturation
effects. One idea is to adjust the length of the slot opening depending on the level
of estimated saturation [24]. In [25], Chalmers introduce saturation factors similar
to those commonly used to account for skin-effect, showing good agreement with
measurements.

The tooth tip saturation will affect the distribution of all space harmonics.
However, the complete analysis of these effects is beyond the scope of this thesis.
In this work the influence of the leakage path saturation on the fundamental current
is investigated. In order to simplify the analysis a combined analytical and finite
element model is used. The method defines impedance correction factors depending
on the fundamental current. These equations are then solved with initial values
from a FEM-simulated locked rotor test.
4.2. SATURATION OF THE LEAKAGE PATHS 49

4.2.1 Model used to account for saturation


During an online start the rotor impedance is much smaller than the magnetizing
impedance. The magnetizing branch in the equivalent circuit can therefore be
neglected at low rotor speeds. The resulting circuit is shown in Figure 4.9. The
stator and rotor resistances are usually quite small. The starting current is therefore
mainly determined by the value of the leakage inductances. The saturated reactance
is defined as:
1
Xsat = Xunsat (4.13)
ksat

The large fundamental current is the reason for the leakage path saturation. The
model is developed based on the assumption that the level of leakage paths satu-
ration is proportional to the stator current. If the saturation factor is assumed to
have the following variation with the rotor slip;

A
ksat (s) = 1 + 1 (4.14)
B+ s

the validity of the model will depend on the definition of the constants A and B.
The first condition used for the calculation of these constants is obtained from a
finite element simulation of a locked-rotor test. From this simulation the saturation
factors can be obtained giving the boundary condition as zero speed.

ksat (s = 1) = kF EM (s = 1) (4.15)

Based on the results from the FEM-simulation, the stator current is studied as
function of the terminal voltage. Figures 4.10(a) and 4.10(b) show the normalized
current for the aluminum rotor and the copper rotor, respectively. The red lines
showing the starting current obtained if saturation is neglected. Du to saturation
effects the starting current is increased by as much as 37 % for the copper motor

I1 R1 jX1

+
R21
s
U1

jX21

Figure 4.9: Equivalent circuit during online start.


50 CHAPTER 4. EFFECTS DURING A DIRECT-ON-LINE START

8 9
FEM - IF EM FEM - IF EM
7 8
Linear extrapolation - Ik Linear extrapolation - Ik


In

In
I

I
7


6

Normalized stator current

Normalised stator current


6
5
5
4
4
3
3
2
2
1 1

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalized stator voltage UUn Normalized stator voltage UUn
(a) Aluminium rotor. (b) Copper rotor.

Figure 4.10: FEM-simulated locked rotor test.

and 34 % for the aluminium motor. The saturation factor at zero speed is defined
as:
IF EM
kF EM (s = 1) = (4.16)
Ik

The stator current remains quite high during the rotor acceleration, up to the
speed where the peak-torque occurs, above this speed it decreases rapidly. The
value of the current at this point is therefore used when defining the second condi-
tion for the calculation of the constants A and B. The stator current at peak-torque
Ip , can be estimated from the locked rotor current by the use of a circle diagram.
According to Alm in [26], when neglecting the magnetizing current and the iron
losses, this current is obtained as shown in Figure 4.11.

The phase displacement between the stator voltage and current φk is, for the
studied machine, close to 60◦ , for both the aluminium rotor and the copper rotor. In
this case, the graphical solution of the problem gives the following relation between
the starting current and the current at peak torque:
1
Ip = √ Ik (4.17)
3
Thus, the saturation factor is decreased by the same amount giving:
kF EM (s = 1) − 1
ksat (s = sp ) = 1 + √ (4.18)
3
Neglecting the stator resistance, the corresponding speed is found as:
R21
sp = (4.19)
X21
4.2. SATURATION OF THE LEAKAGE PATHS 51

Ik
Ip

ϕk

Figure 4.11: Simplified circle diagram for the induction motor defining the starting
current and the current at break-down torque.

Based on Equation 4.14, with the conditions defined by Equations 4.16 and 4.18,
the saturation factor can be obtained as a function of fundamental rotor slip. The
method ensures that the saturation factor is roughly proportional to the stator
current during a start. Figure 4.12 shows the saturation factors for the studied
machine as a function of rotor speed. As the peak-torque occurs at a higher speed
for the copper rotor, the two saturation curves becomes a bit different. In the

1.4

1.35

1.3
Saturation factor

1.25

1.2

1.15

1.1
Al-rotor
1.05 Cu-rotor

1
-4 -2 0 2 4 6 8 10 12 14 16
Rotor speed [× 100 rpm]

Figure 4.12: Saturation factors as a function of rotor speed.


52 CHAPTER 4. EFFECTS DURING A DIRECT-ON-LINE START

forthcoming analysis, this method is used when calculating the fundamental stator
current during a direct-online-start.

4.2.2 Iron losses due to leakage flux


The large leakage flux during start gives rise to additional iron losses in the stator
and rotor teeth. In the equivalent circuit, these losses can be modeled by introduc-
ing additional resistances in parallel with the stator and rotor leakage reactances
[22], as shown in Figure 4.13.

As a result, the equivalent leakage reactance is reduced and the equivalent


resistance is increased. Referred to the stator side, the increase in resistance is
given by; √
∆R(s) = kLR (X1 + X21 ) s (4.20)
and the reduction of the leakage reactance is given by;

∆X(s) = kLX (X1 + X21 ) s (4.21)
Where X1 is the stator leakage reactance and X21 is the rotor leakage reactance,
referred to the stator side. The coefficients kLR and kLR are empirical constants
that are obtained from short-circuit tests on a large number of rotors.
These values can be used when correcting the analytical starting current ac-
cording to measurements.
RF e
+∆R −∆X X
X ⇔

Figure 4.13: Additional resistance taking iron losses into account.

4.3 Rotor losses and starting torque


In Chapter 3, the rotor bar currents were derived as a function of stator current and
inter-bar resistivity. The case study of the 4-pole machine having 36 stator slots
and 44 rotor slots showed the effects of a finite inter-bar resistance on bar current
distribution. In this section the corresponding torque is calculated, showing the
effects of inter-bar currents on the starting performance.

4.3.1 Rotor losses during a start


According to the model used, the losses created within the rotor cage is derived
from three different regions. The rotor bars, the short-circuit rings and the contact
4.3. ROTOR LOSSES AND STARTING TORQUE 53

region between bar and core. This defines the total loss created by a bar current of
order n as:
P2n = Pbn + Prn + Ptn (4.22)
Where the bar losses are;
Z ℓ
2
Pbn = Qr Rbn Ibn (x)2 dx (4.23)
− 2ℓ

and the end ring losses, derived through the ring currents becomes:

Ibn ( 2ℓ )2 + Ibn ( −ℓ
2 )
2
Prn = Qr Ran  jnpπ
2 (4.24)
1 − e Qr

Finally the inter-bar current losses created in the contact region.


Z ℓ
2
Ptn = Qr Rn Jtn (x)2 dx (4.25)
− 2ℓ

The airgap power, which is the power transferred through the airgap, is given
by:
ωTn
Pδn = np (4.26)
2
Together with the equation describing the relation between airgap power and cage
losses,
P2n = sn Pδn (4.27)
the developed shaft torque is obtained as:
P2n np
Tn = (4.28)
sn ω 2
The total shaft torque is given by the sum of all the harmonic torques.

X
T = Tn (4.29)
n=1

If the torque of order n at slip sn is positive, it contributes to the acceleration. If


it is negative it counteracts the accelerating torque, reducing the acceleration.

4.3.2 Effects of a finite inter-bar resistance on starting torque


In Chapter 3 it was indicated that the inter-bar currents are of minor importance
in the unskewed rotor. This is further reinforced by Figure 4.14(a), showing the
obtained starting torque. The total torque is more or less the same for all the
studied inter-bar resistivities. Although fundamental inter-bar currents are present
54 CHAPTER 4. EFFECTS DURING A DIRECT-ON-LINE START

350

5 · 10−6 ≤ Rtn ≤ 5 · 10−2


300

250

Torque [Nm] 200

150

100

50

0
-400 -200 0 200 400 600 800 1000 1200 1400
Speed [rpm]
(a) Total torque.

250

200 n=1
n = −5
150 n=7
n = −11
Torque [Nm]

100
n = 13
n = −17
50
n = 19

-50

-100

-150
-400 -200 0 200 400 600 800 1000 1200 1400
Speed [rpm]
(b) Torque components.

Figure 4.14: Starting torque of Motor A with unskewed rotor.


4.3. ROTOR LOSSES AND STARTING TORQUE 55

at the rotor ends at zero speed, the inter-bar resistivity is too small to create any
torque.

As expected for an unskewed rotor, there are large asynchronous torques during
start-up. The torque components considered are shown separately in Figure 4.14(b).
Apart from the fundamental component, the dominating torques are those created
by the first order slot harmonics. These torques reduce the rotor acceleration during
the start, and at rated speed they create additional losses. It should be noted that
these torques do not become zero at fundamental synchronous speed. A small
breaking torque at this high speed can create considerable additional losses. Again,
at no-load when the main flux path usually is saturated, these braking torques are
increased even further due to the increased fundamental current. This effect is not
included in this simulation.

However, these effects are one of the reasons why rotors generally are skewed.
Figure 4.15 shows the starting torque when the rotor is skewed by one stator slot
pitch. When the inter-bar resistivity is high the asynchronous torques caused by
the slot harmonics are efficiently suppressed. When the inter-bar resistivity gets
lower the starting torque decreases rapidly. And at a certain level the machine
might not even be able to start.

400
Rtn = 5 · 10−2 Ωm

350 Rtn = 5 · 10−3 Ωm


Rtn = 5 · 10−4 Ωm
300 Rtn = 5 · 10−5 Ωm
Rtn = 5 · 10−6 Ωm
250
Torque [Nm]

200

150

100

50

0
-400 -200 0 200 400 600 800 1000 1200 1400
Speed [rpm]

Figure 4.15: Starting torque of Motor A with a rotor skewed by one stator slot
pitch.
56 CHAPTER 4. EFFECTS DURING A DIRECT-ON-LINE START

275

250

225

200

175

Torque [Nm] 150

125
Rtn = 5 · 10−2 Ωm
100 Rtn = 5 · 10−3 Ωm
75 Rtn = 5 · 10−4 Ωm
50 Rtn = 5 · 10−5 Ωm

25 Rtn = 5 · 10−6 Ωm

0
-400 -200 0 200 400 600 800 1000 1200 1400
Speed [rpm]
(a) Fundamental torque.

120

100 Rtn = 5 · 10−2 Ωm

80 Rtn = 5 · 10−3 Ωm

60 Rtn = 5 · 10−4 Ωm

40 Rtn = 5 · 10−5 Ωm
Torque [Nm]

Rtn = 5 · 10−6 Ωm
20

-20

-40

-60

-80

-100
-400 -200 0 200 400 600 800 1000 1200 1400
Speed [rpm]
2Qs
(b) Torque caused by slot harmonics, dashed lines are representing n = 1 − p
2Qs
and solid lines are representing n = 1 + p
.

Figure 4.16: Main torque components for Motor A with a rotor skewed by one
stator slot pitch.
4.4. SUMMARY 57

As there are inter-bar currents having fundamental rotor frequency, the fun-
damental torque is increased. This can be seen in Figure 4.16(a), showing the
fundamental components at different inter-bar resistivities, indicating that a con-
siderable increase of the fundamental torque can be obtained. However, for this
machine, in the motoring region the braking torques are obviously larger than this
increase. The torques created by the first pair of slot harmonics are the main
cause of this effect, shown in Figure 4.16(b). For low inter-bar resistivities, the
asynchronous torques are much larger than in the unskewed rotor, and their mag-
nitudes remain large even at speeds well above their synchronous speed. This can
result in a considerable decrease of the pull-out torque. This machine will most
likely have large stray-load losses unless the cage is insulated.

4.4 Summary
When a cage induction motor is started directly against the grid, difficulties may
arise in the calculation of the starting current. FEM-simulations of the studied ma-
chine have shown that the starting current is increased by 34% with an aluminum
rotor and 37% with a copper rotor due to the presence of leakage path saturation.

A numerical model used to account for the skin-effect has been verified with
FEM-simulations. A combined analytical and finite element model has been devel-
oped for the calculation of the fundamental starting current, taking saturation of
the slot- and differential leakage paths into account. A method to introduce the
effects of additional iron losses during a start has been introduced.

It have been shown that the effects of the inter-bar currents on the starting
torque in unskewed rotors are of minor importance. While in skewed rotors, the
inter-bar currents can have a considerable effect on the motor starting performance.
The inter-bar currents having fundamental frequency, contribute to the fundamen-
tal torque, i.e. they create useful torque and should not be considered as a source
of losses. It have been shown that the inter-bar currents created by the slot har-
monics can cause large asynchronous torques and are still very large even at speeds
well above their synchronous speed, resulting in a reduced pull-out torque. In some
cases the machine might not even be able to start.
Chapter 5

Simulation results and measurements

In this chapter, the 11 kW machine having 36 stator slots and 28 rotor slots is
simulated with either an aluminium- or a copper cage rotor, skewed by one stator
slot pitch. The differences in starting performance between the copper and the alu-
minium rotor is studied. In the analysis, the starting torque is calculated using the
values of inter-bar resistivities obtained from measurements on the corresponding
rotors. The additional rotor losses at rated speed, caused by the inter-bar currents,
are also calculated as a function of the inter-bar resistivity.
In order to verify the results, measurements of the starting current and torque
have been performed.

5.1 Simulation results


Based on the models described in the previous chapters, the machine performance
is calculated at different rotor slips s and inter-bar resistivities Rtn . In the anal-
ysis, space harmonics of order n, up to the first pair of stator slot harmonics are
considered. The calculation procedure is described in the flow-chart in Figure 5.1.

5.1.1 Starting torque


The torque speed characteristics for the two rotor concepts are calculated when
starting direct-on-line at rated voltage, neglecting the line impedance.
First the torque is calculated assuming insulated rotor bars. Secondly the mea-
sured inter-bar resistivities are used for the study of inter-bar effects. Figure 5.2
shows the torque speed characteristics in the case of insulated and uninsulated rotor
bars for the two motor concepts.

In the case of insulated rotor bars, the stator slot harmonics are sufficiently sup-
pressed by the rotor skew. Since the 5th and the 7th space harmonics have larger
wave lengths than the slot harmonics, they still cause asynchronous torques. As

59
60 CHAPTER 5. SIMULATION RESULTS AND MEASUREMENTS

Start
For: Rtn

For: n

For: s

If n = 1 If n 6= 1

Calculate coefficients accounting for;


Calculate skin-effect coefficients.
skin-effect, saturation and additional iron losses.

Calculate stator current


from equivalent circuit. From the fundamental stator
current, calculate; rotor- ring,
bar and inter-bar currents.
Calculate the rotor- ring, bar
and inter-bar currents.

Calculate the terminal voltage from Estimate a new value


the stator and rotor fluxes. of the stator current.

No
Is the terminal voltage correct?

Yes

Calculate the rotor losses and torque.

End

Figure 5.1: Procedure for the calculation of motor performance at different speeds
and inter-bar resistivities.
5.1. SIMULATION RESULTS 61

250

200

150
Torque [Nm]

100

50
Al
Cu

0
-400 -200 0 200 400 600 800 1000 1200 1400 1600
Speed [rpm]
(a) Calculated for insulated rotor bars.

250

200

150
Torque [Nm]

100

50
Al with Rtn = 8, 0 µΩm
Cu with Rtn = 0, 35 µΩm

0
-400 -200 0 200 400 600 800 1000 1200 1400 1600
Speed [rpm]
(b) Calculated at measured values of inter-bar resistivity.

Figure 5.2: Simulated starting torque for the studied aluminium and copper rotors
skewed by one stator slot pitch.
62 CHAPTER 5. SIMULATION RESULTS AND MEASUREMENTS

275
250
225
200 n=1
175 n = −5
150

Torque [Nm]
n=7
125
n = −11
100
n = 13
75
n = −17
50
n = 19
25
0
-25
-50
0 500 1000 1500
Speed [rpm]
(a) Aluminium rotor with Rtn = 8, 0 µΩm.

275
250
225
200 n=1
175 n = −5
150
Torque [Nm]

n=7
125
n = −11
100
n = 13
75
n = −17
50
n = 19
25
0
-25
-50
0 500 1000 1500
Speed [rpm]
(b) Copper rotor with Rtn = 0, 35 µΩm.

Figure 5.3: Starting torque components for the studied aluminium and copper
rotors calculated with measured values of inter-bar resistivity.
5.1. SIMULATION RESULTS 63

there are no inter-bar currents, the pull-out torques were expected to be the same
for the two motor concepts. However, the short-circuit ring of the copper rotor is
somewhat smaller than the short-circuit ring of the aluminium rotor. This, com-
bined with the fact that the copper motor is slightly more saturated in the leakage
paths during a start, results in a somewhat higher pull-out torque of the copper
rotor.

With uninsulated rotor bars, the skewing is no longer as effective. Large asyn-
chronous torques occurs due to the first order slot harmonics, indicating inter-bar
current flow. Contrary to the results obtained for the motor having 44 rotor slots
simulated in chapter 4, the inter-bar currents seem to increase the pull-out torque
of the aluminum rotor. The pull-out torque is now 4,5 % higher in the aluminium
rotor than in the copper rotor. This is most likely due to a more suitable slot num-
ber combination, reducing the influence of the slot harmonics on the starting torque.

The different torque components contributing to the starting torque in the alu-
minium and the copper rotor are shown in Figure 5.3(a) and Figure 5.3(b), re-
spectively. Due to the very low inter-bar resistivity in the copper rotor, the slot
harmonic torques becomes quite narrow, resulting in a lower braking torque at high
speeds than the the corresponding torques in the aluminium rotor. From these re-
sults it can be concluded that, for the studied machine, the higher pull-out torque
of the aluminium rotor is rather due to an increase of the fundamental torque than
due to the braking torques from the space harmonics in the copper rotor.

5.1.2 Rotor losses at rated speed


At rated speed the airgap space harmonics create high frequency rotor cage losses.
If the rotor bars are insulated these losses can be suppressed by rotor skewing. In-
vestigation of the starting torque characteristics has however indicated that unin-
sulated rotor bars increase the high frequency currents in the rotor circuit. The
dependency of the rotor stray-load losses is therefore studied as a function of the
inter-bar resistivity at rated power.
As the torque depends on the inter-bar resistivity the slip is recalculated to
maintain the shaft power. In order to get a realistic value of the rotor slip, the
simulation is performed at a motor temperature of 75 ℃, and measured values
of friction- and iron losses are included in the analysis. Since the temperature
dependency of the inter-bar resistivity is unknown, no correction is performed on
this value. Figure 5.4 shows the high frequency cage losses obtained in the studied
machines, the rotors being either skewed by one stator slot pitch or unskewed.

The rotor without skew is not affected by inter-bar currents, as long as the
impedance of the short circuit ring to inter-bar resistance ratio is low. When the
rotor is skewed by one stator slot pitch and the rotor bars are insulated, the volt-
ages induced by the stator slot harmonics are efficiently suppressed. The resulting
64 CHAPTER 5. SIMULATION RESULTS AND MEASUREMENTS

120 120
αs = 1 αs = 1
100 αs = 0 100 αs = 0

Additional rotor losses [W]

Additional rotor losses [W]


Rtn Al-1 Rtn Cu-1
80 Rtn Al-2 80 Rtn Cu-2
Rtn Cu-3
60 60

40 40

20 20

0 0
10−8 10−6 10−4 10−2 100 10−8 10−6 10−4 10−2 100
Inter-bar resistivity Rtn [Ωm] Inter-bar resistivity Rtn [Ωm]
(a) Aluminium rotor (b) Copper rotor.

Figure 5.4: High frequency cage losses as a function of inter-bar resistivity Rtn at
75 ℃.

losses are mainly due to the lower order phase belt harmonics.

As the bar to core resistance decreases from a very large value, the high fre-
quency losses start to increase rapidly. The maximum value of these losses is
strongly influenced by the number of stator and rotor slots [15]. According to the
model used, for the studied machine, these losses can be as large as 1 % of the
output power. If the bar to core resistivity is reduced even further, the additional
losses starts to decrease, and reaches the value of losses equal to the losses of the
unskewed rotor. The higher bar to core resistivity in the aluminium rotor is, in this
case, resulting in higher inter-bar current losses than the equivalent copper rotor.

5.2 Measurements of starting torque


The torque during start-up was measured on the rotors having 28 slots skewed by
one stator slot pitch. One aluminum rotor and one copper rotor have been tested.
The torque was measured dynamically during start when the machine was loaded
with a flywheel. The unfiltered torque signal was sampled at a high frequency
and the signal noise in the sampled data was suppressed by the use of a low-pass
butterworth filter of the 5th order with a cut-off frequency selected sufficiently high.

5.2.1 Measurement setup


A rotating torque transducer was used to measure the torque. The torque trans-
ducer, Magtrol TM-312 rated at 200 Nm, was mounted between the motor shaft
and the flywheel, according to Figure 5.2. The flywheel, having a moment of inertia
5.2. MEASUREMENTS OF STARTING TORQUE 65

Flywheel
Torque
transducer Motor

Figure 5.5: Setup for measurements of starting torque.

0.4
of 1.6 kgm2 , reduces the rotor acceleration making it possible to capture the torque
signal.

First, the 4-pole motor was accelerated in reverse direction to 1000 rpm. Then,
the torque was measured after shifting two phases of the sinusoidal supply voltage,
forcing the machine to accelerate in the opposite direction. By doing this, the in-
fluence of the asynchronous torques could be measured with minimum distortion
from the switching transient.

300 1600
Measured Measured
250 Filtered 1200 Filtered

200
800
Torque [Nm]

Speed [rpm]

150
400
100
0
50

0 -400

-50 -800
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Time [s] Time [s]
(a) Torque (b) Speed.

Figure 5.6: Measured and filtered torque and speed as a function of time.
66 CHAPTER 5. SIMULATION RESULTS AND MEASUREMENTS

Using this measuring procedure, resulted in an acceleration time of approxi-


mately 3 seconds for the copper rotor at rated voltage. Figure 5.6 shows a typical
result of the measured torque and speed as a function of time. The red lines show
the corresponding values obtained after the filtering process. One can see that the
there are huge fluctuations in the torque when starting direct-on-line. Therefore,
the use of a low-pass filter is necessary in order to validate the static model used
in the analysis.

5.2.2 Results from measurements


The RMS-value of the starting current, obtained after postprocessing of the mea-
sured sinusoidal current, is shown in Figure 5.7(a) and Figure 5.7(b) for the alu-
minium and the copper rotor, respectively. The current has been adjusted linearly
to account for the voltage drop during start. For a comparison, the corresponding
analytical values are calculated in three different ways. Firstly, they are calculated
by only considering the skin effect in the rotor bars. Secondly, the effect of satura-
tion of the leakage paths is introduced. And finally, the effect of iron losses in the
leakage paths is included.
From these results it can be concluded that the high level of leakage flux during
a start has considerable impact on the starting current due to saturation effects
and iron losses. These effects have to be considered when calculating a machines
starting performance.

Figure 5.8 shows the measured starting torques for the two rotor concepts. The
torque has been adjusted in quadratic relation to the voltage to account for the
voltage drop during the start. Large asynchronous torques are caused by the first
order stator slot harmonics. This verifies the prediction from the analytical model,

200 220
180 200
160 180

140 160
140
Current [A]

Current [A]

120
120
100
100
80
Measured 80 Measured
60 60
Skin-effect Skin-effect
40 Skin-effect + saturation 40 Skin-effect + saturation
20 Skin-effect + saturation + iron-losses 20 Skin-effect + saturation + iron-losses
0 0
-400 -200 0 200 400 600 800 1000 1200 1400 1600 -400 -200 0 200 400 600 800 1000 1200 1400 1600
Speed [rpm] Speed [rpm]
(a) Aluminium rotor (b) Copper rotor.

Figure 5.7: Simulated and measured starting currents.


5.2. MEASUREMENTS OF STARTING TORQUE 67

250

200

150
Torque [Nm]

100

50
Al
Cu

0
-400 -200 0 200 400 600 800 1000 1200 1400 1600
Speed [rpm]

Figure 5.8: Measured torque for the aluminium and the copper rotor when starting
direct-on-line at rated voltage.

which showed that the inter-bar currents are counteracting the effect of rotor skew,
and that the pull-out torque of the aluminum rotor is higher than for the copper
rotor. The measured pull-out torque of the aluminium rotor is 7 % higher than the
pull-out torque of the copper rotor. This is even larger than expected theoretically.

After the introduction of the coefficients accounting for additional iron losses in
the leakage paths, the simulated starting current correlated well with the measure-
ments. By comparing the corresponding starting torques, one can get an idea of
the accuracy of the analytical model used to calculate the rotor losses. Figure 5.9
shows a comparison between the measured and the simulated starting torques for
the two rotor concepts.

It should be mentioned that, since the simulations are based on a static model,
it is difficult to model rapid changes in the rotor acceleration. That is probably
the reason for the overestimated pull-out torques. However, given that the machine
is heavily saturated during start, the overall torque speed characteristics show ac-
ceptable correlation, except for the asynchronous torque caused by the 5th space
harmonic, which is overestimated by the analytical model. Regarding the asyn-
chronous torque caused by the 7th space harmonic, it is difficult to draw conclusions
as a synchronous torque is present at the same speed.
As a result of the calculated rotor losses caused by the inter-bar currents, the
68 CHAPTER 5. SIMULATION RESULTS AND MEASUREMENTS

250

200

150

Torque [Nm]

100

50
Simulated
Measured

0
-400 -200 0 200 400 600 800 1000 1200 1400 1600
Speed [rpm]
(a) Aluminium rotor

250

200

150
Torque [Nm]

100

50
Simulated
Measured

0
-400 -200 0 200 400 600 800 1000 1200 1400 1600
Speed [rpm]
(b) Copper rotor.

Figure 5.9: Simulated and measured torques for the aluminium and the copper
rotor when starting direct-on-line at rated voltage.
5.3. SUMMARY 69

torque speed characteristic of the slot harmonic torques was predicted to be dif-
ferent for the copper- and the aluminium rotors. This result is verified by the
measurements, showing a similar behavior, indicating that the analytical model
seems to give a reasonable estimate of the inter-bar current losses.

Note that the rotor slot harmonics are neglected in the analysis, which results
in less distortion of the simulated torque, especially since the rotor slots are semi-
closed and not closed.

5.3 Summary
Good agreement has been demonstrated between simulated and measured starting
characteristics for Motor B, for both aluminium and copper rotors.
Simulations have shown that the pull-out torque of the studied aluminium rotor
is higher than that for the equivalent copper rotor. This is rather due to an increase
of the fundamental starting torque of the aluminium rotor, than due to the braking
torques from the space harmonics in the copper rotor.
Measurements have shown that the difference between the pull-out torques is
even larger than calculated from the model. The measured pull-out torque of the
studied aluminium rotor was 7 % higher than for the equivalent copper rotor.
Thereby, it can be concluded that the inter-bar currents have a considerable effect
on motor starting performance.
At rated speed the braking torques are larger in the aluminium rotor than in
the copper rotor. This is seen as increased harmonic joule losses in the rotor cage.
Simulations have shown, that these losses can be as large as 1 % of the output
power for the studied machine.
Chapter 6

Conclusions and Future work

6.1 Conclusions

A numerical model used to account for the skin-effect has been verified with FEM-
simulations. A combined analytical and finite element model has been developed
for the calculation of the fundamental starting current, taking saturation of the
slot- and differential leakage paths into account. A method to introduce the effects
of additional iron losses during a start has been introduced. FEM-simulations of
the studied machine have shown that the starting current is increased with approx-
imately 35 % due to the presence of leakage path saturation and skin-effect.

A test-rig has been built for the measuring of rotor voltages, from which the
inter-bar resistivity has been calculated. Measurements have shown that the inter-
bar resistivity is as much as 10 times higher in cast aluminum than in cast copper
rotors. The aluminium rotors showed results indicating an unevenly distribution
of the inter-bar resistivity, while the copper rotors where indicating a more evenly
distributed inter-bar resistivity.

A model to include the effects of inter-bar currents in cage induction motors


has been derived. Simulations have shown that the inter-bar current density is
increasing rapidly with decreasing inter-bar resistivity in skewed machines. This
affects the bar current distribution and phase angle in such a way that the skewing
is made ineffective, these results are consistent with the findings in [18].

It have been shown that in skewed rotors, the inter-bar currents can have a
considerable effect on the motor starting performance. The inter-bar currents hav-
ing fundamental frequency, contribute to the fundamental torque, i.e. they create
useful torque and should not be considered as a source of losses. It have been
shown that the inter-bar currents created by the slot harmonics can cause huge
asynchronous torques and are still very large even at speeds well above their syn-

71
72 CHAPTER 6. CONCLUSIONS AND FUTURE WORK

chronous speed, resulting in a reduced pull-out torque. For Motor A, in some cases
the machine might not even be able to start. These effects are strongly dependent
on the combination of number of stator and rotor slots.

Good agreement has been demonstrated between simulated and measured start-
ing characteristics for Motor B with both aluminium and copper casted rotors.
Simulations have shown that the pull-out torque is 4,5% higher for the aluminium
rotor than for the equivalent copper rotor. This is rather due to an increase of the
fundamental starting torque of the aluminium rotor, than due to braking torques
from the space harmonics in the copper rotor. Measurements have, however, shown
that the difference between the pull-out torques is even larger than calculated from
the model. The measured pull-out torque of the studied aluminium rotor was 7%
higher than for the equivalent copper rotor. Thereby, it can be concluded that the
inter-bar currents have a considerable effect on motor starting performance.
At rated speed the braking torques are larger in the aluminium rotor than in
the copper rotor. This is seen as increased harmonic joule losses in the rotor cage.
Simulations have shown, that these losses can be as large as 1% of the output power
for the studied machine.

6.2 Future work guidelines


Based on the results obtained in this thesis, for a cast copper rotor having a suitable
number of slots skewed by one stator slot pitch, one can expect similar performance
as for an equivalent rotor without skewing. This could be verified by measurements
on prototype machines. In case of a cast aluminium rotor, suitable measurements
should be made to verify if the stray-load losses are larger than in cast copper rotors
or not. It could also be favorably to study the starting torque and efficiency of a
machine having an unsuitable slot number combination, increasing the inter-bar
current flow, making the verification of the analytical model much easier.
In general, unskewed rotors are said to create larger noise levels than skewed
rotors. As the inter-bar currents seem to counteract the rotor skew, it is of interest
to measure and compare the noise levels of skewed and unskewed rotors having
low inter-bar resistivities. Further, in order to suppress both noise and inter-bar
currents, the concept of asymmetrical rotor slots studied among others in [27],
should be evaluated by measurements on a prototype machine.
Bibliography

[1] A.R. Hagen, A. Binder, M. Aoulkadi, T. Knopik, and K. Bradley. Comparison


of measured and analytically calculated stray load losses in standard cage
induction machines. 18th International Conference on Electrical Machines,
pages 1–6, 2008.

[2] H. Nishizawa, K. Itomi, S. Hibino, and F. Ishibashi. Study on reliable reduction


of stray load losses in three-phase induction motor for mass production. IEEE
Transactions on Energy Conversion, EC-2:489–495, 1987.

[3] A. Boglietli, A. Cavagnino, M. Lazzari, and A. Pastorelli. Induction motor


efficiency measurements in accordance to ieee 112-b, iec 34-2 and jec 37 inter-
national standards. IEEE International Electric Machines and Drives Confer-
ence, 3:1599–1605, 2003.

[4] A.A. Jimoh, R.D. Findlay, and M. Poloujadoff. Stray losses in induction ma-
chines: Part i, definition, origin and measurement. IEEE Transactions on
Power Apparatus and Systems, PAS-104:1500–1505, 1985.

[5] Y.N. Feng, J. Apsley, S. Williamson, A.C. Smith, and D.M. Ionel. Reduced
losses in die-cast machines with insulated rotors. IEEE International Electric
Machines and Drives Conference, pages 57–64, 2009.

[6] A. M. Odok. Stray-load losses and stray torques in induction machines. Power
Apparatus and Systems, Transactions of the American Institute of Electrical
Engineers, 77(3):43–53, 1958.

[7] P. L. Alger. Induction Machines: Their Behavior and Uses. Taylor & Francis,
1995.

[8] A. Harson, P.H. Mellor, and D. Howe. Design considerations for induction
machines for electric vehicle drives. Seventh International Conference on Elec-
trical Machines and Drives, pages 16–20, 1995.

[9] K. Dabala. Modified method to determine rotor bar-iron resistance in three-


phase copper casted squirrel-cage induction motors. Proceedings of ICEM,
pages 231–234, 2006.

73
74 BIBLIOGRAPHY

[10] P. L. Alger. Induced high-frequency currents in squirrel-cage windings. Power


Apparatus and Systems, Transactions of the American Institute of Electrical
Engineers, 76(3):724 – 729, 1957.
[11] K. Dabala. A new experimental-computational method to determine rotor
bar-iron resistance. Proceedings of ICEM, 2:69–72, 1996.
[12] O. Aglén. Calorimetric Measurements of Losses in Induction Motors. Licen-
tiate thesis, Royal Institute of Technology, Stockholm, Sweden, 1995.
[13] A.H. Bonnett and T. Albers. Squirrel-cage rotor options for ac induction
motors. IEEE Transactions on Industry Applications, pages 1197–1209, 2001.
[14] A. Behdashti and M. Poloujadoff. A new method for the study of inter-bar
currents in polyphase squirrel-cage induction motors. IEEE Transactions on
Power Apparatus and Systems, PAS-98(3):902–911, 1979.
[15] B. Heller and V. Hamata. Harmonic Field Effects in Induction Machines.
Elsevier Science Ltd, 1977.
[16] M. Ivanes and M. Bourmault. Etudes des pertes supplementaires dans les
moteurs asynchrones. Technical report, Cie Electro-Mecanique, October 1968.
[17] A. Behdashti. Contribution a l’etude des pertes supplementaires des machines
asynchrones dans une tres large zone de fonctionnement. PhD thesis, L’Institut
national polytechnique de Grenoble, June 1975.
[18] A.C. Smith, S. Williamson, and C.Y. Poh. Distribution of inter-bar currents
in cage induction machines. Second International Conference on Power Elec-
tronics, Machines and Drives, 1:297–302, 2004.
[19] P. L. Alger and J. H. Wray. Double and triple squirrel cages for polyphase in-
duction motors. Power Apparatus and Systems, Transactions of the American
Institute of Electrical Engineers, 72(2):637 – 645, 1953.
[20] M. Liwschitz-Garik. Skin-effect bars of squirrel-cage rotors. Power Apparatus
and Systems, Transactions of the American Institute of Electrical Engineers,
73(1):255 – 258, 1954.
[21] S. Williamson and D. R. Gersh. Finite element calculation of double-cage
rotor equivalent circuit parameters. IEEE Transactions on Energy Conversion,
11(1):41–48, 1996.
[22] C. Sadarangani. Electrical machines - design and analysis of induction and
permanent magnet motors. KTH Hogskoletryckeriet, 2000.
[23] P. D. Agarwal and P. L. Alger. Saturation factors for leakage reactance of in-
duction motors. Power Apparatus and Systems, Transactions of the American
Institute of Electrical Engineers, 79(3):1037–1042, 1960.
BIBLIOGRAPHY 75

[24] H. M. Norman. Induction motor locked saturation curves. Transactions of the


American Institute of Electrical Engineers, 53(4):536 – 541, 1934.
[25] B. J. Chalmers and R. Dodgson. Saturated leakage reactances of cage induction
motors. Proceedings IEE, 116(8):1395–1404, 1969.
[26] E. Alm. Elektroteknik, Band 3, Del 2B, Elektromaskinlära - Asynkronmaski-
nens teori, driftegenskaper och beräkning. Alb. Bonniers boktryckeri, 1931.
[27] R. Chitroju. Improved Performance Characteristics of Induction Machines
with Non-Skewed Asymmetrical Rotor Slots. Licentiate thesis, Royal Institute
of Technology, Stockholm, Sweden, 2009.
List of Figures

1.1 Stray-load loss components (0,2-37 kW induction motors) [2]. . . . . . . 2

2.1 Rotor test setup for measurements of inter-bar resistance. . . . . . . . . 9


2.2 Conducting washer between test-plate and rotor short-circuit ring. . . . 10
2.3 Equivalent circuit of the rotor used for the calculation of the inter-bar
resistivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Calculated voltages for the aluminium rotor at a total current of 200 A. 12
2.5 Calculated voltages for the copper rotor at a total current of 200 A. . . 12
2.6 Two of the studied aluminium and copper rotors. . . . . . . . . . . . . . 13
2.7 Measured and calculated voltage UAX for rotor Al 2 at a total current
of 200 A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.8 Measured and calculated voltage UAX for rotor Cu 2 at a total current
of 100 A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.1 Definition of bar- and inter-bar currents in a small element of the rotor
circuit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2 Phase displacement between inter-bar currents. . . . . . . . . . . . . . . 18
3.3 Definitions of inter-bar resistances. . . . . . . . . . . . . . . . . . . . . . 19
3.4 Airgap flux density due to current in phase a, Model A. . . . . . . . . . 21
3.5 Airgap flux density due to current in phase a, Model B. . . . . . . . . . 22
3.6 Comparison between analytical and FEM-simulated airgap flux density
at no-load. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.7 Position of a wave of order n at time t = 0 in the rotor reference frame. 25
3.8 Current in bar number k and the corresponding MMF in the airgap. . . 27
3.9 Permeance variation along the airgap circumference as defined by Model
A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.10 Rotor currents at the boundary x = − 2ℓ . . . . . . . . . . . . . . . . . . . 31
3.11 Change in rotor bar current due to the interaction with inter-bar currents. 32
3.12 Magnitude of the fundamental locked rotor inter-bar current density and
the resulting angle β1 , for Motor A with unskewed rotor. . . . . . . . . 33
3.13 Magnitude of fundamental locked rotor bar current and the correspond-
ing angle γ1 , for Motor A with unskewed rotor. . . . . . . . . . . . . . . 34

76
List of Figures 77

3.14 Magnitude of the currents in the unskewed rotor caused by the funda-
mental and the first pair of slot space harmonics when Rtn = 5 · 10−5 Ωm. 35
3.15 Magnitude of the fundamental locked-rotor inter-bar current density and
the resulting angle β1 for Motor A with skewed rotor. . . . . . . . . . . 36
3.16 Magnitude of fundamental locked-rotor bar current and the correspond-
ing angle γ1 , for Motor A with skewed rotor. . . . . . . . . . . . . . . . 37
3.17 Magnitude of the currents in the rotor skewed by one stator slot pitch,
caused by the fundamental and the first pair of slot space harmonics
when Rtn = 5 · 10−2 Ωm. . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.18 Magnitude of the currents in the rotor skewed by one stator slot pitch,
caused by the fundamental and the first pair of slot space harmonics
when Rtn = 5 · 10−5 Ωm. . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.1 Different types of rotor bars. . . . . . . . . . . . . . . . . . . . . . . . . 42


4.2 The v th section of the rotor slot. . . . . . . . . . . . . . . . . . . . . . . 44
4.3 Procedure for the calculation of the impedance correction factors kr and
kx that accounts for skin- effect. . . . . . . . . . . . . . . . . . . . . . . 45
4.4 Locked rotor bar current density for the aluminium rotor. . . . . . . . . 46
4.5 Locked rotor bar current phase angle relative to the bottom of the bar
for the aluminium rotor. . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.6 Locked rotor bar current density for the copper rotor. . . . . . . . . . . 47
4.7 Locked rotor bar current phase relative to the bottom of the bar for the
copper rotor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.8 Skin-effect correction factors for the studied rotor slot. . . . . . . . . . . 48
4.9 Equivalent circuit during online start. . . . . . . . . . . . . . . . . . . . 49
4.10 FEM-simulated locked rotor test. . . . . . . . . . . . . . . . . . . . . . . 50
4.11 Simplified circle diagram for the induction motor defining the starting
current and the current at break-down torque. . . . . . . . . . . . . . . 51
4.12 Saturation factors as a function of rotor speed. . . . . . . . . . . . . . . 51
4.13 Additional resistance taking iron losses into account. . . . . . . . . . . . 52
4.14 Starting torque of Motor A with unskewed rotor. . . . . . . . . . . . . . 54
4.15 Starting torque of Motor A with a rotor skewed by one stator slot pitch. 55
4.16 Main torque components for Motor A with a rotor skewed by one stator
slot pitch. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.1 Procedure for the calculation of motor performance at different speeds


and inter-bar resistivities. . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.2 Simulated starting torque for the studied aluminium and copper rotors
skewed by one stator slot pitch. . . . . . . . . . . . . . . . . . . . . . . . 61
5.3 Starting torque components for the studied aluminium and copper rotors
calculated with measured values of inter-bar resistivity. . . . . . . . . . 62
5.4 High frequency cage losses as a function of inter-bar resistivity Rtn at
75 ℃. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.5 Setup for measurements of starting torque. . . . . . . . . . . . . . . . . 65
78 List of Figures

5.6 Measured and filtered torque and speed as a function of time. . . . . . . 65


5.7 Simulated and measured starting currents. . . . . . . . . . . . . . . . . . 66
5.8 Measured torque for the aluminium and the copper rotor when starting
direct-on-line at rated voltage. . . . . . . . . . . . . . . . . . . . . . . . 67
5.9 Simulated and measured torques for the aluminium and the copper rotor
when starting direct-on-line at rated voltage. . . . . . . . . . . . . . . . 68

You might also like