You are on page 1of 100

EXAMENSARBETE INOM ENERGI OCH MILJÖ,

AVANCERAD NIVÅ, 30 HP
STOCKHOLM, SVERIGE 2018

Design and comparison of


PMaSynRM versus PMSM for
pumping applications

VIKTOR BRIGGNER

KTH
SKOLAN FÖR ELEKTROTEKNIK OCH DATAVETENSKAP
TRITA TRITA-EECS-EX-2018:496

www.kth.se
Design and comparison of PMaSynRM versus PMSM for
pumping applications

VIKTOR BRIGGNER

Master of Science Thesis in Electrical Energy Conversion


at the School of Electrical Engineering and Computer Science
KTH Royal Institute of Technology
Stockholm, Sweden, August 2018.

Supervisors: Tanja Hedberg and Øystein Krogen


Examiner: Oskar Wallmark

TRITA-EECS-EX-2018:496
Design and comparison of PMaSynRM versus PMSM for pumping applications
VIKTOR BRIGGNER

c VIKTOR BRIGGNER, 2018.

School of Electrical Engineering and Computer Science


Department of Electric Power Engineering and Energy Systems
Kungliga Tekniska högskolan
SE–100 44 Stockholm
Sweden
Abstract
This master thesis aimed to design a permanent magnet assisted synchronous reluctance
machine (PMaSynRM) rotor for pump applications which were to be implemented in
an existing Induction Machine stator. The machine were to be compared with a similar
permanent magnet synchronous machine (PMSM) with similar torque production in terms
of cost and performance.
This thesis goes through the theory of the Synchronous Reluctance Machine and
the Permanent Magnet assistance. A rotor was designed by utilizing existing design ap-
proaches and simulation of performance by use of finite element analysis. A demagneti-
zation study was conducted on the added permanent magnets in order to investigate the
feasiblity of the design.
The final design of the PMaSynRM was thereafter compared to the equivalent
surface-mounted PMSM in terms of performance and cost. The performance parameters
was torque production, torque ripple, efficiency and power factor. Due to the lower torque
density of the PMaSynRM, for equal torque production the PMSM had a 40% shorter
lamination stack than the PMaSynRM.
The economic evaluation resulted in that when utilizing ferrite magnets in the PMa-
SynRM it became slightly cheaper than the PMSM, up to 20%. However, due to the fluc-
tuating prices of NdFeB magnets, there exist breakpoints below which the PMaSynRM is
in fact more expensive than the PMSM or where the price reduction of the PMaSynRM
is not worth the extra length of the motor. However, it was shown that the PMaSynRM
is very insensitive to magnet price fluctuations and thereby proved to be a more secure
choice than the PMSM

Keywords: Demagnetization, economic evaluation, permanent magnet assistance,


synchronous reluctance machine.

ii
Sammanfattning
Detta examensarbete avsåg att designa en rotor till en permanentmagnetsassisterad
synkron reluktansmaskin (PMaSynRM) för pumpapplikationer, vilken skulle implement-
eras i en befintlig asynkronmaskin (IM) stator. Maskinen jämfördes ekonomiskt och pre-
standamässigt med en liknande synkronmaskin med permanentmagneter (PMSM) med
jämförbar vridmomentsproduktion.
Uppsatsen avhandlar teorin bakom synkrona reluktansmaskiner och konceptet kring
permanentmagnetassistans. Rotorn designades genom användandet av befintliga design-
metoder och simulering genom finit elementanalys (FEA). En avmagnetiseringsstudie
utfördes på de adderade magneterna för att undersöka rimligheten kring designen
Den slutgiltiga designen av PMaSynRMen jämfördes därefter mot den jämlika
PMSMen i termer om prestanda och kostnad. De undersökta prestandaparameterarna var
vridmoment, vridmomentsrippel, verkningsgrad och effektfaktor. Eftersom vridmoments-
densiteten i en PMaSynRM är lägre än hos en PMSM så visade sig PMSMen ha en 40%
kortare lamineringskropp än PMaSynRMen vid jämnlik vridmomentsproduktion.
Den ekonomiska utvärderingen resulterade i att vid användandet av ferritmagneter
i PMaSynRMen så blev den något billigare än PMSMen, upp till 20%. På grund av fluk-
tuerande priser hos NdFeB magneter, så finns det brytpunkter där PMaSynRMen faktiskt
blir dyrare än PMSMen eller då kostnadsreduktionen för PMaSynRMen kan bedömas att
vara för låg med tanke på den ökade längden och vridmomentsrippel. Däremot visades
det att PMaSynRMen är väldigt okänslig för prisvariationer och därför visades vara ett
kostnadsmässigt tryggare val än PMSMen.

Nyckelord: Avmagnetisering, ekonomisk utvärdering, permanentmagnetassistans,


synkron reuktansmaskin.

iii
Acknowledgements
This master thesis has been carried out at the department of Research and Development
for electrical motors at Xylem Water Solutions in Stockholm, Sweden.

I would like to thank Xylem Water Solutions for giving me the opportunity to do my
master thesis for them and for the great experience that it has entailed. I would especially
like to thank Tanja Hedberg and Øystein Krogen for their supervision and help throughout
the duration of the project. Furthermore would I like to thank my co-workers at Xylem
Water Solutions for making my stay there even more enjoyable with their company.

I would also like to express a special thanks to Associate Professor Oskar Wallmark for
sparking my interest in electrical machines and for inspiring me to pursue this field of en-
gineering. Additionally I would like to thank him for acting as my examiner for this thesis.

I also want to give thanks to all of my friends here in Stockholm who has made my
years at KTH unforgettable to say the least. Thank you for all the memories and for your
friendship. Even if we eventually find ourselves in different parts of the world, I know
that we will always stay in touch.

Finally, I would like to give my deepest gratitude to my parents and my sister who always
have supported me and helped me whenever I needed it. I would also like to especially
thank my girlfriend, Saga Kubulenso, for her never-ending patience with me when my
studies has gotten the best of me and for always being there for me no matter what. Thank
you so much.

Viktor Briggner
Stockholm, Sweden
August 2018

iv
Contents

Abstract ii

Sammanfattning iii

Acknowledgements iv

Contents v

Acronyms 1

Nomenclature 3

1 Introduction 6
1.1 Background and objectives . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2 Synchronous Reluctance Machine and Permanent Magnet assistance 9


2.1 Concept of reluctance torque . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Synchronous reluctance machine . . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 Definition of axes . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Governing equations . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Saliency and performance . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Iron saturation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5 Permanent magnet assistance . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.1 PM flux magnitude . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.6 Geometry and performance of PMaSynRM . . . . . . . . . . . . . . . . 21
2.6.1 Parameterization of PMaSynRM . . . . . . . . . . . . . . . . . . 21
2.6.2 Insulation ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.6.3 Number of flux barriers . . . . . . . . . . . . . . . . . . . . . . . 23
2.6.4 Torque ripple and rotor slots . . . . . . . . . . . . . . . . . . . . 24
2.6.5 Air-gap length . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.6.6 Radial and tangential ribs . . . . . . . . . . . . . . . . . . . . . . 25

v
Contents

2.6.7 Magnet dimensions and placement . . . . . . . . . . . . . . . . . 28


2.6.8 Stator and rotor steel . . . . . . . . . . . . . . . . . . . . . . . . 29
2.7 Permanent magnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.7.1 Demagnetization . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.8 Theoretical foundation of design approach . . . . . . . . . . . . . . . . . 32
2.8.1 Rotor barrier end angles . . . . . . . . . . . . . . . . . . . . . . 33
2.8.2 d/q-axis MMF and barrier sizing . . . . . . . . . . . . . . . . . . 33

3 Method of analysis 37
3.1 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.1.1 Performance parameters . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Initial dimensions and target PMSM . . . . . . . . . . . . . . . . . . . . 38
3.2.1 Stator selection for PMaSynRM . . . . . . . . . . . . . . . . . . 39
3.2.2 Target PMSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Design procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.1 Parametric study . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3.2 SynRM base-line design . . . . . . . . . . . . . . . . . . . . . . 42
3.4 SynRM design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4.1 Insulation ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4.2 Air-gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4.3 Rotor barrier end angles . . . . . . . . . . . . . . . . . . . . . . 44
3.4.4 Choice of barriers . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4.5 Radial ribs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.5 PMaSynRM design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.5.1 Magnet sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.5.2 Steel grade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.5.3 Demagnetization . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.6 Performance comparison and economic analysis . . . . . . . . . . . . . . 47
3.6.1 Comparing the machines . . . . . . . . . . . . . . . . . . . . . . 47

4 Results 50
4.1 Rotor design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.1.1 q-axis insulation ratio . . . . . . . . . . . . . . . . . . . . . . . . 50
4.1.2 d-axis insulation ratio . . . . . . . . . . . . . . . . . . . . . . . . 51
4.1.3 Air-gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.1.4 Rotor barrier end angles . . . . . . . . . . . . . . . . . . . . . . 54
4.1.5 Radial ribs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.1.6 Final SynRM rotor geometry . . . . . . . . . . . . . . . . . . . . 57
4.2 PMaSynRM design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.2.1 Magnet addition . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.2.2 Without magnet in outermost barrier . . . . . . . . . . . . . . . . 59

vi
Contents

4.2.3 NdFeB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.2.4 Steel types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.3 Demagnetization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.4 PMSM versus PMaSynRM . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.4.1 Performance comparison . . . . . . . . . . . . . . . . . . . . . . 66
4.4.2 Cost comparison . . . . . . . . . . . . . . . . . . . . . . . . . . 66

5 Conclusions and discussion 71


5.1 Performance of PMaSynRM . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2 Economic feasibility of PMaSynRM . . . . . . . . . . . . . . . . . . . . 72
5.3 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

A General calculations 74
A.1 Derivation of expression for IPF . . . . . . . . . . . . . . . . . . . . . . 74
A.2 Center of gravity of rotor segments . . . . . . . . . . . . . . . . . . . . . 75

B Data sheets 77
B.1 Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
B.1.1 M400-50A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
B.1.2 M600-50A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
B.1.3 M800-50A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
B.2 Magnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
B.2.1 Ferrite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
B.2.2 NdFeB - N33EH . . . . . . . . . . . . . . . . . . . . . . . . . . 83

C Results 84
C.1 Demagnetization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

References 87

vii
Acronyms

Notation Description Page


List
ALA axially laminated anisotropy 11

CSPR constant power speed range 11

FEA finite element analysis 7, 21


FFT fast fourier transform 38
FSCW fractional slot concentrated winding 40

IM induction machine 6
IPF internal power factor 15
IPMSM interior PM synchronous machine 11

LSPM line start PMSM 7

MOOA multi-objective optimization algorithms 73


MTPA maximum torque per ampere 16
MTPkVA maximum torque per kVA 15

NdFeB neodymium-iron-boron 6

PF power factor 15
PM permanent magnet 10
PMaSynRM permanent magnet assisted synchronous machine 7
PMSM permanent magnet synchronous machine 6

SP salient pole 11
SynRM synchronous reluctance machine 7, 9

TLA transversally laminated anisotropy 11

1
Acronyms

Notation Description Page


List

VFD variable frequency drive 7

2
Nomenclature

Notation Description Page


List
AF e Area of rotor segments 27
e Complex valued emf 12
vs Complex valued stator voltage 12
Br Remanent flux density 30
Fc Centrifugal force acting on the rotor material 26
fd,i Average MMF seen by the i:th iron segment due 34
to sinusoidal d-axis MMF
fq,i Average MMF seen by the i:th iron segment due 34
to sinusoidal q-axis MMF
fq,i Average q-axis MMF difference across the i:th 35
flux barrier
g Air-gap length 25
Hc Coercivity 30
Hc,b Normal coercivity 31
Hc,i Intrinsic coercivity 31
Hk Value of magnetic field at the knee of the intrinsic 31
curve
ic Complex valued iron loss current 12
i Complex valued stator current vector after iron 12
losses
id d-axis stator current 13
Im Magnetic polarization 31
iq q-axis stator current 13
Is Magnitude of current vector in dq-frame 16
is Complex valued stator current vector in dq-frame 12
kmag,a Magnet fill factor for center barrier 46
kmag,b Magnet fill factor for barrier arms 46
kw,d Insulation ratio in d-axis 21
Ld d-axis inductance 14

3
Nomenclature

Notation Description Page


List
Wk,q Height of barrier k in q-axis 21
Lq d-axis inductance 14
Lstk Stack length of the machine 26
nr Number of rotor barrier slots per pole 24
ns Number of stator slots per pole 24
p Number of poles 13
ps Stator slot pitch 23
R1 Rotor radius 35
Rc Equivalent iron loss resistance 12
RG Center of gravity of rotor segments 27
Rs Stator resistance 12
R1 Shaft radius 35
Sh,q Height of iron segment h in q-axis 21
✓b,k Rotor barrier end angle of barrier k 21
vd d-axis stator voltage 13
vq d-axis stator voltage 13
kw,q Insulation ratio in q-axis 21
Wk,d Height of barrier k in d-axis 21
wma,i Width of barriers placed in center of flux barrier 28
wmb,i Width of barriers placed in arms of flux barrier 28
wr,i Width of radial ribs 26
wt,i Width of tangential ribs 26
wtooth Stator tooth width 23

↵i Angle between flux barrier arm and center of flux 21


barrier
↵m Rotor slot pitch angle 33
Torque angle 13
s Rotor slot displacement angle 33
Load angle 13
⌘ Machine efficiency 38
Current angle from d-axis 13
⌫rib Safety factor for dimensioning of radial barrier 26
ribs
!e Electrical angular frequency 12
!m Mechanical angular rotor frequency 27
' Power factor angle 13
'i Internal power factor angle 13

4
Nomenclature

Notation Description Page


List
Complex valued flux linkage in dq-frame 12
d d-axis flux linkage 13
PM Permanent magnet flux linkage 18
q q-axis flux linkage 13
⇢lam Steel mass density 27
r Tensile strength of steel lamination 26
⌧em Electromagnetic torque 13
⌧P M PM torque 19
⌧rel Reluctance torque 19
⇠ Saliency ratio 15

5
Chapter 1

Introduction

1.1 Background and objectives


Electrical machines is an ever-present piece of equipment found in numerous industrial
and household applications. In fact, it is estimated that the energy usage by electrical
machines correspond to approximately 40% of the total electrical power consumption in
industrialized countries and up to 65% of the industrial energy consumption. Addition-
ally, in the EU approximately 22% of the energy consumed by electrical machines used
in industry found its usage for pumping applications [1, 2]. Therefore, an increase in effi-
ciency of these machines would turn out to be hugely important in the strive for reducing
overall energy consumption in the world.
The vast majority of the electrical machines on the market today are induction
machines (IMs). In Sweden it has been estimated that 90 % of all electrical machines
of power ratings between 0.75-375 kW are IMs [3]. These machines have a relatively
poor power factor and efficiency compared to permanent magnet synchronous machine
(PMSM), keeping power rating and pole numbers equal. However, the main benefit of
the IM is the sheer simplicity and reliability of the machine and its long history in indus-
trial applications. It can be line-started without the need for any power electronic drives
and doesn’t have any magnetic components which are expensive and are at risk of being
demagnetized.
However, as the price of power electronics continue to decline [4], synchronous
machine-based drives increase in popularity. This because synchronous machines are eas-
ier to control and generally has a higher or comparable torque density and higher effi-
ciency compared to IMs [5]. Furthermore, in order to keep the power ratings, and thus
costs, of the power electronic components down for these drives, the power factor plays a
significant role when evaluating these kinds of machines.
High-efficiency PMSMs utilize rare-earth magnets such as neodymium-iron-boron
(NdFeB) magnets which are not only relatively expensive compared to non rare-earth
magnets but also pose with quite poor price-stability as shown over the last few years

6
1.1. Background and objectives

(a) PMaSynRM (b) PMSM

Fig. 1.1: Example of PMaSynRM and surface-mounted PMSM

[6]. Furthermore, the extraction process of rare-earth magnets entails both a hazardous
environment for workers and for the people living nearby the extraction and refinement
plants, as well as the process used being very environmentally harmful [7].
synchronous reluctance machines (SynRMs), which utilize the anisotropy of the ro-
tor to produce torque known as reluctance torque, has been shown to perform better than
induction machines in terms of efficiency [8] while however falling short of equivalent
PMSMs by a large margin both in terms of efficiency and power factor, but also torque
density [9]. However, lately permanent magnet assisted synchronous machines (PMaSyn-
RMs) has been a source of interest in order to find a feasible competitor to the PMSMs.
The permanent magnets utilized in a PMaSynRM are either of far lesser quantities of
rare-earth magnets or alternatively of weaker, more abundant, and cheaper magnets such
as ferrite magnets [9]. High-efficiency PMSMs also generally perform better than PMa-
SynRM in terms of efficiency and torque density [5, 9], however this difference might be
small enough that the economical benefit can outweigh the reduction in performance. In
Fig. 1.1 a PMaSynRM and a surface-mounted PMSM is shown.
In the present thesis, a PMaSynRM will be designed and analyzed based on torque
density, efficiency and power factor and thereafter compared to an equivalent high-efficiency
surface-mounted PMSM. The machines are simulated and analyzed by means of finite el-
ement analysis (FEA). The work conducted in this thesis is in part based on the work
conducted by Adrian Ortega Dulanto as a master thesis [10]. However, the investigated
dimensions of the machines has been increased as to see if the PMaSynRM might dis-
play a relative increase in performance at greater dimensions as well as investigate the
scalability of the design developed in [10].
Furthermore, the choice to compare the PMaSynRM to a PMSM can be argued to be
a better comparison rather than to a line start PMSM (LSPM) or IM as both PMaSynRM
and PMSM requires a variable frequency drive (VFD) to operate as opposed to the other

7
1.2. Thesis outline

two and therefore their applications are more similar.

1.2 Thesis outline


The thesis consist of 5 chapters and will hold the following structure

• Chapter 1: Introduction, background and justification of the thesis

• Chapter 2: Theoretical foundation of PMaSynRM design and analysis

• Chapter 3: Description of the analysis and design process

• Chapter 4: Results and comparison of the developed PMaSynRM and PMSM

• Chapter 5: Conclusions and discussions regarding further work

8
Chapter 2

Synchronous Reluctance Machine and


Permanent Magnet assistance

As was briefly stated in Chapter 1 the synchronous reluctance machine (SynRM) and
PMaSynRM relied on the anisotropy of the rotor in order to produce torque. In this chapter
the theory of the SynRM will be discussed and how utilizing magnets to further improve
the operation of SynRM influence the operation, thus producing a PMaSynRM.

2.1 Concept of reluctance torque


Reluctance torque relies on, as the name would suggest, the reluctance of an object. More
specifically, it relies on the difference in reluctance in different directions. This differ-
ence means that depending on the how the object is placed in a magnetic field relative
to the direction of the field, different magnetic behaviour is displayed, i.e. the specimen
is anisotropic. This anisotropy is easiest achieved by altering the geometry of a magnetic
material specimen. The torque produced is dependent on the interaction of the specimen
and an applied magnetic field. By defining an object-orientated coordinate system, with
the direct (d) axis aligned with the lowest reluctance and the quadrature (q) axis along the
path of highest reluctance we can begin to define the torque produced.
To illustrate this, assume that the anisotropic specimen is subjected to a homoge-
neous magnetic field and there is an angle between the d-axis of the specimen and the
magnetic field. This angle means that a distortion in the field is introduced and hence the
curl of the field will be non-zero. This in turn creates a force which does not cancel out
and hence a torque is produced. In Fig. 2.1 object a) is completely isotropic and thus dis-
plays equal reluctance in all directions in the plane of the magnetic field B and therefore
no torque is exerted on it and consequently it remains unaffected by the field. However,
object b) is anisotropic with an angle to the field and therefore experience a net torque.
The angle is known as the load angle and it is the angle which determines the magnitude
of the torque since the system always aim to keep equal to zero [11].

9
2.2. Synchronous reluctance machine

F
d F d
B
q q

F
F

a) b) c)

Fig. 2.1: Conceptual description of reluctance torque. a) an isotropic object, b) an anisotropic


object, c) a rotor-like anisotropic object.

Object c) is added in Fig. 2.1 as to illustrate how the concept is transferred to rotat-
ing machines. Anisotropy is achieved by introducing air gaps, or flux barriers, in a rotor
structure and a torque is produced. However, note that in a SynRM the magnetic field will
be directed radially and rotating in order to produce a continuous torque, but the same
concept applies.
From this qualitative description of the reluctance torque it can be derived that the
torque production in a SynRM (and PMaSynRM) is dependent on and that there has to
be an optimal angle if it is sought to maximize the torque. The main flux in these machines
is induced by the current and therefore by controlling the current, the torque is controlled.
How the flux, current and reluctance is related is expanded on in the coming sections.

2.2 Synchronous reluctance machine


The theory behind torque production from salient pole machines has been well-established
since the 1920s. Following the development of inverter technology in the 1970s the thought
of commercializing these types of synchronous machines became a source of growing in-
terest [9]. However, given the development of high-energy permanent magnets (PMs)
such as NdFeB the interest for machines driven purely by reluctance torque faded as they
couldn’t compete with the high-energy magnet machines.
The drawbacks of the SynRM is that inherently has a high torque ripple, lower
torque density and low power factor compared to equivalent PMSMs. This can in part be
remedied by the addition of permanent magnets, giving a higher torque production and
power factor albeit still worse than the equivalent PMSM [5].

10
2.2. Synchronous reluctance machine

The main advantage of the SynRM when compared to a PMSM is generally the
lower price range as it doesn’t utilize expensive rare-earth magnets. However, there are
more advantages over the PMSM of the SynRM and PMaSynRM as outlined in [9] and
[12]. To name a few we have
• The SynRM is not as vulnerable to short-circuit conditions as the lack of magnets
means that no current is induced.

• The constant power speed range (CSPR) is very good for SynRM and especially
PMaSynRM

• The rotor saliency provides with easy rotor position detection at stand-still
The design of the SynRM rotor as it looks today is still conceptually based on the
work done by Kostko in 1923 where the rotor is divided into different segments with
flux barriers in order to achieve a high saliency [9, 13] as seen in Fig. 2.2b and c. Salient
pole machines can be constructed in a few different ways. First, there is the conventional
salient pole (SP) rotor, the axially laminated anisotropy (ALA) rotor and the transversally
laminated anisotropy (TLA) rotor [14] and these types can be seen in Fig. 2.2.
However, the SP design configuration has been shown to be sub-optimal for SynRM
drives and is more suitable for wound rotor synchronous machines. The ALA is more
theoretically appealing and is believed to be able to provide a higher saliency ratio than the
TLA configuration. However, the TLA configuration is much easier to mass produce as it
utilizes the same punching and assembly procedure as traditional electrical machines [15]
and therefore will be the focus of this thesis.
The geometry of a TLA SynRM is similar to that of object c) in Fig. 2.1. As was ex-
plained in Section 2.1 the SynRM produces its torque by differences in reluctance around
the rotor. Generally several flux barriers are introduced and it has been shown that the pole
number should be kept as low as possible where four poles are held as the most suitable
pole number. In [11] a thorough investigation regarding pole numbers are presented. Fig.
2.3 displays a common four-pole SynRM rotor design with 3 flux barriers.

2.2.1 Definition of axes


Due to the inherent ansiotropy of the SynRM rotor, an analysis in a stator reference frame
is difficult. The inductance of the machine is not only current dependent, but also depen-
dent on the position of the rotor. In order to derive useful expressions and to perform a
proper analysis, a rotor-based (synchronous) reference frame needs to be established, this
coordinate system can be seen in Fig. 2.3.
The direct axis (d) is defined along the path which displays the highest inductance
and conversely the quadrature axis (q) axis is defined along the path which has the low-
est inductance. Note that this definition differs from the convention of interior PM syn-
chronous machines (IPMSMs) and surface-mounted PMSMs where the d-axis is defined

11
2.2. Synchronous reluctance machine

Fig. 2.2: Different rotor designs for rotor saliency. a) Conventional salient pole. b) Axially lami-
nated anisotropy. c) Transversally laminated anisotropy. From [14]

along the flux from the PM, this distinction will prove important when discussing perma-
nent magnet assistance since the consequence will be that the axes are reversed in terms
of permanent magnet flux.

2.2.2 Governing equations


A model over the SynRM can be formulated in accordance with the equivalent circuit
seen in Fig. 2.4 [16], where all bold-faced quantities are complex-valued. is the flux
linkage of the machine, vs is the stator voltage, is is the stator current, i is the current not
contributing to the iron losses, ic is the iron loss current, Rs is the stator resistance, Rc is
the equivalent iron loss resistance and !e is the angular frequency. Here e represents the
emf as

d
e= + j!e (2.1)
dt
Neglecting iron losses, the dynamics of the synchronous machine in dq-frame can
be described by

12
2.2. Synchronous reluctance machine

Flux barriers

Iron segments

d
Fig. 2.3: Cross-section of SynRM with defined dq-reference frame.

d d
v d = R s id + !e q (2.2a)
dt
d q
v q = R s iq + + !e d (2.2b)
dt
where vd and vq are the stator voltages, id and iq are the stator currents, d and q are the
machine flux linkages [9].
Fig. 2.4 can be utilized to draw a phasor diagram in the dq-plane as seen in Fig. 2.5.
Here the is the load angle, is the torque angle, is the current angle from the d-axis,
' and 'i are the power factor angle and internal power factor angle respectively.
The torque, ⌧em , can be described as
3
⌧em = p( d iq q id ) (2.3)
4
where p is the number of poles [9].
The d- and q-axis flux linkages are both very dependent on the operating point and

13
2.2. Synchronous reluctance machine

j!e
Rs is i

+
vs d
e Rc L
dt
ic

Fig. 2.4: Circuit diagram including iron losses.

q
Rs id
vs !e Lq iq
e j!e Ld id

is ic
iq
i
Rs iq
'
'i jLq iq

Ld id
id d
Fig. 2.5: Phasor diagram for SynRM.

experience cross-coupling from currents in the adjacent axis [9, 11], i.e.
(
d = d (id , iq )
(2.4)
q = q (id , iq )

This cross-coupling occurs since the q-axis current cause a flux component in the d-axis
and vise verse. This not only contributes to the total flux in the respective axis but it also
affect the saturation level of the iron in the respective axes. Hence, when rewriting the
flux linkages as current times inductances it is very important to note that the inductances
(Ld , Lq ) are indeed not constant [9, 11]
(
d = Ld (id , iq )id
(2.5)
q = Lq (id , iq )iq

14
2.3. Saliency and performance

Note that this definition of the inductances is a simplification and for a more thor-
ough discussion see Chapter 4 in [9]. Note also that the inductances includes not only
magnetizing inductances but also the leakage inductances, which are not significantly in-
fluenced by the aforementioned cross-coupling and saturation [16].

2.3 Saliency and performance


The saliency ratio, ⇠ is of great significance for the performance of the SynRM. The
saliency ratio is defined as
Ld
⇠= (2.6)
Lq
and in a SynRM, ⇠ generally do not exceed values much higher than 10 [9].
Consider the phasor diagram in Fig. 2.5. The following relationships hold for the
defined angles

+ = + 'i (2.7)
2
= + (2.8)

Utilizing these relationships we find that we can write the internal power factor (IPF) as

⇣⇡ ⌘ ⇠ 1
IPF = cos 'i = cos + =r (2.9)
2 1 1
⇠2 +
sin2 cos2
for the derivation of this expression refer to Appendix A.1. And thus, it becomes obvious
that the saliency ratio influences the IPF heavily, as seen in Fig. 2.6 where the it is plotted
for different values of ⇠ as a function of the current vector angle. It is important to note
that IPF is not the same thing as power factor (PF) but they are, however, related and
a high value of IPF leads to a high value of PF since the difference between these two
are only governed by Rs and Rc which can be verified by looking at Fig. 2.5 and Fig.
2.4. Therefore, IPF is discussed here as it is quite straightforward to derive an analytical
expression from current angle and saliency.
From equation (2.9) it is obvious that for any given value of ⇠ there exist a value
of the current vector angle which allows for the optimal IPF. It can be shown that this
p ⇠ 1
occurs when tan = ⇠ and then the IPF is equal to . This operating point is often
⇠+1
called maximum torque per kVA (MTPkVA) [16] and thus correspond to the operating
point when the least amount of reactive power is required by the supply.
With the definition of flux in equation (2.5) and dropping the parentheses for sim-
plicity we find that equation (2.3) can be rewritten as

15
2.3. Saliency and performance

0.9 ⇠=2
⇠=5
0.8 ⇠=10
⇠=15
0.7
⇠=20
0.6 ⇠=100
IPF
0.5

0.4

0.3

0.2

0.1

0
0 10 20 30 40 50 60 70 80 90
[deg]

Fig. 2.6: IPF as function of current angle for different values of ⇠.

3
⌧em = p(Ld Lq )id iq (2.10)
4
Thus, the torque is proportional to the difference in inductance between the two
axes. It is interesting to note that while the IPF is governed by the saliency ratio, the
torque production is dependent on the saliency difference. Even though these parameters
are related, it can prove difficult to maximize both parameters simultaneously when de-
veloping a rotor design [10]. This is due to the non-linear dependency of d- and q-axis
inductances on the rotor geometry [17].
Another conflict occurs when operating a SynRM and it becomes apparent when
rewriting equation (2.10) utilizing the phasor quantities presented in Fig. 2.5 in steady
state, i.e.
3
⌧em = p(Ld Lq )Is2 sin 2 (2.11)
4
where Is is the stator current magnitude in steady state. Here we see that for given (con-
stant) inductances and current magnitude, the maximum torque is achieved for a current
vector angle of 45 degrees. This operating point is referred to as maximum torque per
ampere (MTPA). Again, looking at Fig. 2.6 we see that this current vector angle does not
coincide with the angle which maximizes the IPF for moderately high values of ⇠.

16
2.4. Iron saturation

It should also be noted that saturation affect both the inductance difference and
saliency ratio such that in reality ⇠ and inductance difference decrease with increasing
current, which affect both the torque production and power factor. The effect of iron sat-
uration will be expanded upon below.

2.4 Iron saturation


As has been stated above the performance of SynRMs and PMaSynRMs is affected by
the saturation of the iron. The most prominent effect of this is that the inductances change
as a function of current level. This in turn contributes to a different behaviour in terms of
torque and PF between low and high levels of current. In Fig. 2.7 the two inductances are
plotted as functions of the applied current modulus. Since the d-axis is defined along the
path of lowest reluctance and therefore the path which contains the greatest amount of iron
it is quite straightforward to understand that it is also affected by the saturation to a greater
extent. In fact, to saturate the q-axis is desirable because it allows for ⇠ and the inductance
difference to increase. However, the decreased values of Ld results in undesirable effects.
As the torque is proportional to the inductance difference it becomes obvious that
the torque does not vary with the square of the current level when considering saturation.
The saturation of the iron leads to a shift towards higher values of optimal current vector
angle for maximum torque (MTPA) as the current increases. The same tendencies can
be seen when analyzing the PF curve. Saturation implies that ⇠ decrease in value for
greater current levels. Again, referring to Fig. 2.6 we see that as ⇠ decrease, so does the
internal power factor and therefore the power factor. However, some saturation effects
actually benefits the PF when operating in the maximum torque (MTPA) operating point
as discussed in [18]. This was attributed to the fact that the current angle for maximized
power factor (MTPkVA) and MTPA current angle values were shifted closer together and
thereby resulting in a higher PF.
Moreover, as was stated previously the d- and q-axis inductances are not only de-
pendent on the d- and q-axis currents respectively but cross-saturation also occurs to vary-
ing degrees. In Fig. 2.8 the flux is plotted against different values of id and iq . Here we
see difference in flux when the opposite axis current is non-zero.

2.5 Permanent magnet assistance


Reviewing Fig. 2.6 we see that in order to for an ordinary SynRM to have a IPF greater
than 0.9 a saliency ratio beyond 20 is needed, which is an unrealistic value [16]. There-
fore, in order to make the SynRM feasible compared to PMSMs some modifications need
to be done. Such a modification is to add magnets in the rotor barriers, thus making a
PMaSynRM. This addition slightly alters the characteristics of the machine. The magnets

17
2.5. Permanent magnet assistance

100
Ld
90 Lq
80

70
Inductance [mH]
60

50

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Current magnitude [A]

Fig. 2.7: Ld and Lq as function of current modulus for machine with 26 A rated current (RMS).

add a flux linkage component in the q-axis and thus equation (2.5) becomes
(
d = Ld (id , iq )id
(2.12)
q = Lq (id , iq )iq PM

where P M is the permanent magnet flux linkage addition. As can be seen in Fig. 2.9 the
phasor diagram is altered due to this addition. The effect of the PM flux is highlighted
in red. The voltage phasor is rotated towards the current vector effectively increasing the
power factor. Additionally, since the PM-flux is largely directed in the q-axis direction
this flux also help saturate the iron in the q-axis which reduces the q-axis inductance and
thus increases the saliency ratio [9].
To derive an analytical expression for the IPF as a function of the saliency ratio of a
PMaSynRM is not as straightforward as it were for the SynRM due to it being dependent
on the q-axis PM flux linkage aswell. However, when comparing the two phasor diagrams
in Fig. 2.5 and Fig. 2.9 we notice that the difference is the aforementioned rotation of
the voltage vector. This rotation occur because the added PM flux counteracts the q-axis
stator flux linkage and thus shifts the flux vector away from the current vector. For both
machines in the steady state, the emf (e) will be perpendicular to the flux vector ( ) and
therefore the emf will be rotated towards the current vector which increases the IPF and
hence PF.

18
2.5. Permanent magnet assistance

2.5

1.5

1
Flux linkage [Wb]
0.5

0
d (id , iq = 0)
0.5 d (id , iq = 50)
d (id , iq = 100)
1 q (iq , id = 0)
q (iq , id = 50)
1.5
q (iq , id = 100)
2

2.5
100 80 60 40 20 0 20 40 60 80 100
id or iq [A]

Fig. 2.8: d and q as function of different values of id and iq , for machine with rated current of
26 A (RMS).

The added PM flux also significantly alter the expression for the torque production
and equation (2.10) becomes
3
⌧em = p P M id + (Ld Lq )id iq (2.13)
4
i.e. we get two torque components, the PM induced torque (⌧P M ) and reluctance torque
(⌧rel ). Note that this equation is similar to that of a salient-pole PMSM, but the key dif-
ferences being that the torque contribution of the PM-flux is lower compared to the the
reluctance torque and that the reference frame is rotated.
Rewriting equation (2.13) in the same fashion as equation (2.11) we get
3
⌧em = p P M Is cos + (Ld Lq )Is2 sin 2 (2.14)
4
This implies that the PM torque and reluctance torque does not display coinciding
maxima with respect to . Hence, the optimal value of depends on the ratio between the
two but typically lies around 40 [9] when neglecting saturation.
In addition to the shifting of flux vectors and PM torque production, another feature
of permanent magnet assistance benefits the operation of the PMaSynRM. The PM flux
helps saturate the ribs within the rotor structure (which are further discussed in Section

19
2.5. Permanent magnet assistance

q
!e Lq iq R s id
j! PM vs
e
j!e Ld id
is
ic
iq
i
Rs iq
'
'i
jLq iq
id L d id
d
j PM

Fig. 2.9: Phasor diagram for PMaSynRM including iron losses.

2.6.6). This hence helps reduce the flux between the iron segments in the rotor which
further reduces the q-axis inductance [18].

2.5.1 PM flux magnitude


As was discussed previously adding PMs to an ordinary SynRM improves its performance
in more than one way. However, the amount of added PM flux linkage needs to be deter-
mined in order to know the amount and type of magnet to utilize in the design. The limit
between ”permanent magnet assisted machine” and ”permanent magnet machine” is not
clearly defined and it can be discussed when the machine stops being assisted. In [18]
an analytical expression describing the PM flux linkage magnitude has been derived and
corresponding FE analysis has been carried out. It was shown that for maximum torque
production for lower saliency ratios the required PM flux linkage were far greater than
50% of the nominal flux and at that point it can be argued that the machine is no longer
just PM assisted since such a large part of the flux is supplied by the magnets. Addition-
ally this amount of flux linkage implies the use of high-energy rare-earth PMs and the
economical benefits vanishes as it closes in on being a IPMSM. However, this shows that
there are no real upper limit in terms of PM flux magnitude when utilizing ferrite PMs or
similar lower energy magnets.
Additionally, another property shown in [18] was that the torque curve and PF curve
as function of the PM flux linkage had a very flat maxima, meaning that even quite low
values of PM flux linkage would generate acceptable levels of power factor and torque
production and even a small PM flux linkage contributed to reverse the q-axis flux linkage.

20
2.6. Geometry and performance of PMaSynRM

In fact, it is stated in [18] that the PM flux should be in the vicinity of 25-35% of the
nominal flux. This also implies that ferrite PMs might be sufficient in order to reach the
desired operating point.

2.6 Geometry and performance of PMaSynRM


The geometry of a SynRM rotor is inherently complex and as the operation of this ma-
chine depends on the saliency of the rotor, it is of utmost importance that the geometry is
well defined in terms of parameters to analyze. Due to the complexity and non-linearity of
these machines, it is very difficult and of questionable use to derive an analytical model of
the machine and optimize. Instead, it is better to combine analytical expressions which de-
termines some key geometrical parameters and thereafter conduct the performance analy-
sis utilizing computer-aided finite element analysis (FEA) as is done in [11], [19] and [10].
The parameterization in this thesis is based on the work conducted in [15] and [11] and
its theoretical relevance is expanded upon in Section 2.8.

2.6.1 Parameterization of PMaSynRM


In order to properly describe the rotor of the PMaSynRM it is crucial to define the ge-
ometry through comprehensible parameters. In Fig. 2.10 a pole of a PMaSynRM rotor
with 3 barriers is shown. This parameterization follows that conducted in papers such
as [15], [19].

• The barrier height of barrier k is described by Wk,q in the q-direction and Wk,d in
the d-direction.

• The height of the iron segment h between barriers is described in the q-direction as
Sh,q

• The angle that the barrier k makes at the periphery of the rotor with the d-axis is
defined as ✓b,k

• The angle between flux barrier arm and center of flux barrier is defined as ↵i

Often, the q-axis position of a rotor barrier along the q-axis is of interest. The dis-
tance to the n:th barrier is defined as
n
X n 1
X
D0,n = Sh,q + Wk,q (2.15)
h=1 k=1

Furthermore, in order to give an indication as to how much air versus iron there is in
the rotor in both the q- and d-direction the insulation ratios kw,q and kw,d are qualitatively

21
2.6. Geometry and performance of PMaSynRM

↵3

S4,q
↵2
W3,q

S3,q
↵1 ✓b,3
W3,d
W2,q

S2,q
W2,d ✓b,2
W1,q

S1,q
W1,d ✓b,1

Rsh
R1

Fig. 2.10: Parameterization of PMaSynRM rotor.

defined as
Amount of air
kw,q = (2.16)
Amount of iron q axis
Amount of air
kw,d = (2.17)
Amount of iron d axis

which gives that a value below 1 of these ratios means that there is more iron than air in
the respective direction and conversely a value above 1 means that there is more air than
iron. Note that the path of calculation for the q-axis is easily defined along the axis. For
the d-axis it is a somewhat more complicated. The expressions for the insulation ratios
are given in Section 2.8.

22
2.6. Geometry and performance of PMaSynRM

2.6.2 Insulation ratios


The aforementioned insulation ratios have a great influence on the torque production as
well as the power factor. The reason of this is most easily explained by the fact that these
ratios determine the reluctance of the machine. This, in turn, affects the flux linkage inside
the machine and therefore the saturation level of the iron and thus the inductance is altered.
More air introduced in the q-axis direction lowers the q-axis inductance, thus allowing
for increasing the inductance difference and saliency ratio. Therefore it is common to
optimize the insulation ratio in terms of torque production. However, the relationship of
the insulation ratios are not linear towards either ⇠ or inductance difference (for fixed
values of and current). It can be shown that there exist an optimum in terms of kw,q and
kw,d when it comes to inductance difference and saliency ratio [11, 20]. These optimum
values does not necessarily coincide and therefore a design trade-off has to be made in
terms of power factor and torque production. Furthermore, an upper limit of the rotor
insulation ratios has been suggested, which is related to the stator insulation ratio. This
value is described by
ps wtooth
kw,s = (2.18)
ps
where ps is the stator slot pitch and wtooth is the stator tooth width. It is desirable to choose
the q-axis insulation ratio of the rotor to a value close to or below the value of kw,s [9,15].
This is true since the insulation ratios in large determines the flux density magnitude in
the stator and rotor and hence the saturation levels in rotor and stator. Therefore, if kw,q <
kw,s the stator teeth experience a greater saturation flux than the rotor and conversely if
kw,q > kw,s is true the opposite applies. This affects the flux linkage magnitudes at higher
current levels and therefore the torque production. This was shown in [9] and there it was
concluded that a lower value for kw,q was preferable to a higher in terms of flux linkage
and torque production. Following the same reasoning, it can be concluded that the d-axis
insulation ratio kw,d should be equal or less than the value of kw,q , i.e. the amount of iron
in the d-axis should be higher than in the q-axis [15, 21].

2.6.3 Number of flux barriers


While it can be shown that the insulation ratios mostly govern the torque production the
number of flux barriers also impacts the performance of the SynRM. Generally speaking,
a greater amount of flux barriers has a positive impact on both torque production and
power factor. However, when it comes to aspects such as torque ripple the interaction
between stator and rotor is of great importance and therefore the number of barriers will
have a great influence on these parameters [10, 15, 19, 22], albeit there are no simple rules
for how many are optimal.
The choice of number of barriers is non-trivial and a simple analytical expression
is difficult to derive. However, in order to minimize the torque ripple a general rule was

23
2.6. Geometry and performance of PMaSynRM

presented in [23] as

nr = ns ± 4 (2.19)

where nr is the number of rotor barrier slots per pole and ns is the number of stator slots
per pole. Whether the equation should be treated with a plus or a minus is determined by
the feasibility of the structure albeit it is stated that +4 generates better results.
In [22] a thorough analysis of the behavior of different numbers of stator slots and
barriers is presented. There it was derived that different number of stator slots perform
at its best for different number of rotor barriers For instance was it shown that for a 48
slot machine, torque production was maximized for 4 or 6 number of barriers whereas
efficiency was maximized for 4 barriers and torque ripple was minimized for 6 barriers.
Whereas in [10] it was determined that for the 36 slot machine 3 barriers generated the
overall best performance in terms of torque production and power factor.

2.6.4 Torque ripple and rotor slots


High torque ripple is an inherent problem that plagues the SynRM and other machines
such as IPMSMs [24]. It has already been determined that the interaction between stator
and rotor influences the performance of the machine and most importantly the torque
ripple. Therefore it becomes obvious that the positioning of the rotor barrier ends, or rotor
barrier slots, impacts the operation of the machine.
The ripple that occurs in a SynRM is due to the variation in reluctance that occur
when a rotor slot passes a stator slot [19, 23–25]. Therefore, it is of interest to place the
rotor slots such that the torque ripple is minimized. A great deal of research has been
made in the field of reducing the torque ripple of these types of machines. Torque ripple
reduction can be achieved in a variety of ways. The method discussed above presented
in [23] focuses on equally distributed rotor slots along the circumference and focus more
on number of barriers rather than the placement of the rotor slots, [25] discusses the
possibility of asymmetrically placed stator slot openings as a means of combating torque
ripple, and [24] proposes asymmetrically shifting the rotor slots between every other or
several laminations. A rule of thumb is presented in [19], where it is stated that the rotor
barrier should be placed such that when one of the two barrier slots enters below a stator
slot, the other barrier slot should enter below a stator tooth. Fig. 2.11 illustrates how the
stator and rotor slots can be aligned. In this case the innermost barrier might be properly
aligned since the top slot it is about to pass below a stator slot while the right slot is about
to pass below a tooth. However, the middle barrier is more problematic since both slots
are situated below the middle of both teeth which cause discrepancies in reluctance and
thereby torque ripple.
However, for pumping applications torque ripple is not a serious issue as the load
torque typically is proportional to the square of the rotor speed, and therefore the torque

24
2.6. Geometry and performance of PMaSynRM

Fig. 2.11: PMaSynRM stator and rotor interaction.

ripple won’t correspond to any major speed ripple. Hence, to optimize torque ripple is not
the focus of this paper but measures in terms of altering the placement of the rotor slots
will be made as long as it does not negatively impacts the torque production.

2.6.5 Air-gap length


It was shown in [17] that the air-gap length, g, of a SynRM greatly affects the d-axis
inductance while leaving the q-axis inductance virtually unaffected. This thus decreases
the inductance difference and saliency ratio and consequently the torque production and
power factor. This is due to the air-gap being the only air that the d-axis inductance sees
while in the q-axis the air-gap is a small fraction of the total amount of air for the q-
axis inductance due to the rotor flux barriers [15]. Hence, it is preferable to maintain the
air-gap length as low as possible in order to get as large torque production as possible.
However, there is a manufacturing gain of increasing the air-gap as the tolerances can be
kept larger. Additionally, a bigger air-gap yields lower torque ripple and lower iron losses
in the rotor and therefore the effect of the air-gap length is of interest.

2.6.6 Radial and tangential ribs


Due to the shape of the flux barriers of a SynRM rotor, structural weaknesses is something
that needs to be addressed. This means that the rotational forces might compromise the
structural integrity of the rotor as the radial forces can cause the structure to break down.
This can be remedied by adding radial ribs to some, or all, barriers as to reinforce the
structure. [10, 15, 26]. Another structural issue is that of the flux barrier ends towards the
air gap. Ideally, there would be no steel between the rotor barrier ends - or rotor slots -
and the air gap. But as has been mentioned previously the chosen manufacturing process,

25
2.6. Geometry and performance of PMaSynRM

wr,3
wt,3

wr,2
wt,2

wr,1
wt,1

Fig. 2.12: Illustration of radial and tangential ribs in SynRM.

TLA, means that the rotor is punched and therefore require a continuous sheet of metal.
The width of this rib is in part determined by the tolerance of the punching machine, but
also by the expected tangential forces from torque ripple or load variations. However, the
calculation of the thickness of the tangential ribs are determined to be outside of the scope
of this project. In Fig. 2.12, these ribs are illustrated, and the parameters describing the
widths defined, wr,i is the widths of the radial ribs and wt,i is the width of the tangential
ribs.
Not all machines require radial ribs and it is rather a question of size of the rotor,
radial positioning of the flux barriers, and maximum allowable speed of the machine
which determines the need and widths of them. The width of the radial ribs can calculated
by diving the rotor into i segments and calculate the rotational force exerted on each
segment [26]. This is done as
Fc,i
wr,i = ⌫rib (2.20)
r Lstk

where Fc is the centrifugal force acting on the rotor, r is the tensile strength of the
material, Lstk is the total stack length of the rotor and ⌫rib is a safety factor usually in the

26
2.6. Geometry and performance of PMaSynRM

1 2 3

D0,3

D0,2

D0,1

Fig. 2.13: Areas for radial rib calculation.

range between 2 and 3 [26]. The centrifugal force can be calculated as


2
Fc,i = ⇢lam !m RG,i AF e,i Lstk (2.21)

where AF e is the area of the relevant rotor segment, RG is the center of gravity of the
rotor segment, !m is the mechanical angular frequency of the rotor and ⇢lam is the mass
density of the steel.
An overestimation of the radial ribs can be achieved by simplifying the geometry
as shown in Fig. 2.13 and use these sections to calculate the force that each rib will expe-
rience. Note that each area contains all steel within its boundaries, i.e. Area 1 includes 2
and 3 and so on. This is an overestimation since the air in the rotor structure is neglected
which means that the force in reality is lower provided that the center of gravity is not
increased dramatically. This also holds when magnets are added to the flux barriers since
the density of ferrite magnets is lower than the steel and NdFeB magnets are about com-
parable with the steel while they will likely not fill up the entirety of the flux barriers. For
the equations to derive the center of gravity refer to Appendix A.2.
Introduction of radial ribs in the barrier structure leads to an unwanted flux path
in the rotor which contributes to increase the q-axis inductance and therefore leads to a
torque reduction. In Fig. 2.14, the influence of the ribs on the q-axis flux is visible when
looking at the span -5 A to 5 A, the slope of the flux, and thus inductance, is significantly
higher in that span than elsewhere. This increased inductance is due to the flux path pro-

27
2.6. Geometry and performance of PMaSynRM

2.5

1.5

1
Flux linkage [Wb]
0.5

0.5

1
d (id , iq
= 0)
1.5
(i ,
q q d i = 0)
2

2.5
60 50 40 30 20 10 0 10 20 30 40 50 60
id or iq [A]

Fig. 2.14: Influence of radial ribs on flux, for machine with 26 A rated current.

vided by the radial ribs. However, it is also visible that for greater values of the current,
the ribs saturate and start to behave as air. In [15] it was shown that this reduction is in the
magnitude of a few percent of the nominal torque. In [27] an expression to estimate of the
magnitude of the torque reduction is presented where the torque reduction is proportional
to the number of poles in square times the width of the ribs assuming constant width of
the ribs. In [11] further analysis of the influence and design of radial ribs is conducted.

2.6.7 Magnet dimensions and placement


When adding magnets to the SynRM design they will be placed in the flux barriers as
shown in Fig. 2.15. The width of the magnets in the center barrier is given by the param-
eter wma,i and the width of the magnets in the barrier arms are given by wmb,i .
The placement of the magnets is quite important with regards to the performance
of the machine. [28] showed that when keeping the total volume of the magnets constant
it was more suitable in terms of torque ripple, and to some extent torque production, to
distribute the magnets in all of the center barriers instead of filling some barriers with
magnets while leaving some empty. Likewise, in [29] it was shown that when placing
magnets only in the barrier arms and leaving the center barriers empty the torque produc-
tion increased but the torque ripple became much greater compared to distributed magnets

28
2.6. Geometry and performance of PMaSynRM

wma,3

wma,2
wmb,3
wma,1

wmb,2

wmb,1
Fig. 2.15: Magnet variables and magnet placement in rotor.

in the center barriers.


Furthermore, to keep the majority of the magnet volume deep within the rotor also
helps to protect against demagnetization due to stator flux [28], a property which is very
desirable when utilizing weaker magnets such as ferrite.

2.6.8 Stator and rotor steel


The grade of the stator and rotor steel primarily determines the iron losses of the machine
as the iron losses in a machine is dependent on the flux density variation in the material
and the material specific hysteresis curves of the steel. For a through description of cal-
culation of iron losses, refer to [30]. Therefore, it is important to take the steel grade into
consideration as it can have a significant impact on the efficiency of the machine. How-
ever, for synchronous machines the iron losses in the rotor are generally lower in the rotor
than the stator as the rotor follows the fundamental of the magnetic field and therefore
experience only flux variations in terms of harmonic content.
In [19], it was shown through simulations that lower-loss steel grades affected iron
losses and output power of a SynRM greatly. In [19], when only varying the steel type

29
2.7. Permanent magnets

B
Br

Hc
H

Fig. 2.16: Typical B-H curve for a hard magnetic material.

for a machine in the range of 12 kW output power, the efficiency saw an increase of 9
percentage units between the lowest and highest loss steel grade. At the same time, the
output power increased by 8 percent when using the low loss grade steel compared to the
higher loss grade.

2.7 Permanent magnets


Magnetic materials are in general characterized by the hysteresis loop of the B-H curve
which describes how the magnetic flux density varies when an external magnetic field
is applied. A magnetic material is largely defined by the remanent flux density, Br , and
the coercivity, Hc . Br defines the flux density in the material when no external H-field
is applied and Hc describes the H-field required in order to bring the flux density inside
the material to zero. Magnetic materials can be divided into two major groups, hard and
soft magnetic materials. Hard magnetic materials are defined by large values of Br and
Hc while soft magnetic materials have lower values [31, 32]. Fig. 2.16 illustrates a typical
curve for a hard magnetic material.
Permanent magnets are hard magnetic materials and are, as the name suggests,
permanently magnetized. This magnetization, M, relates to the magnetic flux density as
[32]

B = µ0 (H + M) (2.22)

30
2.7. Permanent magnets

B or Im

Br

Hc,i Hc,b
H

Fig. 2.17: Typical normal (dashed) and intrisic (line) curve for permanent magnet.

2.7.1 Demagnetization
Typically, permanent magnets do not lose its magnetization when the flux density is re-
duced to zero, i.e. when the coercivity is reached. Utilizing the above stated relationship
one can define the magnetic polarization, Im , as

I m = µ0 M = B µ0 H (2.23)

Both Im and B can be plotted in the same graph as is done in Fig. 2.17. The Im -H
plot is often referred to as the intrinsic curve and the B-H plot is called normal curve.
In these plots two coercivities appear, the intrinsic and normal coercivity Hc,i and Hc,b .
Analogous to the definition of Hc , Hc,i is the value at which the magnetization is forced
to zero and beyond this point the magnetization will start to shift polarity [31]. Note that
Hc,b = Hc as defined previously.
Demagnetization of the permanent magnet occurs when the magnetic field intensity
approaches the intrinsic coercivity. In fact, when the magnetic field passes the value close
to the knee of the intrinsic curve partial demagnetization start to occur. For most practical
situations in electrical machines, only the second quadrant of the hysteresis loops are of
interest. One can define the magnetic field knee value in the second quadrant as Hk and
when that value is exceeded and thereafter reduced to below that value again, the magne-
tization of the magnet will be reduced and therefore also the remanent flux density. It can
be shown that when Hk is exceeded, the new intrinsic curve follows the so called recoil
lines shown in Fig. 2.18. The slope of the recoil lines are similar to that of the slope of

31
2.8. Theoretical foundation of design approach

Br

Bk
Hc,i Hk

Fig. 2.18: Demagnetization curve for permanent magnet with recoil lines.

the original intrinsic curve when the magnetic field is zero [33]. This reduction of mag-
netization is what is referred to as demagnetization of the magnets and the consequences
of this is that the maximum energy product of the magnets is reduced which lowers the
magnetic torque and saturation flux of the magnets and is therefore undesirable.
As with all materials, magnetic materials are temperature sensitive where the tem-
perature of the magnet alter the magnetic characteristics. NdFeB magnets experience a
reduction in both remanent flux and coercivity for higher temperatures [31], meaning that
rated magnet values, which often are given at room-temperature, are slightly misleading
since the operating temperature tends to be higher. This, in turn, means that the NdFeB
magnets are more sensitive to demagnetization for higher temperatures. Ferrite magnets
are also affected by temperature differences, but as opposed to NdFeB the remanent flux
actually increases with decreasing temperature while the coercivity decreases [31]. This
means that the ferrite magnets are the most sensitive to demagnetization at lower temper-
atures.

2.8 Theoretical foundation of design approach


In a previous section a range of geometric variables were defined. In this section the
relationship between these are explained and the relevant design parameters are defined
to allow for manageable design variables. The design approach is expanded on in [15] and

32
2.8. Theoretical foundation of design approach

it relies on a number of key assumptions which are

• Saturation effects are neglected

• Stator slotting effects are neglected

• Magnetic scalar potential drop in the iron is neglected

• The stator winding is assumed to be ideal

• Distribution effects of the MMF is disregarded

2.8.1 Rotor barrier end angles


The rotor barrier end angles are are distributed along the rotor periphery with the constant
rotor slot pitch angle, ↵m , between them according to

(2h 1)↵m
✓b,h = (2.24)
2
This is done as this part of the design is based on the work conducted in [23] which
advocates constant rotor slot pitch. However, in order to allow for a greater degree of free-
dom in order to minimize the torque ripple and other unwanted side-effects of rotor/stator
slotting as expanded upon in Section 2.6.4 the point (B) is introduced on the periphery of
the rotor which allows for altering the position of the rotor slots. The outermost barrier
slots is shifted further from the q-axis with the displacement angle s . Fig. 2.19 displays
the situation in a machine with three flux barriers. The addition of the displacement angle
entails that all rotor slots are shifted from each-other with equal angles except the slots
closest to the q-axis.
Hence, the rotor slot pitch angle can be calculated as

s
2p
↵m = (2.25)
1
k+
2
where k is the number of rotor barriers and s can be regarded as a design variable.

2.8.2 d/q-axis MMF and barrier sizing


The applied magnetomotive force (MMF) is assumed to be sinusoidal in d- and q-axis.
When this MMF is applied in either the q- or d-axis direction it is possible to derive a step
function where the steps has the value of the average MMF seen by the corresponding
iron segment as seen in Fig. 2.20 [15, 23].

33
2.8. Theoretical foundation of design approach

s
(B)
(C)

↵m

↵m

↵m

↵m /2
d

Fig. 2.19: Rotor barrier angles in design approach.

Hence, assuming that there are k barriers the average values seen by the iron seg-
ments for a d- and q-axis MMF, fd,i and fq,i , represented by the steps in Fig. 2.20 can in
per unit be expressed as
✓Z
b,i+1
1 sin ✓b,i+1 sin ✓b,i
fd,i = cos ✓d✓ = i = 0, ..., k 1 (2.26a)
✓b,i ✓b,i+1 ✓b,i
✓b,i
✓Z
b,i+1
1 cos ✓b,i cos ✓b,i+1
fq,i = sin ✓d✓ = i = 0, ..., k 1 (2.26b)
✓b,i ✓b,i+1 ✓b,i
✓b,i

Where the angles are defined as in Fig. 2.10. Note that ✓b,0 =-✓b,1 since the iron segment
aligned with the d-axis is shared with another pole.
It was shown in [15] that if the ratio of the permeances across each flux barrier were
assumed to be constant for any barriers, the following relationship for the flux barrier

34
2.8. Theoretical foundation of design approach

q-axis MMF d-axis MMF


1 1

0.5 0.5

0 0
0 10 20 30 40 0 10 20 30 40
Rotor periphery angle Rotor periphery angle

Fig. 2.20: Sinusoidal MMF as function rotor periphery, sinusoidal and averaged step function.

widths in the q-axis direction applied


!2
Wi,q fq,i
= (2.27)
Wj,q fq,j

where i and j denotes any barriers in the structure and fq,i is the difference in q-axis
MMF across the i:th flux barrier. Note here that other assumptions can be made with
regards to the permeance ratio which would lead to other relationships, however the con-
stant permeance ratio was utilized as this would allow for a sinusoidal flux distribution in
the air gap [15].
However, if there are k barriers we find that equations (2.27) only gives k 1
equations. Recalling from previous section, the insulation ratio is defined as the ratio
between air and iron in the different directions. Hence, in order to complete the system of
equations we define the following relationship
k
X (R1 Rsh )
Wh,q = (2.28)
1
h=1 1+
kw,q
where R1 and R1 are the rotor radius and shaft radius respectively, as seen in Fig. 2.10.
Furthermore, in order to allow for a near-constant flux density in each iron segment
the iron segment width was set to be proportional to the average d-axis MMF seen by the
segment according to
2S1,q fd,1
= (2.29a)
S2,q fd,2
Si,q fd,i
= i = 2, ..., k (2.29b)
S(i+1),q fd,(i+1)
where the first equation is a product of the angle definition in equation (2.26a).

35
2.8. Theoretical foundation of design approach

Similarly as with (2.27) we have k + 1 unknowns and k equations and in the same
fashion we define the last equation from kw,q according to
k+1
X (R1 Rsh )
Sh,q = (2.30)
h=1
1 + kw,q

The d-axis flux barrier widths are determined by assuming them to be proportional
to their corresponding q-axis widths as
Wi,d Wi,q
= i = 1, ..., k 1 (2.31)
W(i+1),d W(i+1),q

Once again we find the system of equations to be under-determined with k 1


equations and k unknowns. However, the path along which to calculate the d-axis insula-
tion ratio is defined by introducing a point (C) on the rotor periphery as seen in Fig. 2.19.
↵m
Point (C) is defined as point (B) if s = . In [15] point (C) is said to be positioned
2
on the ”conventional path” meaning defined as if s was not variable. Therefore, the in-
sulation ratio in the d-axis is calculated from point (C) along the dashed line in Fig. 2.19
towards the d-axis. With this in place it is possible to define the last equation to complete
the equation system corresponding to equation (2.29a) as
0
⇡ 3↵m
k
X R1 sin ( )
2p 4
Wh,d = (2.32)
1
h=1 1+
kw,d

where ↵m is the angle derived from equation (2.25) if s = ↵m /2 and has the value

2p(k + 1)
Hence, the rotor flux barrier placement and geometry of the SynRM is defined by
the three input variables kw,q , kw,d and s . It is important to note here that the displacement
angle affects each individual barrier width and which barrier is the thickest, this becomes
obvious when looking at the above equations and this should be kept in mind when alter-
ing s such that no barrier becomes unfeasible. Note also that this design approach only
fully defines the rotor geometry if the barrier arm angles ↵i are defined as constant or
considered known (see Fig. 2.10).

36
Chapter 3

Method of analysis

In this chapter, the design process is expanded upon based on the theory described in
the previous chapter. The parametric study is explained and assumptions are justified.
Additionally, the economic evaluation is expanded upon.

3.1 Modeling
The machines, both PMaSynRM and PMSM, were modelled utilizing the finite element
software FLUX. All models created were verified utilizing the analytical/finite element
software SPEED which provides a lesser accuracy at a much faster computational time.
When simulating in FLUX, all models were calculated with imposed pure sinusoidal cur-
rent.

3.1.1 Performance parameters


The performance of the machines was evaluated based on four key indicators at the speed
1500 RPM. These were

• Average torque

• Torque ripple

• Power factor

• Efficency

Average torque production of a machine is the most obvious and crucial indicator
as, for a given speed, it indicates the power output of the machine and shows how much
load the machine can handle. The average torque calculated over the span of one elec-
trical period since over that period the rotor will see all possible rotor/stator slot relative
positions.

37
3.2. Initial dimensions and target PMSM

The torque ripple is another important aspect in terms of operation. While not a key
factor for pumping applications, as stated earlier, it is an undesirable aspect which should
be minimized if possible The torque ripple is calculated as the peak-peak torque ripple,
given in percent of the average torque.
The power factor is another important factor to control as a high power factor of the
motor can keep the VFD at lower rating and thereby reduce costs. In FLUX, it is possible
to add the end-winding inductances to the model, this in turn allows for the possibility
of retrieving the voltage as seen from the motor terminals from the time-derivative of the
flux. The fundamental voltage and angle was retrieved via FFT and since the current angle
was known, the power factor could be calculated.
The efficiency is an important factor since it determines how great the losses of the
machine is. It is calculated according to its definition as
Pmech
⌘= (3.1)
PM ech + PIron + PM agn + PF ric + PCu

where ⌘ is the efficiency. PM ech is the mechanical output power, PM agn is the losses in
the magnets, PF ric is the frictional losses from bearings and such, PCu is the copper
losses in the stator winding. The iron losses are a sum of three parts, namely Physt , Peddy
and Pstray which correspond to the hysteresis losses, eddy current losses and stray losses
respectively.
The hysteresis and eddy losses were calculated by use of a built-in function in
FLUX, which utilizes the Loss Surface (LS) model. The accuracy of the LS model is quite
good and can handle complex sinusoidal waveforms, but do require prior knowledge of
the material [34]. The hysteresis model utilized by the LS model is expanded upon in [35].
The stray losses, which are very difficult to predict analytically, were assumed constant
for all simulations. Therefore, based on data on stray losses from the utilized IM stator the
SynRM/PMaSynRM stray losses were set to be equal to 140 W. Similarly, for the PMSM
the stray losses were assumed to be 80 W.
Furthermore, the magnet losses in the PMaSynRM was assumed to be zero due to
the fact that the magnets would be mostly be buried in the rotor while the losses in the
surface mounted magnets on the PMSM were calculated in FLUX. The frictional losses
were based on the data from the IM and assumed equal for both the PMaSynRM and the
PMSM and was set to 50 W.

3.2 Initial dimensions and target PMSM


The aim of this thesis was to investigate the possible economic gain of utilizing a PMa-
SynRM for certain applications instead of a surface-mounted PMSM and to test this on
larger machines than in [10]. Therefore, the outer diameter of the machines was defined
to be 250 mm which is greater than the dimensions utilized in [10] and corresponded to

38
3.2. Initial dimensions and target PMSM

the largest PM machine available at Xylem.


Based on the design algorithm defined in Chapter 2, a parametric study was con-
ducted in order to derive the PMaSynRM rotor. The influence of the design parameters
were analyzed and the best overall performance for each design parameters was kept for
the final design.

3.2.1 Stator selection for PMaSynRM


To begin with the design procedure of the PMaSynRM, some target values needed to
be defined. In analogy with [10], a standard IM stator were utilized such that only the
rotor was to be designed. The stator utilized for the PMaSynRM had the dimensions and
winding data given in Table 3.1. Since it was assumed that the stator and rotor of the
PMaSynRM are punched from the same sheet of steel, both stator and rotor are assumed
to have the same steel grade throughout this thesis.

Table 3.1: Data for PMaSynRM

Number of slots 48
Outer diameter [mm] 250
Length [mm] 262
Number of turns 9
Number of parallel strands 10
Strand diameter [mm] 0.9
End-winding inductance [mH] 0.34
Stator steel grade M800-50

The current was set to a rated value of 26 A as this was the rated value for the
intended IM and provided a suitable current density value which corresponded to what
was found in literature [30] and the experience at Xylem.

3.2.2 Target PMSM


The PMSM used for comparison was based on an existing model developed by Xylem.
However, since the torque density for a PMaSynRM is lower than for a PMSM, the origi-
nal sizing of the PMSM would be wrong in terms of torque production. Instead, the length
of the machine was adjusted after the design of the PMaSynRM was finished such that
the torque in the MTPA operating point of the PMSM matched that of the PMaSynRM.
The relevant end-winding inductance and stator resistance for the adjusted length was
calculated with the software SPEED. A cross-section of the PMSM can be seen in Fig.
3.1. As can be seen, the PMSM had 12 slots and 10 poles and the stator was divided

39
3.3. Design procedure

Fig. 3.1: Cross-section of target PMSM.

up into segments. This is because it was of the type fractional slot concentrated wind-
ing (FSCW), and in this case it thus has single-tooth windings which gives it a very low
leakage inductance.
The primary rated parameters of the PMSM can be seen in Table 3.2

Table 3.2: Preliminary PMSM parameters

Outer radius [mm] 250


Rated current [A] 44
Number of poles 10
Number of slots 12
Number of turns 24
Number of parallel strands 7
End-winding inductance [mH] 0.042
Stator steel grade M600-50
Rotor steel grade M800-50

3.3 Design procedure


The design procedure utilized in this paper is based on the algorithm outlined in [15]
as it provided a good theoretical foundation and also was the procedure followed in the

40
3.3. Design procedure

Pre-defined Base-line Electrical


dimensions model limitations

Barrier
Insulation center PF
Torque
ratios magnet Torque
width
Torque
Air-gap Barrier arm
Torque ripple PF
magnet
Torque
width
Rotor
Torque ripple Efficiency
slot pitch
Steel types (Torque)
(PF)
3 or 4
barriers

Torque Radial ribs

Fig. 3.2: Flowchart for design procedure.

previous thesis [10]. Since the primary objective of this thesis were not to achieve an
optimal design for an PMaSynRM but to investigate the feasibility of replacing a PMSM
with a PMaSynRM this algorithm was deemed sufficient to pose as a basis for the design
and comparison. Based on this theory, a parametric study was conducted where the inputs
were varied in order to find the best results. The performance was evaluated at the MTPA
operating point, which shifted in terms of current angle when the geometry changed.
First the SynRM rotor was developed and when the performance of this was satisfactory,
magnets were added in order to further improve its performance. Fig. 3.2 displays the
design procedure in a flowchart together with the most relevant performance parameters
equipped to the different design variables, the blue boxes describes the SynRM design
while the green boxes describe the PMaSynRM design.

3.3.1 Parametric study


As has been shown in the Chapter 2, the utilized design algorithm allows for 3 degrees of
freedom when it comes to determining a basic suitable rotor design, these three parameters

41
3.3. Design procedure

were
• q-axis insulation ratio, kw,q

• d-axis insulation ratio, kw,d

• Rotor slot displacement angle, s

These three parameters were deemed sufficient to determine a basic design of a


SynRM-rotor. However, more aspects needed to be considered in order to derive a finished
PMaSynRM rotor design such as
• Air-gap length, g

• Radial rib width, wr,i

• Magnet widths, wma,i and wm,b


The study was conducted and the performance was primarily operationalized from
the torque production, as the torque density of the PMaSynRM generally is lower than the
PMSM and the most important factor for pumping applications. However, other aspects
of the machine was also taken into consideration and were torque ripple, power factor and
efficiency.

3.3.2 SynRM base-line design


In order to initiate the design procedure a starting point needed to be defined. Given the
discussion regarding the number of barriers it was determined that two different designs
were to be developed initially, one with three barriers and one with four barriers, and
then compared as to which one displayed the best performance. The pole number of the
machine was set to be four, following the discussion in Section 2.2.
Hence, initial guesses needed to be made with regards to the input values in order
to have a design starting point where non-swept values were given a default value.
Additionally, some values were out of the scope of the analysis and assigned a fixed
value. These values were
• The angle ↵i between barrier arm and center part of barrier (see Fig. 2.10) was to
be kept constant at 135 as it simplified the analysis and was shown to be suitable
in [11].

• The tangential ribs were kept constant at 1 mm as their influence was out of the
scope of the analysis and kept at the minimum allowable tolerance from the manu-
facturer at 1 mm
The two base-line designs were based on the same input-values and can be seen in
Table 3.3. Note that the base-line value for s was 5.625 for three barriers and 4.5 for
four barriers.

42
3.4. SynRM design

Table 3.3: Base-line design inputs

Input Description Value


kw,q Insulation ratio, q-axis 0.7
kw,d Insulation ratio, q-axis 0.2
s Rotor slot displacement angle [deg] ↵m /2
g Air-gap length [mm] 0.6
↵i Flux barrier angle [deg] 135
wt,i Tangential rib width [mm] 1

Fig. 3.3: Impact of q-axis insulation ratio, 0.4 (left) and 0.9 (right).

3.4 SynRM design


The above stated design parameters was utilized in order to derive a satisfactory SynRM
rotor design. Below a brief explanation of each design parameter is given and the intervals
that were investigated.

3.4.1 Insulation ratios


Given the importance of the insulation ratios for the performance of the machine as ex-
panded on in Section 2.6.2, these were the first parameters to be investigated and set.
Since the q-axis insulation ratio in large determine the upper limit of the d-axis insulation
ratio this was the first to be investigated. For the utilized stator the stator insulation ratio
was calculated as 0.78. Therefore, the upper limit of kw,q was set to 0.9, a value slightly
above the stator insulation ratio and the lower limit was set to 0.4. From this simulation
the value for kw,q was considered set and the d-axis insulation ratio was investigated. A
representation of how the selected interval of kw,q affects the rotor geometry is illustrated
in Fig. 3.3.
The upper limit of the d-axis insulation ratio was set to be equal to that of the chosen
q-axis insulation ratio and the lower value was set to a value reasonably lower than that
of the upper limit. Fig. 3.4 illustrates how kw,d impacts the rotor dimensions.

43
3.4. SynRM design

Fig. 3.4: Impact of d-axis insulation ratio, 0.2 (left) and 0.7 (right).

Fig. 3.5: Impact of rotor slot displacement angle, 0 (left) and 20.625 (right).

3.4.2 Air-gap
While the torque production and power factor benefits of a shorter air-gap length, the
torque ripple increases with lower lengths. Therefore, the air-gap length was investigated
to find the most suitable value. While the mechanical limit was set to 0.6 mm for the
stator due to manufacturing limitations, the investigated interval was set to 0.5 mm to 1
mm in order to investigate how the performance was affected by a lower value for future
references.

3.4.3 Rotor barrier end angles


The rotor barrier end angles were, as mentioned in Chapter 2, determined by the rotor
displacement angle s . The displacement angle was set to be varied by between 0 and
20.625 degrees with an increment of one fourth of a stator slot pitch. This interval was
chosen as preliminary studies showed that for values of s exceeding 20 degrees would
not result in any feasible barrier positions and barrier thicknesses. The impact of s on
the rotor geometry can be seen in Fig. 3.5, here we see how the the displacement angle
changes the thicknesses of each individual barrier.

44
3.5. PMaSynRM design

3.4.4 Choice of barriers


After the aforementioned parameter investigations given previously in this section, it was
assumed that enough data on whether three or four barriers was the best choice in terms of
performance. Therefore, it was determined sufficient to only investigate one design when
adding PMs, radial ribs and such.

3.4.5 Radial ribs


It was expected that introducing the radial ribs in the flux barrier structure would corre-
spond to a decrease in torque and therefore it was deemed desirable to keep the ribs as
small as possible. However, in the manufacturing process the punching does not allow for
small ribs. As the case with the tangential ribs, the minimum size of the radial ribs was
deemed to be 1 mm based on previous experience.
As stated above, the ribs were desirable to keep as small as possible, the safety
factor was put in place in order to ensure that the ribs do not break as that would have
devastating consequences. The sizes of the radial ribs that were tested was calculated
based on equation (2.20) utilizing different values on the safety factor ⌫ with restrictions
on the minimum size due to the manufacturing aspects. This analysis was made in order
to see how much of a detrimental effect the ribs would have depending on their thickness.
However, given the overestimation of applied force to the rotor segments sufficiently small
ribs was neglected as it could be assumed that it was not necessary. Hence, the limitations
were set to be

• No ribs were allowed to be thinner than 1 mm due to manufacturing limitations

• Any calculated rib size smaller than 0.2 mm were neglected and no ribs was utilized
in that barrier

The values for the safety factor to be utilized was set to be ⌫ = {1.5, 2, 2.5, 3}
where the 1.5 value was motivated by the overestimation of the rib sizes.

3.5 PMaSynRM design


Two different types of magnets were used in the simulations of the PMaSynRM, one
ferrite magnet and one NdFeB. The ferrite magnet data was supplied by NORDMAG,
without specified grade standard, and the NdFeB was the grade N33EH. The characteris-
tics of the two magnets are summarized in Table 3.4 and the data-sheets can be seen in
Appendix B.2.

45
3.5. PMaSynRM design

Table 3.4: Rated magnet type characteristics at 20 C

Ferrite NdFeB
Br [T] 0.430 1.165
Hc,b [kA/m] 310 867
BHmax [kJ/m3 ] 35 267

Fig. 3.6: Magnet fill procedure.

3.5.1 Magnet sizes


The magnet widths were determined by introducing the magnet fill factors for the barrier
center and arms, kmag,a and kmag,b , which describes the amount of magnets in the barrier
segments. This was done as it followed the distributed magnet approach and at the same
time reduced the number of variables.
As ferrite magnets possess a lower energy product than NdFeB magnets and thus
posed as the limiting factor in terms of PM air-gap flux density the design started with
ferrite magnets. First the center barriers was filled with magnets with kmag,a = [0, 1] and
thereafter with kmag,b = [0, 1]. Fig. 3.6 illustrates the magnet filling procedure.
After the ferrite magnets, the NdFeB magnets were tested. However, no sweep was
made for the NdFeB since it is obvious that NdFeB would prove a better performance.
Instead, the NdFeB magnet amount was determined such that the flux linkage from the
NdFeB corresponded to that of the ferrite magnets as this would allow for a fair economic
comparison with regards to which magnet type would be cheapest. This was done by
assuming an equal distribution of all barriers, both center and arms. Thereafter the fraction
of magnets in the barriers were either reduced or increased depending on if the flux linkage
was greater or lower than the ferrite flux linkage.

3.5.2 Steel grade


The influence of utilizing different steel types was tested. The types analyzed were M800,
which was the the base-line steel-type, M600 and M400. All steel-types had the same
thickness of 0.50 mm and thus only the grade was different. All geometrical parameters

46
3.6. Performance comparison and economic analysis

were kept constant. For datasheets of the steels refer to Appendix B.1.

3.5.3 Demagnetization
In order to investigate demagnetization of the different magnets, the demagnetization
model as described in Section 2.7.1 was utilized in a built-in macro in FLUX where it
reduces the remanent flux when Hk was exceeded. The threshold for any part of the mag-
net being considered demagnetized was set to be 90% of the nominal remanent flux and
after the simulation the fraction of each magnet that violated that threshold was calculated.
In order to aptly account for as many demagnetization scenarios as possible the
current was set to be 1, 1.5, 2 and 2.5 times the nominal current levels. 2.5 was chosen
to be the upper limit because the drives normally utilized for machines of the intended
purpose has a current limit of around two times the nominal current.
Furthermore, the current angle also influences the demagnetization as the stator flux
should counteract the magnetization of the magnets. Since the magnets are magnetized in
different directions, different current angles were tested. Preliminary tests showed that
the most detrimental current angles were in the span 70 to 110 degrees. While these are
not commonplace current angles, it is important to take them into consideration since
the motor is intended for pumping applications which might lead to unintended clogging
which abruptly stops the machine and these angles might occur for short periods of time.
The demagnetization was tested at an operating temperature at which the magnets
was most at risk for experience demagnetization. For the ferrite magnets, this was deter-
mined to be at 0 C which was deemed the lowest feasible magnet temperature given a
start-up sequence in cold water. For the NdFeB the maximum operating temperature was
set to 120 C.

3.6 Performance comparison and economic analysis


The intended purpose of the developed PMaSynRM was to pose as a more cost-effective
alternative for the applications of a surface-mounted PMSM, where a power electronic
drive was utilized for speed control. Therefore, it was important to compare these two
types of machines in order to investigate whether or not the developed PMaSynRM were
comparable to the PMSM.

3.6.1 Comparing the machines


The performance comparison was conducted based on equal torque production as the
spatial requirements for the intended application was not of great concern. Since the two
machines displayed different behaviour in terms of torque production and power fac-
tor related to current angle, the performance comparison was made based on the torque

47
3.6. Performance comparison and economic analysis

production curves as function of current angle at rated current in order to assess the char-
acteristics of the machines. The machines were run at 1500 RPM. Any speed range tests
along with high-speed analysis was not deemed to be of great importance as the speed
range is in part limited by the DC link voltage which for this thesis was not a constraint.
Field-weakening operation was not of great relevance as the load torque for pumping ap-
plications tend to increase with the square of the rotational speed, and therefore to some
extent impractical in CPSR.
The economic analysis were made based on material costs and some production
costs of the stators due to the differences in production methods between the two ma-
chines. The costs of the machines are an important factor in this assessment as it com-
bined with the performance of the machines will help determine for which applications
the PMaSynRM is feasible.
For the economic analysis the following assumptions are made

• The PMaSynRM stator and rotor was assumed to be punched from a single sheet
of steel and excess steel is considered waste. The total steel usage thus is a square
with sides equal to the outer diameter of the machine.

• The PMSM stator segments are smaller and easier to punch and the shape of the
segments allows for less waste and therefore was calculated by assuming that a
rectangular segment with one side as wide as a stator segment and one side 50%
longer than one stator segment has space for two stator segment as exemplified in
Fig. 3.7

• The PMSM rotor is assumed to be punched from a single sheet of steel, given the
consumed material equal to the square of the diameter of the rotor.

• Only the cost per weight of the magnets were considered without taking the press-
ing, cutting or transport of the magnets into consideration.

• Only the manufacturing costs of the stators were taken into consideration as the
rotors are manufactured by a third party and thereby the manufacturing cost is diffi-
cult to estimate. Furthermore, the greatest difference manufacturing-wise is deemed
to be located to the stator. The manufacturing costs was given by the economics de-
partment at Xylem.

As magnets corresponded to a significant part of the cost of a machine, which is es-


pecially true when NdFeB magnets were utilized, and given the fluctuating nature of these
types of magnet, it was investigated if there occurred any breaking points with regards to
the NdFeB machines and the ferrite PMaSynRM.
In this thesis, given the sensitive nature of material pricing, all costs was given in
per unit (pu) cost as opposed to real currency. The base cost was set to be the ferrite PMa-
SynRM with the steel M800-50A and magnet price in the middle of the given interval.

48
3.6. Performance comparison and economic analysis

Fig. 3.7: Punching layout of PMSM stator segments.

49
Chapter 4

Results

In this chapter, all the results are presented. All relevant plots and figures are given and
discussed to some extent. The values which were carried over are given and the final
design is displayed.

4.1 Rotor design


As was described in Chapter 3, the design process was divided up into several parts where
the result from the previous simulations were added on to the baseline design. In this
section, the results from this analysis will be described.

4.1.1 q-axis insulation ratio


The q-axis insulation ratio was varied between 0.4 and 0.9, utilizing all base-line values
as seen in Table 4.1.
Table 4.1: Simulation parameters for kw,q

kw,q kw,d g s
Variable Baseline Baseline Baseline

In Fig. 4.1 the results for torque, torque ripple, power factor and efficiency can be
seen respectively. It is useful to note that the torque doesn’t vary particularly much over
this interval, which can be attributed to the fact that the interval was too narrow to see the
detrimental effects of too low or too high values, and only the maxima was achieved.
We can also note that in terms of torque and power factor, the four barrier design
displayed better performance than three barriers whilst generally displaying worse perfor-
mance in terms of torque ripple. However, the torque ripple wasn’t the key performance
indicator for kw,q since the ripple was shown to be mostly governed by the location of
the rotor barrier ends. Thus, the optimum values for both barriers were deemed to be

50
4.1. Rotor design

160
Three barriers Three barriers

Average torque [Nm]


30
Four barriers Four barriers

Torque ripple [%]


140
20
120
10
100

0
0.4 0.5 0.6 0.7 0.8 0.9 0.4 0.5 0.6 0.7 0.8 0.9
kw,q kw,q

0.8 0.96
Three barriers Three barriers
0.75 Four barriers 0.955 Four barriers
Power factor

Efficency
0.7 0.95

0.65 0.945

0.6 0.94
0.4 0.5 0.6 0.7 0.8 0.4 0.5 0.6 0.7 0.8 0.9
kw,q kw,q

Fig. 4.1: Simulation results for sweep over kw,q .

0.7 and utilized in the upcoming simulation, since this value displayed the highest torque
production and acceptable levels of power factor and efficiency

4.1.2 d-axis insulation ratio


Given the choice of kw,q , the d-axis insulation ratio was altered between 0.2 to 0.7 with
the parameters seen in Table. 4.2. The results can be seen in Fig. 4.1.

Table 4.2: Simulation parameters for kw,d

kw,q kw,d g s
0.7 Variable Baseline Baseline

Here we see that the torque is maximized for both three and four barriers at kw,d =
0.3 and the detrimental impact on torque production of a too large insulation ratio becomes
apparent. Furthermore, we see that the torque ripple is affected negatively by increasing

51
4.1. Rotor design

160
Three barriers

Average torque [Nm]


30
Four barriers

Torque ripple [%]


140
20
120
10
100 Three barriers
Four barriers
0
0.2 0.3 0.4 0.5 0.6 0.7 0.2 0.3 0.4 0.5 0.6 0.7
kw,d kw,d

0.8 0.96
Three barriers Three barriers
0.75 Four barriers 0.955 Four barriers
Power factor

Efficency
0.7 0.95

0.65 0.945

0.6 0.94
0.2 0.3 0.4 0.5 0.6 0.7 0.2 0.3 0.4 0.5 0.6 0.7
kw,d kw,d

Fig. 4.2: Simulation results for sweep over kw,d .

insulation ratio. Once again, torque ripple wasn’t the key performance indicator for the
insulation ratio and since the increase in ripple is so low for the optimal value for torque
the slight increase is allowed. The same thing applies to the power factor, the maxima for
both power factors are situated at kw,d = 0.5 but since the variation is so small, especially
for the four barrier configuration this non-optimal value is allowed. Furthermore, we see
that the efficiency is negatively affected by increasing insulation ratio for four barriers
while higher values affect the efficiency negatively for three barriers. This can probably
be attributed to increased rotor saturation, which also can explain the decrease in torque
production for greater values of kw,d .

4.1.3 Air-gap
The input data for the air-gap sweep is given in Table 4.3. The air-gap sweep was con-
ducted between 0.4 mm to 1 mm. As can be seen from Fig. 4.3, the torque production
declines almost linearly with the air-gap length. Also, as expected the power factor re-
duces greatly. The efficiency declines, most likely as a product of the reduced mechanical

52
4.1. Rotor design

160
Three barriers Three barriers

Average torque [Nm]


30
Four barriers Four barriers

Torque ripple [%]


140
20
120
10
100

0
0.4 0.6 0.8 1 0.4 0.6 0.8 1
Air-gap length [mm] Air-gap length [mm]

0.8
Three barriers Three barriers
Four barriers 0.96 Four barriers
Power factor

0.7

Efficency
0.94
0.6

0.5 0.92
0.4 0.6 0.8 1 0.4 0.6 0.8 1
Air-gap length [mm] Air-gap length [mm]

Fig. 4.3: Simulation results for sweep over g.

power. The torque ripple reduction is quite modest considering the drastic reduction of
torque production. Thus, it is visible that it is imperative to maintain the air-gap as short
as possible and a value below 0.6 mm would be beneficial. Given the steep decline of the
torque production and the low decrease in torque ripple, the minimum allowable air-gap
height was chosen, i.e. g = 0.6 mm.

Table 4.3: Simulation parameters for g

kw,q kw,d g s
0.7 0.3 Variable Baseline

53
4.1. Rotor design

160 80
Three barriers Three barriers

Average torque [Nm]


Four barriers

Torque ripple [%]


60 Four barriers
140

120 40

20
100

0
0 5 10 15 20 0 5 10 15 20
s [deg] s [deg]

0.8 0.96
Three barriers Three barriers
0.75 Four barriers 0.955 Four barriers
Power factor

Efficency
0.7 0.95

0.65 0.945

0.6 0.94
0 5 10 15 20 0 5 10 15 20
s [deg] s [deg]

Fig. 4.4: Simulation results for sweep over s.

4.1.4 Rotor barrier end angles


The rotor slot displacement angle, s , was swept between 0 and 22.5 degrees with the
determined simulation parameters set as given in Table 4.4.

Table 4.4: Simulation parameters for s

kw,q kw,d g s
0.7 0.3 0.6 mm Variable

As can be seen from Fig. 4.4 the torque production remains quite constant up until
the higher values of s where it starts to decline. As for the torque ripple however, we
observe a very pronounced influence. As was discussed in Chapter 2 the position of the
rotor slots are very important for the torque ripple which becomes very apparent here. The
power factor displays a slight variance in value, but not significantly so.
Note that the optimum value for three and four barriers are widely separated here,
for three barriers the optimum value is for 9.375 and for four barriers the optimum value

54
4.1. Rotor design

was 0 when considering both torque production and torque ripple. These differeces in
rotor angles are reasonable since the actual rotor slot positions are different for the same
value of s . With previous simulations, the basic SynRM design is concluded with the
performance as seen in Table. 4.5.

Table 4.5: Performance for best preliminary 3 and 4 barrier design

3 barriers 4 barriers
Torque 117.56 119.96
Torque ripple 10.32 6.85
Power factor 0.649 0.650
Efficiency 0.952 0.953

After these simulations it could be concluded that for this 48 slot stator, four bar-
riers were the best performing configuration in all regards. This resulted in the choice of
continuing the simulations with only the four-barrier alternative with the design parameter
values which can be seen in Table. 4.6

Table 4.6: Basic design parameters for four-barrier SynRM rotor

kw,q kw,d g s
0.7 0.3 0.6 mm 0

4.1.5 Radial ribs


The geometry from the previous simulations was used to calculate the rib-sizes for the
radial ribs following the restrictions described in Chapter 2. The resulting values can be
seen in Table. 4.7. The constraints set on the ribs with the minimum size affected all ribs
sizes indicated by an asterisk giving an over-dimensioning of the two middle ribs and a
under-dimensioning of the outer barrier.

Table 4.7: Radial rib sizes for different values of safety factors

Safety factor, ⌫ wr,1 [mm] wr,2 [mm] wr,3 [mm] wr,4 [mm]
1.5 1.4986 1.0000* 1.0000* 0*
2 1.9981 1.1797 1.0000* 0*
2.5 2.4976 1.4747 1.0000* 0*
3 2.9971 1.7696 1.0000* 0*

The results can be seen in Fig.4.5. As expected, the torque production is affected by
the introduction of ribs and we also see a decrease in power factor. The torque reduction

55
4.1. Rotor design

10

Average torque [Nm]

Torque ripple [%]


120 8

4
100
2

0
No ribs 1.5 2 2.5 3 No ribs 1.5 2 2.5 3
⌫ ⌫

0.7 0.96

0.68 0.958
Power factor

Efficency
0.66 0.956

0.64 0.954

0.62 0.952

0.6 0.95
No ribs 1.5 2 2.5 3 No ribs 1.5 2 2.5 3
⌫ ⌫

Fig. 4.5: Simulation results for different values of ⌫ on four barrier rotor.

and PF reduction from the case where no ribs were utilized as a function of safety factor
can be seen in Tab. 4.8.

Table 4.8: Torque reduction due to introduction of ribs

Safety factor, ⌫ Torque reduction [%] PF reduction [%]


1.5 -2.75 -1.47
2 -3.41 -4.13
2.5 -4.20 -4.99
3 -5.01 -5.85

For the forthcoming simulations it was determined that the safety factor of 1.5 was
sufficient, following the discussion with regards to the overestimation of the rib width.

56
4.1. Rotor design

Fig. 4.6: Final design of SynRM rotor.

4.1.6 Final SynRM rotor geometry


With the above conducted analysis, the rotor design corresponding to the SynRM-design
can be regarded as set. The final design can be seen in Fig. 4.6 and corresponding rotor
dimensions can be seen in Table. 4.9. It can be seen that the outermost barrier became
quite small, only 1.25 millimeters. This might cause difficulties to add magnets to that
barrier, which will be discussed further in the next section.

57
4.1. Rotor design

Table 4.9: Final rotor dimensions of SynRM rotor

Parameter Value Description


W1,q 10.37 Barrier 1 q-axis height [mm]
W2,q 8.02 Barrier 2 q-axis height [mm]
W3,q 4.42 Barrier 3 q-axis height [mm]
W4,q 1.25 Barrier 4 q-axis height [mm]
W1,d 5.20 Barrier 1 d-axis height [mm]
W2,d 4.02 Barrier 2 d-axis height [mm]
W3,d 2.21 Barrier 3 d-axis height [mm]
D01 31.97 Barrier 1 radial position [mm]
D02 53.54 Barrier 2 radial position [mm]
D03 70.70 Barrier 3 radial position [mm]
D04 81.08 Barrier 4 radial position [mm]
✓b,1 5 Barrier 1 end angle [ ]
✓b,2 15 Barrier 2 end angle [ ]
✓b,3 25 Barrier 3 end angle [ ]
✓b,4 35 Barrier 4 end angle [ ]
wr,1 1.50 Barrier 1 radial rib width [mm]
wr,2 1 Barrier 2 radial rib width [mm]
wr,3 1 Barrier 3 radial rib width [mm]
wr,4 0 Barrier 4 radial rib width [mm]
R1 84.40 Rotor radius [mm]

58
4.2. PMaSynRM design

4.2 PMaSynRM design


With the rotor design complete, the magnets were ready to be added. First, the ferrite
PM addition was simulated, thereafter the NdFeB magnet amount was calculated and
simulated. Additionally, the effects of leaving the outer barrier empty was investigated.
Last, the influence of utilizing different types of steel was analyzed.

4.2.1 Magnet addition


As stated in Chapter 3, the center barriers was filled with ferrite magnets and when this
was done the barrier arms were filled. In Fig. 4.7, the center barrier data can be seen.
As expected, the torque production and power factor experience a great increase. The
torque increases from 118 Nm to 143 Nm and the power factor goes from 0.64 to 0.79.
The torque ripple remains constant between 10-11 % for all values except the last where
it goes down to 8.5%. However, looking at the numerical values we see that the torque
ripple maintains a value between 11.5-14.5 Nm. From this data the choice fell quite easily
on completely filling the center barriers with magnets.
Next the barrier arms were filled with magnets, with all barrier centers filled with
magnets. The results can be seen in Fig. 4.8. Here we see that the torque goes from 143
Nm to 157 Nm while the power factor goes from 0.79 to 0.85. Interestingly, the torque
ripple reduces until 60 % of the arms are filled and thereafter it increases. However, these
values are not deemed to be worrisome given the application and therefore the entirely
filled barrier arm was chosen to define the final design of the rotor.
And thus, with the magnets in place the ferrite PM flux of the machine corresponded
to 40% of the nominal flux, a value which was in acceptable agreement with the discussion
with regards to the magnitude of PM flux.

4.2.2 Without magnet in outermost barrier


The final design of the rotor resulted in a quite small outer barrier (1.25 mm thick) as
seen in Table 4.9. Even though it might be possible to add ferrite magnets to this barrier
after discussion with the supplier it would result in very many small magnets since the
manufacturing process does not allow for magnet dimensions where one side is much
larger than the others for those dimensions. Therefore, the impact of leaving that barrier
filled with air was investigated. The results can be seen in Table. 4.10 where it is compared
to the entirely magnet-filled rotor.
Here we see that the torque reduction is around 2 Nm while the torque ripple also
decrease slightly and the other values are left virtually unaffected. Hence, it has been
shown that if it proves unpractical to add magnets to the outermost barrier the performance
reduction is within reasonable values.

59
4.2. PMaSynRM design

160

Average torque [Nm]


30

Torque ripple [%]


140
20
120
10
100

0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Fraction of center barrier filled Fraction of center barrier filled

0.9 0.97

0.965
Power factor

0.8

Efficency
0.96
0.7
0.955

0.6 0.95
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Fraction of center barrier filled Fraction of center barrier filled

Fig. 4.7: Simulation data when adding magnets to center barrier.

4.2.3 NdFeB
The NdFeB magnets were filled in all barriers including the outermost, as that topology
would provide the base for the economic calculations. The tolerance for the NdFeB flux
linkage versus the ferrite flux linkage was 1%. The magnet fill factor for the center barrier,
kmag,a , was set to 0.4 while for the barrier arms, kmag,b was set to 0.3625. The results of
the simulations can be seen in Table. 4.11
It is quite noticeable that the torque production is lower for the NdFeB, while the
torque ripple is higher. This can probably be attributed to the distribution of the magnets in
the barriers and the fact that the magnet flux does not affect all parts of the iron segments
in the same way since there are large portions of air in the flux barriers with NdFeB
magnets in the rotor. However, these results were deemed to be sufficient as basis for
the economic comparison since the performance of the two machines are similar to each
other.

60
4.3. Demagnetization

160

Average torque [Nm]


30

Torque ripple [%]


140
20
120
10
100

0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Fraction of center barrier filled Fraction of center barrier filled

0.9 0.97

0.965
Power factor

0.8

Efficency
0.96
0.7
0.955

0.6 0.95
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Fraction of center barrier filled Fraction of center barrier filled

Fig. 4.8: Simulation data when adding magnets to barrier arms.

4.2.4 Steel types


The different steel-types tested were M800, M600 and M400. The results of the simula-
tions can be seen in Table 4.12.
As can be seen, the torque production actually decrease as the steel grade improves.
However, we see that the power factor does get affected by the steel grade. This can prob-
ably be attributed to the steel grades yielding different inductances and having slightly
different B-H curves and saturation levels which affected the d-q inductance difference
as well as the saliency ratio. Additionally, while the output mechanical power slightly de-
crease, so does the iron losses leaving the efficiency largely unaffected to any significant
number.

4.3 Demagnetization
The demagnetization was studied for a variety of current magnitudes and angles. The basis
for this simulation was the ferrite rotor with all barriers filled. The current levels analyzed

61
4.3. Demagnetization

Table 4.10: Performance of rotor without magnets in outer barrier

Filled rotor No magnet in Barrier 4


Torque [Nm] 156.80 154.89
Torque ripple [%] 7.97 7.43
Power factor 0.85 0.85
Efficiency 0.962 0.962

Table 4.11: Performance of NdFeB magnets versus ferrite

Ferrite NdFeB
Torque [Nm] 156.80 152.82
Torque ripple [%] 7.97 10.45
Power factor 0.85 0.86
Efficiency 0.962 0.962

were I = {In , 1.5In , 2In , 2.5In } and for every current value the current angles 70 , 80 ,
90 , 100 and 110 were investigated. The knee values utilized for the demagnetization
studies can be seen in Table 4.13
Due to limitations in FLUX, all magnet could not be investigated. Therefore only
half of the magnets in the center barriers were analyzed by assuming that the demagneti-
zation of these would be low due to their position and that any noticeable effect would be
fairly symmetrical. The analyzed magnets and their names can be seen in Fig. 4.9.
In Fig. 4.10 the amount of demagnetization is plotted for the percentually most
affected magnet for each current and angle. This figure serves as an indication for how
the demagnetization affected the machine, it is however not very descriptive for the state
of demagnetization throughout the machine as the higher current values demagnetized
more magnets a significant amount compared to the lower.
As it is hinted in Fig. 4.10, the demagnetization at nominal current was negligible
with values well below 1% demagnetization for all magnets. For the 1.5 times the nominal
current the situation changed somewhat but here the demagnetization really only affected
two magnets, namely L3 and R3 for all current angles except 90 . All demagnetization
values for nominal and 1.5 times nominal current can be seen in Table 4.14 where the
highest values for each current level is highlighted. However, the demagnetized areas for
the relevant magnets were not particularly worrisome since the values are not especially
high and the demagnetization is localized to the edges of the magnets as seen in Appendix
C, Fig. C.1, note that in these plots every color other than the main color signifies a value
below the threshold value.
Similarly, the demagnetization values for 2 and 2.5 times the nominal current can be
seen in Table 4.15 where the greatest values are highlighted. For these values a noticeable
trend appears which is that the most affected magnets are R3 and L3 and mostly around

62
4.3. Demagnetization

Table 4.12: Performance for different steel grades

M800-50 M600-50 M400-50


Torque [Nm] 156.80 156.00 153.99
Torque ripple [%] 7.97 8.04 7.93
Power factor 0.852 0.896 0.899
Efficiency 0.962 0.962 0.968
Rotor iron losses (only hysteresis and eddy) [W] 103.4 99.2 93.8
Stator iron losses (only hysteresis and eddy) [W] 211.2 209.7 205.4

Table 4.13: Magnet demagnetization values

Ferrite NdFeB
Br [T] 0.45 1.025
Hk [kA/m] 360 1195

100 for R3 and 70 for L3 . This is reasonable since R3 and L3 are positioned such that the
stator flux direction is mostly counteracting the remanent flux of the magnets. Similarly,
it can be noticed that the demagnetization for barriers 2 and 1 increases with the angular
distance from 90 . Additionally we see that the center barriers are well protected from de-
magnetization as was stated in Chapter 2. For these two values, the demagnetization began
to be significant. However, utilizing the same argument as earlier, the demagnetization for
2 times the nominal current was deemed to be insignificant for producing a noticeable
performance reduction. For 2.5 times the nominal current, however, the demagnetization
of especially barrier 3 and 2 begin to be of significant magnitude. The demagnetization
for these two current levels can be seen in Appendix C, Fig. C.2.
A similar analysis were conducted on the NdFeB rotor aswell, and as expected
almost no demagnetization occurred as can be seen in Table. 4.16 where only the highest
current is displayed. The low demagnetization can be easily explained by the high values
of both remanent flux and coercive force that characterizes the NdFeB magnets.

63
4.3. Demagnetization

L1 L2 L3
CL4

CL3

CL2

CL1
R3

R2

R1

Fig. 4.9: Naming of magnets in rotor.

14 = 70
= 80
Amount demagnetized [%]

12
= 90
10 = 100
= 110
8

0
In 1.5In 2In 2.5In

Fig. 4.10: Amount of demagnetization for worst magnet.

64
4.3. Demagnetization

Table 4.14: Percent of ferrite magnet demagnetized below 0.41 T for In and 1.5In

Current In 1.5In
Angle 70 80 90 100 110 70 80 90 100 110
L1 0.39 0.29 0.00 0.02 0.05 1.05 0.77 0.31 0.19 0.36
L2 0.35 0.35 0.35 0.12 0.00 1.45 1.45 0.35 0.12 0.25
L3 0.00 0.00 0.00 0.00 0.00 3.89 3.89 0.00 0.00 0.00
R1 0.06 0.02 0.00 0.26 0.43 0.40 0.26 0.32 0.91 1.07
R2 0.00 0.11 0.33 0.33 0.33 0.35 0.45 0.67 1.73 2.08
R3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.88 3.88
CL1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
CL2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
CL3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
CL4 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Table 4.15: Percent of ferrite magnet demagnetized below 0.41 T for 2In and 2.5In

Current 2In 2.5In


Angle 70 80 90 100 110 70 80 90 100 110
L1 1.35 1.26 0.65 0.60 0.93 2.14 1.56 1.07 0.98 1.50
L2 2.14 2.14 0.70 0.97 0.85 3.74 3.03 1.91 2.07 2.08
L3 3.89 4.15 4.15 0.26 0.26 9.20 13.61 5.84 3.02 5.49
R1 0.92 0.66 0.62 1.22 1.48 1.58 1.04 1.31 1.64 2.38
R2 0.84 0.95 0.77 2.18 2.08 2.07 1.73 2.18 4.06 4.06
R3 1.70 1.70 5.57 5.57 4.14 5.76 4.10 5.83 13.59 9.45
CL1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
CL2 0.00 0.00 0.00 0.00 0.00 0.02 0.02 0.02 0.02 0.02
CL3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
CL4 0.00 0.00 0.00 0.00 0.36 0.00 0.00 0.98 0.98 0.98

Table 4.16: Percent of NdFeB magnet demagnetized below 0.9 T

Current 2.5In
Angle 80 90 100 110
L1 0 0 0 0
L2 0.3221 0 0 0
L3 0 0 0 0
R1 0 0 0 0
R2 0 0 0.3273 0.3273
R3 0 0 0 0

65
4.4. PMSM versus PMaSynRM

4.4 PMSM versus PMaSynRM


The finished design of the PMaSynRM was compared to the PMSM in terms of both
performance and economic parameters. Below the results of the simulations conducted
on both PMaSynRM and PMSM is presented along with the calculations of the cost of
both machines.

4.4.1 Performance comparison


When it comes to comparing the two machines in terms of performance, there are no
straight-forward method to do this. Table 4.17 displays the results of the simulations at
MTPA for both machines. We see that the torque production of the PMSM was slightly
higher than the PMaSynRM but within a reasonable range. The torque ripple of the PMSM
was as expected significantly lower than the PMaSynRM. The efficiency of the PMSM
became slightly higher than the PMaSynRM.

Table 4.17: Performance of PMaSynRM versus PMSM at MTPA

PMaSynRM PMSM
Torque [Nm] 156.80 163.03
Torque ripple [%] 7.97 1.19
Efficiency 0.962 0.966

In Fig. 4.11 the torque and power factor of the machines are plotted as function of
the current vector angle. We see that the PMaSynRM maximum torque (MTPA operating
point) occurs at about 47 while the PMSM occurs at about 0 , even though none of the
machines display any particularly good power factor values at this point it is interesting
to note that the PMaSynRM in fact has a better value at this point. Both machines dis-
play a peak PF (MTPkVA operating point) of very close to 1. However, it is worth noting
that none of these operating points pose with a particularly high torque value. However,
it is very noticeable that the PF-curve is much flatter for the PMSM than for the PMa-
SynRM, meaning that for most torque values in these charts the PMSM is better than the
PMaSynRM.

4.4.2 Cost comparison


The cost of the machines was calculated in terms of material costs and production costs
of the stators. The price in pu of the materials are given in Table 4.18. The PMSM and
PMaSynRM utilized different types of copper wire and thereby they obtained different
prices.
The magnet material price of NdFeB were given in the span of 0.17-0.34 pu, and
for ferrite magnets about a factor of 10 lower meaning 0.017-0.034 pu. Setting the magnet

66
4.4. PMSM versus PMaSynRM

160 1
140
120 0.8

Power Factor
Torque [Nm]
100
0.6
80
60 0.4
40
0.2
20
0 0
0 20 40 60 80
[deg]
(a) PMaSynRM

160 1
140
120 0.8

Power Factor
Torque [Nm]

100
0.6
80
60 0.4
40
0.2
20
0 0
0 20 40 60 80
[deg]
(b) PMSM

Fig. 4.11: Torque and power factor as function of current vector angle .

67
4.4. PMSM versus PMaSynRM

Table 4.18: Material prices

Material Price [pu/kg]


Steel, M800 1.52·10 3
Steel, M600 1.56·10 3
Steel, M400 1.58·10 3
Copper, PMSM 11.58·10 3
Copper, PMaSynRM 11.74·10 3

prices to the middle values of both these intervals, we find that the cost is divided up as
seen in Table. 4.19

Table 4.19: Cost breakdown of PMaSynRM and PMSM in middle of magnet cost intervals
[pu]

PMaSynRM PMSM
Magnets Ferrite NdFeB NdFeB
Steel M800 M600 M400 M800 M600 M400 M800/M600
Steel 0.188 0.193 0.195 0.188 0.193 0.195 0.088
Magnets 0.091 0.091 0.091 0.279 0.279 0.279 0.474
Copper 0.281 0.281 0.281 0.281 0.281 0.281 0.155

Misc. mat. 0.177 0.177 0.177 0.177 0.177 0.177 0.210


Assembly 0.263 0.263 0.263 0.263 0.263 0.263 0.118

Total 1.000 1.005 1.007 1.188 1.193 1.195 1.046

where the category ”Misc. mat.” is miscellaneous materials such as tape, zip ties,
stator housing and slot insulation. In the ”Assembly” category costs for operations such
as winding procedures, impregnation, insulation mounting etc are represented. As can be
seen from this table, the ferrite PMaSynRM with M800 steel became only slightly cheaper
than the PMSM whereas the NdFeB PMaSynRM proved to be more expensive than the
PMSM throughout the cost interval and was therefore deemed not cost-effective.
However, given that the prices for the magnets are by no means fixed and tend to
fluctuate this had to be accounted for in order to see if there were any breaking points
in terms of price differences. In Fig. 4.12 the price for both ferrite and NdFeB magnets
are varied for the M800 ferrite PMaSynRM and PMSM. Here we see that the PMSM is
in fact cheaper than the PMaSynRM for lower prices of NdFeB. Additionally, we notice
that the PMaSynRM cost is not particularly sensitive to price fluctuations of the ferrite
magnets as it only varies between 0.97-1.03 pu. Conversely, the PMSM is very dependent
on the magnet price as it varies between 0.89-1.20 pu. Thereby it can be seen that the

68
4.4. PMSM versus PMaSynRM

1.4
PMaSynRM
PMSM
1.2

Total price [pu]


1

0.8

0.6
0.1722 0.2066 0.2410 0.2755 0.3099 0.3443
NdFeB price [pu/kg]

0.01722 0.02066 0.02410 0.02755 0.03099 0.03443


Ferrite price [pu/kg]

Fig. 4.12: PMaSynRM and PMSM cost as function of magnet costs.

breaking point occurs somewhere around an NdFeB price of 0.23 pu, depending on the
cost of the ferrite magnets. If the price of ferrite and NdFeB would follow each-other,
the PMaSynRM proves to be 14% cheaper than the PMSM at the maximum of the price
interval. If ferrite would be at its cheapest and NdFeB at its most expensive, the PMa-
SynRM would be almost 20% cheaper than the PMSM. On the other hand, if it were to
be the other way around, that NdFeB is at its cheapest and ferrite at its most expensive the
PMaSynRM would actually be 16% more expensive than the PMSM.
However, it is interesting to note that these prices are mostly due to the higher
production cost of the stator utilized for the PMaSynRM. As seen in Table 4.19, the cost
of assembly for the PMaSynRM stator is almost double that of the PMSM which is an
important factor. Related to this, in Table. 4.20 a cost breakdown in percent for each
machine is given when the magnets are at their most expensive. The magnets in the PMSM

Table 4.20: Cost breakdown for PMaSynRM and PMSM at peak magnet price [%]

PMaSynRM PMSM
Steel 18.77 8.41
Magnets 9.09 45.35
Copper 28.11 14.80
Misc. mat. 17.68 20.12
Assembly 26.34 11.32

correspond to almost half of the total cost of the machine in contrast to only 9% for the

69
4.4. PMSM versus PMaSynRM

PMaSynRM. For the PMaSynRM, the significant cost could be found mainly in the copper
usage and assembly cost. The copper cost is reasonable since the long end-windings and
longer motor adds to the overall cost.
The assembly is an interesting aspect to analyze, due to the PMSM having a single-
tooth winding with stator segments rather than a stator plate allows for much cheaper
assembly and winding cost.

70
Chapter 5

Conclusions and discussion

5.1 Performance of PMaSynRM


In this thesis it were shown that the performance of the PMaSynRM could compete with
the PMSM in terms of performance. However, the PMaSynRM was about 40% longer in
stack length than the PMSM at approximately equal torque production. For certain appli-
cations, the requirements on volume and weight are not particularly harsh and therefore
the machines are readily interchangeable. It was shown that changing the steel grade did
not have a significant impact on the performance of the PMaSynRM, and the torque pro-
duction actually declined for better grades. It was shown that the maximum current level
for the PMaSynRM operating in cold conditions should be limited to approximately two
times times the rated current in order to avoid demagnetization.
Furthermore, the PMaSynRM has a better power factor than the PMSM at the
MTPA operating point. Assuming that both machines operate at the MTPA operating
point and display equal active power consumption, this means that the kVA-rating of the
power electronic drive can be lower for the PMaSynRM, which is preferable from an
economic point of view.
Additionally, the torque ripple is lower for the PMSM than the PMaSynRM, yield-
ing in another selling point on behalf for the PMSM if the application is sensitive to such
behaviour. However, as has already been stated this is not the case for pumping applica-
tions.
Hence, it can be concluded that in terms of torque production and efficiency the
PMaSynRM can pose as a viable alternative to the PMSM if the demands on machine
length and torque ripple are not especially rigid.
Furthermore, it must be mentioned that the iron loss calculations and hence the
efficiency is the parameter which pose as greatest uncertainty as the stray losses are not
calculated through FEA, but are added as assumptions.

71
5.2. Economic feasibility of PMaSynRM

5.2 Economic feasibility of PMaSynRM


Following the economic analysis conducted in this report we were able to find that the
ferrite-based PMaSynRM was cheaper than the corresponding PMSM with NdFeB mag-
nets, given that the NdFeB price was high enough, whereas it would not be economically
feasible to utilize NdFeB magnets in the PMaSynRM.
It was shown that the PMaSynRM would not be particularly sensitive to fluctuations
in the ferrite magnet price as opposed to the PMSM which could fluctuate 35% in cost.
It was shown that the PMSM could actually become cheaper than the ferrite PMaSynRM
when the cost of NdFeB was low enough. This combined with the better performance and
lower volume of the PMSM means that it was a better choice when NdFeB magnets are
cheap. However, if one seeks to develop a machine which is robust towards magnet price
fluctuations the ferrite PMaSynRM proved to be a very good alternative as the magnet
price didn’t correspond to a significant part of the machine price and varied vary little
when changing the ferrite cost.
The price of the assembly is a significant factor in the cost calculations which are
unaffected by material prices. The cost of assembly shift the overall cost of the machine,
independent of different material prices. If the PMaSynRM stator could be manufactured
in such a way that the price is reduced, the overall savings of the PMaSynRM compared
to the PMSM would be greater.
Had the price of each machine only depended on the steel, copper and magnets the
price outlook would have differed greatly. If that would have been the case, the machines
would have had the same price when the magnet prices were at the lower end of the price
interval. Furthermore, the cost reduction of utilizing the PMaSynRM when both magnets
were at their most expensive would have been 30% along with the best-case scenario with
NdFeB in the expensive end of the interval and ferrite in the cheaper end which would
have yielded the PMaSynRM to be 40% cheaper.
The discrepancy in the assembly cost can be attributed to the difference in assembly
procedure. For instance, the stator laminations for the PMaSynRM are punched in as a
solid part whereas the PMSM is assembled as segments which is possible to do since the
PMSM is a single-tooth FSCW that allows each stator segment to be wound individually
before insertion into the stator house in contrast to how the PMaSynRM is wound which
is more complicated. Additionally, the copper cables differ to a certain degree as to how
they are insulated and impregnated which simplified the process to a great deal for the
PMSM. Overall, it can be stated that PMSM stator is much easier to assemble than the
PMaSynRM stator.
Additionally, the one uncertainty left out of this calculation is the cost of assembly
for the rotor. They were neglected in these calculations since no useful data with regards
to this could be retrieved. It can be assumed that the production cost of these two rotors
are similar, with a possible chance of lowered assembly cost for the PMaSynRM because

72
5.3. Future work

of the lower complexity of handling low-energy magnets as opposed to high-energy mag-


nets.

5.3 Future work


The first thing that should be done in terms of the continuation of this project is to build
the prototype after the design derived in this thesis in order to verify the models created
and find eventual discrepancies.
When it comes to design of the PMaSynRM, this thesis makes no claims of hav-
ing derived an optimal design for the PMaSynRM. As with all design issues, trade-offs
and simplifications were made in conjunction with assumptions in order to stream-line
the design process. These issues are especially true when it comes to the design of the
PMaSynRM-rotor as there are so many factors to take into account while there also are
many restrictions and design objects. This combined with cumbersome FE simulations
means that there also are time restrictions in the design process. Therefore, before decid-
ing on a final design to take to production it is suggested to perform an analysis utilizing
multi-objective optimization algorithms (MOOA) [9], in order to derive an optimized ro-
tor.
Furthermore, while there can be made an argument in terms of simplicity and vol-
ume production to utilize a preexisting stator for the PMaSynRM, there are no guarantees
that it is the best choice to be made in order to make the rotor perform at its best. Ad-
ditionally, the price of assembly of the PMaSynRM stator proved to be significant and
therefore a different stator should be investigated. A possible path forward could be to
design both a rotor and stator in order to further increase the performance and cost effi-
ciency. Related to stator design, the windings of the stator influences the PMaSynRM as
well. The star/delta-winding configuration might prove to give an additional performance
increase to the PMaSynRM at virtually unchanged copper weight [19]. Related to this, an
estimation of difference in rotor cost of assembly should be investigated more thoroughly
since this can prove to give the PMaSynRM an additional cost-reduction compared to the
PMSM.

73
Appendix A

General calculations

A.1 Derivation of expression for IPF


Consider the angular relationships given in Section 2.3, repeated here for simplicity


+ = + 'i (A.1)
2
= + (A.2)

In the second equality in equation (2.9) we have that the IPF can be described as

cos ( + ). Consider the following expression
2
Ld id iq
+
⇡ Lq iq id
tan ( + ) = cot ( )= (A.3)
2 Ld
1
Lq
where the last equality holds because
cot cot + 1
cot ( )= (A.4)
cot cot
Hence, we find that the internal power factor can be described as
0 1
Ld id iq !
+
B Lq iq id C
IPF = cos arctan @ B C (A.5)
Ld A
1
Lq
where the definition of the angles in along with equation (2.5) has been used. Using the
relation that for any argument x we have that
1
cos (arctan x) = p (A.6)
1 + x2

74
A.2. Center of gravity of rotor segments

along with the definition of ⇠ and


iq
= tan (A.7)
id
we can thus write that
2 ⇣ ⌘2 3 1/2
1
6 ⇠ + tan 7
tan
IPF = 6
41 + ⇣ ⌘2 7
5 (A.8)
⇠ 1

and hence, expanding this expression and using that


1 1
1+ = (A.9)
tan2 sin2
1
1 + tan2 = (A.10)
cos2
we have proven that the internal power factor can be written as
⇠ 1
IPF = r (A.11)
1 1
⇠2 +
sin2 cos2

A.2 Center of gravity of rotor segments


The center of gravity of a surface V can be calculated according to
R
rdm
Rc = RV (A.12)
V
dm

r is a point in the volume, dm is the infinitesimal mass of the volume. Since the rotor is
cylindrical and assuming that the mass density is constant (since we neglect the air in the
barriers) we find that we can write the following expression

RR
r2 cos ✓drd✓ num
Rc = RR = (A.13)
rdrd✓ den

where ✓ is the segment angles. Note here that the cos ✓-factor is because of the
symmetry of the segments, meaning that the center of gravity is along the q-axis.

75
A.2. Center of gravity of rotor segments

Hence we find that, by reffering to Fig. 2.13

Z✓b,i ZR1
2 3
num = 2 r2 cos ✓drd✓ = R sin ✓b,i 3
D0,i tan ✓b,i (A.14)
3 1
0 D0,i
cos ✓

Z ZR1
b,i

den = 2 rdrd✓ = R12 ✓i 2


D0,i tan ✓b,i (A.15)
0 D0,i
cos ✓
and thus we can calculate the center of gravity for the i:th segment as

2 R13 sin ✓b,i D0,i


3
tan ✓b,i
Rc,i = 2 2
(A.16)
3 R1 ✓b,i D0,i tan ✓b,i

76
Appendix B

Data sheets

B.1 Steel
In this Appendix, all the steel data-sheets is presented. All data-sheets are retrieved from
[36].

77
B.1.1 M400-50A

Typical data for SURA® M400-50A

I L$`\ K6$`\ 6$b L$`\ L$`\ L$`\ L$`\ L$`\


Vi*%=o Vi*%=o Vi*%=o Vi&%%=o Vi'%%=o Vi)%%=o Vi&%%%=o Vi'*%%=o
%!& %!%' %!%, ('!+ %#%, %#&+ %!)- '#&' -#+)
%!' %!%. %!&- )(!* %#'+ %#+) &!-% ,#). (%#&
%!( %!&. %!(( *%!- %#*) &#(* (!,, &*#( +'#,
%!) %!(& %!*% *,!' %#-- '#'* +!'. '*#, &%.
%!* %!)+ %!+. +(!) &#', (#(( .!(, (.#% &,'
%!+ %!+' %!.& +.!. &#,( )#*- &(!& *+#& '*+
%!, %!-& &!&+ ,,!( '#') +#%( &,!* ,,#& (+,
%!- &!%& &!)+ -+!% '#-% ,#+- ''!, &%(#& *%.
%!. &!') &!-& .,!' (#)) .#*- '-!- &(*#% +-*
&!% &!). '!'( &&(!' )#&* &&#, (*!. &,(#( -..
&!& &!,+ '!,. &(,!- )#.* &)#' ))!' '&-#- &&**
&!' '!%. (!+% &-%!' *#-* &,#% *(!- ','#) &)*(
&!( '!)+ *!%, '+.!* +#-- '%#' +)!. (()#+ &,.(
&!) '!.+ -!-% *&+!- -#&- '(#- ,,!) )%*#+ '&(%
&!* (!*, '&!+ &(%, .#-' '-#( .&!, )--#)
&!+ )!(- *,!' (&-%
&!, *!%' &'- +(+&
&!- *!), ')( &%-.%

AdhhVi&#*I!*%=o!L$`\ (!*,
AdhhVi&#%I!*%=o!L$`\ &!).
6c^hdigdend[adhh! -

BV\cZi^XedaVg^oVi^dcVi*%=o
=2'*%%6$b!I &!*.
=2*%%%6$b!I &!+-
=2&%%%%6$b!I &!,.

8dZgX^k^in98!6$b *%
GZaVi^kZeZgbZVW^a^inVi&#*I &%*%
GZh^hi^k^in!¥7Xb )'

N^ZaYhigZc\i]!C$bb ('*
IZch^aZhigZc\i]!C$bb )+*
Ndjc\»hbdYjajh!G9!C$bb '%%%%%
Ndjc\»hbdYjajh!I9!C$bb '&%%%%
=VgYcZhh=K*KEC &+*

RD represents the rolling direction


TD represents the transverse direction
Values for yield strength (0.2 % proof strength)
and tensile strength are given for the rolling direction
Values for the transverse direction are approximately 5% higher
Oct 2009
B.1.2 M600-50A

Typical data for SURA® M600-50A

I L$`\ K6$`\ 6$b


Vi*%=o Vi*%=o Vi*%=o
%!& %!%) %!&( +*!+
%!' %!&+ %!(* -(!-
%!( %!(( %!*. .)!&
%!) %!*) %!-, &%(
%!* %!,- &!&- &&%
%!+ &!%) &!*& &&-
%!, &!(' &!-- &',
%!- &!+( '!'- &(+
%!. &!.+ '!,( &),
&!% '!() (!'* &*.
&!& '!,+ (!-+ &,,
&!' (!'( )!+' '%*
&!( (!,- *!+- '**
&!) )!)( ,!** (,%
&!* *!&, &'!- ,&-
&!+ *!.) (&!, &-)%
&!, +!+% -'!+ )(,%
&!- ,!%+ &,+ -((%

AdhhVi&#*I!*%=o!L$`\ *!&,
AdhhVi&#%I!*%=o!L$`\ '!()
6c^hdigdend[adhh! +

BV\cZi^XedaVg^oVi^dcVi*%=o
=2'*%%6$b!I &!+(
=2*%%%6$b!I &!,'
=2&%%%%6$b!I &!-(

8dZgX^k^in98!6$b -*
GZaVi^kZeZgbZVW^a^inVi&#*I &++%
GZh^hi^k^in!¥7Xb (%

N^ZaYhigZc\i]!C$bb '-*
IZch^aZhigZc\i]!C$bb )%*
Ndjc\»hbdYjajh!G9!C$bb '&%%%%
Ndjc\»hbdYjajh!I9!C$bb ''%%%%
=VgYcZhh=K*KEC &'*

RD represents the rolling direction


TD represents the transverse direction
Values for yield strength (0.2 % proof strength)
and tensile strength are given for the rolling direction
Values for the transverse direction are approximately 5% higher
June 2008
B.1.3 M800-50A

Typical data for SURA® M800-50A

I L$`\ K6$`\ 6$b


Vi*%=o Vi*%=o Vi*%=o
%!& %!%* %!&, -)!*
%!' %!&- %!)) &%,
%!( %!)( %!,+ &'&
%!) %!,% &!&' &((
%!* &!%& &!*' &)*
%!+ &!(* &!., &*+
%!, &!,' '!)+ &+-
%!- '!&( (!%% &-%
%!. '!*+ (!+% &.)
&!% (!%* )!', '%.
&!& (!*. *!%) ''-
&!' )!'% *!.+ '*)
&!( )!.& +!.* (%)
&!) *!,% -!). )%'
&!* +!+% &&!. ++%
&!+ ,!*) '*!' &)-%
&!, -!(% +,!) (,&%
&!- -!-( &*& ,(%%

AdhhVi&#*I!*%=o!L$`\ +!+%
AdhhVi&#%I!*%=o!L$`\ (!%*
6c^hdigdend[adhh! *

BV\cZi^XedaVg^oVi^dcVi*%=o
=2'*%%6$b!I &!+*
=2*%%%6$b!I &!,)
=2&%%%%6$b!I &!-*

8dZgX^k^in98!6$b &%%
GZaVi^kZeZgbZVW^a^inVi&#*I &-&%
GZh^hi^k^in!¥7Xb '(

N^ZaYhigZc\i]!C$bb (%%
IZch^aZhigZc\i]!C$bb )&*
Ndjc\»hbdYjajh!G9!C$bb '&%%%%
Ndjc\»hbdYjajh!I9!C$bb ''%%%%
=VgYcZhh=K*KEC &(%

RD represents the rolling direction


TD represents the transverse direction
Values for yield strength (0.2 % proof strength)
and tensile strength are given for the rolling direction
Values for the transverse direction are approximately 5% higher
June 2008
B.2. Magnets

B.2 Magnets
In this section, all the data-sheets for the magnets are presented. The ferrite data-sheet is
retrieved via contact with the supplier [37]. The NdFeB data-sheet is retrieved from [38].

81
B.2. Magnets

B.2.1 Ferrite

82
B.2.2 NdFeB - N33EH

N33EH
Sintered Neodymium-Iron-Boron Magnets

These are also referred to as "Neo" or NdFeB magnets. They offer a


combination of high magnetic output at moderate cost. Please contact Arnold Characteristic Units C // C
for additional grade information and recommendations for protective coating.
Reversible Temperature Coefficients (1)
Assemblies using these magnets can also be provided.

Thermal Properties
of Induction, α(Br) %/ºC -0.120
of Coercivity, α(Hcj) %/ºC -0.420
Coefficient of Thermal Expansion (2)
-6
Characteristic Units min. nominal max. ∆L/L per ºCx10 7.5 -0.1
Gauss 11,300 11,650 12,000 Thermal Conductivity W / (m • K) 7.6
Br, Residual Induction (3)
mT 1130 1165 1200 Specific Heat J / (kg • K) 460
Magnetic Properties

Oersteds 10,300 10,900 11,500 Curie Temperature, Tc ºC 310


HcB, Coercivity
kA/m 820 867 915 psi 41,300
Flexural Strength

Properties
Oersteds 30,000 MPa 285
HcJ, Intrinsic Coercivity

Other
kA/m 2,388 Density g/cm3 7.5
MGOe 31 34 36 Hardness, Vickers Hv 620
BHmax, Maximum Energy Product
kJ/m3 247 267 287 Electrical Resistivity, • cm 180
Notes: (1) Coefficients measured between 20 and 200 ºC
(2) Between 20 and 200 ºC (3) Between 20 and 140 ºC

kG Tesla
Pc = B H
0.5
Material: N33EH 0.75 1 1.5 2 3 5
15

14 1.4

13

12 1.2
-40°C
11
20°C

Flux Density, B
10 1.0

0.3 9

8 0.8

7
60°C

Polarization, J
6 0.6

80°C 5

100°C 4 0.4

120°C
0.1 3

150°C
2 0.2
180°C
200°C
220°C
1

0 0
30
kOe 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0

kA/m 2230 2070 1910 1750 1590 1430 1275 1115 955 795 640 475 320 160 0
Demagnetizing Field, H
1 kA/m = 12.566 Oe 1 kOe = 79.577 kA/m

Notes The material data and demagnetization curves shown above represent typical properties that may vary due to product shape and size.
Magnets can be supplied thermally stabilized or magnetically calibrated to customer specifications.
Additional grades are available. Please contact the factory for information.

© Arnold Magnetic Technologies Corp.


770 Linden Avenue, Rochester, NY 14625 E-mail: info@arnoldmagnetics.com
Rev. 151021a Ph: (+1) 585-385-9010 www.arnoldmagnetics.com
Appendix C

Results

C.1 Demagnetization

84
C.1. Demagnetization

(a) L1 ,L2 and L3 at In \70 .

(b) R1 ,R2 and R3 at 1.5In \110

Fig. C.1: Demagnetization of ferrite magnets for In and 1.5In .

85
C.1. Demagnetization

(a) R1 ,R2 and R3 at 2In \100 .

(b) R1 ,R2 and R3 at 2.5In \100

Fig. C.2: Demagnetization of ferrite magnets for 2In and 2.5In .

86
References

[1] R. Saidur, “A review on electrical motors energy use and energy savings,” Renewable
and Sustainable Energy Reviews, vol. 14, no. 3, pp. 877 – 898, 2010. [Online].
Available: http://www.sciencedirect.com/science/article/pii/S1364032109002494

[2] P. Waide and C. U. Brunner, “Energy-efficiency policy opportunities for electric


motor-driven systems,” may 2011. [Online]. Available: https://doi.org/10.1787/
5kgg52gb9gjd-en

[3] Energimyndigheten, June 2015, [Accessed 2018-03-13]. [Online]. Avail-


able: http://www.energimyndigheten.se/energieffektivisering/produkter-med-krav/
produktgrupper/mapp-for-produkter/elmotorer/

[4] K. O. Armstrong, S. Das, and J. Cresko, “Wide bandgap semiconductor


opportunities in power electronics,” in 2016 IEEE 4th Workshop on Wide Bandgap
Power Devices and Applications (WiPDA). IEEE, nov 2016. [Online]. Available:
https://doi.org/10.1109/wipda.2016.7799949

[5] J.-R. Riba, C. López-Torres, L. Romeral, and A. Garcia, “Rare-earth-free


propulsion motors for electric vehicles: A technology review,” Renewable and
Sustainable Energy Reviews, vol. 57, pp. 367–379, May 2016. [Online]. Available:
https://doi.org/10.1016/j.rser.2015.12.121

[6] J. Rowlatt, “Rare earths: Neither rare, nor earths,” BBC News, Mar 2014, [Accessed
2018-03-13]. [Online]. Available: http://www.bbc.com/news/magazine-26687605

[7] C. Bontron, “Rare-earth mining in china comes at a heavy cost for local villages,”
The Guardian, Aug 2012, [Accessed 2018-03-13]. [Online]. Available: https://www.
theguardian.com/environment/2012/aug/07/china-rare-earth-village-pollution

[8] A. T. de Almeida, F. J. T. E. Ferreira, and G. Baoming, “Beyond induction


motors—technology trends to move up efficiency,” IEEE Transactions on Industry
Applications, vol. 50, no. 3, pp. 2103–2114, may 2014. [Online]. Available:
https://doi.org/10.1109/tia.2013.2288425

87
References

[9] G. Pellegrino, T. M. Jahns, N. Bianchi, W. Soong, and F. Cupertino,


The Rediscovery of Synchronous Reluctance and Ferrite Permanent Magnet
Motors. Springer International Publishing, 2016. [Online]. Available: https:
//doi.org/10.1007/978-3-319-32202-5

[10] A. O. Dulanto”, “”design of a synchronous reluctance motor assisted with perma-


nent magnets for pump applications”,” ”Master thesis”, ”KTH Royal Institute of
Technology”, ”Stockholm”, ”2015”.

[11] R. Moghaddam, “Synchronous reluctance machine (synrm) in variable speed drives


(vsd) applications,” Ph.D. dissertation, KTH Royal Institute of Technology, Stock-
holm, 2011, doctoral thesis.

[12] N. Bianchi, E. Fornasiero, and W. Soong, “Optimal selection of PM flux


linkage in a PM assisted synchronous reluctance machine,” in 2014 International
Conference on Electrical Machines (ICEM). IEEE, sep 2014. [Online]. Available:
https://doi.org/10.1109/icelmach.2014.6960356

[13] J. K. Kostko, “Polyphase reaction synchronous motors,” Journal of the American


Institute of Electrical Engineers, vol. 42, no. 11, pp. 1162–1168, nov 1923.
[Online]. Available: https://doi.org/10.1109/joaiee.1923.6591529

[14] T. Fukami, M. Momiyama, K. Shima, R. Hanaoka, and S. Takata, “Steady-state


analysis of a dual-winding reluctance generator with a multiple-barrier rotor,”
IEEE Transactions on Energy Conversion, vol. 23, no. 2, pp. 492–498, jun 2008.
[Online]. Available: https://doi.org/10.1109/tec.2008.918656

[15] R. Rajabi Moghaddam, “Synchronous reluctance machine (synrm) design,” KTH,


Skolan for elektro- och systemteknik (EES), Elektrisk energiomvandling, 2007.

[16] R. Moghaddam, F. Magnussen, C. Sadarangani, and H. Lendenmann, “New theoret-


ical approach to the synchronous reluctance machine behavior and performance,” in
Electrical Machines, 2008. ICEM 2008. 18th International Conference on. IEEE,
September 2008, pp. 1–6.

[17] M. Kamper, “Effect of rotor dimensions and cross magnetisation on ld and lq


inductances of reluctance synchronous machine with cageless flux barrier rotor,”
IEE Proceedings - Electric Power Applications, vol. 141, no. 4, p. 213, 1994.
[Online]. Available: https://doi.org/10.1049/ip-epa:19941261

[18] N. Bianchi, E. Fornasiero, and W. Soong, “Selection of pm flux linkage for max-
imum low-speed torque rating in a pm-assisted synchronous reluctance machine,”
IEEE Transactions on Industry Applications, vol. 51, no. 5, pp. 3600–3608, Sept
2015.

88
References

[19] M. N. F. Ibrahim, “Design aspects of high performance synchronous reluctance ma-


chines with and without permanent magnets,” Ph.D. dissertation, Ghent University,
2017.

[20] T. Matsuo and T. Lipo, “Rotor design optimization of synchronous reluctance


machine,” IEEE Transactions on Energy Conversion, vol. 9, no. 2, pp. 359–365, jun
1994. [Online]. Available: https://doi.org/10.1109/60.300136

[21] R.-R. Moghaddam and F. Gyllensten, “Novel high-performance SynRM design


method: An easy approach for a complicated rotor topology,” IEEE Transactions on
Industrial Electronics, vol. 61, no. 9, pp. 5058–5065, sep 2014. [Online]. Available:
https://doi.org/10.1109/tie.2013.2271601

[22] M. Palmieri, M. Perta, F. Cupertino, and G. Pellegrino, “Effect of the


numbers of slots and barriers on the optimal design of synchronous reluctance
machines,” in 2014 International Conference on Optimization of Electrical
and Electronic Equipment (OPTIM). IEEE, may 2014. [Online]. Available:
https://doi.org/10.1109/optim.2014.6850947

[23] A. Vagati, M. Pastorelli, G. Francheschini, and S. Petrache, “Design of low-


torque-ripple synchronous reluctance motors,” IEEE Transactions on Industry
Applications, vol. 34, no. 4, pp. 758–765, 1998. [Online]. Available: https:
//doi.org/10.1109/28.703969

[24] M. Sanada, K. Hiramoto, S. Morimoto, and Y. Takeda, “Torque ripple improvement


for synchronous reluctance motor using an asymmetric flux barrier arrangement,”
IEEE Transactions on Industry Applications, vol. 40, no. 4, pp. 1076–1082, jul
2004. [Online]. Available: https://doi.org/10.1109/tia.2004.830745

[25] M. Xu, G. Liu, W. Zhao, and N. Aamir, “Minimization of torque ripple


in ferrite-assisted synchronous reluctance motors by using asymmetric stator,”
AIP Advances, vol. 8, no. 5, p. 056606, may 2018. [Online]. Available:
https://doi.org/10.1063/1.5006114

[26] C. Babetto, G. Bacco, and N. Bianchi, “Analytical approach to determine the power
limit of high-speed synchronous reluctance machines,” in 2017 IEEE International
Electric Machines and Drives Conference (IEMDC). IEEE, may 2017. [Online].
Available: https://doi.org/10.1109/iemdc.2017.8002316

[27] A. Vagati, G. Franceschini, I. Marongiu, and G. Troglia, “Design criteria of


high performance synchronous reluctance motors,” in Conference Record of the
1992 IEEE Industry Applications Society Annual Meeting. IEEE, 1992. [Online].
Available: https://doi.org/10.1109/ias.1992.244463

89
References

[28] K. Khan, O. Wallmark, and M. Leksell, “Design aspects on magnet placement


in permanent-magnet assisted synchronous reluctance machines,” in 5th IET
International Conference on Power Electronics, Machines and Drives (PEMD
2010). Institution of Engineering and Technology, 2010. [Online]. Available:
https://doi.org/10.1049/cp.2010.0203

[29] K. S. Khan, “Design of a permanent-magnet assisted synchronous reluctance ma-


chine for a plug-in hybrid electric vehicle,” KTH, Skolan for elektro- och sys-
temteknik (EES), Elektrisk energiomvandling, Stockholm, 2011.

[30] O. Wallmark, AC Machine Analysis. KTH Royal Institute of Technology, Depart-


ment of Electric Power and Energy Systems, 2017.

[31] S. Sjokvist, “Demagnetization studies on permanent magnets: Comparing fem sim-


ulations with experiments,” Uppsala universitet, Teknisk-naturvetenskapliga veten-
skapsomradet, Tekniska sektionen, Institutionen for teknikvetenskaper, Elektricitet-
slara, Uppsala, 2014.

[32] D. K. Cheng, Field and Wave Electromagnetics (Addison-Wesley series in electrical


engineering). Addison-Wesley Educational Publishers Inc, 1983.

[33] P. Zhou, D. Lin, Y. Xiao, N. Lambert, and M. A. Rahman, “Temperature-dependent


demagnetization model of permanent magnets for finite element analysis,” IEEE
Transactions on Magnetics, vol. 48, no. 2, pp. 1031–1034, feb 2012. [Online].
Available: https://doi.org/10.1109/tmag.2011.2172395

[34] A. Krings and J. Soulard, “Overview and comparison of iron loss models for electri-
cal machines,” Journal of Electrical Engineering, vol. 10, no. 3, pp. 162–169, 2010.

[35] T. Chevalier, A. Kedous-Lebouc, B. Cornut, and C. Cester, “A new dynamic


hysteresis model for electrical steel sheet,” Physica B: Condensed Matter, vol.
275, no. 1-3, pp. 197–201, jan 2000. [Online]. Available: https://doi.org/10.1016/
s0921-4526(99)00768-1

[36] “Download catalogues, datasheets and iso certificates,” Accessed 2018-06-19.


[Online]. Available: https://cogent-power.com/downloads

[37] “NORDMAG.” [Online]. Available: http://www.nordmag.fi/#contact

[38] “Neodymium magnets (NdFeB),” Accessed 2018-06-19. [Online]. Available:


http://www.arnoldmagnetics.com/products/neodymium-iron-boron-magnets/

90

You might also like