Marko Rilla

DESIGN OF SALIENT POLE PM SYNCHRONOUS

MACHINES FOR A VEHICLE TRACTION
APPLICATION ANALYSIS AND IMPLEMENTATION
Acta Universitatis
Lappeenrantaensis 497
Thesis for the degree of Doctor of Science (Technology) to be presented with
due permission for public examination and criticism in the Auditorium 1383
at Lappeenranta University of Technology, Lappeenranta, Finland on
the 8th of December, 2012, at noon.
Marko Rilla
DESIGN OF SALIENT POLE PM SYNCHRONOUS
MACHINES FOR A VEHICLE TRACTION
APPLICATION ANALYSIS AND IMPLEMENTATION
Acta Universitatis
Lappeenrantaensis 497
Thesis for the degree of Doctor of Science (Technology) to be presented with
due permission for public examination and criticism in the Auditorium 1383
at Lappeenranta University of Technology, Lappeenranta, Finland on
the 8th of December, 2012, at noon.


Super%isor .rofessor /uha .yrh0nen
Department of 1lectrical 1ngineering
2nstitute of 1nergy Technology
#appeenranta $ni%ersity of Technology
#appeenranta& 'inland


Re%iewers Associate .rofessor .a%ol Rafa3dus
$ni%ersity of 4ilina
Slo%ak Republic

D*Sc*& Ad3unct .rofessor /anne 566n6nen
The 7erggren 8roup
9elsinki& 'inland


:pponent Associate .rofessor .a%ol Rafa3dus
$ni%ersity of 4ilina
Slo%ak Republic

D*Sc*& Ad3unct .rofessor /anne 566n6nen
The 7erggren 8roup
9elsinki& 'inland












2S7; ,-"<,=(<(>=<!!><,
2S7; ,-"<,=(<(>=<!!-<> (.D')
2SS; +=><++,

#appeenrannan teknillinen yliopisto
?liopistopaino ()(


Super%isor .rofessor /uha .yrh0nen
Department of 1lectrical 1ngineering
2nstitute of 1nergy Technology
#appeenranta $ni%ersity of Technology
#appeenranta& 'inland


Re%iewers Associate .rofessor .a%ol Rafa3dus
$ni%ersity of 4ilina
Slo%ak Republic

D*Sc*& Ad3unct .rofessor /anne 566n6nen
The 7erggren 8roup
9elsinki& 'inland


:pponent Associate .rofessor .a%ol Rafa3dus
$ni%ersity of 4ilina
Slo%ak Republic

D*Sc*& Ad3unct .rofessor /anne 566n6nen
The 7erggren 8roup
9elsinki& 'inland












2S7; ,-"<,=(<(>=<!!><,
2S7; ,-"<,=(<(>=<!!-<> (.D')
2SS; +=><++,

#appeenrannan teknillinen yliopisto
?liopistopaino ()(


Abstract
Marko Rilla
Design of salient pole .M synchronous machines for a %ehicle traction application @ Analysis and
2mplementation
#appeenranta ()(
=- p*
Acta $ni%ersitatis #appeenrantaensis +,-
Diss* #appeenranta $ni%ersity of Technology
2S7; ,-"<,=(<(>=<!!><,& 2S7; ,-"<,=(<(>=<!!-<> (.D')& 2SS; +=><++,

This doctoral thesis presents a study on the de%elopment of a liAuid<cooled frame salient pole
permanent<magnet<exited traction machine for a four<wheel<dri%en electric car* The emphasis of
the thesis is put on a radial flux machine design in order to achie%e a light<weight machine
structure for traction applications* The design features combine electromagnetic and thermal
design methods& because traction machine operation does not ha%e a strict operating point*
Arbitrary load cycles and the flexible supply reAuire special attention in the design process*
2t is shown that accurate modelling of the machine magnetic state is essential for high<performance
operation* The saturation effect related to the cross<saturation has to be taken carefully into
account in order to achie%e the desired operation*
Two prototype machines ha%e been designed and built for testingB one totally enclosed machine
with a special magnet module pole arrangement and another through<%entilated machine with a
more traditional embedded magnet structure* 7oth structures are built with magnetically salient
structures in order to increase the torAue production capability with the reluctance torAue
component* 7oth machine structures show potential for traction usage* 9owe%er& the traditional
embedded magnet design turns out to be mechanically the more secure one of these two machine
options*



CeywordsB .ermanent magnet synchronous motor& radial flux& .MSM& salient pole& traction&
thermal analysis*
$DD >(*!!E*!!!B>(,*)("



Abstract
Marko Rilla
Design of salient pole .M synchronous machines for a %ehicle traction application @ Analysis and
2mplementation
#appeenranta ()(
=- p*
Acta $ni%ersitatis #appeenrantaensis +,-
Diss* #appeenranta $ni%ersity of Technology
2S7; ,-"<,=(<(>=<!!><,& 2S7; ,-"<,=(<(>=<!!-<> (.D')& 2SS; +=><++,

This doctoral thesis presents a study on the de%elopment of a liAuid<cooled frame salient pole
permanent<magnet<exited traction machine for a four<wheel<dri%en electric car* The emphasis of
the thesis is put on a radial flux machine design in order to achie%e a light<weight machine
structure for traction applications* The design features combine electromagnetic and thermal
design methods& because traction machine operation does not ha%e a strict operating point*
Arbitrary load cycles and the flexible supply reAuire special attention in the design process*
2t is shown that accurate modelling of the machine magnetic state is essential for high<performance
operation* The saturation effect related to the cross<saturation has to be taken carefully into
account in order to achie%e the desired operation*
Two prototype machines ha%e been designed and built for testingB one totally enclosed machine
with a special magnet module pole arrangement and another through<%entilated machine with a
more traditional embedded magnet structure* 7oth structures are built with magnetically salient
structures in order to increase the torAue production capability with the reluctance torAue
component* 7oth machine structures show potential for traction usage* 9owe%er& the traditional
embedded magnet design turns out to be mechanically the more secure one of these two machine
options*



CeywordsB .ermanent magnet synchronous motor& radial flux& .MSM& salient pole& traction&
thermal analysis*
$DD >(*!!E*!!!B>(,*)("











Acknowledgements
The preparation for the doctoral dissertation started in August ())> right after the Masters degree
in 1lectrical engineering* The actual work for the dissertation was carried out during ()), @ ()(*
2 would like to thank .rofessor /uha .yrh0nen& and Sami Ruotsalainen of Metropolia $ni%ersity of
applied sciences for the possibility to work with such a leading edge topic in electrical
engineering*
2 would like to express my gratitude to Doctors /anne ;erg and Markku ;iemel6* ;ergs ad%ices in
scientific methods during the years and his contribution in the de%elopment of the prototype
machines is highly appreciated* The long<term guidance of ;erg and ;iemel6& especially during
this last candidate year& has made the completion of the thesis possible*
2 also need to express my gratitude to Associate .rofessor .a%ol Rafad3us and D*Sc Ad3unct
professor /anne 566n6nen for the thorough pre<examination work and the %aluable comments
which ha%e helped me to impro%e the scientific le%el of the dissertation*
Special thanks goes also to /ouni Ryh6nen and Martti #indh for the technical implementation of
the test setups*
2Fm also in gratitude to 9anna ;imel6 for the proofreading of the thesis and the impro%ement of
the 1nglish language to make the dissertation fluently readable*
2 also appreciate the financial support of South@Carelian 'und of 'innish Dultural 'oundation&
#auri and #ah3a 9otinen 'und& 'innish 'oundation for Technology .romotion and Research
'oundation of #appeenranta $ni%ersity of Technology during the candidate years*
2 would also like to thank all the friends and colleagues that 2 ha%e met along the years for the
prolonged coffee breaks& late nights and early mornings*
And to my 'amily& Thank ?ou for the support and understanding during the years*

#appeenranta ;o%ember (=& ()(
Marko Rilla






Acknowledgements
The preparation for the doctoral dissertation started in August ())> right after the Masters degree
in 1lectrical engineering* The actual work for the dissertation was carried out during ()), @ ()(*
2 would like to thank .rofessor /uha .yrh0nen& and Sami Ruotsalainen of Metropolia $ni%ersity of
applied sciences for the possibility to work with such a leading edge topic in electrical
engineering*
2 would like to express my gratitude to Doctors /anne ;erg and Markku ;iemel6* ;ergs ad%ices in
scientific methods during the years and his contribution in the de%elopment of the prototype
machines is highly appreciated* The long<term guidance of ;erg and ;iemel6& especially during
this last candidate year& has made the completion of the thesis possible*
2 also need to express my gratitude to Associate .rofessor .a%ol Rafad3us and D*Sc Ad3unct
professor /anne 566n6nen for the thorough pre<examination work and the %aluable comments
which ha%e helped me to impro%e the scientific le%el of the dissertation*
Special thanks goes also to /ouni Ryh6nen and Martti #indh for the technical implementation of
the test setups*
2Fm also in gratitude to 9anna ;imel6 for the proofreading of the thesis and the impro%ement of
the 1nglish language to make the dissertation fluently readable*
2 also appreciate the financial support of South@Carelian 'und of 'innish Dultural 'oundation&
#auri and #ah3a 9otinen 'und& 'innish 'oundation for Technology .romotion and Research
'oundation of #appeenranta $ni%ersity of Technology during the candidate years*
2 would also like to thank all the friends and colleagues that 2 ha%e met along the years for the
prolonged coffee breaks& late nights and early mornings*
And to my 'amily& Thank ?ou for the support and understanding during the years*

#appeenranta ;o%ember (=& ()(
Marko Rilla



















Dontents
Abstract
Acknowledgements
Dontents
;omenclature
2ntroduction ***************************************************************************************************************************** !
* :b3ecti%es of the study ************************************************************************************************** +
*( Traction machine for %ehicle propulsion *********************************************************************** >
*(* 'ield of traction applications ************************************************************************************* ()
*(*( 1lectric traction systems ******************************************************************************************** ((
*(*! 9istory of electric traction ***************************************************************************************** (=
*! Synchronous reluctance<torAue<assisted .M machine *************************************************** (>
*!* 'undamentals of electrical machine torAue production ********************************************* (-
*!*( .ull<out torAue maximiGation ************************************************************************************ !
*!*! 1ffect of saliency and leakage on the dri%e performance ****************************************** !!
*!*+ .ermanent magnet materials************************************************************************************** !"
*!*= TorAue Auality *********************************************************************************************************** +(
*+ :utline of the work ******************************************************************************************************** +(
*= Scientific contribution of the work ******************************************************************************** +!
*> #ist of publications ******************************************************************************************************** ++
*>* #ist of related publications **************************************************************************************** ++
*>*( #ist of supporting publications ********************************************************************************** ++
( Cey design areas of a traction machine *********************************************************************************** +-
(* 1lectromagnetic design ************************************************************************************************** +-
(** 7asic design rules ****************************************************************************************************** +-
(**( Hindings ******************************************************************************************************************* =
(**! Magnetic circuit ********************************************************************************************************* ==
(*( #oss e%aluation ************************************************************************************************************** ="
(*(* /oule losses **************************************************************************************************************** ="
(*(*( 2ron losses ****************************************************************************************************************** >
(*(*! Dontact friction losses in bearings ***************************************************************************** >+



Dontents
Abstract
Acknowledgements
Dontents
;omenclature
2ntroduction ***************************************************************************************************************************** !
* :b3ecti%es of the study ************************************************************************************************** +
*( Traction machine for %ehicle propulsion *********************************************************************** >
*(* 'ield of traction applications ************************************************************************************* ()
*(*( 1lectric traction systems ******************************************************************************************** ((
*(*! 9istory of electric traction ***************************************************************************************** (=
*! Synchronous reluctance<torAue<assisted .M machine *************************************************** (>
*!* 'undamentals of electrical machine torAue production ********************************************* (-
*!*( .ull<out torAue maximiGation ************************************************************************************ !
*!*! 1ffect of saliency and leakage on the dri%e performance ****************************************** !!
*!*+ .ermanent magnet materials************************************************************************************** !"
*!*= TorAue Auality *********************************************************************************************************** +(
*+ :utline of the work ******************************************************************************************************** +(
*= Scientific contribution of the work ******************************************************************************** +!
*> #ist of publications ******************************************************************************************************** ++
*>* #ist of related publications **************************************************************************************** ++
*>*( #ist of supporting publications ********************************************************************************** ++
( Cey design areas of a traction machine *********************************************************************************** +-
(* 1lectromagnetic design ************************************************************************************************** +-
(** 7asic design rules ****************************************************************************************************** +-
(**( Hindings ******************************************************************************************************************* =
(**! Magnetic circuit ********************************************************************************************************* ==
(*( #oss e%aluation ************************************************************************************************************** ="
(*(* /oule losses **************************************************************************************************************** ="
(*(*( 2ron losses ****************************************************************************************************************** >
(*(*! Dontact friction losses in bearings ***************************************************************************** >+


(*(*+ Additional losses ******************************************************************************************************* >=
(*! 9eat transfer ****************************************************************************************************************** >>
(*!* 'undamentals of heat transfer *********************************************************************************** >>
(*!*( Donduction **************************************************************************************************************** >-
(*!*! Radiation ******************************************************************************************************************* >"
(*!*+ Don%ection **************************************************************************************************************** >,
(*!*= #umped<parameter model ***************************************************************************************** -
(*+ Summary of the key design areas ********************************************************************************** -!
! 2mplementation of the design methods************************************************************************************ -=
!* 7ackground of the machine de%elopment ********************************************************************** -=
!** Testing and mechanical aspects ********************************************************************************* "=
!**( Second generation ***************************************************************************************************** ">
!*( Steady<state analysis ****************************************************************************************************** "-
!*(* 7ack electromoti%e force ******************************************************************************************* ""
!*(*( Synchronous inductances ****************************************************************************************** ,
!*(*! TorAue production ***************************************************************************************************** ,>
!*(*+ .erformance o%er speed range ********************************************************************************* )=
!*! Time stepping analysis ************************************************************************************************ ),
!*!* 5oltage<fed dynamic model ************************************************************************************ )
!*!*( Short<circuit endurance *******************************************************************************************
!*+ Thermal network ********************************************************************************************************** !
!*+* 'rame region and end<winding space ********************************************************************** +
!*+*( Teeth and slots ********************************************************************************************************* >
!*+*! Air gap region ********************************************************************************************************** "
!*+*+ Rotor region************************************************************************************************************* ,
!*= Results of the thermal modelling ********************************************************************************* ()
!*=* Steady<state analysis************************************************************************************************ ()
!*=*( Transient analysis **************************************************************************************************** (+
!*=*! Sensiti%ity of the de%eloped model ************************************************************************* (=
!*> Donclusions on the machine design ***************************************************************************** (>
+ Analysis of the experimental studies ************************************************************************************* (,
+* Test set<up******************************************************************************************************************** (,
+** #oad analysis *********************************************************************************************************** !)
+**( 1fficiency and loss analysis************************************************************************************* !!


(*(*+ Additional losses ******************************************************************************************************* >=
(*! 9eat transfer ****************************************************************************************************************** >>
(*!* 'undamentals of heat transfer *********************************************************************************** >>
(*!*( Donduction **************************************************************************************************************** >-
(*!*! Radiation ******************************************************************************************************************* >"
(*!*+ Don%ection **************************************************************************************************************** >,
(*!*= #umped<parameter model ***************************************************************************************** -
(*+ Summary of the key design areas ********************************************************************************** -!
! 2mplementation of the design methods************************************************************************************ -=
!* 7ackground of the machine de%elopment ********************************************************************** -=
!** Testing and mechanical aspects ********************************************************************************* "=
!**( Second generation ***************************************************************************************************** ">
!*( Steady<state analysis ****************************************************************************************************** "-
!*(* 7ack electromoti%e force ******************************************************************************************* ""
!*(*( Synchronous inductances ****************************************************************************************** ,
!*(*! TorAue production ***************************************************************************************************** ,>
!*(*+ .erformance o%er speed range ********************************************************************************* )=
!*! Time stepping analysis ************************************************************************************************ ),
!*!* 5oltage<fed dynamic model ************************************************************************************ )
!*!*( Short<circuit endurance *******************************************************************************************
!*+ Thermal network ********************************************************************************************************** !
!*+* 'rame region and end<winding space ********************************************************************** +
!*+*( Teeth and slots ********************************************************************************************************* >
!*+*! Air gap region ********************************************************************************************************** "
!*+*+ Rotor region************************************************************************************************************* ,
!*= Results of the thermal modelling ********************************************************************************* ()
!*=* Steady<state analysis************************************************************************************************ ()
!*=*( Transient analysis **************************************************************************************************** (+
!*=*! Sensiti%ity of the de%eloped model ************************************************************************* (=
!*> Donclusions on the machine design ***************************************************************************** (>
+ Analysis of the experimental studies ************************************************************************************* (,
+* Test set<up******************************************************************************************************************** (,
+** #oad analysis *********************************************************************************************************** !)
+**( 1fficiency and loss analysis************************************************************************************* !!


+*( Thermal characteristics ************************************************************************************************ !>
4..! "teady#state temperatures ************************************************************************************ !>
+*(*( Domparison with the dynamic model ********************************************************************** !,
+*(*! 1stimation of the magnet temperature ********************************************************************* +
+*(*+ Dooling efficiency *************************************************************************************************** +!
+*! Donclusions ***************************************************************************************************************** +=
= Donclusions *************************************************************************************************************************** +-
=* Dontributions of the doctoral thesis ***************************************************************************** +-
=*( .rospects of the future work **************************************************************************************** +"
A..1;D2I 2B Test setup
A..1;D2I 22B Thermal network
A..1;D2I 222B Test machine dimensions






+*( Thermal characteristics ************************************************************************************************ !>
4..! "teady#state temperatures ************************************************************************************ !>
+*(*( Domparison with the dynamic model ********************************************************************** !,
+*(*! 1stimation of the magnet temperature ********************************************************************* +
+*(*+ Dooling efficiency *************************************************************************************************** +!
+*! Donclusions ***************************************************************************************************************** +=
= Donclusions *************************************************************************************************************************** +-
=* Dontributions of the doctoral thesis ***************************************************************************** +-
=*( .rospects of the future work **************************************************************************************** +"
A..1;D2I 2B Test setup
A..1;D2I 22B Thermal network
A..1;D2I 222B Test machine dimensions






$omenclature
%reek variables

coefficient
absorpti%ity
air gap

r
relati%e permitti%ity
emissi%ity
efficiency
coefficient

r
relati%e permeability
number of harmonic order
electrical resisti%ity JKmL
electrical conducti%ity JSL
pitch factor
angular %elocity JradEsL
mechanical angular %elocity JradEsL
magnetic flux J5sL
current linkage JAL
magnetic flux linkage J5sL
specific loss
bearing load
constant friction coefficient

&oman variables

a number of parallel branches
A linear current density JAEmL
a
%
acceleration JmEs
(
L
B flux density J5sEm
(
L
b breadth JmL
C friction coefficient
d depth JmL
D diameter JmL
E back electromoti%e %oltage J5L
f freAuency JEsL
h height JmL
H magnetic field strength JAEmL
I current JAL
k thermal conducti%ity JMDEHL
k
D
Darter coefficient
k
d
winding distribution factor
k
p
winding pitch factor
k
sA
winding skew factor
k
w
winding factor of fundamental wa%e
L inductance J9L
l length JmL
m mass JkgL
n rotating speed Jmin
<
L
N
s
phase turn number
Nu ;usselt number


$omenclature
%reek variables

coefficient
absorpti%ity
air gap

r
relati%e permitti%ity
emissi%ity
efficiency
coefficient

r
relati%e permeability
number of harmonic order
electrical resisti%ity JKmL
electrical conducti%ity JSL
pitch factor
angular %elocity JradEsL
mechanical angular %elocity JradEsL
magnetic flux J5sL
current linkage JAL
magnetic flux linkage J5sL
specific loss
bearing load
constant friction coefficient

&oman variables

a number of parallel branches
A linear current density JAEmL
a
%
acceleration JmEs
(
L
B flux density J5sEm
(
L
b breadth JmL
C friction coefficient
d depth JmL
D diameter JmL
E back electromoti%e %oltage J5L
f freAuency JEsL
h height JmL
H magnetic field strength JAEmL
I current JAL
k thermal conducti%ity JMDEHL
k
D
Darter coefficient
k
d
winding distribution factor
k
p
winding pitch factor
k
sA
winding skew factor
k
w
winding factor of fundamental wa%e
L inductance J9L
l length JmL
m mass JkgL
n rotating speed Jmin
<
L
N
s
phase turn number
Nu ;usselt number


p pole number
P power JHL
q slots per pole and phase
Q slot number
r radius JmL
S area Jm
(
L
T torAue J;mL
U %oltage J5L
w width JmL
:
N
parallel conductors in a slot

"ubscripts

eA eAui%alent
air gap
leakage
a%g a%erage
7 bearing
d tooth
e eddy current
f fluid (con%ection)
'e iron
h hysteresis
# line
## line<to<line
m magnetiGing
max maximum
n nominal
p pole
A Auadrature
r radial& relati%e& rotor& relati%e
rr rolling
s stator
tan tangential
u slot
% drag
w wheel&winding
ins insulation
impr impregnation
w fundamental freAuency

Acronyms

AD alternating current
DD direct current
1M embedded manget
1M' electromoti%e force
1<RA 1lectric<Race About
2D1 2nternal combustion engine
21A 2nternational 1nergy Agency
2M 2nduction machine
MM magnet module
p*u* per unit


p pole number
P power JHL
q slots per pole and phase
Q slot number
r radius JmL
S area Jm
(
L
T torAue J;mL
U %oltage J5L
w width JmL
:
N
parallel conductors in a slot

"ubscripts

eA eAui%alent
air gap
leakage
a%g a%erage
7 bearing
d tooth
e eddy current
f fluid (con%ection)
'e iron
h hysteresis
# line
## line<to<line
m magnetiGing
max maximum
n nominal
p pole
A Auadrature
r radial& relati%e& rotor& relati%e
rr rolling
s stator
tan tangential
u slot
% drag
w wheel&winding
ins insulation
impr impregnation
w fundamental freAuency

Acronyms

AD alternating current
DD direct current
1M embedded manget
1M' electromoti%e force
1<RA 1lectric<Race About
2D1 2nternal combustion engine
21A 2nternational 1nergy Agency
2M 2nduction machine
MM magnet module
p*u* per unit


.M permanent magnet
.HM pulse width modulation
RD radio controlled
RMS Root<mean<sAuare
SRM Switched reluctance machine
SMD Soft magnetic composite
SynRa.M Synchronous<reluctance<assisted .M machine
SynRM Synchronous reluctance machine
T1 totally enclosed


$atural constants

g specific gra%ity O,*" JmEs
(
L
relation of circle perimeter to diameter O!*+>

)
permitti%ity of %acuum O "*"=+)
<(
J'EmL

)
permeability of %acuum O*(=>)
<>
J5sEAmL










.M permanent magnet
.HM pulse width modulation
RD radio controlled
RMS Root<mean<sAuare
SRM Switched reluctance machine
SMD Soft magnetic composite
SynRa.M Synchronous<reluctance<assisted .M machine
SynRM Synchronous reluctance machine
T1 totally enclosed


$atural constants

g specific gra%ity O,*" JmEs
(
L
relation of circle perimeter to diameter O!*+>

)
permitti%ity of %acuum O "*"=+)
<(
J'EmL

)
permeability of %acuum O*(=>)
<>
J5sEAmL








!


! Introduction
1lectromechanical energy con%ersion processes are among the key elements in the de%elopment of
modern society* There are two kinds of primary energy sourcesB renewable and non<renewable*
The ma3ority of the current primary energy consumption is still satisfied with non<renewable
energy sources* According to the 2nternational 1nergy Agency (21A)& !*( P of the total ( --
Mtoe ( Mtoe Q +(8/ Q *- MHh) of primary energy supply was produced with renewable
sources in ()) (21A& ()()* The renewable primary energy supply consists mainly of hydro
energy and energy from combustible renewables and waste* The rest& )*, P units of the renewable
primary energy supply& come from geothermal& wind& solar and similar energy sources* A
comparison of the primary energy supply in ,-! and ()), is presented in 'ig* **

'ig* ** Horld primary energy supply in ,-! and ()) (21A& ()()*
8lobally& the estimated final energy consumption is about ">-- Mtoe& which is about ">--E(=)
Q >" P of the total primary energy supply* About =!> Mtoe of this energy is consumed as
electricity& of which the industry takes about +*= P and transportation only about *> P* The rest
is left for residential use& non<specific other use and commercial and public ser%ices* :n the other
hand& !=-) Mtoe of energy is consumed as oil& of which >*= P is consumed in transportation* As
a result of the increasing oil price and awareness of en%ironmental issues& the de%elopment of
transportation systems is heading towards cleaner propulsion technologies with electrical machines
in %ehicles* The consumption of primary energy in transportation can be significantly reduced with
electric traction systems& because the energy con%ersion efficiency is significantly higher in large
power plants running at steady<state power compared with %arying load cycle efficiencies of
!


! Introduction
1lectromechanical energy con%ersion processes are among the key elements in the de%elopment of
modern society* There are two kinds of primary energy sourcesB renewable and non<renewable*
The ma3ority of the current primary energy consumption is still satisfied with non<renewable
energy sources* According to the 2nternational 1nergy Agency (21A)& !*( P of the total ( --
Mtoe ( Mtoe Q +(8/ Q *- MHh) of primary energy supply was produced with renewable
sources in ()) (21A& ()()* The renewable primary energy supply consists mainly of hydro
energy and energy from combustible renewables and waste* The rest& )*, P units of the renewable
primary energy supply& come from geothermal& wind& solar and similar energy sources* A
comparison of the primary energy supply in ,-! and ()), is presented in 'ig* **

'ig* ** Horld primary energy supply in ,-! and ()) (21A& ()()*
8lobally& the estimated final energy consumption is about ">-- Mtoe& which is about ">--E(=)
Q >" P of the total primary energy supply* About =!> Mtoe of this energy is consumed as
electricity& of which the industry takes about +*= P and transportation only about *> P* The rest
is left for residential use& non<specific other use and commercial and public ser%ices* :n the other
hand& !=-) Mtoe of energy is consumed as oil& of which >*= P is consumed in transportation* As
a result of the increasing oil price and awareness of en%ironmental issues& the de%elopment of
transportation systems is heading towards cleaner propulsion technologies with electrical machines
in %ehicles* The consumption of primary energy in transportation can be significantly reduced with
electric traction systems& because the energy con%ersion efficiency is significantly higher in large
power plants running at steady<state power compared with %arying load cycle efficiencies of
+


indi%idual internal combustion engines (2D1) in %ehicles* 2n the case of hybrid dri%es& the 2D1 unit
can be downsiGed and used in a more efficient way to significantly increase the o%erall
performance* Thus& the research and de%elopment of electric traction machine technology is still
highly important& e%en though it already has a history of more than hundred years* ;ew
technologies such as high energy product permanent magnets and efficient power electronics gi%e
rise to the de%elopment of electric traction systems*
1lectrical machine dri%es make it possible to change the direction of energy con%ersion between
electric energy and mechanical work& depending on the need* 1lectrical machine dri%es can be
di%ided into se%eral different categories according to the torAue and speed range characteristics of
the load* Some applications& such as con%eyor belt systems and compressors& reAuire constant
torAue output o%er the speed range* .ump and fan systems& on the other hand& reAuire the torAue
output to be directly proportional to the second power of the rotating speed* Thirdly& there is a load
type that reAuires a large breakaway torAue and a constant torAue range up to a certain base speed
and a broad constant power range beyond this point* #oad characteristics of this kind are typical
for traction applications& such as high<speed gearless ele%ators& high<performance con%eyors and
mobile %ehicles& which is the application field focused on in this thesis*

!.! 'b(ectives of the study
This doctoral thesis presents a study on the de%elopment of a liAuid<cooled frame salient pole
permanent<magnet<exited traction machine for a four<wheel<dri%en electric car in the 1lectric Race
About (1<RA) pro3ect of 9elsinki Metropolia $ni%ersity of Applied Sciences (later referred to as
Metropolia)*


'ig* *(* 1<RA concept car of 9elsinki Metropolia $ni%ersity of Applied Sciences
(Metropolia& ()),)*
The study concentrates on promoting a machine design for %ehicle traction& the key features and
ad%antages of which are light weight and high performance o%er a broad rotating speed range
+


indi%idual internal combustion engines (2D1) in %ehicles* 2n the case of hybrid dri%es& the 2D1 unit
can be downsiGed and used in a more efficient way to significantly increase the o%erall
performance* Thus& the research and de%elopment of electric traction machine technology is still
highly important& e%en though it already has a history of more than hundred years* ;ew
technologies such as high energy product permanent magnets and efficient power electronics gi%e
rise to the de%elopment of electric traction systems*
1lectrical machine dri%es make it possible to change the direction of energy con%ersion between
electric energy and mechanical work& depending on the need* 1lectrical machine dri%es can be
di%ided into se%eral different categories according to the torAue and speed range characteristics of
the load* Some applications& such as con%eyor belt systems and compressors& reAuire constant
torAue output o%er the speed range* .ump and fan systems& on the other hand& reAuire the torAue
output to be directly proportional to the second power of the rotating speed* Thirdly& there is a load
type that reAuires a large breakaway torAue and a constant torAue range up to a certain base speed
and a broad constant power range beyond this point* #oad characteristics of this kind are typical
for traction applications& such as high<speed gearless ele%ators& high<performance con%eyors and
mobile %ehicles& which is the application field focused on in this thesis*

!.! 'b(ectives of the study
This doctoral thesis presents a study on the de%elopment of a liAuid<cooled frame salient pole
permanent<magnet<exited traction machine for a four<wheel<dri%en electric car in the 1lectric Race
About (1<RA) pro3ect of 9elsinki Metropolia $ni%ersity of Applied Sciences (later referred to as
Metropolia)*


'ig* *(* 1<RA concept car of 9elsinki Metropolia $ni%ersity of Applied Sciences
(Metropolia& ()),)*
The study concentrates on promoting a machine design for %ehicle traction& the key features and
ad%antages of which are light weight and high performance o%er a broad rotating speed range
=


within the boundaries of the gi%en application* :ne of these is the direct dri%e principle& which is
not %ery common in electric %ehicles*
The key target of the machine design& besides the direct dri%e& is to achie%e a light structure with
an enhanced torAue output capability with inherent saliency by a suitable pole design* The
reluctance torAue helps in the pull<out torAue production and boosts the machine torAue properties
throughout the speed range* The main part of the torAue is still produced by permanent magnet
excitation* 9ence& the machine is here designated as a synchronous& reluctance<torAue<assisted
permanent magnet (SynRa.M) machine in contrast to synchronous reluctance machines with an
additional .M excitation* 'urther& the design process includes certain thermal endurance aspects*
A thermal analysis is essential in these kinds of machine structures where the loss distribution is
dominated by /oule losses*
The rated power of (= kH per motor at a rotating speed of ))) min
<
in constant operation was
originally set as a basis for the machine design according to the known race car track data pro%ided
by Metropolia* The constant operation at (= kH for the rotating speed of ))) min
<
yields a
torAue of (+) ;m* The target set for the peak torAue of the machine was ))) ;m* The dri%e
propulsion system consists of an accumulator connected to four three<phase in%erter bridges with
each one dri%ing a .M traction motor connected to a wheel %ia a dri%e shaft* Thus& the combined
continuous power of the traction system is )) kH at ))) min
<
* The motors are& howe%er& also
capable of producing ()) kH of combined continuous power at ())) min
<
because of the low
iron loss generation and impro%ed cooling of the through<%entilated design* The accumulator
package consists of two sets of +! series<connected cells resulting in a )) Ah capacity* The
indi%idual battery cell cut<off %oltage as a function of cell charge is presented in 'ig* *!*

'ig* *!* 2ndi%idual cell cut<off %oltage le%els of Altairnano =) Ah #ithium titanate battery
cell as a function of cell charge for charge currents of =) A& (=) A and =)) A
(Altairnano& ()),)*
=


within the boundaries of the gi%en application* :ne of these is the direct dri%e principle& which is
not %ery common in electric %ehicles*
The key target of the machine design& besides the direct dri%e& is to achie%e a light structure with
an enhanced torAue output capability with inherent saliency by a suitable pole design* The
reluctance torAue helps in the pull<out torAue production and boosts the machine torAue properties
throughout the speed range* The main part of the torAue is still produced by permanent magnet
excitation* 9ence& the machine is here designated as a synchronous& reluctance<torAue<assisted
permanent magnet (SynRa.M) machine in contrast to synchronous reluctance machines with an
additional .M excitation* 'urther& the design process includes certain thermal endurance aspects*
A thermal analysis is essential in these kinds of machine structures where the loss distribution is
dominated by /oule losses*
The rated power of (= kH per motor at a rotating speed of ))) min
<
in constant operation was
originally set as a basis for the machine design according to the known race car track data pro%ided
by Metropolia* The constant operation at (= kH for the rotating speed of ))) min
<
yields a
torAue of (+) ;m* The target set for the peak torAue of the machine was ))) ;m* The dri%e
propulsion system consists of an accumulator connected to four three<phase in%erter bridges with
each one dri%ing a .M traction motor connected to a wheel %ia a dri%e shaft* Thus& the combined
continuous power of the traction system is )) kH at ))) min
<
* The motors are& howe%er& also
capable of producing ()) kH of combined continuous power at ())) min
<
because of the low
iron loss generation and impro%ed cooling of the through<%entilated design* The accumulator
package consists of two sets of +! series<connected cells resulting in a )) Ah capacity* The
indi%idual battery cell cut<off %oltage as a function of cell charge is presented in 'ig* *!*

'ig* *!* 2ndi%idual cell cut<off %oltage le%els of Altairnano =) Ah #ithium titanate battery
cell as a function of cell charge for charge currents of =) A& (=) A and =)) A
(Altairnano& ()),)*
>


According to the manufacturer data presented in 'ig* *!& the cell cut<off %oltages are O*= 5
DD
at
the lower limit and O(*, 5
DD
at the upper limit* Thus& the battery package is capable of producing
%oltage le%els between +!R*= 5
DD
S (+*= 5
DD
and +!R(*, 5
DD
S ++*- 5
DD
* 7efore the ") P
depth of the discharge le%el is reached& the normal maximum operating %oltage is around !-+ 5
DD
&
which leads to a maximum AD line<to<line %oltage of (>+ 5
RMS
* The cut<off line<to<line AD
%oltage le%el is =*> 5
RMS*
The a%ailable %oltage is intended to be used at the top speed without
entering deeply into the field weakening region* Hith this %oltage le%el& the machine is capable of
achie%ing a maximum speed of ())) min
<
in normal operation& which allows top speeds beyond
()) kmEh with a suitable wheel configuration* Hith the cut<off %oltage& the motor can achie%e
=) min
<
without entering the field weakening region* ;e%ertheless& field weakening operation is
mandatory at higher speeds& since the accumulator %oltage drops with the increasing load current
and the decreasing charge as presented in 'ig* *!*

!. )raction machine for vehicle propulsion
Dommon for traction applications is the high torAue reAuirement o%er a broad operating speed
range* Traction applications usually reAuire a torAue to speed cur%e that resembles the cur%es
presented in 'ig *+*

'ig* *+* 1xamples of per unit torAue to speed cur%es of .MSMs (red) and per unit
constant power cur%es (black) for traction applications* L
d
SL
A
S)*( pu* The
permanent magnet flux linkage is ,
.M
S p*u*
>


According to the manufacturer data presented in 'ig* *!& the cell cut<off %oltages are O*= 5
DD
at
the lower limit and O(*, 5
DD
at the upper limit* Thus& the battery package is capable of producing
%oltage le%els between +!R*= 5
DD
S (+*= 5
DD
and +!R(*, 5
DD
S ++*- 5
DD
* 7efore the ") P
depth of the discharge le%el is reached& the normal maximum operating %oltage is around !-+ 5
DD
&
which leads to a maximum AD line<to<line %oltage of (>+ 5
RMS
* The cut<off line<to<line AD
%oltage le%el is =*> 5
RMS*
The a%ailable %oltage is intended to be used at the top speed without
entering deeply into the field weakening region* Hith this %oltage le%el& the machine is capable of
achie%ing a maximum speed of ())) min
<
in normal operation& which allows top speeds beyond
()) kmEh with a suitable wheel configuration* Hith the cut<off %oltage& the motor can achie%e
=) min
<
without entering the field weakening region* ;e%ertheless& field weakening operation is
mandatory at higher speeds& since the accumulator %oltage drops with the increasing load current
and the decreasing charge as presented in 'ig* *!*

!. )raction machine for vehicle propulsion
Dommon for traction applications is the high torAue reAuirement o%er a broad operating speed
range* Traction applications usually reAuire a torAue to speed cur%e that resembles the cur%es
presented in 'ig *+*

'ig* *+* 1xamples of per unit torAue to speed cur%es of .MSMs (red) and per unit
constant power cur%es (black) for traction applications* L
d
SL
A
S)*( pu* The
permanent magnet flux linkage is ,
.M
S p*u*
-


The reAuirements for electrical machine performance can be e%aluated by comparing the
acceleration force with the force reAuired to maintain a constant operating speed* The torAue acting
on a wheel forms a force parallel to the surface* There are three forces acting against this force*
They are rolling friction& air friction and the force reAuired to alter the momentum* Hith no
ele%ation taken into account& the force balance eAuation takes the form
I
r
w
= mo + C
rr
mg +
1
2
p
aIr
:
2
SC
v
*

where
< m is the total mass of the %ehicle JkgL&
< a is the acceleration JmEs
(
L&
< g the gra%itational constant JmEs
(
L&
<
air
is the air density JkgEm
!
L&
< S is the %ehicle cross<sectional area towards mo%ing direction&
< C
rr
is the rolling friction factor&
< C
%
is the drag coefficient&
< T is the wheel torAue J;mL&
< r
w
is the wheel radius JmL and
< v is the %elocity of the %ehicle JmEsL
To maintain a constant speed& the machine wheel must produce torAue eAual to the drag and rolling
friction* The desired acceleration rate determines the reAuired maximum torAue output*
Hith a practical example& the proportions of different drag types are easier to percei%e* An a%erage
passenger car weighs about =)) kg& and with a carrying capacity of +=) kg& the total mass eAuals
,=) kg* A typical drag coefficient is around )*! (%arying between )*( and )*>) and a typical
rolling friction coefficient is around )*)& %arying between )*))- and )*)+ (TR7& ())>)* The tyre
radius for a common ()=<==<R> wheel is around )*!> m* The dri%ing direction cross<sectional
area of a typical car can be approximated to be (*> m
(
* Hith these assumptions& a passenger car
reAuires
I = [u.u1 19Su 9.81 +
1
2
1.22S 27.8
2
2.16 u.S u.S16 = 1S7.2 J;mL

of the total torAue to maintain a speed of )) kmEh with the specified wheel arrangement& which
eAuals an a%erage power of !*" kH yielding an energy consumption of )*!" kHhEkm* This is
close to the measured results of the 1<RA car presented in 'ig* *=*
-


The reAuirements for electrical machine performance can be e%aluated by comparing the
acceleration force with the force reAuired to maintain a constant operating speed* The torAue acting
on a wheel forms a force parallel to the surface* There are three forces acting against this force*
They are rolling friction& air friction and the force reAuired to alter the momentum* Hith no
ele%ation taken into account& the force balance eAuation takes the form
I
r
w
= mo + C
rr
mg +
1
2
p
aIr
:
2
SC
v
*

where
< m is the total mass of the %ehicle JkgL&
< a is the acceleration JmEs
(
L&
< g the gra%itational constant JmEs
(
L&
<
air
is the air density JkgEm
!
L&
< S is the %ehicle cross<sectional area towards mo%ing direction&
< C
rr
is the rolling friction factor&
< C
%
is the drag coefficient&
< T is the wheel torAue J;mL&
< r
w
is the wheel radius JmL and
< v is the %elocity of the %ehicle JmEsL
To maintain a constant speed& the machine wheel must produce torAue eAual to the drag and rolling
friction* The desired acceleration rate determines the reAuired maximum torAue output*
Hith a practical example& the proportions of different drag types are easier to percei%e* An a%erage
passenger car weighs about =)) kg& and with a carrying capacity of +=) kg& the total mass eAuals
,=) kg* A typical drag coefficient is around )*! (%arying between )*( and )*>) and a typical
rolling friction coefficient is around )*)& %arying between )*))- and )*)+ (TR7& ())>)* The tyre
radius for a common ()=<==<R> wheel is around )*!> m* The dri%ing direction cross<sectional
area of a typical car can be approximated to be (*> m
(
* Hith these assumptions& a passenger car
reAuires
I = [u.u1 19Su 9.81 +
1
2
1.22S 27.8
2
2.16 u.S u.S16 = 1S7.2 J;mL

of the total torAue to maintain a speed of )) kmEh with the specified wheel arrangement& which
eAuals an a%erage power of !*" kH yielding an energy consumption of )*!" kHhEkm* This is
close to the measured results of the 1<RA car presented in 'ig* *=*
"



'ig* *=* 1<RA constant speed energy consumption measurement results from the Dhelsea
.ro%ing 8round (D.8) test area (1<RA report& ()()*
Hith the same assumptions& the reAuired torAue for a constant speed of ()) kmEh would be ++-*=
;m* Hhen the effect of acceleration is taken into account& the need for torAue rapidly increases* A
ten<second constant acceleration to )) kmEh reAuires a total wheel torAue of -( ;m in addition
to the constant speed torAue reAuirement* The torAue reAuired for the acceleration is almost
times the torAue of the constant speed operation reAuirement* This example sets the basis for the
machine performance in traction applications* 2t should be beneficial to use a motor& the efficiency
of which is high in constant speed operation& and which is capable of producing a high torAue
during acceleration* To achie%e a high efficiency in constant speed operation& the machine should
not be hea%ily o%erdimensioned* This means that the machine should be designed closer to the
needed a%erage power instead of the peak power reAuirement*
2n synchronous machines& the torAue is about in%ersely proportional to the synchronous inductance
(see 1A* *() if the %oltage is kept constant* Therefore& a machine with a low synchronous
inductance is suitable for acceleration* This fa%ours a permanent magnet machine design with a
large magnetic air gap* The field weakening performance instead takes ad%antage of the high
synchronous inductance %alues* 9igh inductance %alues do not necessarily cause a problem&
because the torAue production can be impro%ed with the a%ailable %oltage reser%e at low speeds&
where the highest acceleration rates are needed*
2f the constant operating speed torAue of )) kmEh is selected as the rated %alue and there is a
limited %oltage reser%e in the system& the synchronous inductance should be less than )* per unit
to achie%e the times per unit peak torAue for the ) s acceleration to )) kmEh* Depending on
the application& the reAuired maximum torAue may lead to a significantly larger machine
construction than the constant operation would reAuire* This is the case in 2D1<powered %ehicles&
where the engine is o%erdimensioned to get good acceleration& because the engine o%erload
capability is poor* This generally results in poor efficiency in partial loads reAuired for constant
operation* 2n electric dri%es& howe%er& it is possible to achie%e peak torAues se%eral times the rated
"



'ig* *=* 1<RA constant speed energy consumption measurement results from the Dhelsea
.ro%ing 8round (D.8) test area (1<RA report& ()()*
Hith the same assumptions& the reAuired torAue for a constant speed of ()) kmEh would be ++-*=
;m* Hhen the effect of acceleration is taken into account& the need for torAue rapidly increases* A
ten<second constant acceleration to )) kmEh reAuires a total wheel torAue of -( ;m in addition
to the constant speed torAue reAuirement* The torAue reAuired for the acceleration is almost
times the torAue of the constant speed operation reAuirement* This example sets the basis for the
machine performance in traction applications* 2t should be beneficial to use a motor& the efficiency
of which is high in constant speed operation& and which is capable of producing a high torAue
during acceleration* To achie%e a high efficiency in constant speed operation& the machine should
not be hea%ily o%erdimensioned* This means that the machine should be designed closer to the
needed a%erage power instead of the peak power reAuirement*
2n synchronous machines& the torAue is about in%ersely proportional to the synchronous inductance
(see 1A* *() if the %oltage is kept constant* Therefore& a machine with a low synchronous
inductance is suitable for acceleration* This fa%ours a permanent magnet machine design with a
large magnetic air gap* The field weakening performance instead takes ad%antage of the high
synchronous inductance %alues* 9igh inductance %alues do not necessarily cause a problem&
because the torAue production can be impro%ed with the a%ailable %oltage reser%e at low speeds&
where the highest acceleration rates are needed*
2f the constant operating speed torAue of )) kmEh is selected as the rated %alue and there is a
limited %oltage reser%e in the system& the synchronous inductance should be less than )* per unit
to achie%e the times per unit peak torAue for the ) s acceleration to )) kmEh* Depending on
the application& the reAuired maximum torAue may lead to a significantly larger machine
construction than the constant operation would reAuire* This is the case in 2D1<powered %ehicles&
where the engine is o%erdimensioned to get good acceleration& because the engine o%erload
capability is poor* This generally results in poor efficiency in partial loads reAuired for constant
operation* 2n electric dri%es& howe%er& it is possible to achie%e peak torAues se%eral times the rated
,


torAue le%el for high acceleration and still achie%e a small and efficient machine for normal
operation*
2n our application& the torAue production task is di%ided between all four wheels* The torAue
reAuired per wheel is about +) ;m at )) kmEh* The rated torAue of (+) ;m per motor would
yield a synchronous inductance around (+)E+(" S )*=> to achie%e the -(E+ S+(" ;m torAue for
the ) s acceleration from ) to )) kmEh if considering a non<salient pole permanent magnet
machine with a back electromoti%e force per unit and a constant UEf S control up to the desired
speed* #ow per unit %alues of direct<axis synchronous inductances are typical for permanent
magnet machines* 2n field<current<controlled synchronous machines& the L
d
can be as high as (*)
per unit*
The next task is twofold* To accomplish the desired performance& the supply %oltage and current
ratings ha%e to be selected appropriately* The industrial %oltage le%el (>,) 5 line<to<line) would
fa%our smaller machine dimensions and significantly smaller cabling dimensions than the %oltage
le%els used in the 1<RA* The higher %oltage ratings lead to a smaller coil turn cross<sectional area&
which allows a better control of the end winding o%erhang in the machine design* Smaller cabling
dimensions help to reduce the dri%e system o%erall weight*
The industrial %oltage le%el would reAuire a ))) 5 DD %oltage le%el in battery<supplied
applications* :wing to the stability issues related to the battery management system and the
a%ailability of battery capacity %alues& these %oltage le%els are a%oided in systems of this kind* The
reason lies mainly in the lithium ion battery technology used in the study* #ithium<ion batteries
reAuire the battery management systems to pre%ent indi%idual cell charge differences& that is& a
State<of<Dharge (SoD) mismatch& which is a common problem in series<connected lithium<ion
cells* Thus& lower %oltage le%els are preferred in lithium<ion battery solutions* Still& lithium<ion
batteries are practically the standard solution for energy storages in mo%ing electric %ehicles
because of their high energy storage capability compared with other cell types a%ailable* The
lithium ion battery also has Auite a low self<discharging rate& which supports the usage of this
battery technology in %ehicle applications* The practical upper limit of the DD %oltage seems to
settle around -=)@")) 5& but ratings of !))@+))5 are much more common* The lower %oltage
rating reAuires more current handling capability& which& on the other hand& reAuires hea%ier power
electronics*
The %oltage& flux linkage and rotating speed ha%e to be matched together* 2n practice& the machine
nominal operating speed has to be selected according to the highest rotating speed at which the
loading capability is needed* The end of the operating speed range depends on the field weakening
region* 2f the field weakening of the machine is not allowed& the usable %oltage should be
consumed in this point* .ermanent magnet machines reAuire extra attention in the field weakening
operation because of the demagnetiGing armature reaction from the stator and the temperature
sensiti%ity of the magnet material*
The speed range of the maximum torAue should also be considered& because the torAue production
is proportional to the cross product of the stator flux linkage and current %ectors* Hith a suitable
reser%e at the supply %oltage le%el& the torAue can be boosted by forcing additional flux into the
machine* At low speed and high torAue& the a%ailable %oltage reser%e can be used to increase the
,


torAue le%el for high acceleration and still achie%e a small and efficient machine for normal
operation*
2n our application& the torAue production task is di%ided between all four wheels* The torAue
reAuired per wheel is about +) ;m at )) kmEh* The rated torAue of (+) ;m per motor would
yield a synchronous inductance around (+)E+(" S )*=> to achie%e the -(E+ S+(" ;m torAue for
the ) s acceleration from ) to )) kmEh if considering a non<salient pole permanent magnet
machine with a back electromoti%e force per unit and a constant UEf S control up to the desired
speed* #ow per unit %alues of direct<axis synchronous inductances are typical for permanent
magnet machines* 2n field<current<controlled synchronous machines& the L
d
can be as high as (*)
per unit*
The next task is twofold* To accomplish the desired performance& the supply %oltage and current
ratings ha%e to be selected appropriately* The industrial %oltage le%el (>,) 5 line<to<line) would
fa%our smaller machine dimensions and significantly smaller cabling dimensions than the %oltage
le%els used in the 1<RA* The higher %oltage ratings lead to a smaller coil turn cross<sectional area&
which allows a better control of the end winding o%erhang in the machine design* Smaller cabling
dimensions help to reduce the dri%e system o%erall weight*
The industrial %oltage le%el would reAuire a ))) 5 DD %oltage le%el in battery<supplied
applications* :wing to the stability issues related to the battery management system and the
a%ailability of battery capacity %alues& these %oltage le%els are a%oided in systems of this kind* The
reason lies mainly in the lithium ion battery technology used in the study* #ithium<ion batteries
reAuire the battery management systems to pre%ent indi%idual cell charge differences& that is& a
State<of<Dharge (SoD) mismatch& which is a common problem in series<connected lithium<ion
cells* Thus& lower %oltage le%els are preferred in lithium<ion battery solutions* Still& lithium<ion
batteries are practically the standard solution for energy storages in mo%ing electric %ehicles
because of their high energy storage capability compared with other cell types a%ailable* The
lithium ion battery also has Auite a low self<discharging rate& which supports the usage of this
battery technology in %ehicle applications* The practical upper limit of the DD %oltage seems to
settle around -=)@")) 5& but ratings of !))@+))5 are much more common* The lower %oltage
rating reAuires more current handling capability& which& on the other hand& reAuires hea%ier power
electronics*
The %oltage& flux linkage and rotating speed ha%e to be matched together* 2n practice& the machine
nominal operating speed has to be selected according to the highest rotating speed at which the
loading capability is needed* The end of the operating speed range depends on the field weakening
region* 2f the field weakening of the machine is not allowed& the usable %oltage should be
consumed in this point* .ermanent magnet machines reAuire extra attention in the field weakening
operation because of the demagnetiGing armature reaction from the stator and the temperature
sensiti%ity of the magnet material*
The speed range of the maximum torAue should also be considered& because the torAue production
is proportional to the cross product of the stator flux linkage and current %ectors* Hith a suitable
reser%e at the supply %oltage le%el& the torAue can be boosted by forcing additional flux into the
machine* At low speed and high torAue& the a%ailable %oltage reser%e can be used to increase the
()


stator flux linkage* A constant flux operation in traction systems is not a necessity* 1specially in
battery<powered applications& the machine should be designed to a suitably low %oltage le%el to
ensure torAue output in conditions where the battery %oltage decreases under load* A high current
supply makes possible an increase in the flux linkage as the %oltage reser%e is consumed* The
higher flux le%el can also enhance reluctance torAue production* ;e%ertheless& the use of a flux
linkage boost reAuires suitably loose dimensioning of the magnetic iron circuit in order to fully
exploit the %oltage reser%e in the torAue production*

!..! *ield of traction applications
As the basis of this doctoral thesis is application specific& the work concentrates on the permanent
magnet machine technology* The direct connection to the wheels reAuires a large torAue output as
presented in the pre%ious section*
Hith a flat battery& the field weakening operation reAuires about =) P of the stator flux linkage
le%el*
Deep field weakening operation was not reAuired in the machine performance characteristics* This
leads to the selection of a .M machine design with a suitable synchronous inductance le%el* ;o
other machine type has the efficiency and torAue capability of the permanent magnet machine*
1%en more importantly& when comparing for instance with an induction machine& the .MSM
machine offers more freedom in machine parameters& such as pole number& dimensions and slots*
;e%ertheless& other machine topologies ha%e successfully been applied to hybrid and full electric
propulsion systems* Depending on the reAuired performance characteristics and cost issues& there
are numerous alternati%es for the propulsion machine*
2n general& electrical machines represent well<known technology* Durrently& the discussion is
intense on the suitability of different machine topologies for hybrid and fully electric propulsions*
There are se%eral types of electrical machines that can be harnessed to the needs of hybrid and
electric systems* The main categories are asynchronous and synchronous AD machines& e%en
though DD machines are also used in low<le%el applications* The DD technology is not suitable for
modern higher<le%el propulsion applications& but in the transition period toward cost<efficient AD
dri%e systems& the low cost and easy control of the DD machines ha%e fa%oured the structure in
small %ehicle propulsion& such as motorcycles& mopeds and all<terrain %ehicles (AT5s)* :ne of the
problems of the DD system is the need for freAuent maintenance because of the mechanical
commutator circuit* The same problem also affects the traditional AD synchronous machine e%en
though it probably has the best properties for high torAue output and field weakening* 7ecause of
its complicated structure and high cost& it has to be left out of the comparison*
The control of freAuency con%erter AD dri%es is& in theory& more complicated than the control of
DD machines& but the efficiency of the AD machines is much higher compared with brushed DD
machine dri%e systems* As the %olumes increase& the AD dri%e systems will be more appealing as
the costs will decrease* As a result& the options for hybrid and full electric dri%e systems are down
()


stator flux linkage* A constant flux operation in traction systems is not a necessity* 1specially in
battery<powered applications& the machine should be designed to a suitably low %oltage le%el to
ensure torAue output in conditions where the battery %oltage decreases under load* A high current
supply makes possible an increase in the flux linkage as the %oltage reser%e is consumed* The
higher flux le%el can also enhance reluctance torAue production* ;e%ertheless& the use of a flux
linkage boost reAuires suitably loose dimensioning of the magnetic iron circuit in order to fully
exploit the %oltage reser%e in the torAue production*

!..! *ield of traction applications
As the basis of this doctoral thesis is application specific& the work concentrates on the permanent
magnet machine technology* The direct connection to the wheels reAuires a large torAue output as
presented in the pre%ious section*
Hith a flat battery& the field weakening operation reAuires about =) P of the stator flux linkage
le%el*
Deep field weakening operation was not reAuired in the machine performance characteristics* This
leads to the selection of a .M machine design with a suitable synchronous inductance le%el* ;o
other machine type has the efficiency and torAue capability of the permanent magnet machine*
1%en more importantly& when comparing for instance with an induction machine& the .MSM
machine offers more freedom in machine parameters& such as pole number& dimensions and slots*
;e%ertheless& other machine topologies ha%e successfully been applied to hybrid and full electric
propulsion systems* Depending on the reAuired performance characteristics and cost issues& there
are numerous alternati%es for the propulsion machine*
2n general& electrical machines represent well<known technology* Durrently& the discussion is
intense on the suitability of different machine topologies for hybrid and fully electric propulsions*
There are se%eral types of electrical machines that can be harnessed to the needs of hybrid and
electric systems* The main categories are asynchronous and synchronous AD machines& e%en
though DD machines are also used in low<le%el applications* The DD technology is not suitable for
modern higher<le%el propulsion applications& but in the transition period toward cost<efficient AD
dri%e systems& the low cost and easy control of the DD machines ha%e fa%oured the structure in
small %ehicle propulsion& such as motorcycles& mopeds and all<terrain %ehicles (AT5s)* :ne of the
problems of the DD system is the need for freAuent maintenance because of the mechanical
commutator circuit* The same problem also affects the traditional AD synchronous machine e%en
though it probably has the best properties for high torAue output and field weakening* 7ecause of
its complicated structure and high cost& it has to be left out of the comparison*
The control of freAuency con%erter AD dri%es is& in theory& more complicated than the control of
DD machines& but the efficiency of the AD machines is much higher compared with brushed DD
machine dri%e systems* As the %olumes increase& the AD dri%e systems will be more appealing as
the costs will decrease* As a result& the options for hybrid and full electric dri%e systems are down
(


to four different machine types and their %ariations* These are the permanent magnet synchronous
machine (.MSM)& synchronous reluctance machine (SynRM)& induction machine (2M) and the
switched reluctance machine (SRM)* The basic ideas of these machine structures are presented in
'ig* *>*

'ig* *>* Radial flux machine principle topologies where a) and b) represent salient pole
.M synchronous machines& c) switched reluctance machine (SRM)& d) non<
salient pole .M synchronous machine& e) SynRM machine and f) asynchronous
sAuirrel cage machine*
Donsidering the synchronous machine types& the permanent magnet synchronous machine
(.MSM) and the synchronous reluctance machine (SynRM) are good ri%als* A benefit of a
synchronous reluctance machine is its low cost structure& although its torAue density is not as high
as that of a .M machine* The efficiency of the SynRM structure competes head<to<head with that
of the 2M* The suitability of SynRMs for (9)15 applications has been studied for instance in
(Arkadan et al*& ())-)* The drawback is that the most effecti%e topology in SynRMs is the four<
pole design& because this arrangement guarantees a high inductance difference in the SynRMs*
This feature makes the machine type suitable for geared high<speed operation and& therefore&
inappropriate for direct<dri%e applications reAuiring a light machine construction* Some studies
about the suitability and design aspects of the SRM technology ha%e been reported in (Nionghua et
al& ())!)& (Hu et al*& ())()& (Ramamurthy& ())) and (:hyama et al*& ())>)* An SRM reAuires a
control of its own& but the structure of the machine is as robust as with the 2M and SynRM
technology* 7oth the SRM and SynRM structures are based on reluctance torAue production& but
the difference is that the SRM has always different pole arrangements in the rotor and the stator*
Thus& the torAue Auality produced in the SRM structure is poor compared with actual AD
machines*
An asynchronous machine could be a good choice because of its robust and low<cost structure*
The problem is that the use of an asynchronous machine in a propulsion dri%e system reAuires the
use of gears* The reason for this is that asynchronous machines are more suitable for higher<speed
applications& because the fa%ourable pole number of these machines is two or four* 2n
asynchronous machines& increasing the pole number causes a reduction in the power factor* The
(


to four different machine types and their %ariations* These are the permanent magnet synchronous
machine (.MSM)& synchronous reluctance machine (SynRM)& induction machine (2M) and the
switched reluctance machine (SRM)* The basic ideas of these machine structures are presented in
'ig* *>*

'ig* *>* Radial flux machine principle topologies where a) and b) represent salient pole
.M synchronous machines& c) switched reluctance machine (SRM)& d) non<
salient pole .M synchronous machine& e) SynRM machine and f) asynchronous
sAuirrel cage machine*
Donsidering the synchronous machine types& the permanent magnet synchronous machine
(.MSM) and the synchronous reluctance machine (SynRM) are good ri%als* A benefit of a
synchronous reluctance machine is its low cost structure& although its torAue density is not as high
as that of a .M machine* The efficiency of the SynRM structure competes head<to<head with that
of the 2M* The suitability of SynRMs for (9)15 applications has been studied for instance in
(Arkadan et al*& ())-)* The drawback is that the most effecti%e topology in SynRMs is the four<
pole design& because this arrangement guarantees a high inductance difference in the SynRMs*
This feature makes the machine type suitable for geared high<speed operation and& therefore&
inappropriate for direct<dri%e applications reAuiring a light machine construction* Some studies
about the suitability and design aspects of the SRM technology ha%e been reported in (Nionghua et
al& ())!)& (Hu et al*& ())()& (Ramamurthy& ())) and (:hyama et al*& ())>)* An SRM reAuires a
control of its own& but the structure of the machine is as robust as with the 2M and SynRM
technology* 7oth the SRM and SynRM structures are based on reluctance torAue production& but
the difference is that the SRM has always different pole arrangements in the rotor and the stator*
Thus& the torAue Auality produced in the SRM structure is poor compared with actual AD
machines*
An asynchronous machine could be a good choice because of its robust and low<cost structure*
The problem is that the use of an asynchronous machine in a propulsion dri%e system reAuires the
use of gears* The reason for this is that asynchronous machines are more suitable for higher<speed
applications& because the fa%ourable pole number of these machines is two or four* 2n
asynchronous machines& increasing the pole number causes a reduction in the power factor* The
((


power factor decreases along with the magnetiGing inductance& which is in%ersely proportional to
the second power of the pole pair number J9eikkil6& ())(L* The total cost of the systems can
usually be kept at a reasonable le%el& because industrial gears are typically inexpensi%e*
At the moment& the .M technology is the most studied of all machine types* A .M machine does
not suffer from a high pole number& which allows a design of lightweight multiple pole
arrangements e%en for direct traction dri%es*
The .M technology pro%ides the highest torAue density compared with any other machine type*
This results in smaller and lighter machine structures* There is not necessarily a need for a gear or
a complex transmission arrangement& which makes the dri%e system simple and more robust* The
direct connection also helps in reducing noise& as high<speed dri%e gears are usually not %ery Auiet*
A direct dri%e also reAuires a high motor torAue& which results in a large rotor %olume and high
magnet material consumption compared with high<speed motor dri%es*
Thus& a weak point of any direct dri%e arrangement is the high material cost caused by the high
price of rare<earth<based .M materials* The operating speed range is also limited& because the
magnets are typically sensiti%e to temperature and demagnetiGation* The interest in radial flux .M
machine optimiGation and suitability for (9)15 purposes is demonstrated in numerous articles
(/ung et al*& ())-)& (Iingming et al*& ()),)& ('aiG& ())>)& (7arcaro et al*& ())")& (Dhau et al*&
())-)& (Iia& ,,")& (Dhan& ,,>) and (Dhau et al*& ())")* The proposed machine topologies are
mainly radial flux machines& but some approaches to the .M axial flux technology ha%e been
proposed for instance in (D%etko%ski et al*& ())>) and (Daricchi et al*& ,,+)*
Different propulsion systems ha%e been compared in (Teraoulia et al*& ())=)& (2111 Doll*& ,,!)&
(;eudorfer et al*& ())") and (Iiong et al*& ())") with 2M& SRM and .M* The main conclusion is
that induction machines are competing head<to<head with .M machinesU both solutions ha%e their
ad%antages o%er another* $sually& the factors tipping the scales are the 2M cost %ersus the .M
machine torAue density* Although the energy storage system is the ma3or issue in the market
penetration of the hybrid and electric systems& the electrical machine is still the key factor in the
production of energy efficient propulsion systems to replace the current internal combustion
engine (2D1) technology*

!.. +lectric traction systems
The main categoriGation of electric traction systems is based on the di%ision between hybrid and
full electric traction systems* The common point of the systems is the application of the electrical
machine dri%e in the dri%e system* The term VmachineF refers both to the generating and motoring
capability of an electrical machine* 9ybrid dri%e systems as well as full electric dri%es can be
di%ided into se%eral subcategories* The hybrid dri%e systems can be di%ided into series and parallel
topologies and the combination of these two according to the electrical machine position in the
dri%e train*
((


power factor decreases along with the magnetiGing inductance& which is in%ersely proportional to
the second power of the pole pair number J9eikkil6& ())(L* The total cost of the systems can
usually be kept at a reasonable le%el& because industrial gears are typically inexpensi%e*
At the moment& the .M technology is the most studied of all machine types* A .M machine does
not suffer from a high pole number& which allows a design of lightweight multiple pole
arrangements e%en for direct traction dri%es*
The .M technology pro%ides the highest torAue density compared with any other machine type*
This results in smaller and lighter machine structures* There is not necessarily a need for a gear or
a complex transmission arrangement& which makes the dri%e system simple and more robust* The
direct connection also helps in reducing noise& as high<speed dri%e gears are usually not %ery Auiet*
A direct dri%e also reAuires a high motor torAue& which results in a large rotor %olume and high
magnet material consumption compared with high<speed motor dri%es*
Thus& a weak point of any direct dri%e arrangement is the high material cost caused by the high
price of rare<earth<based .M materials* The operating speed range is also limited& because the
magnets are typically sensiti%e to temperature and demagnetiGation* The interest in radial flux .M
machine optimiGation and suitability for (9)15 purposes is demonstrated in numerous articles
(/ung et al*& ())-)& (Iingming et al*& ()),)& ('aiG& ())>)& (7arcaro et al*& ())")& (Dhau et al*&
())-)& (Iia& ,,")& (Dhan& ,,>) and (Dhau et al*& ())")* The proposed machine topologies are
mainly radial flux machines& but some approaches to the .M axial flux technology ha%e been
proposed for instance in (D%etko%ski et al*& ())>) and (Daricchi et al*& ,,+)*
Different propulsion systems ha%e been compared in (Teraoulia et al*& ())=)& (2111 Doll*& ,,!)&
(;eudorfer et al*& ())") and (Iiong et al*& ())") with 2M& SRM and .M* The main conclusion is
that induction machines are competing head<to<head with .M machinesU both solutions ha%e their
ad%antages o%er another* $sually& the factors tipping the scales are the 2M cost %ersus the .M
machine torAue density* Although the energy storage system is the ma3or issue in the market
penetration of the hybrid and electric systems& the electrical machine is still the key factor in the
production of energy efficient propulsion systems to replace the current internal combustion
engine (2D1) technology*

!.. +lectric traction systems
The main categoriGation of electric traction systems is based on the di%ision between hybrid and
full electric traction systems* The common point of the systems is the application of the electrical
machine dri%e in the dri%e system* The term VmachineF refers both to the generating and motoring
capability of an electrical machine* 9ybrid dri%e systems as well as full electric dri%es can be
di%ided into se%eral subcategories* The hybrid dri%e systems can be di%ided into series and parallel
topologies and the combination of these two according to the electrical machine position in the
dri%e train*
(!


9ybrid electric systems are slightly more complicated than fully electric systems* A hybrid system
combines an internal combustion engine with an electrical machine* The focus in the hybrid
electric systems is to increase the system efficiency and impro%e the o%erall performance of the
system* 2n most cases& both targets are reached& but the weighing between the targeted properties
%aries depending on which characteristic is emphasiGed by the application*
9ybrid electric solutions can be di%ided into two categories or topologiesU series and parallel
hybrids* The series hybrid solution is closest to a fully electric dri%e& because the reAuired
mechanical work is always produced through an electric motor* $sually& these hybrid electric
systems include an energy storage (battery)& in addition to a fuel tank*
Systems with a combination of these two are also possible* 'ull electric propulsion systems can be
di%ided into two subcategories* Dompared with the direct connection applications& the difference
in an indirect connection is a transmission or a reduction gear included in the system* The
subcategories of the hybrid electric systems can also be di%ided into categories similar to the full
electric applications*
The full electric and hybrid solutions can be presented in a component block chart as shown in 'ig*
*-*

'ig* *-* Schematic of a hybrid and full electric propulsion system* Depending on the
application& the units presented by a dashed line are optional* #ightning bolts
denote electric coupling and cogwheels mechanical coupling*
2n general& a full electric or hybrid propulsion system usually includes a combination of an energy
source or a storage& a freAuency con%erter& at least one electrical machine and a load* Depending
on the system& other parts are optional* Hithout the 2D1 and generator blocks& the schematic
represents a full electric system with or without a gear coupling* 2f the 2D1 and generator blocks
are included in the system& we ha%e a series hybrid topology* Again& if the generator block is
omitted& the system represents a parallel hybrid solution* 2t is also possible to ha%e a combined
series parallel system containing all of the blocks of 'ig* *-*
A %ast ma3ority of modern efficient dri%e systems are three<phase systems with freAuency
con%erters implemented by ad%anced control methods* Modern permanent magnet electrical
machines in this power range can reach up to ,> P efficiency ratings in constant operation* The
energy source& usually a battery or a fuel cell& pro%ides the power for the electric motor through a
freAuency con%erter* Some systems are designed with a super capacitor in parallel connection with
the battery to absorb or discharge fast and powerful energy pulses*
(!


9ybrid electric systems are slightly more complicated than fully electric systems* A hybrid system
combines an internal combustion engine with an electrical machine* The focus in the hybrid
electric systems is to increase the system efficiency and impro%e the o%erall performance of the
system* 2n most cases& both targets are reached& but the weighing between the targeted properties
%aries depending on which characteristic is emphasiGed by the application*
9ybrid electric solutions can be di%ided into two categories or topologiesU series and parallel
hybrids* The series hybrid solution is closest to a fully electric dri%e& because the reAuired
mechanical work is always produced through an electric motor* $sually& these hybrid electric
systems include an energy storage (battery)& in addition to a fuel tank*
Systems with a combination of these two are also possible* 'ull electric propulsion systems can be
di%ided into two subcategories* Dompared with the direct connection applications& the difference
in an indirect connection is a transmission or a reduction gear included in the system* The
subcategories of the hybrid electric systems can also be di%ided into categories similar to the full
electric applications*
The full electric and hybrid solutions can be presented in a component block chart as shown in 'ig*
*-*

'ig* *-* Schematic of a hybrid and full electric propulsion system* Depending on the
application& the units presented by a dashed line are optional* #ightning bolts
denote electric coupling and cogwheels mechanical coupling*
2n general& a full electric or hybrid propulsion system usually includes a combination of an energy
source or a storage& a freAuency con%erter& at least one electrical machine and a load* Depending
on the system& other parts are optional* Hithout the 2D1 and generator blocks& the schematic
represents a full electric system with or without a gear coupling* 2f the 2D1 and generator blocks
are included in the system& we ha%e a series hybrid topology* Again& if the generator block is
omitted& the system represents a parallel hybrid solution* 2t is also possible to ha%e a combined
series parallel system containing all of the blocks of 'ig* *-*
A %ast ma3ority of modern efficient dri%e systems are three<phase systems with freAuency
con%erters implemented by ad%anced control methods* Modern permanent magnet electrical
machines in this power range can reach up to ,> P efficiency ratings in constant operation* The
energy source& usually a battery or a fuel cell& pro%ides the power for the electric motor through a
freAuency con%erter* Some systems are designed with a super capacitor in parallel connection with
the battery to absorb or discharge fast and powerful energy pulses*
(+


Modern lithium ion batteries ha%e reported efficiencies up to ,= P in the discharge mode& and the
same efficiency can also be applied to the recharge operation (Shidore et al*& ())-)* The efficiency
ratings of freAuency con%erters are also high& usually up to ,- P and abo%e in static operation
(H18& ()),)* (Rooks& ())+)* The use of a freAuency con%erter is mandatory since it enables a fast
response and controllability of the electrical machine in %arying load conditions and makes it
possible to ha%e different operation modes depending on the application*
The efficiency of an electrical machine is considerably higher compared with internal combustion
engines* A modern 2D1 unit (diesel) can reach up to +) P efficiency at best& but this only applies
to a %ery limited range of speed and load* 9igher efficiencies can be achie%ed& but only with %ery
large engine siGes and sophisticated auxiliary systems* A study (TR7& ())>) of the passenger car
fuel economy indicates that energy flows in a standard<siGed passenger car are di%ided as shown in
Table **
Table* ** 1nergy loss of an 2D1<powered medium siGed passenger car (TR7& ())>).
#oss type
#oss percentage JPL of total ))
$rban dri%ing 9ighway dri%ing
1ngine >( >,
Standby - +
Dri%e line > =
7raking > (
Rolling + -
Aerodynamic loss (Air friction) !
Accessories ( (

According to Table *& only - P of the total energy is used in the actual mo%ing process
(aerodynamics& rolling) in urban dri%ing conditions and " P in highway dri%ing* The 2D1 units
consume most of the energy a%ailable* The a%erage 2D1 efficiency rating in constant operation&
including standby losses& is thus ( P in urban dri%ing and (- P in highway operation*
Hhen comparing the efficiency of an electric traction system with that of an 2D1 unit (Table *)&
the difference is clear* 2n order to achie%e the total efficiency of the system& the indi%idual
efficiencies of the dri%etrain ha%e to be multiplied with each other*
Hhen considering constant operation and neglecting additional dri%e line mechanical losses& the
efficiency in the motoring mode for a full electric dri%e system with efficiencies of )*,= for the
motor& )*," for the freAuency con%erter and )*,= for the battery package would be
p
MotorIng
= u.9S u.98 u.9S = u.88
The total efficiency decreases significantly e%en though the indi%idual efficiencies are high* 2f the
system has a simple reduction gear& the system efficiency is slightly lower* A simple reduction
gear may ha%e an efficiency up to )*,"* According to the %alues gi%en in Table *& the efficiency
of a transmission in an a%erage passenger car is around )*,+* 7y introducing this to the pre%ious
eAuation& the total efficiency reduces to
(+


Modern lithium ion batteries ha%e reported efficiencies up to ,= P in the discharge mode& and the
same efficiency can also be applied to the recharge operation (Shidore et al*& ())-)* The efficiency
ratings of freAuency con%erters are also high& usually up to ,- P and abo%e in static operation
(H18& ()),)* (Rooks& ())+)* The use of a freAuency con%erter is mandatory since it enables a fast
response and controllability of the electrical machine in %arying load conditions and makes it
possible to ha%e different operation modes depending on the application*
The efficiency of an electrical machine is considerably higher compared with internal combustion
engines* A modern 2D1 unit (diesel) can reach up to +) P efficiency at best& but this only applies
to a %ery limited range of speed and load* 9igher efficiencies can be achie%ed& but only with %ery
large engine siGes and sophisticated auxiliary systems* A study (TR7& ())>) of the passenger car
fuel economy indicates that energy flows in a standard<siGed passenger car are di%ided as shown in
Table **
Table* ** 1nergy loss of an 2D1<powered medium siGed passenger car (TR7& ())>).
#oss type
#oss percentage JPL of total ))
$rban dri%ing 9ighway dri%ing
1ngine >( >,
Standby - +
Dri%e line > =
7raking > (
Rolling + -
Aerodynamic loss (Air friction) !
Accessories ( (

According to Table *& only - P of the total energy is used in the actual mo%ing process
(aerodynamics& rolling) in urban dri%ing conditions and " P in highway dri%ing* The 2D1 units
consume most of the energy a%ailable* The a%erage 2D1 efficiency rating in constant operation&
including standby losses& is thus ( P in urban dri%ing and (- P in highway operation*
Hhen comparing the efficiency of an electric traction system with that of an 2D1 unit (Table *)&
the difference is clear* 2n order to achie%e the total efficiency of the system& the indi%idual
efficiencies of the dri%etrain ha%e to be multiplied with each other*
Hhen considering constant operation and neglecting additional dri%e line mechanical losses& the
efficiency in the motoring mode for a full electric dri%e system with efficiencies of )*,= for the
motor& )*," for the freAuency con%erter and )*,= for the battery package would be
p
MotorIng
= u.9S u.98 u.9S = u.88
The total efficiency decreases significantly e%en though the indi%idual efficiencies are high* 2f the
system has a simple reduction gear& the system efficiency is slightly lower* A simple reduction
gear may ha%e an efficiency up to )*,"* According to the %alues gi%en in Table *& the efficiency
of a transmission in an a%erage passenger car is around )*,+* 7y introducing this to the pre%ious
eAuation& the total efficiency reduces to
(=


p
MotorIng,gcar
= u.9S u.98 u.9S u.94 = u.8S
DonseAuently& the efficiency of a hybrid system is significantly lower because of the 2D1 unit in
the system* 2n some applications& the parallel hybrid solution makes it possible to dri%e the system
without an 2D1 similarly as in series hybrid applications* 9ere& the efficiencies are theoretically
eAual to full electric systems* 2n series hybrids& the 2D1 is run at a constant load when power
generation is needed* 2n this situation& the 2D1 unit achie%es its maximum efficiency& assumed here
+) P* Thus& the efficiency in a series hybrid dri%e system with and without a reduction gear&
considering the efficiencies of )*,= for the generator& )*," for the freAuency con%erter& )*,= for the
battery package& )*,= for the motor& )*!+= for the 2D1 unit and )*,+ for the transmission& would be
p
MotorIng
= u.S4S u.9S u.98 u.9S u.98 u.9S = u.28
p
MotorIng,gcar
= u.S4S u.9S u.98 u.9S u.98 u.9S u.94 = u.27
.arallel hybrids can also be operated this way& but usually& the 2D1 load %aries according to the
rotating speed& and thus& the 2D1 efficiency is lower than at a constant load* According to the
efficiency e%aluation& the full electric application seems to be the most energy efficient solution
e%en though only +) P of the electric production efficiency and ,> P of the transmission
efficiency are taken into account* Hhen comparing the efficiencies& it has to be noted that the
production of electric energy in power plants is not lossless* The benefit in full electric
applications is the localiGation of emissions& which are easier to handle than pollution that is
scattered*
The introduction of a reduction gear or a transmission in the dri%e system should also be carefully
inspected* The use of a gear depends more on the total cost of the dri%e system than on the total
efficiency or machine siGe in the dri%e system* The gear also makes a difference in the machine
type selection* ;aturally& actual systems are much more complicated by natureU howe%er& the
examples gi%e an idea of differences in the traction systems*

!.., -istory of electric traction
As an introduction to hybrid and full electric applications in mo%ing machinery& a brief glance is
taken at the history of these applications* The history of hybrid electric and electric systems is
actually Auite long* The spur for the de%elopment of these systems is the need to tra%el fast and
efficiently* The dri%ing force was @ of course @ the automoti%e industry*
At the beginning of the history of automoti%e technology& the electric& hybrid electric and internal
combustion engine technologies were competing on eAual terms with each other* The 5oltaic cell
was de%eloped by Alessandro 5olta in "))& which was the stepping stone for the mo%ing
electrical machinery as it enabled a portable energy storage* The first electric %ehicles ha%e been
reported to be built around "!)s* After that& the de%elopment was so rapid that electric %ehicles
were preferred in taxi and pri%ate cars (#arminie& ())!)* 1%en though the internal combustion
engine was designed in ")> by 2* Ri%aG& the production of petrol started only fifty years later* The
(=


p
MotorIng,gcar
= u.9S u.98 u.9S u.94 = u.8S
DonseAuently& the efficiency of a hybrid system is significantly lower because of the 2D1 unit in
the system* 2n some applications& the parallel hybrid solution makes it possible to dri%e the system
without an 2D1 similarly as in series hybrid applications* 9ere& the efficiencies are theoretically
eAual to full electric systems* 2n series hybrids& the 2D1 is run at a constant load when power
generation is needed* 2n this situation& the 2D1 unit achie%es its maximum efficiency& assumed here
+) P* Thus& the efficiency in a series hybrid dri%e system with and without a reduction gear&
considering the efficiencies of )*,= for the generator& )*," for the freAuency con%erter& )*,= for the
battery package& )*,= for the motor& )*!+= for the 2D1 unit and )*,+ for the transmission& would be
p
MotorIng
= u.S4S u.9S u.98 u.9S u.98 u.9S = u.28
p
MotorIng,gcar
= u.S4S u.9S u.98 u.9S u.98 u.9S u.94 = u.27
.arallel hybrids can also be operated this way& but usually& the 2D1 load %aries according to the
rotating speed& and thus& the 2D1 efficiency is lower than at a constant load* According to the
efficiency e%aluation& the full electric application seems to be the most energy efficient solution
e%en though only +) P of the electric production efficiency and ,> P of the transmission
efficiency are taken into account* Hhen comparing the efficiencies& it has to be noted that the
production of electric energy in power plants is not lossless* The benefit in full electric
applications is the localiGation of emissions& which are easier to handle than pollution that is
scattered*
The introduction of a reduction gear or a transmission in the dri%e system should also be carefully
inspected* The use of a gear depends more on the total cost of the dri%e system than on the total
efficiency or machine siGe in the dri%e system* The gear also makes a difference in the machine
type selection* ;aturally& actual systems are much more complicated by natureU howe%er& the
examples gi%e an idea of differences in the traction systems*

!.., -istory of electric traction
As an introduction to hybrid and full electric applications in mo%ing machinery& a brief glance is
taken at the history of these applications* The history of hybrid electric and electric systems is
actually Auite long* The spur for the de%elopment of these systems is the need to tra%el fast and
efficiently* The dri%ing force was @ of course @ the automoti%e industry*
At the beginning of the history of automoti%e technology& the electric& hybrid electric and internal
combustion engine technologies were competing on eAual terms with each other* The 5oltaic cell
was de%eloped by Alessandro 5olta in "))& which was the stepping stone for the mo%ing
electrical machinery as it enabled a portable energy storage* The first electric %ehicles ha%e been
reported to be built around "!)s* After that& the de%elopment was so rapid that electric %ehicles
were preferred in taxi and pri%ate cars (#arminie& ())!)* 1%en though the internal combustion
engine was designed in ")> by 2* Ri%aG& the production of petrol started only fifty years later* The
(>


first petrol<powered 2D1 was used in an automobile by Carl 7enG in ""=& and he is usually
considered the father of the petrol<powered automobile* A short while later& in ,)& one of the
first hybrid electric cars was built by 'erdinand .orsche& the later creator of 5olkswagen 7eetle
(1ckermann& ()))*
The de%elopment of the hybrid electric and electric technology stopped because of the in%ention of
the electric starter motor at the beginning of the ()
th
century and the in%ention of car mass
production by 9enry 'ord* The electric and hybrid electric systems could not compete with the
2D1s because of the slow progress in the energy storage systems* :nly after the oil crisis& at the
beginning of the ,-)s& the interest in hybrid and electric technology has been arising because of
the increasing cost of oil& and nowadays& also because of en%ironmental reasons*
There are many hybrid and full electric solutions that often remain unnoticed* ;owadays& most of
the railway locomoti%es ha%e either hybrid or full electric propulsion systems* #ocal trams are
usually full electric as are underground trains* 2ntercity trains without an electrical line supply are
usually eAuipped with series hybrid technology& which applies a diesel generator that pro%ides
power for the wheel motors of the railway engine& which are located in the engine or railcar axles*
Modern %essels are also more and more often eAuipped with similar solutions& the difference being
that the electric motor dri%es the propeller*
The need for high energy efficiency and control o%er emissions has led to mushrooming of hybrid
and full electric solutions in industrial and automoti%e %ehicles* ;owadays& almost e%ery leading
manufacturer of automoti%e and industrial solutions has introduced hybrid or full electric models
besides the traditional solely internal combustion engine (2D1) powered units* The alternati%e
topologies are numerous*
So far& no branch of de%elopment in hybrid and full electric systems has pro%ided superior
performance o%er others* The push towards cleaner propulsion ine%itably leads to petrol<free
operation& but substitute fuels are emerging strongly* 7io diesel& ethanol& hydrogen and natural
gases are ri%al substitutes for petrol* The idea of full electric propulsion is thus a faraway dream*
9ybrid and full electric systems will de%elop together& and it seems that no propulsion system will
take o%er the transportation market similarly as 2D1s ha%e done*

!., "ynchronous reluctance#tor.ue#assisted PM machine
As presented in Section *(*& a permanent magnet machine belongs to the synchronous machine
category of electrical machines* This means that the rotor follows the stator field with a
synchroniGed speed without sustained slip* The machine excitation is inherent because of the
permanent magnets in the rotor* $sually& .M machines are inner rotor machines& but external rotor
structures are also found* .ermanent magnet machines can be di%ided into two different
categoriesB salient and non<salient pole machines* The difference is illustrated in 'ig* *"*
(>


first petrol<powered 2D1 was used in an automobile by Carl 7enG in ""=& and he is usually
considered the father of the petrol<powered automobile* A short while later& in ,)& one of the
first hybrid electric cars was built by 'erdinand .orsche& the later creator of 5olkswagen 7eetle
(1ckermann& ()))*
The de%elopment of the hybrid electric and electric technology stopped because of the in%ention of
the electric starter motor at the beginning of the ()
th
century and the in%ention of car mass
production by 9enry 'ord* The electric and hybrid electric systems could not compete with the
2D1s because of the slow progress in the energy storage systems* :nly after the oil crisis& at the
beginning of the ,-)s& the interest in hybrid and electric technology has been arising because of
the increasing cost of oil& and nowadays& also because of en%ironmental reasons*
There are many hybrid and full electric solutions that often remain unnoticed* ;owadays& most of
the railway locomoti%es ha%e either hybrid or full electric propulsion systems* #ocal trams are
usually full electric as are underground trains* 2ntercity trains without an electrical line supply are
usually eAuipped with series hybrid technology& which applies a diesel generator that pro%ides
power for the wheel motors of the railway engine& which are located in the engine or railcar axles*
Modern %essels are also more and more often eAuipped with similar solutions& the difference being
that the electric motor dri%es the propeller*
The need for high energy efficiency and control o%er emissions has led to mushrooming of hybrid
and full electric solutions in industrial and automoti%e %ehicles* ;owadays& almost e%ery leading
manufacturer of automoti%e and industrial solutions has introduced hybrid or full electric models
besides the traditional solely internal combustion engine (2D1) powered units* The alternati%e
topologies are numerous*
So far& no branch of de%elopment in hybrid and full electric systems has pro%ided superior
performance o%er others* The push towards cleaner propulsion ine%itably leads to petrol<free
operation& but substitute fuels are emerging strongly* 7io diesel& ethanol& hydrogen and natural
gases are ri%al substitutes for petrol* The idea of full electric propulsion is thus a faraway dream*
9ybrid and full electric systems will de%elop together& and it seems that no propulsion system will
take o%er the transportation market similarly as 2D1s ha%e done*

!., "ynchronous reluctance#tor.ue#assisted PM machine
As presented in Section *(*& a permanent magnet machine belongs to the synchronous machine
category of electrical machines* This means that the rotor follows the stator field with a
synchroniGed speed without sustained slip* The machine excitation is inherent because of the
permanent magnets in the rotor* $sually& .M machines are inner rotor machines& but external rotor
structures are also found* .ermanent magnet machines can be di%ided into two different
categoriesB salient and non<salient pole machines* The difference is illustrated in 'ig* *"*
(-



'ig* *"* a) ;on<salient and b) salient pole (embedded magnets) .M structure
The non<salient pole design is more efficient in the permanent magnet usage* The non<salient pole
composition usually means that the magnets are placed in the surface position in the rotor
geometry as shown in 'ig* *"a* The salient pole design ('ig* *"b) with embedded magnets loses
some of the .M energy as a result of the short<circuiting of the .M poles in the rotor* The salient
pole structure is& on the other hand& mechanically more stable* The non<salient pole design is
usually more beneficial in low<speed applications& but the magnet poles can be retained with fibre
or steel banding*

!.,.! *undamentals of electrical machine tor.ue production
The high torAue output capability of permanent magnet machines is a result of the long eAui%alent
air gap in the pole direction& which is due to the low permeability of the .M material* Depending
on the dimensioning of the machine& the pull<out torAue can& in practice& be e%en !@+ times the
nominal torAue* The low inductance may also cause problems in the machine control in freAuency<
con%erter<operated systems* 'rom the industrial point of %iew& the .M technology is Auite new*
The full potential of the .M technology could not ha%e been reached without modern 287T
switches& which were first de%eloped in the late ,")s* After that& the .M technology gained
ground in ser%o dri%e applications* This technology has also enabled the use of the .M machine
technology in traction applications* The rotor in salient pole .M structures is normally of
embedded magnet type*
The principle of the pole arrangement comes usually from the four<pole SynRM type& as presented
in 'ig *,*
(-



'ig* *"* a) ;on<salient and b) salient pole (embedded magnets) .M structure
The non<salient pole design is more efficient in the permanent magnet usage* The non<salient pole
composition usually means that the magnets are placed in the surface position in the rotor
geometry as shown in 'ig* *"a* The salient pole design ('ig* *"b) with embedded magnets loses
some of the .M energy as a result of the short<circuiting of the .M poles in the rotor* The salient
pole structure is& on the other hand& mechanically more stable* The non<salient pole design is
usually more beneficial in low<speed applications& but the magnet poles can be retained with fibre
or steel banding*

!.,.! *undamentals of electrical machine tor.ue production
The high torAue output capability of permanent magnet machines is a result of the long eAui%alent
air gap in the pole direction& which is due to the low permeability of the .M material* Depending
on the dimensioning of the machine& the pull<out torAue can& in practice& be e%en !@+ times the
nominal torAue* The low inductance may also cause problems in the machine control in freAuency<
con%erter<operated systems* 'rom the industrial point of %iew& the .M technology is Auite new*
The full potential of the .M technology could not ha%e been reached without modern 287T
switches& which were first de%eloped in the late ,")s* After that& the .M technology gained
ground in ser%o dri%e applications* This technology has also enabled the use of the .M machine
technology in traction applications* The rotor in salient pole .M structures is normally of
embedded magnet type*
The principle of the pole arrangement comes usually from the four<pole SynRM type& as presented
in 'ig *,*
("



'ig* *,* Synchronous reluctance<assisted permanent magnet topology with an embedded
magnet structure* The notation d refers to the direct axis and A to the Auadrature
axis*
'ig *, presents a typical starting point of the machine performance e%aluation* The torAue output
of an electrical machine can be estimated by a two<axis model& which obser%es the machine
beha%iour in the direct and Auadrature components* The direct<axis component is aligned with the
direction of the permanent magnet main flux path*
The principle of saliency is to create flux barriers with embedded air gaps and to enhance flux
orientation with permanent magnet patterns to achie%e a maximum reluctance difference between
the direct and Auadrature axes* The flux is forced to tra%el in the A<axis direction& as will be shown
later in this work* 2n a permanent magnet construction& the reluctance of the direct axis is greater
compared with the Auadrature<axis reluctance& because the .M material is magnetically close to
air* Thus& the direct<axis inductance is low compared with the Auadrature<axis inductance* The
deduction is Auite simple because the in%erse product of reluctance& the magnetic conducti%ity
called permeance& has the same unit as the inductance* The effect of the difference in inductances
can be e%aluated by a per unit %alue examination* The torAue eAuation to be used is the load angle
eAuation for synchronous machines& based on the two<axis theorem
I = Sp
uE

s
2
I
d
sino + Spu
2
I
d
- I
q

s
2
2I
d
I
q
sin2o *(
where
< L
d&A
are the synchronous inductances in the direct and Auadrature axes&
respecti%ely&
< U is the RMS %alue of the supply phase %oltage&
< E is the RMS %alue of the back electromoti%e phase %oltage and
< o is the load angle between phasors U and E*
'ig* *) shows a characteristic load angle eAuation cur%e for the inductance ratio L
d
EL
A
S )*=* The
supply %oltage U& the back electromoti%e force (1M') E and the rotating speed c
s
are set to p*u*
("



'ig* *,* Synchronous reluctance<assisted permanent magnet topology with an embedded
magnet structure* The notation d refers to the direct axis and A to the Auadrature
axis*
'ig *, presents a typical starting point of the machine performance e%aluation* The torAue output
of an electrical machine can be estimated by a two<axis model& which obser%es the machine
beha%iour in the direct and Auadrature components* The direct<axis component is aligned with the
direction of the permanent magnet main flux path*
The principle of saliency is to create flux barriers with embedded air gaps and to enhance flux
orientation with permanent magnet patterns to achie%e a maximum reluctance difference between
the direct and Auadrature axes* The flux is forced to tra%el in the A<axis direction& as will be shown
later in this work* 2n a permanent magnet construction& the reluctance of the direct axis is greater
compared with the Auadrature<axis reluctance& because the .M material is magnetically close to
air* Thus& the direct<axis inductance is low compared with the Auadrature<axis inductance* The
deduction is Auite simple because the in%erse product of reluctance& the magnetic conducti%ity
called permeance& has the same unit as the inductance* The effect of the difference in inductances
can be e%aluated by a per unit %alue examination* The torAue eAuation to be used is the load angle
eAuation for synchronous machines& based on the two<axis theorem
I = Sp
uE

s
2
I
d
sino + Spu
2
I
d
- I
q

s
2
2I
d
I
q
sin2o *(
where
< L
d&A
are the synchronous inductances in the direct and Auadrature axes&
respecti%ely&
< U is the RMS %alue of the supply phase %oltage&
< E is the RMS %alue of the back electromoti%e phase %oltage and
< o is the load angle between phasors U and E*
'ig* *) shows a characteristic load angle eAuation cur%e for the inductance ratio L
d
EL
A
S )*=* The
supply %oltage U& the back electromoti%e force (1M') E and the rotating speed c
s
are set to p*u*
(,



'ig* *)* TorAue production according to the load angle eAuation with L
d
EL
A
S )*= and
scaled to the peak torAue of p*u* The red cur%e illustrates the synchronous
torAue part and the blue cur%e the reluctance part* The black cur%e is the
superposition of these two* The low inductance ratio (W) results in breako%er
%alues abo%e ,) degrees*
7y considering a nominal operation at E S U S p*u* and c
s
S p*u*& the proportion of reluctance
torAue T
rel
to total torAue T
tot
can be reducd to a form

I =
sin2o
y
rcI
sino + sin2o

*!
where
y
rcI
=
2
I
d
I
q
- 1
.
*+

The eAuation also applies to torAue production* The percentage %alue of the reluctance torAue
compared with the total torAue T at different load angles is presented in 'ig* ** The normal
operation of salient pole .M machines is found in the upper left corner because a low inductance
ratio (W) results in breako%er torAue %alues at load angles abo%e ,) electrical degrees*
(,



'ig* *)* TorAue production according to the load angle eAuation with L
d
EL
A
S )*= and
scaled to the peak torAue of p*u* The red cur%e illustrates the synchronous
torAue part and the blue cur%e the reluctance part* The black cur%e is the
superposition of these two* The low inductance ratio (W) results in breako%er
%alues abo%e ,) degrees*
7y considering a nominal operation at E S U S p*u* and c
s
S p*u*& the proportion of reluctance
torAue T
rel
to total torAue T
tot
can be reducd to a form

I =
sin2o
y
rcI
sino + sin2o

*!
where
y
rcI
=
2
I
d
I
q
- 1
.
*+

The eAuation also applies to torAue production* The percentage %alue of the reluctance torAue
compared with the total torAue T at different load angles is presented in 'ig* ** The normal
operation of salient pole .M machines is found in the upper left corner because a low inductance
ratio (W) results in breako%er torAue %alues at load angles abo%e ,) electrical degrees*
!)



'ig* ** Reluctance torAue T
rel
to the total torAue T
tot
according to the load angle eAuation
for synchronous machines* The horiGontal axis refers to the inductance ratio
L
d
EL
A
and the %ertical axis to the load angle in degrees* The blue contour line
%alues represent the T
rel
ET
tot
with E S U S per unit*
Hith the set boundaries E S U S per unit& the ratio of the reluctance torAue to the total torAue
produced at a certain load angle depends only on the inductance ratio between the direct< and
Auadrature<axis inductances* Hith further modification of 1A* *+& the inductance ratio y
cI
is let to
grow to infinity* Thus&
y
cI
= lim
L
d
L
q
-0
2
I
d
I
q
- 1
= -2
the maximum proportion of reluctance torAue is achie%ed in the point where coso reaches the
minimum& which is @ at o S X* 1Auation *! transforms to
I
%
=
1
1 -[
1
cos o

which yields
I
rcI
=
1
1 -[
1
cos

= u.S
!)



'ig* ** Reluctance torAue T
rel
to the total torAue T
tot
according to the load angle eAuation
for synchronous machines* The horiGontal axis refers to the inductance ratio
L
d
EL
A
and the %ertical axis to the load angle in degrees* The blue contour line
%alues represent the T
rel
ET
tot
with E S U S per unit*
Hith the set boundaries E S U S per unit& the ratio of the reluctance torAue to the total torAue
produced at a certain load angle depends only on the inductance ratio between the direct< and
Auadrature<axis inductances* Hith further modification of 1A* *+& the inductance ratio y
cI
is let to
grow to infinity* Thus&
y
cI
= lim
L
d
L
q
-0
2
I
d
I
q
- 1
= -2
the maximum proportion of reluctance torAue is achie%ed in the point where coso reaches the
minimum& which is @ at o S X* 1Auation *! transforms to
I
%
=
1
1 -[
1
cos o

which yields
I
rcI
=
1
1 -[
1
cos

= u.S
!


Relying on the theory of the load angle eAuation *( for electrical machines& the proportion of the
maximum reluctance torAue of the total torAue T at the gi%en boundaries of E S U S p*u* is =) P
of the total torAue output T*

!.,. Pull#out tor.ue ma/imi0ation
9ybrid and full electric propulsion systems reAuire high pull<out torAue capability& which is
usually considerably higher than the a%erage rating at which the machine is usually thermally
dimensioned* The analytical e%aluation of machine performance pro%ides numerous %ariables to
be manipulated in order to affect the machine torAue output* 1A* *( gi%es a good start for the
maximiGation of the machine pull<out torAue*
The pull<out torAue production can be di%ided into two categories* According to 1A* *(& the torAue
production depends on the machine inductances and the supply*
The torAue in a salient pole structure consists of two parts* The first part in 1A* *( is the
synchronous part& which applies to all synchronous machines* The maximum torAue I
`
for the first
part is
I
`
=
uE
I
d
=
u

PM
I
d

which shows that the maximum torAue can be increased by raising the %oltage to speed ratio UEc
s
Q

s
or by decreasing the direct<axis synchronous inductance L
d
* 2n traction applications where the
%oltage reser%e is high at the lowest speeds& it is easier to increase the %oltage to speed ratio&
within the limits of saturation& which makes the minimiGation of the direct<axis synchronous
inductance less important and allows a potential increase in the field weakening performance*
The second term forms the reluctance torAue part& which is a special feature for salient pole
machines* The reluctance part responds to the excess %oltage because of the second power term*
The allowable amount of excess %oltage has to be confirmed by finite element methods because of
the saturation of the inductances* #oose dimensioning of the iron circuit with the excess supply
%oltage assists in pull<out torAue generation& although it increases the machine dimensions and
thereby the weight*
To build up the machine reluctance torAue& the machine dimensions can be ad3usted to fa%our
reluctance torAue generation* The analytical determination of synchronous inductances facilitates
the design task of increasing the difference between the direct< and Auadrature<axis magnetic
conducti%ity* According to (5ogt& ,"!) the eAuation
I
m
=
m
2
2
n
p
0
1
2p
4
n

p
o
l'(k
ws1
N
s
)
2
*=
where
!


Relying on the theory of the load angle eAuation *( for electrical machines& the proportion of the
maximum reluctance torAue of the total torAue T at the gi%en boundaries of E S U S p*u* is =) P
of the total torAue output T*

!.,. Pull#out tor.ue ma/imi0ation
9ybrid and full electric propulsion systems reAuire high pull<out torAue capability& which is
usually considerably higher than the a%erage rating at which the machine is usually thermally
dimensioned* The analytical e%aluation of machine performance pro%ides numerous %ariables to
be manipulated in order to affect the machine torAue output* 1A* *( gi%es a good start for the
maximiGation of the machine pull<out torAue*
The pull<out torAue production can be di%ided into two categories* According to 1A* *(& the torAue
production depends on the machine inductances and the supply*
The torAue in a salient pole structure consists of two parts* The first part in 1A* *( is the
synchronous part& which applies to all synchronous machines* The maximum torAue I
`
for the first
part is
I
`
=
uE
I
d
=
u

PM
I
d

which shows that the maximum torAue can be increased by raising the %oltage to speed ratio UEc
s
Q

s
or by decreasing the direct<axis synchronous inductance L
d
* 2n traction applications where the
%oltage reser%e is high at the lowest speeds& it is easier to increase the %oltage to speed ratio&
within the limits of saturation& which makes the minimiGation of the direct<axis synchronous
inductance less important and allows a potential increase in the field weakening performance*
The second term forms the reluctance torAue part& which is a special feature for salient pole
machines* The reluctance part responds to the excess %oltage because of the second power term*
The allowable amount of excess %oltage has to be confirmed by finite element methods because of
the saturation of the inductances* #oose dimensioning of the iron circuit with the excess supply
%oltage assists in pull<out torAue generation& although it increases the machine dimensions and
thereby the weight*
To build up the machine reluctance torAue& the machine dimensions can be ad3usted to fa%our
reluctance torAue generation* The analytical determination of synchronous inductances facilitates
the design task of increasing the difference between the direct< and Auadrature<axis magnetic
conducti%ity* According to (5ogt& ,"!) the eAuation
I
m
=
m
2
2
n
p
0
1
2p
4
n

p
o
l'(k
ws1
N
s
)
2
*=
where
!(


< m is the phase number&
< u
)
is the permeability of %acuum&
< p is the pole pair number&
< :
p
is the pole pitch&
< o is the air gap length&
< k
ws
is the fundamental winding factor&
< N
s
is the winding turn number of a phase and
< lF is the effecti%e length of the machine stack&
shows that the inductance is directly dependent on the second power of winding turns N
s
(
& and
in%ersely proportional to the eAui%alent air gap diameter and the second power of the pole pair
number& since

p
=

s
2p
*>
where D
s
is the stator inner diameter* 7y comparing 1As* *( and *=& the method to maximiGe the
torAue output of a .M synchronous machine is easier to understand* ;e%ertheless& finding an
optimal solution is a more complicated task* 1A* *= is for the magnetiGing inductance only and
does not include the leakage term* The direct< and Auadrature<axis synchronous inductances in 1A*
*( consist of the magnetiGing inductance and the leakage term according to the eAuation
I
d,q
= I
md,mq
+ I
o
*-
in which L
md&mA
refers to the direct< or Auadrature<axis magnetiGing inductance and L
Y
to the
leakage component* The leakage affects the inductance ratio between the direct< and Auadrature<
axis components* A lower leakage component results in a higher inductance ratio& which is more
beneficial both to the reluctance torAue production and the synchronous torAue production*
9owe%er& leakage can be useful in extending the machine operating speed range to the field
weakening region in the .M machine case*
The direct<axis synchronous inductance of a .M machine is inherently low because of the long
effecti%e air gap in the direct<axis direction& caused by the low<permeability magnet material* To
obtain a lower inductance ratio& the use of tall and narrow magnets in the flux path is preferable to
a thin and wide magnet shape*
7ased on 1A* *(& the decrease in the direct<axis synchronous inductance increases the breako%er
power output of the machine* The use of the reluctance part reAuires different actions* According
to 1A* *(& the inductance of the A<axis has to be maximiGed in order to increase the reluctance
torAue component* The inductance in the A<axis direction is increased by adding iron to the A<axis
flux path* This is the reason why embedded magnet topologies ha%e inherent saliency*
The abo%e<described idea can be extended by obser%ing 1A* *=* The eAui%alent air gap %alue can
be ad3usted between the direct and Auadrature axis* The inductance ratio would benefit from a
smaller eAui%alent air gap in the A<axis direction* The direct<axis inductance can also be kept low
!(


< m is the phase number&
< u
)
is the permeability of %acuum&
< p is the pole pair number&
< :
p
is the pole pitch&
< o is the air gap length&
< k
ws
is the fundamental winding factor&
< N
s
is the winding turn number of a phase and
< lF is the effecti%e length of the machine stack&
shows that the inductance is directly dependent on the second power of winding turns N
s
(
& and
in%ersely proportional to the eAui%alent air gap diameter and the second power of the pole pair
number& since

p
=

s
2p
*>
where D
s
is the stator inner diameter* 7y comparing 1As* *( and *=& the method to maximiGe the
torAue output of a .M synchronous machine is easier to understand* ;e%ertheless& finding an
optimal solution is a more complicated task* 1A* *= is for the magnetiGing inductance only and
does not include the leakage term* The direct< and Auadrature<axis synchronous inductances in 1A*
*( consist of the magnetiGing inductance and the leakage term according to the eAuation
I
d,q
= I
md,mq
+ I
o
*-
in which L
md&mA
refers to the direct< or Auadrature<axis magnetiGing inductance and L
Y
to the
leakage component* The leakage affects the inductance ratio between the direct< and Auadrature<
axis components* A lower leakage component results in a higher inductance ratio& which is more
beneficial both to the reluctance torAue production and the synchronous torAue production*
9owe%er& leakage can be useful in extending the machine operating speed range to the field
weakening region in the .M machine case*
The direct<axis synchronous inductance of a .M machine is inherently low because of the long
effecti%e air gap in the direct<axis direction& caused by the low<permeability magnet material* To
obtain a lower inductance ratio& the use of tall and narrow magnets in the flux path is preferable to
a thin and wide magnet shape*
7ased on 1A* *(& the decrease in the direct<axis synchronous inductance increases the breako%er
power output of the machine* The use of the reluctance part reAuires different actions* According
to 1A* *(& the inductance of the A<axis has to be maximiGed in order to increase the reluctance
torAue component* The inductance in the A<axis direction is increased by adding iron to the A<axis
flux path* This is the reason why embedded magnet topologies ha%e inherent saliency*
The abo%e<described idea can be extended by obser%ing 1A* *=* The eAui%alent air gap %alue can
be ad3usted between the direct and Auadrature axis* The inductance ratio would benefit from a
smaller eAui%alent air gap in the A<axis direction* The direct<axis inductance can also be kept low
!!


by a low phase turn number* This leads to longer machines& but because of the second power of the
winding turns& the significance of the winding turn number is greater than the machine length in
order to obtain low inductance %alues*
The pole pitch and the pole pair number are strongly connected to the physical siGe of the machine
along with the machine length* A low pole pitch %alue can be achie%ed by increasing the pole pair
number* This is good also for decreasing the machine siGe*
The problem with a small pole pitch and a large pole pair number is that the direct and Auadrature
axes are geometrically close to each other* 2f the pole dimensions are scaled down eAui%alently&
the inductance difference should be close to the original pole arrangement* The difference comes
from the air gap length* Hhen the pole dimensions are scaled down in multiple pole arrangements&
the air gap should also be decreased eAui%alently* The minimum %alue of the air gap is within a
few millimetres for mechanical and electric reasons* Too short an air gap may cause
manufacturing problems as well as increased friction losses and harmonic losses on the rotor
surface*
Depending on the design case& the degrees of freedom in changing the machine dimensions %ary*
The machine outer dimensions can be strictly specified& which causes restrictions to the machine
design* The operating speed range and machine performance reAuirements also affect the allowed
dimensions* 7ecause of the large number of %ariables and multiplier effects in the machine design&
the applied solutions ha%e to be weighed %ery carefully to achie%e an optimal machine structure
for the gi%en application*
Rough estimates on the machine performance can be obtained by swift analytical calculations& but
the feasibility of the decisions has to be tested in a finite element en%ironment*

!.,., +ffect of saliency and leakage on the drive performance
To obtain a broad field weakening speed range& the direct<axis inductance has to be designed large
enough* Hith a large direct<axis inductance& the field weakening reAuires less demagnetiGing I
d

current at the same operating speed compared with a small direct<axis inductance* The drawback
in large L
d
%alues is the reduction in the pull<out torAue as the pull<out torAue of a .M machine is
in%ersely proportional to the electrical machine direct<axis inductance as presented in 1A* *(* A
suitable amount of leakage helps achie%ing higher speeds in field weakening operations*
Depending on the application& the weighing between the power factor& the operating speed range
and the torAue output characteristics ha%e to be altered according to the desired beha%iour*
A phasor diagram presentation gi%es more information about the machine operation in different
conditions* This section concentrates on the effect of leakage and saliency on the electrical
machine performance*
An I
d
S ) control is good for rotor surface permanent magnet applications as it does not stress the
magnets with a counter<magnetiGing field* This control method also gi%es the highest torAue to
!!


by a low phase turn number* This leads to longer machines& but because of the second power of the
winding turns& the significance of the winding turn number is greater than the machine length in
order to obtain low inductance %alues*
The pole pitch and the pole pair number are strongly connected to the physical siGe of the machine
along with the machine length* A low pole pitch %alue can be achie%ed by increasing the pole pair
number* This is good also for decreasing the machine siGe*
The problem with a small pole pitch and a large pole pair number is that the direct and Auadrature
axes are geometrically close to each other* 2f the pole dimensions are scaled down eAui%alently&
the inductance difference should be close to the original pole arrangement* The difference comes
from the air gap length* Hhen the pole dimensions are scaled down in multiple pole arrangements&
the air gap should also be decreased eAui%alently* The minimum %alue of the air gap is within a
few millimetres for mechanical and electric reasons* Too short an air gap may cause
manufacturing problems as well as increased friction losses and harmonic losses on the rotor
surface*
Depending on the design case& the degrees of freedom in changing the machine dimensions %ary*
The machine outer dimensions can be strictly specified& which causes restrictions to the machine
design* The operating speed range and machine performance reAuirements also affect the allowed
dimensions* 7ecause of the large number of %ariables and multiplier effects in the machine design&
the applied solutions ha%e to be weighed %ery carefully to achie%e an optimal machine structure
for the gi%en application*
Rough estimates on the machine performance can be obtained by swift analytical calculations& but
the feasibility of the decisions has to be tested in a finite element en%ironment*

!.,., +ffect of saliency and leakage on the drive performance
To obtain a broad field weakening speed range& the direct<axis inductance has to be designed large
enough* Hith a large direct<axis inductance& the field weakening reAuires less demagnetiGing I
d

current at the same operating speed compared with a small direct<axis inductance* The drawback
in large L
d
%alues is the reduction in the pull<out torAue as the pull<out torAue of a .M machine is
in%ersely proportional to the electrical machine direct<axis inductance as presented in 1A* *(* A
suitable amount of leakage helps achie%ing higher speeds in field weakening operations*
Depending on the application& the weighing between the power factor& the operating speed range
and the torAue output characteristics ha%e to be altered according to the desired beha%iour*
A phasor diagram presentation gi%es more information about the machine operation in different
conditions* This section concentrates on the effect of leakage and saliency on the electrical
machine performance*
An I
d
S ) control is good for rotor surface permanent magnet applications as it does not stress the
magnets with a counter<magnetiGing field* This control method also gi%es the highest torAue to
!+


current ratio in non<salient pole machine designs* Hith salient pole structures& a small amount of
negati%e d<axis current is reAuired for the maximum torAue output*
'igures *( and *! represent salient and non<salient pole designs& respecti%ely* 2n addition& the
power factor cur%e is presented in the first Auadrant of the coordinate system& and the specific
power factor corresponding to the presented phasor arrangement is highlighted with a %ertical line
crossing the power factor cur%e* 7oth of the phasor diagrams are presented for motor operation&
where the rotation direction of the machine is counterclockwise* 7oth cases represent under<
magnetiGed machine operation& which results in a lagging power factor& that is& an inducti%e
connection to the supply* To show the difference between salient and non<salient pole machines&
the inductance %alues for 'igs* *( and *! are selected as L
d
S L
A
S )*( p*u* for the non<salient
pole design and L
d
S )*( p*u* L
A
S )*+ p*u* for the salient pole case* The permanent magnet flux and
the supply current I
s
are p*u* The leakage component is )* p*u in both cases*

'ig* *(* .hasor diagram presentation of the steady<state operation of a non<salient pole
machineU I
d
S )* The stator resistance is neglected* The power factor is presented
as a function of per unit current angle from ) to ") degrees*
<)*> <)*+ <)*( ) )*( )*+ )*> )*" *(
)
)*(
)*+
)*>
)*"

*(
*+
$
s
2
s
Z

.M

m
cos
S (*"M
S ,)*)M
!+


current ratio in non<salient pole machine designs* Hith salient pole structures& a small amount of
negati%e d<axis current is reAuired for the maximum torAue output*
'igures *( and *! represent salient and non<salient pole designs& respecti%ely* 2n addition& the
power factor cur%e is presented in the first Auadrant of the coordinate system& and the specific
power factor corresponding to the presented phasor arrangement is highlighted with a %ertical line
crossing the power factor cur%e* 7oth of the phasor diagrams are presented for motor operation&
where the rotation direction of the machine is counterclockwise* 7oth cases represent under<
magnetiGed machine operation& which results in a lagging power factor& that is& an inducti%e
connection to the supply* To show the difference between salient and non<salient pole machines&
the inductance %alues for 'igs* *( and *! are selected as L
d
S L
A
S )*( p*u* for the non<salient
pole design and L
d
S )*( p*u* L
A
S )*+ p*u* for the salient pole case* The permanent magnet flux and
the supply current I
s
are p*u* The leakage component is )* p*u in both cases*

'ig* *(* .hasor diagram presentation of the steady<state operation of a non<salient pole
machineU I
d
S )* The stator resistance is neglected* The power factor is presented
as a function of per unit current angle from ) to ") degrees*
<)*> <)*+ <)*( ) )*( )*+ )*> )*" *(
)
)*(
)*+
)*>
)*"

*(
*+
$
s
2
s
Z

.M

m
cos
S (*"M
S ,)*)M
!=



'ig* *!* .hasor diagram presentation of the steady<state operation of a salient pole
machineU I
d
S )* The stator resistance is neglected* The power factor is presented
as a function of per unit current angle from ) to ") degrees*
The diagrams are based on the torAue production eAuation
I~
sd
i
sq
-
sq
i
sd
*"
The current angle is denoted by o* The stator flux linkage is an integral of the supply %oltage and is
lagging the %oltage phasor by ,) degrees* The leakage term is lost in the energy transfer process
and results in a magnetiGing flux phasor that represents the magnetic tension affecting the rotor
surface* The pole follows the magnetiGing flux by an angle * The strength of the magnetic
coupling is presented with the direct< and Auadrature<axis flux linkages*
Modification of the current angle causes the phasors to tra%el along circular paths& indicated by a
dashed line& the radius of which is determined by the supply current and the direct< and Auadrature<
axis inductances* The increase in current produces larger constant operating circles& and %ice %ersa*
The phasors rotate along the circles in a counterclockwise direction as the current angle increases*
The non<salient pole design operates in the same way& but the phasor paths are elliptical as a result
of the inductance difference in the direct and Auadrature axis* This applies to constant inductance
terms& but in reality& the inductances depend on the magnetiGation state of the machine* ;on<
salient pole machine inductances are Auite steady until the machine reaches saturation and the
<)*> <)*+ <)*( ) )*( )*+ )*> )*" *(
)
)*(
)*+
)*>
)*"

*(
*+
$
s
2
s
Z

.M

m
cos
S *!M
S ,)*)M
!=



'ig* *!* .hasor diagram presentation of the steady<state operation of a salient pole
machineU I
d
S )* The stator resistance is neglected* The power factor is presented
as a function of per unit current angle from ) to ") degrees*
The diagrams are based on the torAue production eAuation
I~
sd
i
sq
-
sq
i
sd
*"
The current angle is denoted by o* The stator flux linkage is an integral of the supply %oltage and is
lagging the %oltage phasor by ,) degrees* The leakage term is lost in the energy transfer process
and results in a magnetiGing flux phasor that represents the magnetic tension affecting the rotor
surface* The pole follows the magnetiGing flux by an angle * The strength of the magnetic
coupling is presented with the direct< and Auadrature<axis flux linkages*
Modification of the current angle causes the phasors to tra%el along circular paths& indicated by a
dashed line& the radius of which is determined by the supply current and the direct< and Auadrature<
axis inductances* The increase in current produces larger constant operating circles& and %ice %ersa*
The phasors rotate along the circles in a counterclockwise direction as the current angle increases*
The non<salient pole design operates in the same way& but the phasor paths are elliptical as a result
of the inductance difference in the direct and Auadrature axis* This applies to constant inductance
terms& but in reality& the inductances depend on the magnetiGation state of the machine* ;on<
salient pole machine inductances are Auite steady until the machine reaches saturation and the
<)*> <)*+ <)*( ) )*( )*+ )*> )*" *(
)
)*(
)*+
)*>
)*"

*(
*+
$
s
2
s
Z

.M

m
cos
S *!M
S ,)*)M
!>


inductances decrease rapidly* The salient pole design is more complicated because the sheet metal
pro%ides alternati%e flux paths depending on the rotor saturation state* The machine inductances in
salient pole designs ha%e to be determined by a careful analysis based on a finite element method*
The purpose of the analysis is to assess the inductance beha%iour in order to predict the machine
performance in %arying load conditions* This was clearly shown in the design process of the
prototype machines*
'ield weakening is problematic in .M machine designs& especially in non<salient pole structures*
2n order to extend the machine operating speed range beyond the supply %oltage limits& the
machine .M excitation has to be weakened to enter the Vo%erspeedF region* At high temperatures&
the .M material BH cur%e is not linear at low B
r
%alues* 2f the .M machine is not designed for
field weakening purposes& it is possible that the countermagnetiGation shifts the operating point
below the permanent magnet demagnetiGation point resulting in irre%ersible demagnetiGation*
Traditional machine design usually aims to minimiGe the leakage& because it degrades the power
factor of the machine and decreases the load capability of the machine* This is true for industrial
machine designs with a fixed operating point* Since the propulsion system reAuires operation o%er
a large rotating speed range& the effect of the leakage term on the machine performance has to be
analysed through the speed range* The benefit from the leakage is found in the field weakening
region*
7y further manipulating 1A* *" to a form
I~
PM
i
sq
+ (I
d
- I
q
)i
sd
i
sq
*,
it is possible to in%estigate the effect of the saliency and the leakage component on the machine
performance characteristics* The results are presented in 'igs* *+ and *=*
!>


inductances decrease rapidly* The salient pole design is more complicated because the sheet metal
pro%ides alternati%e flux paths depending on the rotor saturation state* The machine inductances in
salient pole designs ha%e to be determined by a careful analysis based on a finite element method*
The purpose of the analysis is to assess the inductance beha%iour in order to predict the machine
performance in %arying load conditions* This was clearly shown in the design process of the
prototype machines*
'ield weakening is problematic in .M machine designs& especially in non<salient pole structures*
2n order to extend the machine operating speed range beyond the supply %oltage limits& the
machine .M excitation has to be weakened to enter the Vo%erspeedF region* At high temperatures&
the .M material BH cur%e is not linear at low B
r
%alues* 2f the .M machine is not designed for
field weakening purposes& it is possible that the countermagnetiGation shifts the operating point
below the permanent magnet demagnetiGation point resulting in irre%ersible demagnetiGation*
Traditional machine design usually aims to minimiGe the leakage& because it degrades the power
factor of the machine and decreases the load capability of the machine* This is true for industrial
machine designs with a fixed operating point* Since the propulsion system reAuires operation o%er
a large rotating speed range& the effect of the leakage term on the machine performance has to be
analysed through the speed range* The benefit from the leakage is found in the field weakening
region*
7y further manipulating 1A* *" to a form
I~
PM
i
sq
+ (I
d
- I
q
)i
sd
i
sq
*,
it is possible to in%estigate the effect of the saliency and the leakage component on the machine
performance characteristics* The results are presented in 'igs* *+ and *=*
!-





'ig* *+* ;on<salient pole machine (L
d
S )*( p*u* L
A
S)*( p*u*) torAue production as a
function of current angle* The dashed lines indicate the torAue production cur%e
with a p*u* current* The leakage components are )*)=& )*)& )*= and )*() p*u*
The coloured lines from top to bottom on the right side of the figure depict the
respecti%e armature fluxes in p*u* .M flux is p*u*
The armature flux cur%e in 'ig* *+ is in%ersely proportional to the rotating speed of the machine*
As the leakage component increases& the field weakening region deepens with the same
countermagnetiGing current* A suitable amount of leakage in the machine design seems to fa%our
the field weakening characteristics of a .M machine* The power factor peak also shifts further to
the field weakening region as the leakage component increases*
The saliency itself does not ha%e an influence on the field weakening range* This can be seen in
'ig* *=*

) () +) >) ") )) () +) >) ")
)
)*(
)*+
)*>
)*"

*(
*+
Durrent angle JML
.
e
r

u
n
i
t

s

a
n
d

T
!-





'ig* *+* ;on<salient pole machine (L
d
S )*( p*u* L
A
S)*( p*u*) torAue production as a
function of current angle* The dashed lines indicate the torAue production cur%e
with a p*u* current* The leakage components are )*)=& )*)& )*= and )*() p*u*
The coloured lines from top to bottom on the right side of the figure depict the
respecti%e armature fluxes in p*u* .M flux is p*u*
The armature flux cur%e in 'ig* *+ is in%ersely proportional to the rotating speed of the machine*
As the leakage component increases& the field weakening region deepens with the same
countermagnetiGing current* A suitable amount of leakage in the machine design seems to fa%our
the field weakening characteristics of a .M machine* The power factor peak also shifts further to
the field weakening region as the leakage component increases*
The saliency itself does not ha%e an influence on the field weakening range* This can be seen in
'ig* *=*

) () +) >) ") )) () +) >) ")
)
)*(
)*+
)*>
)*"

*(
*+
Durrent angle JML
.
e
r

u
n
i
t

s

a
n
d

T
!"



'ig* *=*Salient pole machine (L
d
S )*( p*u*) torAue production as a function of current
angle* The dashed lines are the torAue production cur%es with a p*u current*
The coloured lines indicate the corresponding armature fluxes in p*u* from top to
bottom with the L
A
%alues of )*(& )*!& )*+ and )*= per unit& respecti%ely* .M flux
is p*u*
The field weakening speed range in 'ig* *= is the same regardless of the A<axis inductance
component& but the torAue cur%es are different* The saliency is beneficial for the torAue output
capability in the field weakening region compared with non<salient pole designs* As with the
leakage component& the increase in saliency shifts the power factor peak deeper to the field
weakening region as presented in 'ig* *!*

!.,.4 Permanent magnet materials
.ermanent magnetism is a uniAue feature achie%able only in ferro< and ferrimagnetic solid
substances* These materials are called hard magnetic materials because the coerci%e force to
oppose any change in the magnetiGing direction is high compared with soft magnetic materials*
The scientific work of Hilliam 8ilbert& De Magnete (>)))& ga%e rise to the modern theories on
magnetism* 9owe%er& large<scale industrial use of different permanent magnet materials started
only in the late ,>)s* This is because the relationship between electricity and magnetism was
) () +) >) ") )) () +) >) ")
)
)*(
)*+
)*>
)*"

*(
*+
Durrent angle JML
.
e
r

u
n
i
t

s

a
n
d

T
!"



'ig* *=*Salient pole machine (L
d
S )*( p*u*) torAue production as a function of current
angle* The dashed lines are the torAue production cur%es with a p*u current*
The coloured lines indicate the corresponding armature fluxes in p*u* from top to
bottom with the L
A
%alues of )*(& )*!& )*+ and )*= per unit& respecti%ely* .M flux
is p*u*
The field weakening speed range in 'ig* *= is the same regardless of the A<axis inductance
component& but the torAue cur%es are different* The saliency is beneficial for the torAue output
capability in the field weakening region compared with non<salient pole designs* As with the
leakage component& the increase in saliency shifts the power factor peak deeper to the field
weakening region as presented in 'ig* *!*

!.,.4 Permanent magnet materials
.ermanent magnetism is a uniAue feature achie%able only in ferro< and ferrimagnetic solid
substances* These materials are called hard magnetic materials because the coerci%e force to
oppose any change in the magnetiGing direction is high compared with soft magnetic materials*
The scientific work of Hilliam 8ilbert& De Magnete (>)))& ga%e rise to the modern theories on
magnetism* 9owe%er& large<scale industrial use of different permanent magnet materials started
only in the late ,>)s* This is because the relationship between electricity and magnetism was
) () +) >) ") )) () +) >) ")
)
)*(
)*+
)*>
)*"

*(
*+
Durrent angle JML
.
e
r

u
n
i
t

s

a
n
d

T
!,


disco%ered by :ersted in "()& but it took more than )) years to produce the first industrial
permanent magnets materials in the early ()th century (1lliot& ,,!)*
Magnets are nowadays widely used in different e%eryday de%ices* Most of these applications are
so common that the utiliGation of permanent magnet materials remains unnoticed* .ermanent
magnets can be found for example in RD cars& refrigerator door seals& computer hard discs&
household blenders& washing machines& I<ray eAuipment& hybrid %ehicles& credit cards& compasses
and loudspeakers& 3ust to mention but a few*
.article accelerators can also be built with permanent magnets* The largest particle accelerator in
the world is located in D1R;& 'rance& which has " T superconducting niobium<titanium dipole
electromagnets in the #arge 9adron Dollider* The permanent magnets are needed to keep the
particle beam inside the accelerator (Russenschuck& ()))*
Modern transportation techniAues ha%e also ad%anced the use of permanent magnets* The
2ndutrack technology& which uses a 9allbach array for permanent magnet le%itation& is a
completely passi%e technology that operates on rare earth ;d'e7 magnets (.ost& ())))* .rototypes
ha%e been built& and ;ASA in the $SA has considered the 2ndutrack technology also for a rocket
launching application*
7esides their usefulness in commercial products& the main ad%antage of permanent magnets is
found in industrial energy con%ersion solutions*
The relati%e ability of a magnet material to produce work is measured by the maximum energy
product %alue J/Em
!
L* The magnetic material usually operates in the second Auadrant of the BH
cur%e& which illustrates the magnet capability to oppose demagnetiGation* 2n order to efficiently
use the magnet potential& the magnet operating point should be close to the maximum energy
product* Hith a suitable magnet material& the operating range of the material can also be extended
to the third Auadrant of the BH cur%e* A characteristic magnetiGation cur%e of a modern ;d'e7
magnet ;eorem +,!aE=,!a (;eorem Magnets) is presented in 'ig* *>*
The magnetism in ferromagnetic materials occurs when the small magnetic domains are aligned in
the same direction* The non<magnetiGed materials ha%e their elementary magnets arranged in
arbitrary directions* 2n hard magnetic materials& the orientation of the elements is difficult because
of the coerci%e force* 9owe%er& if the field is strong enough& the 7loch areas deform and the
elementary magnets orient according to the field*
!,


disco%ered by :ersted in "()& but it took more than )) years to produce the first industrial
permanent magnets materials in the early ()th century (1lliot& ,,!)*
Magnets are nowadays widely used in different e%eryday de%ices* Most of these applications are
so common that the utiliGation of permanent magnet materials remains unnoticed* .ermanent
magnets can be found for example in RD cars& refrigerator door seals& computer hard discs&
household blenders& washing machines& I<ray eAuipment& hybrid %ehicles& credit cards& compasses
and loudspeakers& 3ust to mention but a few*
.article accelerators can also be built with permanent magnets* The largest particle accelerator in
the world is located in D1R;& 'rance& which has " T superconducting niobium<titanium dipole
electromagnets in the #arge 9adron Dollider* The permanent magnets are needed to keep the
particle beam inside the accelerator (Russenschuck& ()))*
Modern transportation techniAues ha%e also ad%anced the use of permanent magnets* The
2ndutrack technology& which uses a 9allbach array for permanent magnet le%itation& is a
completely passi%e technology that operates on rare earth ;d'e7 magnets (.ost& ())))* .rototypes
ha%e been built& and ;ASA in the $SA has considered the 2ndutrack technology also for a rocket
launching application*
7esides their usefulness in commercial products& the main ad%antage of permanent magnets is
found in industrial energy con%ersion solutions*
The relati%e ability of a magnet material to produce work is measured by the maximum energy
product %alue J/Em
!
L* The magnetic material usually operates in the second Auadrant of the BH
cur%e& which illustrates the magnet capability to oppose demagnetiGation* 2n order to efficiently
use the magnet potential& the magnet operating point should be close to the maximum energy
product* Hith a suitable magnet material& the operating range of the material can also be extended
to the third Auadrant of the BH cur%e* A characteristic magnetiGation cur%e of a modern ;d'e7
magnet ;eorem +,!aE=,!a (;eorem Magnets) is presented in 'ig* *>*
The magnetism in ferromagnetic materials occurs when the small magnetic domains are aligned in
the same direction* The non<magnetiGed materials ha%e their elementary magnets arranged in
arbitrary directions* 2n hard magnetic materials& the orientation of the elements is difficult because
of the coerci%e force* 9owe%er& if the field is strong enough& the 7loch areas deform and the
elementary magnets orient according to the field*
+)



'ig* *>* Dharacteristic magnetiGation cur%e of the ;d'e7 +,!a grade magnet by
;eorem Magnets data sheet (;eorem& ())")*
As it can be seen& the demagnetiGation BH cur%e of ;eorem +,!a in the second Auadrant is linear
in temperatures below about =) MD* The maximum energy product is reached in the middle of the
linear demagnetiGation cur%e* 7ecause the magnetiGation cur%e is linear in the second Auadrant&
the maximum energy product can be calculated easily with the co<energy since it is eAual to the
energy stored into the system* Thus& the maximum energy product W is
w = _
1
2
pE
2
JI *)
'igure *> shows also the material HJ (polariGation) cur%e& which illustrates the magnetiGation of
the material in different temperatures* The line crossings on the B<axis are for the corresponding
temperatures for both H and J*
At present& the ;d'e7 material is the most popular permanent magnetic material because of its
high energy product* 5alues as high as +=)k/Em
!
ha%e been listed (.yrh0nen et al*& ())")* The
drawback of this material is its poor heat endurance& and therefore& the operating temperature must
be low to get the most out of the magnet material* The second best option is the SmDo material&
according to the energy product& but its heat endurance is significantly better compared with
neodymium magnets* Depending on the compound& the SmDo magnets can be used in
temperatures well beyond ()) MD* The third ma3or magnet type is the Al;iDo magnet& which has
the highest heat endurance of the important industrial magnet materials* ;e%ertheless& its energy
products are below )) k/Em
!
* 2n electrical engineering applications& the usable brands are not the
state<of<the<art compounds& and thus& the energy products of ;d'e7& SmDo and Al;iDo magnets
in industrial usage are in the range of (=)@+))& ))@()) and below ") k/Em
!
& respecti%ely*
+)



'ig* *>* Dharacteristic magnetiGation cur%e of the ;d'e7 +,!a grade magnet by
;eorem Magnets data sheet (;eorem& ())")*
As it can be seen& the demagnetiGation BH cur%e of ;eorem +,!a in the second Auadrant is linear
in temperatures below about =) MD* The maximum energy product is reached in the middle of the
linear demagnetiGation cur%e* 7ecause the magnetiGation cur%e is linear in the second Auadrant&
the maximum energy product can be calculated easily with the co<energy since it is eAual to the
energy stored into the system* Thus& the maximum energy product W is
w = _
1
2
pE
2
JI *)
'igure *> shows also the material HJ (polariGation) cur%e& which illustrates the magnetiGation of
the material in different temperatures* The line crossings on the B<axis are for the corresponding
temperatures for both H and J*
At present& the ;d'e7 material is the most popular permanent magnetic material because of its
high energy product* 5alues as high as +=)k/Em
!
ha%e been listed (.yrh0nen et al*& ())")* The
drawback of this material is its poor heat endurance& and therefore& the operating temperature must
be low to get the most out of the magnet material* The second best option is the SmDo material&
according to the energy product& but its heat endurance is significantly better compared with
neodymium magnets* Depending on the compound& the SmDo magnets can be used in
temperatures well beyond ()) MD* The third ma3or magnet type is the Al;iDo magnet& which has
the highest heat endurance of the important industrial magnet materials* ;e%ertheless& its energy
products are below )) k/Em
!
* 2n electrical engineering applications& the usable brands are not the
state<of<the<art compounds& and thus& the energy products of ;d'e7& SmDo and Al;iDo magnets
in industrial usage are in the range of (=)@+))& ))@()) and below ") k/Em
!
& respecti%ely*
+


9ard ferrimagnetic compounds are also useful in some applications* The ferrite compounds are not
%ery strong compared with rare earth and Al;iDo magnets& but their ad%antage is the low cost*
'errites can also be used to produce complex magnetic shapes and structures by a modern molding
techniAue* The energy products do not reach !) k/Em
!
in industrial grades& but the material is
applicable to loudspeakers& magnetic pickup of different instruments& small<scale electric motors
and magnetic holding systems* The physical parameters of different hard magnetic materials are
listed in Table *(*
Table* *(* Typical material parameters for common hard magnetic materials (Miller&
,,+) (.yrh0nen et al*& ())")*
.roperty SmDo ;d'e7 Al;iDo 'errites
H
c
JkAEmL +,!@=,) "))@,,) +)@!) ")@+))
B
r
J5sEm(L )*-@*!= *)@*+ )*>@*!= )*!=@)*+!
o
T
JPEMDL @()*)!@)*)=) @()*)"@)*=) @()*)@)*)() @)*(
JgEcm!L "*(@"*+ -*+@-*> -*! +*=@=*
Durie point JMDL -))@")) !)@!=) "=) +=)
o JMSEmL *> )*==@)*, (*! )*)
u
r
*)(@*)- *)+@* *,@- *)=@*=

Dommon to all ma3or permanent magnet compounds is the brittleness of the materials* The
bending and tensile strengths are usually far lower than a couple of hundred ;ewtons per sAuare
millimetre* 2n return& the compression strength of the magnets is good* The strength of the
materials must be carefully taken into account in the design process when the magnet fixing
procedure is determined* Densities are Auite eAual when comparing rare earth and Al;iDo
magnets* The ferrites are a somewhat lighter& which is due to their porous structure& but on the
other hand& the resisti%ity is significantly higher than in the ferromagnetic magnets*
2n electrical machines& the permanent magnet technology has e%ol%ed Auickly along with the new
permanent magnet materials* .ermanent<magnet<assisted motors ha%e se%eral ad%antages o%er
traditional induction and synchronous machines* .M machines ha%e a high efficiency and torAue
product& especially in low<speed applications* 2n current domestic applications& the ;d'e7 magnet
is the most common magnet type among the .M machine types* Although it has some physical
drawbacks& as mentioned abo%e& it still has two significant benefitsB a high energy product and a
low price* The drawbacks can be a%oided by a careful machine designU the most important aspect
in the design is the correct determination of the operating point of the magnet* 'ortunately& there
are numerous different brands of ;d'e7 magnets& which can be selected according to the desired
characteristics*
The operating point of the magnet can be examined analytically& but the method is rather arduous*
A more con%enient engineering approach is achie%ed with finite element method programs* $nder
load operation& the flux density of the magnet and the corresponding coerci%e force are sol%ed& and
the operating point at a certain temperature<dependent BH cur%e can be drawn in the
+


9ard ferrimagnetic compounds are also useful in some applications* The ferrite compounds are not
%ery strong compared with rare earth and Al;iDo magnets& but their ad%antage is the low cost*
'errites can also be used to produce complex magnetic shapes and structures by a modern molding
techniAue* The energy products do not reach !) k/Em
!
in industrial grades& but the material is
applicable to loudspeakers& magnetic pickup of different instruments& small<scale electric motors
and magnetic holding systems* The physical parameters of different hard magnetic materials are
listed in Table *(*
Table* *(* Typical material parameters for common hard magnetic materials (Miller&
,,+) (.yrh0nen et al*& ())")*
.roperty SmDo ;d'e7 Al;iDo 'errites
H
c
JkAEmL +,!@=,) "))@,,) +)@!) ")@+))
B
r
J5sEm(L )*-@*!= *)@*+ )*>@*!= )*!=@)*+!
o
T
JPEMDL @()*)!@)*)=) @()*)"@)*=) @()*)@)*)() @)*(
JgEcm!L "*(@"*+ -*+@-*> -*! +*=@=*
Durie point JMDL -))@")) !)@!=) "=) +=)
o JMSEmL *> )*==@)*, (*! )*)
u
r
*)(@*)- *)+@* *,@- *)=@*=

Dommon to all ma3or permanent magnet compounds is the brittleness of the materials* The
bending and tensile strengths are usually far lower than a couple of hundred ;ewtons per sAuare
millimetre* 2n return& the compression strength of the magnets is good* The strength of the
materials must be carefully taken into account in the design process when the magnet fixing
procedure is determined* Densities are Auite eAual when comparing rare earth and Al;iDo
magnets* The ferrites are a somewhat lighter& which is due to their porous structure& but on the
other hand& the resisti%ity is significantly higher than in the ferromagnetic magnets*
2n electrical machines& the permanent magnet technology has e%ol%ed Auickly along with the new
permanent magnet materials* .ermanent<magnet<assisted motors ha%e se%eral ad%antages o%er
traditional induction and synchronous machines* .M machines ha%e a high efficiency and torAue
product& especially in low<speed applications* 2n current domestic applications& the ;d'e7 magnet
is the most common magnet type among the .M machine types* Although it has some physical
drawbacks& as mentioned abo%e& it still has two significant benefitsB a high energy product and a
low price* The drawbacks can be a%oided by a careful machine designU the most important aspect
in the design is the correct determination of the operating point of the magnet* 'ortunately& there
are numerous different brands of ;d'e7 magnets& which can be selected according to the desired
characteristics*
The operating point of the magnet can be examined analytically& but the method is rather arduous*
A more con%enient engineering approach is achie%ed with finite element method programs* $nder
load operation& the flux density of the magnet and the corresponding coerci%e force are sol%ed& and
the operating point at a certain temperature<dependent BH cur%e can be drawn in the
+(


demagnetiGation cur%e* This usually reAuires se%eral calculations at different temperatures& but it is
a relati%ely fast and straightforward task to perform* The load line determination is %ery important
already at the beginning of the machine design*

!.,.1 )or.ue .uality
TorAue Auality is one of the concerns in the discussion of the traction machine performance* 2n
2D1<powered systems& the torAue is pulsating but the mechanics& that is& the torAue transmission
line& filters the pulsating torAue& which results in a smooth operation within the rotating speed
limits of the D1 unit* 1lectrical machines also suffer from the pulsating torAue form to some
extent* Slotting causes spatial reluctance differences in the flux path while the rotor is rotating&
which yields a pulsating torAue wa%e form called cogging* Another source is the electromagnetic
distortion caused by the winding space harmonics resulting from the winding scheme and current
wa%e form imperfections& which are generally described as torAue ripple* A torAue ripple that is
less than @( P of the rated torAue can be considered low (Miller& ,,+)* There are multiple ways
to minimiGe the cogging effect and torAue ripple* A modern freAuency con%erter control is usually
fast enough to compensate the torAue pulsation for a smooth output* This is due to the fact that the
mechanical time constants are much larger compared with the control loop update speed* As
mentioned abo%e& winding arrangements with a low space harmonic content can be used to
minimiGe the torAue ripple in the machine design process* The cogging torAue can be decreased
with semi<magnetic slot sticks and skewing* 2ncreasing the number of slots per pole also reduces
cogging (Miller et al*& ,,+)* The tooth geometry can also be modified to decrease %ariation in the
magnetic conducti%ity in the air gap space for instance by using bifurcated tooth tips*

!.4 'utline of the work
This doctoral thesis studies the traction motor design for a performance<oriented road %ehicle
application with a direct four<wheel dri%e* 2n the literature study& no distinct pattern was found for
the methods for traction machine design in hybrid and full electric dri%ing systems* Although the
history of electric traction de%elopment is long& the topic of directly connected permanent magnet
traction machines is Auite new& and numerous alternati%e approaches ha%e been proposed* Despite
the wide experience in different special machine designs& there is no pre%ious experience in the
traction motor de%elopment for a %ehicle en%ironment at the uni%ersity& and therefore& a doctoral
study on the traction machine design was considered an appropriate opening to this research field*
The structure and contents of the doctoral thesis are summariGed as followsB
2hapter ! introduces the topic and ob3ecti%es of the thesis* The study starts with a presentation of
the reAuirements set by the %ehicle en%ironment and the application<specific boundary %alues* A
brief glance is also taken at the history and de%elopment of hybrid and full electric propulsion
systems* 9ybrid and full electric systems are categoriGed& and the differences and similarities
between the propulsion systems are in%estigated including electrical machine type selection and
+(


demagnetiGation cur%e* This usually reAuires se%eral calculations at different temperatures& but it is
a relati%ely fast and straightforward task to perform* The load line determination is %ery important
already at the beginning of the machine design*

!.,.1 )or.ue .uality
TorAue Auality is one of the concerns in the discussion of the traction machine performance* 2n
2D1<powered systems& the torAue is pulsating but the mechanics& that is& the torAue transmission
line& filters the pulsating torAue& which results in a smooth operation within the rotating speed
limits of the D1 unit* 1lectrical machines also suffer from the pulsating torAue form to some
extent* Slotting causes spatial reluctance differences in the flux path while the rotor is rotating&
which yields a pulsating torAue wa%e form called cogging* Another source is the electromagnetic
distortion caused by the winding space harmonics resulting from the winding scheme and current
wa%e form imperfections& which are generally described as torAue ripple* A torAue ripple that is
less than @( P of the rated torAue can be considered low (Miller& ,,+)* There are multiple ways
to minimiGe the cogging effect and torAue ripple* A modern freAuency con%erter control is usually
fast enough to compensate the torAue pulsation for a smooth output* This is due to the fact that the
mechanical time constants are much larger compared with the control loop update speed* As
mentioned abo%e& winding arrangements with a low space harmonic content can be used to
minimiGe the torAue ripple in the machine design process* The cogging torAue can be decreased
with semi<magnetic slot sticks and skewing* 2ncreasing the number of slots per pole also reduces
cogging (Miller et al*& ,,+)* The tooth geometry can also be modified to decrease %ariation in the
magnetic conducti%ity in the air gap space for instance by using bifurcated tooth tips*

!.4 'utline of the work
This doctoral thesis studies the traction motor design for a performance<oriented road %ehicle
application with a direct four<wheel dri%e* 2n the literature study& no distinct pattern was found for
the methods for traction machine design in hybrid and full electric dri%ing systems* Although the
history of electric traction de%elopment is long& the topic of directly connected permanent magnet
traction machines is Auite new& and numerous alternati%e approaches ha%e been proposed* Despite
the wide experience in different special machine designs& there is no pre%ious experience in the
traction motor de%elopment for a %ehicle en%ironment at the uni%ersity& and therefore& a doctoral
study on the traction machine design was considered an appropriate opening to this research field*
The structure and contents of the doctoral thesis are summariGed as followsB
2hapter ! introduces the topic and ob3ecti%es of the thesis* The study starts with a presentation of
the reAuirements set by the %ehicle en%ironment and the application<specific boundary %alues* A
brief glance is also taken at the history and de%elopment of hybrid and full electric propulsion
systems* 9ybrid and full electric systems are categoriGed& and the differences and similarities
between the propulsion systems are in%estigated including electrical machine type selection and
+!


fundamental siGing* The causality between the key design features is presented to establish
understanding between the machine dimensions and performance* The chapter draws the outlines
of the thesis and describes the scientific contribution of the study*
2hapter addresses the electrical machine design areas in more detail including the
electromagnetic and thermal design and the e%aluation of losses and efficiency*
2hapter , introduces the prototype machine design process* The machine structures are presented
in detail& and a performance analysis is carried out with analytical and finite element models*
:ptimiGation of the machine structure is performed in order to determine the boundary %alues of
the dri%e system* The de%eloped thermal model is presented in addition to the model<specific
details*
2hapter 4 introduces the measurement systems* The analytical and finite element results are
%erified by experimental studies* An analytical thermal model of the static and dynamic
performance is applied to the measured data*
2hapter 1 pro%ides conclusions on the doctoral thesis including the key contributions of the work&
achie%ing of the ob3ecti%es and recommendations for future work*

!.1 "cientific contribution of the work
The scientific contributions of the doctoral thesis are summariGed in the order of importanceB
< The doctoral thesis proposes a traction machine design seAuence with
combined electrical and thermal design methods for %ehicle traction within
the application<specific geometrical boundaries*
< The work demonstrates the importance of the cross<saturation effect in the
traction machine design compared with the traditional two<dimensional
approach* The work demonstrates the importance of the coupling of the
electric and thermal design machine designs where 3oule losses are
dominating*
< The thesis demonstrates the effect of the battery<package<related supply
%oltage %ariation on the machine performance o%er a certain speed range*
< The study proposes& by introducing a special magnet module design& a cost<
efficient machine structure with impro%ed saliency* The machine structure is
designed for an increased speed range and torAue density to meet the
performance reAuirements in %ehicle traction systems*
< The work promotes the benefits of a salient pole structure in %ehicle traction*
< The study pro%ides a comparison of a special magnet module machine design
with a traditional embedded magnet machine design by a prototype machine
of each structure*

+!


fundamental siGing* The causality between the key design features is presented to establish
understanding between the machine dimensions and performance* The chapter draws the outlines
of the thesis and describes the scientific contribution of the study*
2hapter addresses the electrical machine design areas in more detail including the
electromagnetic and thermal design and the e%aluation of losses and efficiency*
2hapter , introduces the prototype machine design process* The machine structures are presented
in detail& and a performance analysis is carried out with analytical and finite element models*
:ptimiGation of the machine structure is performed in order to determine the boundary %alues of
the dri%e system* The de%eloped thermal model is presented in addition to the model<specific
details*
2hapter 4 introduces the measurement systems* The analytical and finite element results are
%erified by experimental studies* An analytical thermal model of the static and dynamic
performance is applied to the measured data*
2hapter 1 pro%ides conclusions on the doctoral thesis including the key contributions of the work&
achie%ing of the ob3ecti%es and recommendations for future work*

!.1 "cientific contribution of the work
The scientific contributions of the doctoral thesis are summariGed in the order of importanceB
< The doctoral thesis proposes a traction machine design seAuence with
combined electrical and thermal design methods for %ehicle traction within
the application<specific geometrical boundaries*
< The work demonstrates the importance of the cross<saturation effect in the
traction machine design compared with the traditional two<dimensional
approach* The work demonstrates the importance of the coupling of the
electric and thermal design machine designs where 3oule losses are
dominating*
< The thesis demonstrates the effect of the battery<package<related supply
%oltage %ariation on the machine performance o%er a certain speed range*
< The study proposes& by introducing a special magnet module design& a cost<
efficient machine structure with impro%ed saliency* The machine structure is
designed for an increased speed range and torAue density to meet the
performance reAuirements in %ehicle traction systems*
< The work promotes the benefits of a salient pole structure in %ehicle traction*
< The study pro%ides a comparison of a special magnet module machine design
with a traditional embedded magnet machine design by a prototype machine
of each structure*

++


!.3 4ist of publications
This section lists other publications in which the author of this doctoral thesis has been author or a
co<author and which are rele%ant to the doctoral work* :ther supporting publications related to the
topic are listed separately*

!.3.! 4ist of related publications
< ;erg /*& Rilla M* and .yrh0nen /*& [Thermal Analysis of Radial<'lux
1lectrical machines Hith a 9igh .ower Density&\ IEEE Transactions on
Industrial Electronics& 5ol* ==& 2ssue )& :ctober ())"& pp* !=+!@!==+*

< Rilla M*& ;iemel6 M* and .yrh0nen /*& [Thermal 1%aluation of the 1ffects of
Dimensional .arameters on #ow Speed .ermanent Magnet Machine Design&\
2n the XI International Svmposium on Electric Fields& ())-*

< #indh .*& Rilla M*& /ussila 9*& ;erg /*& Tapia /*A* and .yrh0nen /*& \2nterior
.ermanent Magnet Motors for Traction Application with non<:%erlapping
Doncentrated Hindings and with 2nteger Slot Hindings&\ International
Review of Electrical Engineering (I.R.E.E.)& %ol* >& no* +& August ()*

< ;erg /*& Rilla M*& Ruuskanen 5*& .yrh0nen /*& Ruotsalainen S*& [Design of
Direct Dri%en .ermanent magnet Synchronous motors for an 1lectric Sports
Dar\& 2n the International Conference on Electrical Machines& ()(*

< #indh .*& ;erg /*& .yrh0nen /*& .olikarpo%a M*& /ussila 9* and Rilla M*&
[2nterior permanent magnet motors with non<o%erlapping concentrated
windings or with integral slot windings for traction application\& 2n the XJ
International Svmposium on Electromagnetic Fields& September @!& ()*


!.3. 4ist of supporting publications
< Rilla M*& ;iemel6 M*& .yrh0nen /*& .ekola /* and /6ppinen /*& [Design of a >)
kH& ,))) rpm non<salient pole .M<machine&\ 2n the 18th International
Conference on Electrical Machines& 5ilamoura& September >@,& ())"*

< 2mmonen .*& #aurila #*& Rilla M* and .yrh0nen /*& [Modelling and simulation
of a parallel hybrid dri%e system for mobile work machines&\ 2n EUROCON&
St* .etersburg& May "@(!& ()),& pp* ">-@"-(*

++


!.3 4ist of publications
This section lists other publications in which the author of this doctoral thesis has been author or a
co<author and which are rele%ant to the doctoral work* :ther supporting publications related to the
topic are listed separately*

!.3.! 4ist of related publications
< ;erg /*& Rilla M* and .yrh0nen /*& [Thermal Analysis of Radial<'lux
1lectrical machines Hith a 9igh .ower Density&\ IEEE Transactions on
Industrial Electronics& 5ol* ==& 2ssue )& :ctober ())"& pp* !=+!@!==+*

< Rilla M*& ;iemel6 M* and .yrh0nen /*& [Thermal 1%aluation of the 1ffects of
Dimensional .arameters on #ow Speed .ermanent Magnet Machine Design&\
2n the XI International Svmposium on Electric Fields& ())-*

< #indh .*& Rilla M*& /ussila 9*& ;erg /*& Tapia /*A* and .yrh0nen /*& \2nterior
.ermanent Magnet Motors for Traction Application with non<:%erlapping
Doncentrated Hindings and with 2nteger Slot Hindings&\ International
Review of Electrical Engineering (I.R.E.E.)& %ol* >& no* +& August ()*

< ;erg /*& Rilla M*& Ruuskanen 5*& .yrh0nen /*& Ruotsalainen S*& [Design of
Direct Dri%en .ermanent magnet Synchronous motors for an 1lectric Sports
Dar\& 2n the International Conference on Electrical Machines& ()(*

< #indh .*& ;erg /*& .yrh0nen /*& .olikarpo%a M*& /ussila 9* and Rilla M*&
[2nterior permanent magnet motors with non<o%erlapping concentrated
windings or with integral slot windings for traction application\& 2n the XJ
International Svmposium on Electromagnetic Fields& September @!& ()*


!.3. 4ist of supporting publications
< Rilla M*& ;iemel6 M*& .yrh0nen /*& .ekola /* and /6ppinen /*& [Design of a >)
kH& ,))) rpm non<salient pole .M<machine&\ 2n the 18th International
Conference on Electrical Machines& 5ilamoura& September >@,& ())"*

< 2mmonen .*& #aurila #*& Rilla M* and .yrh0nen /*& [Modelling and simulation
of a parallel hybrid dri%e system for mobile work machines&\ 2n EUROCON&
St* .etersburg& May "@(!& ()),& pp* ">-@"-(*

+=


< .0ll6nen R*& ;erg /*& Rilla M* and .yrh0nen :*& [Transient Thermal Model
for Radial Acti%e Magnetic 7earing&\ 2n the Tenth International Svmposium
on Magnetic Bearings& ())>*






+=


< .0ll6nen R*& ;erg /*& Rilla M* and .yrh0nen :*& [Transient Thermal Model
for Radial Acti%e Magnetic 7earing&\ 2n the Tenth International Svmposium
on Magnetic Bearings& ())>*






+>



+>



+-


5ey design areas of a traction machine
Traditional electrical machine design relies on experimental knowledge of suitable machine
parameters based on data gathered from actual machines* Depending on the cooling arrangement&
machine type and siGe& there are certain guidelines a%ailable to determine the thermal performance
characteristics of the machine*
Hith permanent magnet traction machines& an additional set of design tools is reAuired* 7ecause of
the sensiti%ity of the permanent magnet material and %arying load conditions& more accurate
knowledge of the machine thermal beha%iour is needed for the performance analysis* The machine
temperature beha%iour in %arying load conditions can be estimated by a lumped<parameter thermal
model* The use of the model reAuires knowledge of the loss generation inside the machine and
understanding of the mechanisms of heat transfer*
This chapter pro%ides an analysis of the electrical dimensioning of a traction machine with an
introduction to heat transfer mechanisms& lumped<parameter modeling and the e%aluation of loss
distribution in electrical machines*

.! +lectromagnetic design
The design of an electrical machine starts from the dimensioning of the rotor& after the desired
output torAue has been decided* The application<specific dimensional boundary conditions should
also be taken into account at this point* Analytical presentation of the machine structure and the
electrical properties of an integral slot machine are easiest to pro%ide by a single magnetic path
presentation& which consists of one pole of the machine geometry*

.!.! 6asic design rules
Designing an electrical machine for a mobile solution differs from the design process of a
traditional industrial machine* According to the standard 21D >))!+<& there are industrial limits
for the pull<out torAue depending on the machine type* Synchronous motors with a salient pole
structure should be capable of handling a != P excess torAue* ;on<salient pole structures ha%e a
higher peak torAue reAuirement (=) P)U the same as with cage induction machines with a starting
current less than +*= times the rated current (otherwise >) P)* These limits are for direct<on<line
machines and for a = s continuous load at reAuired peak torAue*
.ower tool and traction applications may easily reAuire se%eral times the rated torAue in transient
operation* This cannot be reached by industrial<type machine dimensioning* Thus& it is challenging
to optimiGe the machine structure according to the performance expectations*
'irst& it is important to determine the initial %alues with care* At least the following %alues are
reAuiredB
+-


5ey design areas of a traction machine
Traditional electrical machine design relies on experimental knowledge of suitable machine
parameters based on data gathered from actual machines* Depending on the cooling arrangement&
machine type and siGe& there are certain guidelines a%ailable to determine the thermal performance
characteristics of the machine*
Hith permanent magnet traction machines& an additional set of design tools is reAuired* 7ecause of
the sensiti%ity of the permanent magnet material and %arying load conditions& more accurate
knowledge of the machine thermal beha%iour is needed for the performance analysis* The machine
temperature beha%iour in %arying load conditions can be estimated by a lumped<parameter thermal
model* The use of the model reAuires knowledge of the loss generation inside the machine and
understanding of the mechanisms of heat transfer*
This chapter pro%ides an analysis of the electrical dimensioning of a traction machine with an
introduction to heat transfer mechanisms& lumped<parameter modeling and the e%aluation of loss
distribution in electrical machines*

.! +lectromagnetic design
The design of an electrical machine starts from the dimensioning of the rotor& after the desired
output torAue has been decided* The application<specific dimensional boundary conditions should
also be taken into account at this point* Analytical presentation of the machine structure and the
electrical properties of an integral slot machine are easiest to pro%ide by a single magnetic path
presentation& which consists of one pole of the machine geometry*

.!.! 6asic design rules
Designing an electrical machine for a mobile solution differs from the design process of a
traditional industrial machine* According to the standard 21D >))!+<& there are industrial limits
for the pull<out torAue depending on the machine type* Synchronous motors with a salient pole
structure should be capable of handling a != P excess torAue* ;on<salient pole structures ha%e a
higher peak torAue reAuirement (=) P)U the same as with cage induction machines with a starting
current less than +*= times the rated current (otherwise >) P)* These limits are for direct<on<line
machines and for a = s continuous load at reAuired peak torAue*
.ower tool and traction applications may easily reAuire se%eral times the rated torAue in transient
operation* This cannot be reached by industrial<type machine dimensioning* Thus& it is challenging
to optimiGe the machine structure according to the performance expectations*
'irst& it is important to determine the initial %alues with care* At least the following %alues are
reAuiredB
+"



< a%erage and maximum torAue& T
a%g
and T
max

< nominal and maximum rotating speed& n
n
and n
max

< energy supply %oltage and current& U
##
and I
#

< limits for the machine length and diameter& l
m
and D
m
and
< cooling method and maximum temperature of the cooling fluid*

Despite the reAuirement for a better total system efficiency& the optimiGation of the system is not
entirely a matter of physics but rather an attempt to reach a tolerable cost efficiency without
ad%ersely affecting the dri%e performance*
The .M rotor can be made thin and long or short and large in diameter depending on the space
limitations* Hith a low rotating speed& a large radius is preferable with a high pole number& since
the .M rotor structure can be made light because of multiple flux paths*
The output torAue of the machine depends on the acti%e surface area of the rotor& the rotor radius&
the air gap flux density and the linear current density* The rotor produces a certain force on the
acti%e surface area according to the tangential stress o
tan
& which is proportional to the product of
the air gap flux density B
]
and the linear current density A* The force acting on the cylindrical
surface area& depending on the pole effecti%e area& produces an amount of torAue directly
proportional to the rotor radius r
r
* The torAue eAuation is thus written as
I = o
tan
r
r
2r
r
l'
r
= o
tan
S
r
r
r
(*
where
< l
r
is the effecti%e length of the rotor stack&
< r
r
is the rotor radius&
< o
tan
is the tangential stress component and
< S
r
is the rotor cylinder

surface acti%e area*

2n traditional machines without cooling ducts& this eAuals the rotor stack geometrical length l
r
plus
two times the air gap length o of the machine (5ogt& ,"!)* 2n .MSMs& howe%er& the length is
more complicated to determine (.yrh0nen et al*& ()))*
2n permanent magnet machines& the tangential stress produced by the armature winding has to be
matched with the permanent magnets* The tangential stress is a multiple of the peak %alue of the
air gap flux and the peak %alue linear current density* The a%erage stress o
tan
is then calculated as
o
tan
=
1
2
B
8
A
`
cos. (*(
7ecause the flux density represents the %oltage in the eAuation and the linear current density the
current of the machine& the eAuation has to take the power factor cos into account& which in this
+"



< a%erage and maximum torAue& T
a%g
and T
max

< nominal and maximum rotating speed& n
n
and n
max

< energy supply %oltage and current& U
##
and I
#

< limits for the machine length and diameter& l
m
and D
m
and
< cooling method and maximum temperature of the cooling fluid*

Despite the reAuirement for a better total system efficiency& the optimiGation of the system is not
entirely a matter of physics but rather an attempt to reach a tolerable cost efficiency without
ad%ersely affecting the dri%e performance*
The .M rotor can be made thin and long or short and large in diameter depending on the space
limitations* Hith a low rotating speed& a large radius is preferable with a high pole number& since
the .M rotor structure can be made light because of multiple flux paths*
The output torAue of the machine depends on the acti%e surface area of the rotor& the rotor radius&
the air gap flux density and the linear current density* The rotor produces a certain force on the
acti%e surface area according to the tangential stress o
tan
& which is proportional to the product of
the air gap flux density B
]
and the linear current density A* The force acting on the cylindrical
surface area& depending on the pole effecti%e area& produces an amount of torAue directly
proportional to the rotor radius r
r
* The torAue eAuation is thus written as
I = o
tan
r
r
2r
r
l'
r
= o
tan
S
r
r
r
(*
where
< l
r
is the effecti%e length of the rotor stack&
< r
r
is the rotor radius&
< o
tan
is the tangential stress component and
< S
r
is the rotor cylinder

surface acti%e area*

2n traditional machines without cooling ducts& this eAuals the rotor stack geometrical length l
r
plus
two times the air gap length o of the machine (5ogt& ,"!)* 2n .MSMs& howe%er& the length is
more complicated to determine (.yrh0nen et al*& ()))*
2n permanent magnet machines& the tangential stress produced by the armature winding has to be
matched with the permanent magnets* The tangential stress is a multiple of the peak %alue of the
air gap flux and the peak %alue linear current density* The a%erage stress o
tan
is then calculated as
o
tan
=
1
2
B
8
A
`
cos. (*(
7ecause the flux density represents the %oltage in the eAuation and the linear current density the
current of the machine& the eAuation has to take the power factor cos into account& which in this
+,


case has to be estimated and corrected later* 8uidelines for the tangential stresses can be found in
the literature* Some guidelines can be found in Tables (* and (*(*
Table* (** Tangential stress %alues for different machine types in rated operation
according to (Miller& ,,+)* The lower %alues are for naturally cooled machines and the
higher ones for external cooling fan applications* T1 stands for Vtotally enclosedF*
Machine type )angential stress component 7kPa8
Small T1 motors ('errite) !*+@>*,
T1 motors (Sintered rare earth or
;b'e7)
>*,@()*>
T1 motors (bonded ;d'e7) )*! typical
Aerospace machines !*"@!+*+
9igh<performance ser%omotors >*,@()*>
#arge liAuid<cooled machines >,*)@)!*+

Table* (*(* Tangential stress and linear current density %alues for different machine types
in rated operation according to (5ogt et al*& ,"!)* The lower %alues are for machines
with natural con%ection and the higher ones for external cooling applications*
"alient pole $on#salient pole
2ooling
method
Air Air 9ater Direct water
A JkAEmL !=@>= !)@") ,)@) =)@())
B
]
J5sEm
(
L )*"@*)= )*"@*)= )*"@*)= )*"@*)=
o
tan& min
Jk.aL (*) -*) =*) "=*)
o
tan& a%g
Jk.aL (*= !>*) >=*= +*=
o
tan& max
Jk.aL !!*) =,*= "*= +"*=

Tangential stress may momentarily ha%e higher peak %alues than those presented in Table (* and
(*( depending on the synchronous inductances* 1%en though the torAue depends on the surface
pole area& the torAue production fa%ours larger rotor diameters* Typical current density limits for
different cooling methods are presented in Table (*!*
Table* (*!* Typical current density %alues for different machine cooling methods* (5ogt&
,"+) (Miller& ,,+)* T1 stands for Vtotally enclosedF*
Machine type 2urrent density 7A:mm

8
Air cooled (T1) ,#1;!.11<
Air<o%er 'an<cooled 1!=
#iAuid cooled (direct water) >!= ;!=,=<

+,


case has to be estimated and corrected later* 8uidelines for the tangential stresses can be found in
the literature* Some guidelines can be found in Tables (* and (*(*
Table* (** Tangential stress %alues for different machine types in rated operation
according to (Miller& ,,+)* The lower %alues are for naturally cooled machines and the
higher ones for external cooling fan applications* T1 stands for Vtotally enclosedF*
Machine type )angential stress component 7kPa8
Small T1 motors ('errite) !*+@>*,
T1 motors (Sintered rare earth or
;b'e7)
>*,@()*>
T1 motors (bonded ;d'e7) )*! typical
Aerospace machines !*"@!+*+
9igh<performance ser%omotors >*,@()*>
#arge liAuid<cooled machines >,*)@)!*+

Table* (*(* Tangential stress and linear current density %alues for different machine types
in rated operation according to (5ogt et al*& ,"!)* The lower %alues are for machines
with natural con%ection and the higher ones for external cooling applications*
"alient pole $on#salient pole
2ooling
method
Air Air 9ater Direct water
A JkAEmL !=@>= !)@") ,)@) =)@())
B
]
J5sEm
(
L )*"@*)= )*"@*)= )*"@*)= )*"@*)=
o
tan& min
Jk.aL (*) -*) =*) "=*)
o
tan& a%g
Jk.aL (*= !>*) >=*= +*=
o
tan& max
Jk.aL !!*) =,*= "*= +"*=

Tangential stress may momentarily ha%e higher peak %alues than those presented in Table (* and
(*( depending on the synchronous inductances* 1%en though the torAue depends on the surface
pole area& the torAue production fa%ours larger rotor diameters* Typical current density limits for
different cooling methods are presented in Table (*!*
Table* (*!* Typical current density %alues for different machine cooling methods* (5ogt&
,"+) (Miller& ,,+)* T1 stands for Vtotally enclosedF*
Machine type 2urrent density 7A:mm

8
Air cooled (T1) ,#1;!.11<
Air<o%er 'an<cooled 1!=
#iAuid cooled (direct water) >!= ;!=,=<

=)


'igure (* presents the dependence of the rotor diameter and length on the torAue production&
based on 1A* (*& at a !> k.a tangential stress& applied to the rotor surface by %arying the rotor
diameter to the rotor length ratio in relation to the actual rotor diameter*

'ig* (** Dontour plot describing the effect of r
r
El

as a function of rotor radius r
r
and
produced torAue* The calculations are carried out with a !> k.a tangential stress
and cos S & and ") P of the rotor air gap surface area is considered
magnetically acti%e* The red lines are contours for the constant rotor acti%e area
in sAuare meters Jm
(
L and the black lines represent r
r
/l proportions* The sAuare
markers indicate the design point of the de%eloped machines with a )*("+ m
rotor diameter& )*)>= m rotor length and (+) ;m of torAue*
The rotor contour cur%e is of Auadratic form* According to the slope shapes in 'ig* (*& the
machine torAue production prefers an increase in the rotor radius to an increase in the rotor length*
A ) P increase in the rotor diameter results approximately in =*" P more torAue compared with a
) P increase in the rotor length* 2n both cases& the surface area of the rotor increases by the same
amount& but the extra torAue comes from the larger radius* This can also be %erified by Table (*+&
which lists the recommended r
r
El %alues for synchronous machines according to the pole number*



0
.
2
5
0
.
2
5
0
.
5
0
.
5
0
.
5
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
7
7
7
7
8
8
8
8
9
9
9
9
10
1
0
1
0
1
0
Rotor radius JmL
T
o
r
A
u
e

J
;
m
L
0
.0
2
5
0
.0
2
5
0
.0
2
5
0
.
0
5
0
.
0
5
0
.
0
5
0
.
0
5
0
.
0
7
5
0
.
0
7
5
0
.
0
7
5
0
.
1
0
.
1
0
.
2
)*)= )* )*= )*( )*(=
=)
))
=)
())
(=)
!))
!=)
+))
+=)
=))
=)


'igure (* presents the dependence of the rotor diameter and length on the torAue production&
based on 1A* (*& at a !> k.a tangential stress& applied to the rotor surface by %arying the rotor
diameter to the rotor length ratio in relation to the actual rotor diameter*

'ig* (** Dontour plot describing the effect of r
r
El

as a function of rotor radius r
r
and
produced torAue* The calculations are carried out with a !> k.a tangential stress
and cos S & and ") P of the rotor air gap surface area is considered
magnetically acti%e* The red lines are contours for the constant rotor acti%e area
in sAuare meters Jm
(
L and the black lines represent r
r
/l proportions* The sAuare
markers indicate the design point of the de%eloped machines with a )*("+ m
rotor diameter& )*)>= m rotor length and (+) ;m of torAue*
The rotor contour cur%e is of Auadratic form* According to the slope shapes in 'ig* (*& the
machine torAue production prefers an increase in the rotor radius to an increase in the rotor length*
A ) P increase in the rotor diameter results approximately in =*" P more torAue compared with a
) P increase in the rotor length* 2n both cases& the surface area of the rotor increases by the same
amount& but the extra torAue comes from the larger radius* This can also be %erified by Table (*+&
which lists the recommended r
r
El %alues for synchronous machines according to the pole number*



0
.
2
5
0
.
2
5
0
.
5
0
.
5
0
.
5
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
7
7
7
7
8
8
8
8
9
9
9
9
10
1
0
1
0
1
0
Rotor radius JmL
T
o
r
A
u
e

J
;
m
L
0
.0
2
5
0
.0
2
5
0
.0
2
5
0
.
0
5
0
.
0
5
0
.
0
5
0
.
0
5
0
.
0
7
5
0
.
0
7
5
0
.
0
7
5
0
.
1
0
.
1
0
.
2
)*)= )* )*= )*( )*(=
=)
))
=)
())
(=)
!))
!=)
+))
+=)
=))
=


Table* (*+* Recommended rotor radius to rotor length ratios (5ogt& ,"!)*
p ( ! + = > - " , )
r
r
E l^
)*!!@

)*,) *) *(- *+( *=> *>" *") *, (*)

According to Table (*+& it is recommended to use a shorter rotor length in relation to the rotor
diameter as the pole number increases (5ogt& ,"!)* The larger pole number most likely promotes
low rotating speeds as the low %alues for one () and two (() pole pair machines are intended for
higher<speed machines*

.!. 9indings
Magnetic design complements the basic machine design and determines the machine acti%e
dimensions* After the definition of initial parameters& the desired air gap flux 1

can be determined
as
1

= o
I
B
`
6
2r
8
2p
l' (*!
where

< ((Xr
]
E(p) is the pole pitch :
p

< o
i
is the per unit a%erage of the flux density on a pole&
< p is the pole pair number and
< l is the effecti%e stack length*

The flux concentration o%er the pole area is considered e%enly distributed* o
i
is (EX for a sinusoidal
flux density distribution& but in the surface .M case& o
i
is replaced by o
.M
& usually considered to be
w
pm
E :
p
& which is normally enough for analytical dimensioning* The length of the machine is an
important parameter& especially at low rotor length to rotor radius ratios* The peak %alue of the air
gap flux density can be selected for example from Table (*(* 2n the .M case& it is ad%isable to keep
the maximum peak %alue below or eAual to )*, T (Miller& ,,+)*
The pole number is an optimiGation issue* As noted abo%e& the pole number itself does not
determine the torAue output of the machine& but it has a significant effect on the iron core
dimensions and the pull<out torAue characteristics& because synchronous inductance is in%ersely
proportional to the sAuare of the pole number which can be deduced by combining 1As* *= and
*> in Section *!*
Small pole numbers result in thicker iron passages as a result of larger flux components* A large
pole number reAuires thinner iron passages& and thus& the rotor dimensions can be increased in the
radial direction to enhance torAue production* The stator yoke can be made thinner because of the
=


Table* (*+* Recommended rotor radius to rotor length ratios (5ogt& ,"!)*
p ( ! + = > - " , )
r
r
E l^
)*!!@

)*,) *) *(- *+( *=> *>" *") *, (*)

According to Table (*+& it is recommended to use a shorter rotor length in relation to the rotor
diameter as the pole number increases (5ogt& ,"!)* The larger pole number most likely promotes
low rotating speeds as the low %alues for one () and two (() pole pair machines are intended for
higher<speed machines*

.!. 9indings
Magnetic design complements the basic machine design and determines the machine acti%e
dimensions* After the definition of initial parameters& the desired air gap flux 1

can be determined
as
1

= o
I
B
`
6
2r
8
2p
l' (*!
where

< ((Xr
]
E(p) is the pole pitch :
p

< o
i
is the per unit a%erage of the flux density on a pole&
< p is the pole pair number and
< l is the effecti%e stack length*

The flux concentration o%er the pole area is considered e%enly distributed* o
i
is (EX for a sinusoidal
flux density distribution& but in the surface .M case& o
i
is replaced by o
.M
& usually considered to be
w
pm
E :
p
& which is normally enough for analytical dimensioning* The length of the machine is an
important parameter& especially at low rotor length to rotor radius ratios* The peak %alue of the air
gap flux density can be selected for example from Table (*(* 2n the .M case& it is ad%isable to keep
the maximum peak %alue below or eAual to )*, T (Miller& ,,+)*
The pole number is an optimiGation issue* As noted abo%e& the pole number itself does not
determine the torAue output of the machine& but it has a significant effect on the iron core
dimensions and the pull<out torAue characteristics& because synchronous inductance is in%ersely
proportional to the sAuare of the pole number which can be deduced by combining 1As* *= and
*> in Section *!*
Small pole numbers result in thicker iron passages as a result of larger flux components* A large
pole number reAuires thinner iron passages& and thus& the rotor dimensions can be increased in the
radial direction to enhance torAue production* The stator yoke can be made thinner because of the
=(


multiplication of the flux paths* The pole number is limited by the increasing supply freAuency and
the reAuired rotating speed range*
Hith the magnitude of the magnetiGing flux it is possible to determine the produced back 1M' of
the machine& which is needed to determine the motor supply current and %oltage when the winding
scheme is selected*
The rotating speed and the supply %oltage ha%e to be matched with the machine flux* The relation
between the %oltage E
m
and the air gap flux linkage Vm is
E
m
=
1
V2
k
w1
N
s
1

=
1
V2
V

m
(*+
where
< c is the electrical angular freAuency (Xf& where f is the supply freAuency
< k
w
is the fundamental winding factor and
< N
s
is the phase winding turn number*

Hith the selection of these parameters& the already fixed pole number and the winding turn number
can be selected* 8uidelines for integer slot machines can be found in Table (*=& which gi%es the
possible combinations for different slot and pole configurations and phase winding turns for a
gi%en slot conductor number*
Table* (*=* Allowed winding turn number and slot number for a slots per pole per phase
qS winding with a gi%en slot conductor number :
N
and a pole pair number* N
s
is the
number of series<connected coil turns in one phase*
p
q :
N

( ! + = >
Q N
s

> ( ! + = >
( ( ( + > " ) (
! " ! > , ( = "
+ (+ + " ( > () (+
= !) = ) = () (= !)
> !> > ( " (+ !) !>
- +( - + ( (" != +(
" +" " > (+ !( +) +"
, =+ , " (- !> += =+
) >) ) () !) +) =) >)

The slot and pole combination can also be selected to produce a fractional slot winding* The
fractional slot design can be used to produce shorter end windings for less leakage* An additional
feature is the decreased axial dimension& which is an attracti%e property of this winding
=(


multiplication of the flux paths* The pole number is limited by the increasing supply freAuency and
the reAuired rotating speed range*
Hith the magnitude of the magnetiGing flux it is possible to determine the produced back 1M' of
the machine& which is needed to determine the motor supply current and %oltage when the winding
scheme is selected*
The rotating speed and the supply %oltage ha%e to be matched with the machine flux* The relation
between the %oltage E
m
and the air gap flux linkage Vm is
E
m
=
1
V2
k
w1
N
s
1

=
1
V2
V

m
(*+
where
< c is the electrical angular freAuency (Xf& where f is the supply freAuency
< k
w
is the fundamental winding factor and
< N
s
is the phase winding turn number*

Hith the selection of these parameters& the already fixed pole number and the winding turn number
can be selected* 8uidelines for integer slot machines can be found in Table (*=& which gi%es the
possible combinations for different slot and pole configurations and phase winding turns for a
gi%en slot conductor number*
Table* (*=* Allowed winding turn number and slot number for a slots per pole per phase
qS winding with a gi%en slot conductor number :
N
and a pole pair number* N
s
is the
number of series<connected coil turns in one phase*
p
q :
N

( ! + = >
Q N
s

> ( ! + = >
( ( ( + > " ) (
! " ! > , ( = "
+ (+ + " ( > () (+
= !) = ) = () (= !)
> !> > ( " (+ !) !>
- +( - + ( (" != +(
" +" " > (+ !( +) +"
, =+ , " (- !> += =+
) >) ) () !) +) =) >)

The slot and pole combination can also be selected to produce a fractional slot winding* The
fractional slot design can be used to produce shorter end windings for less leakage* An additional
feature is the decreased axial dimension& which is an attracti%e property of this winding
=!


technology* 'ractioning can also reduce the machine inherent cogging torAue& which is a
characteristic feature of all full pitch windings* The fractional slot design also gi%es much more
freedom in the selection of the pole and slot configuration* The end winding is at shortest when the
pitch to the next slot is at shortest* The most common solution is a pitch of one or two slots*
'ractional slot windings of q S )*-=& *(=& *= or (*(= meet these criteria (Miller& ,,+)* The
fractional slot windings of this kind can also be wound cost effecti%ely with automatic winding
machines* The drawback of the fractional slot design is the subharmonics that cause additional
rotor losses and unbalanced magnetic pull in certain fractional slot arrangements* Hith the full
pitch winding& the risk of a faulty design is smaller& and the cogging torAue can be reduced by
appropriate skewing*
Hith the selected p& q& :
N
and Q it is possible to determine the winding factors of the machine* The
machine harmonic content can be %aried by a suitable selection of these %alues* 2n motors& the
back 1M' is usually slightly lower than the phase supply because of the resisti%e losses in the
stator* 7efore this& the winding arrangement has to be known to calculate the first harmonic
coefficient for effecti%e winding turns* 'or integer slot windings& the calculation of the winding
factor is a rather straightforward task* The eAuation for determining the winding factor k
wv
for a
.M machine is
k
wv
= k
dv
k
pv
k
sqv
(*=
where k
d%
is the distribution factor defined by the slot angle and the number of slots per pole per
phase& k
p
is the pitch factor for short<pitched windings& which reAuires a two<layer winding& and
k
sA%
is the skewing factor for the rotor or stator skewing* The calculation of winding factor can be&
written open as

k
w
== _
sin

u
2
qsin

u
2
_ _sin
w

2
_ _
sin (

sqv
2
)

sqv
2
_ (*>

where
<
u
is the slot angle p(EQ
< is the harmonic order
< q is the number of slots pre pole per phase
< W is the coil pitch
<
p
is the pole pitch and
<
sA%
is the applied skew angle of rotor or stator
The winding scheme can be used to shape the harmonic content of the produced current linkage
wa%e* As q increases& the coefficient of the fundamental wa%e decreases& but at the same time& the
harmonic effects are suppressed* 'or example& q S ( decreases the =
th
and -
th
harmonics by -+* P
in the distribution factor compared with the slots per pole and phase factor of q S *
=!


technology* 'ractioning can also reduce the machine inherent cogging torAue& which is a
characteristic feature of all full pitch windings* The fractional slot design also gi%es much more
freedom in the selection of the pole and slot configuration* The end winding is at shortest when the
pitch to the next slot is at shortest* The most common solution is a pitch of one or two slots*
'ractional slot windings of q S )*-=& *(=& *= or (*(= meet these criteria (Miller& ,,+)* The
fractional slot windings of this kind can also be wound cost effecti%ely with automatic winding
machines* The drawback of the fractional slot design is the subharmonics that cause additional
rotor losses and unbalanced magnetic pull in certain fractional slot arrangements* Hith the full
pitch winding& the risk of a faulty design is smaller& and the cogging torAue can be reduced by
appropriate skewing*
Hith the selected p& q& :
N
and Q it is possible to determine the winding factors of the machine* The
machine harmonic content can be %aried by a suitable selection of these %alues* 2n motors& the
back 1M' is usually slightly lower than the phase supply because of the resisti%e losses in the
stator* 7efore this& the winding arrangement has to be known to calculate the first harmonic
coefficient for effecti%e winding turns* 'or integer slot windings& the calculation of the winding
factor is a rather straightforward task* The eAuation for determining the winding factor k
wv
for a
.M machine is
k
wv
= k
dv
k
pv
k
sqv
(*=
where k
d%
is the distribution factor defined by the slot angle and the number of slots per pole per
phase& k
p
is the pitch factor for short<pitched windings& which reAuires a two<layer winding& and
k
sA%
is the skewing factor for the rotor or stator skewing* The calculation of winding factor can be&
written open as

k
w
== _
sin

u
2
qsin

u
2
_ _sin
w

2
_ _
sin (

sqv
2
)

sqv
2
_ (*>

where
<
u
is the slot angle p(EQ
< is the harmonic order
< q is the number of slots pre pole per phase
< W is the coil pitch
<
p
is the pole pitch and
<
sA%
is the applied skew angle of rotor or stator
The winding scheme can be used to shape the harmonic content of the produced current linkage
wa%e* As q increases& the coefficient of the fundamental wa%e decreases& but at the same time& the
harmonic effects are suppressed* 'or example& q S ( decreases the =
th
and -
th
harmonics by -+* P
in the distribution factor compared with the slots per pole and phase factor of q S *
=+


Short pitching can also be used to decrease the harmonic content of the supply %oltage* Short
pitching of +E= of the pole pitch remo%es the -
th
harmonic almost completely& and the short
pitching of +E- of the pole pitch suppresses the =th harmonic* 7ut as Table (*= shows& the slot
number of the machines increases as a multiple of six in integer slot machines& and thus& the short
pitching has to be a fraction of three or its multiple depending on the Gone width* 9ence& the
closest short pitches to reduce the -
th
and =
th
harmonics are =E> and (E!& respecti%ely* The pitching
effect on the harmonic content of the current linkage wa%e is presented in 'ig* (*(*


'ig* (*(* .itch factor as a function of short pitching for the
st
& !
rd
& =
th
& -
th
and ,
th

harmonics*
'or scaling purposes& it is more beneficial to choose the slot number as a multiple of ( for integer
slot winding machines in order to ha%e a larger selection of pole pairs* The right side of the table
gi%es the winding turns per one phase of the machine with respect to the slot conductor number*
9owe%er& as we can see& the selection for integer slot machines is Auite limited* The skewing
factor approaches unity when q increases and decreases as the slot number Q increases* The main
function of the skewing is to alle%iate the harmonic effects on the rotor surface*
Additional manipulation can be carried out with a parallel branching techniAue& which is especially
useful in larger machines in order to pro%ide low<%oltage machine structures* Hith two parallel
branches& the pole pairs can be separated to ha%e the same number of slot conductors& which cuts
)*= )*> )*- )*" )*,
<
<)*"
<)*>
<)*+
<)*(
)
)*(
)*+
)*>
)*"

Short pitching
.
i
t
c
h

f
a
c
t
o
r


st
!rd
=th
-th
,th
=+


Short pitching can also be used to decrease the harmonic content of the supply %oltage* Short
pitching of +E= of the pole pitch remo%es the -
th
harmonic almost completely& and the short
pitching of +E- of the pole pitch suppresses the =th harmonic* 7ut as Table (*= shows& the slot
number of the machines increases as a multiple of six in integer slot machines& and thus& the short
pitching has to be a fraction of three or its multiple depending on the Gone width* 9ence& the
closest short pitches to reduce the -
th
and =
th
harmonics are =E> and (E!& respecti%ely* The pitching
effect on the harmonic content of the current linkage wa%e is presented in 'ig* (*(*


'ig* (*(* .itch factor as a function of short pitching for the
st
& !
rd
& =
th
& -
th
and ,
th

harmonics*
'or scaling purposes& it is more beneficial to choose the slot number as a multiple of ( for integer
slot winding machines in order to ha%e a larger selection of pole pairs* The right side of the table
gi%es the winding turns per one phase of the machine with respect to the slot conductor number*
9owe%er& as we can see& the selection for integer slot machines is Auite limited* The skewing
factor approaches unity when q increases and decreases as the slot number Q increases* The main
function of the skewing is to alle%iate the harmonic effects on the rotor surface*
Additional manipulation can be carried out with a parallel branching techniAue& which is especially
useful in larger machines in order to pro%ide low<%oltage machine structures* Hith two parallel
branches& the pole pairs can be separated to ha%e the same number of slot conductors& which cuts
)*= )*> )*- )*" )*,
<
<)*"
<)*>
<)*+
<)*(
)
)*(
)*+
)*>
)*"

Short pitching
.
i
t
c
h

f
a
c
t
o
r


st
!rd
=th
-th
,th
==


the reAuired supply %oltage to half and doubles the current drawn for the same rated power& or the
slot per pole number can be increased to (:
N
to maintain the original current rating* 2t has to be
borne in mind that parallel connection may cause circulating currents as a result of geometrical
asymmetries in the coils& which tend to be harmful to the machine* The principle of parallel
branching techniAue is presented in 'ig* (*!*


'ig* (*!* 2llustration of parallel branches* A series<connected winding (aS) on the leftU a
winding with two parallel branches (aS() on the right*

.!., Magnetic circuit
After the determination of the winding scheme of the machine& the iron circuit dimensions ha%e to
be selected* The iron circuit dimensions also ha%e an impact on the desired current density rating&
since the determination of the yoke and teeth dimensioning has an effect on the slot area* The air
gap flux density in 1A* (*( is used to determine the iron circuit dimensions*
The magnetic circuit& including the permanent magnet itself& has to consume the current linkage
produced by the permanent magnet as AmpereFs law states*
The flux density limit depends on the soft magnetic material used& but generally& the design limit
for the electric sheet flux density is around *= T or below the material<specific knee of the BH
cur%e (Miller& ,,+)* 9igher flux density %alues reAuire a substantial increase in the current
linkage because of the material saturation* 9owe%er& to achie%e the highest possible torAue
densities& the flux densities ha%e to be brought to the material<specific limits*
#ower flux density %alues are ad%isable to gain head for flux boost in the iron circuit in order to
enhance reluctance torAue production* 2t is usually assumed that the flux flows completely through
the tooth at a certain moment* Thus& the peak %alue of the flux density B
`
d
in a tooth can be written&
by neglecting the flux through a slot& which is a fair assumption if saturation does not occur& as
follows

==


the reAuired supply %oltage to half and doubles the current drawn for the same rated power& or the
slot per pole number can be increased to (:
N
to maintain the original current rating* 2t has to be
borne in mind that parallel connection may cause circulating currents as a result of geometrical
asymmetries in the coils& which tend to be harmful to the machine* The principle of parallel
branching techniAue is presented in 'ig* (*!*


'ig* (*!* 2llustration of parallel branches* A series<connected winding (aS) on the leftU a
winding with two parallel branches (aS() on the right*

.!., Magnetic circuit
After the determination of the winding scheme of the machine& the iron circuit dimensions ha%e to
be selected* The iron circuit dimensions also ha%e an impact on the desired current density rating&
since the determination of the yoke and teeth dimensioning has an effect on the slot area* The air
gap flux density in 1A* (*( is used to determine the iron circuit dimensions*
The magnetic circuit& including the permanent magnet itself& has to consume the current linkage
produced by the permanent magnet as AmpereFs law states*
The flux density limit depends on the soft magnetic material used& but generally& the design limit
for the electric sheet flux density is around *= T or below the material<specific knee of the BH
cur%e (Miller& ,,+)* 9igher flux density %alues reAuire a substantial increase in the current
linkage because of the material saturation* 9owe%er& to achie%e the highest possible torAue
densities& the flux densities ha%e to be brought to the material<specific limits*
#ower flux density %alues are ad%isable to gain head for flux boost in the iron circuit in order to
enhance reluctance torAue production* 2t is usually assumed that the flux flows completely through
the tooth at a certain moment* Thus& the peak %alue of the flux density B
`
d
in a tooth can be written&
by neglecting the flux through a slot& which is a fair assumption if saturation does not occur& as
follows

=>


B
`
d
=

u
l'
k
Fc
l'b
d
B

(*-
where
<
u
is the slot pitch and
< k
'e
is the filling factor of the iron core
< b
d
is the tooth width
< l is the effecti%e stack length

which o%erestimates the flux density in the tooth area& because the flux is calculated for a surface
limited by the slot pitch* The %alue must be corrected closer to the actual %alue by subtracting the
flux flowing through the slot by assuming that the field strength at the boundary surface of iron is
uniform with the slot opening area* Thus& the amount of flux density to be remo%ed is
B
`
d,rcaI
= B
`
d
-_
+
u
k
Fc
b
d
-1 p
0
E

d
(*"

This can be sol%ed with the BH cur%e of the electric sheet metal& or if the saturation limit is not
close and the flux density is in the linear region of the BH cur%e& the field strength can be found by
combining the two crossing lines as
E

d
=
B
`
d
p
0
p
r
+_
+
u
k
Fc
b
d
-1 p
0

(*,

9owe%er& as stated abo%e& this eAuation applies to flux densities below the knee of the BH cur%e*
Hith the determined flux& the dimensions for the stator and rotor yoke can be calculated* The flux
of the pole flows through two separate parts of the yoke& and thus& the thickness of the yokes can
be determined

,rs
=

m
2k
Fc
l'B
`
,rs
(*)

where the subscripts s and r denote stator and rotor& respecti%ely* ;ow& to ha%e a full description of
the machine geometry& we only ha%e to determine the slot dimensions of the machine* There are
different geometries a%ailable depending on the winding types used* The exact dimensioning is
easiest with technical drawing programs& but a wedge<type slot pro%ides a practical choice for the
initial guess* The wedge angle should be selected to get straight teeth in order to maximiGe the slot
face area* Since G
N
and I
s
are known& the slot height can be determined with the information about
the desired current density J
s
and the slot number Q
s
*

=>


B
`
d
=

u
l'
k
Fc
l'b
d
B

(*-
where
<
u
is the slot pitch and
< k
'e
is the filling factor of the iron core
< b
d
is the tooth width
< l is the effecti%e stack length

which o%erestimates the flux density in the tooth area& because the flux is calculated for a surface
limited by the slot pitch* The %alue must be corrected closer to the actual %alue by subtracting the
flux flowing through the slot by assuming that the field strength at the boundary surface of iron is
uniform with the slot opening area* Thus& the amount of flux density to be remo%ed is
B
`
d,rcaI
= B
`
d
-_
+
u
k
Fc
b
d
-1 p
0
E

d
(*"

This can be sol%ed with the BH cur%e of the electric sheet metal& or if the saturation limit is not
close and the flux density is in the linear region of the BH cur%e& the field strength can be found by
combining the two crossing lines as
E

d
=
B
`
d
p
0
p
r
+_
+
u
k
Fc
b
d
-1 p
0

(*,

9owe%er& as stated abo%e& this eAuation applies to flux densities below the knee of the BH cur%e*
Hith the determined flux& the dimensions for the stator and rotor yoke can be calculated* The flux
of the pole flows through two separate parts of the yoke& and thus& the thickness of the yokes can
be determined

,rs
=

m
2k
Fc
l'B
`
,rs
(*)

where the subscripts s and r denote stator and rotor& respecti%ely* ;ow& to ha%e a full description of
the machine geometry& we only ha%e to determine the slot dimensions of the machine* There are
different geometries a%ailable depending on the winding types used* The exact dimensioning is
easiest with technical drawing programs& but a wedge<type slot pro%ides a practical choice for the
initial guess* The wedge angle should be selected to get straight teeth in order to maximiGe the slot
face area* Since G
N
and I
s
are known& the slot height can be determined with the information about
the desired current density J
s
and the slot number Q
s
*

=-

d
=
_

s
b
u1
+
b
d
2

(*

The dimensions of the iron circuit can be determined backwards to fit the desired flux densities in
the iron circuit by referring the magnetic %oltages of one magnetic path to the current linkage of
permanent magnets
u
m
= J = E
PM

PM
(*(
'or the calculation of inductances and air gap magnetic %oltages& the determination of the
eAui%alent air gap is mandatory* The eAui%alent magnetic air gap length is somewhat larger in
absolute %alue than the geometrical one& but since the air gap holds the greatest part of the
magnetic energy of the magnetiGing path& the e%aluation of this parameter is important* The
eAui%alent %alue of the air gap is necessary when determining the magnet height in the pole
according to the desired air gap flux density fixed at the beginning of the machine design
procedure* The Darter factor k
D
can be expressed by

k

=

u

u
-s
(*!

where s is the slot opening width
and


=
2

_oron
s
2o
-
2o
s
lo

1 -[
s
2o

2
_ (*+


This gi%es an eAui%alent air gap length of
o' = k

o (*=
This is the end of the basic design procedure as the basic geometry is fixed* The further
de%elopment on the machine model is carried out by the finite element method to determine the
magnet and pole area dimensions to achie%e the desired flux*

=-

d
=
_

s
b
u1
+
b
d
2

(*

The dimensions of the iron circuit can be determined backwards to fit the desired flux densities in
the iron circuit by referring the magnetic %oltages of one magnetic path to the current linkage of
permanent magnets
u
m
= J = E
PM

PM
(*(
'or the calculation of inductances and air gap magnetic %oltages& the determination of the
eAui%alent air gap is mandatory* The eAui%alent magnetic air gap length is somewhat larger in
absolute %alue than the geometrical one& but since the air gap holds the greatest part of the
magnetic energy of the magnetiGing path& the e%aluation of this parameter is important* The
eAui%alent %alue of the air gap is necessary when determining the magnet height in the pole
according to the desired air gap flux density fixed at the beginning of the machine design
procedure* The Darter factor k
D
can be expressed by

k

=

u

u
-s
(*!

where s is the slot opening width
and


=
2

_oron
s
2o
-
2o
s
lo

1 -[
s
2o

2
_ (*+


This gi%es an eAui%alent air gap length of
o' = k

o (*=
This is the end of the basic design procedure as the basic geometry is fixed* The further
de%elopment on the machine model is carried out by the finite element method to determine the
magnet and pole area dimensions to achie%e the desired flux*

="


. 4oss evaluation
The loss e%aluation is one of the most difficult& if not the most difficult phenomenon to e%aluate in
a machine design process* The losses generated in an electrical machine can be di%ided into
< iron losses& which are a combination of hysteresis and eddy current losses&
< /oule losses& which are produced in the windings of the electrical machines&
< mechanical losses& which include friction losses& produced by the %iscous
forces acting on the rotor and stator surfaces& bearing losses and
< additional losses& which include for instance external fan losses and leakage
losses*
2n low< and medium<speed permanent magnet machines& the stator /oule losses are usually the
most significant ones in siGe* 2n the normal speed range& the proportion of friction losses and stator
core losses of the total losses %ary in siGe depending on the machine siGe* Rotor losses are usually
Auite small& but as a result of the magnet sensiti%ity to temperature& this loss component is eAual in
importance to the stator /oule losses in order to achie%e a safely operating machine structure*

..! ?oule losses
The /oule losses in the stator copper are the easiest ones to predict as the only winding
arrangement in a permanent magnet machine is the armature winding* The /oule losses follow the
general rule for heat dissipation in a resisti%e medium affected by current as

ouIc
=

2
, (*>
where R
AD
is the alternating current resistance of the conducting circuit& and I is the RMS current
affecting the conducting path*

The alternating current resistance depends on the winding arrangement and the supply freAuency*
The two different mechanisms that increase the AD resistance %alue are the skin effect and the
proximity effect* 2n the skin effect& the increasing supply freAuency causes the current to
concentrate closer to the conductor surface& which increases the /oule losses as the effecti%e
conductor area decreases* 2n the proximity effect& the currents of ad3acent coils repel or attract each
other depending on whether the flow direction is the same or opposite& respecti%ely*

According to (5ogt& ,"!)& the free<space skin effect can be e%aluated by
o
sIn
=
1

,
(*-
which determines the depth at which the current density is decreased to e
<
of the current density
on the conductor surface* The skin effect for copper is presented in 'ig (*+*
="


. 4oss evaluation
The loss e%aluation is one of the most difficult& if not the most difficult phenomenon to e%aluate in
a machine design process* The losses generated in an electrical machine can be di%ided into
< iron losses& which are a combination of hysteresis and eddy current losses&
< /oule losses& which are produced in the windings of the electrical machines&
< mechanical losses& which include friction losses& produced by the %iscous
forces acting on the rotor and stator surfaces& bearing losses and
< additional losses& which include for instance external fan losses and leakage
losses*
2n low< and medium<speed permanent magnet machines& the stator /oule losses are usually the
most significant ones in siGe* 2n the normal speed range& the proportion of friction losses and stator
core losses of the total losses %ary in siGe depending on the machine siGe* Rotor losses are usually
Auite small& but as a result of the magnet sensiti%ity to temperature& this loss component is eAual in
importance to the stator /oule losses in order to achie%e a safely operating machine structure*

..! ?oule losses
The /oule losses in the stator copper are the easiest ones to predict as the only winding
arrangement in a permanent magnet machine is the armature winding* The /oule losses follow the
general rule for heat dissipation in a resisti%e medium affected by current as

ouIc
=

2
, (*>
where R
AD
is the alternating current resistance of the conducting circuit& and I is the RMS current
affecting the conducting path*

The alternating current resistance depends on the winding arrangement and the supply freAuency*
The two different mechanisms that increase the AD resistance %alue are the skin effect and the
proximity effect* 2n the skin effect& the increasing supply freAuency causes the current to
concentrate closer to the conductor surface& which increases the /oule losses as the effecti%e
conductor area decreases* 2n the proximity effect& the currents of ad3acent coils repel or attract each
other depending on whether the flow direction is the same or opposite& respecti%ely*

According to (5ogt& ,"!)& the free<space skin effect can be e%aluated by
o
sIn
=
1

,
(*-
which determines the depth at which the current density is decreased to e
<
of the current density
on the conductor surface* The skin effect for copper is presented in 'ig (*+*
=,



'ig* (*+* 2ndicati%e copper skin depth as a function of freAuency in free space* The copper
conducti%ity is =*,-R)
-
SEm*
2n an electrical machine& the coils are inside a high<permeability electric sheet& which causes the
flux linkage path to tra%el differently compared with a free space* Hith a large coil cross<sectional
area& the flux linkage tends to concentrate close to the slot opening area* The e%aluation of the AD
resistance le%el is more important in large form<wound machines and at high freAuencies* Smaller
machines benefit from random<wound coils& which naturally produce a thread mixing in the coils&
and smaller coil cross<sectional areas* The AD resistance %alue can be decreased by special
winding materials such as multi<threaded #itG wires or form<wound Roebel bars* Special winding
techniAues such as coil mixing and twisting are also useful (96m6l6inen& ()),)*
Dopper and aluminum are the most common conductor materials in electrical machine windings*
7ecause of the heating of the material& the resistance %alue must be treated with a temperature<
related coefficient o
T
& which depends on the winding material& because the winding temperatures
easily reach the %alue of =) MD and abo%e* Temperature coefficients of resisti%ity for different
metals are presented in Table (*>*


=) )) =) ()) (=) !)) !=) +)) +=) =))
!
+
=
>
-
"
,
)

'reAuency J9GL
S
k
i
n

d
e
p
t
h

J
m
m
L
=,



'ig* (*+* 2ndicati%e copper skin depth as a function of freAuency in free space* The copper
conducti%ity is =*,-R)
-
SEm*
2n an electrical machine& the coils are inside a high<permeability electric sheet& which causes the
flux linkage path to tra%el differently compared with a free space* Hith a large coil cross<sectional
area& the flux linkage tends to concentrate close to the slot opening area* The e%aluation of the AD
resistance le%el is more important in large form<wound machines and at high freAuencies* Smaller
machines benefit from random<wound coils& which naturally produce a thread mixing in the coils&
and smaller coil cross<sectional areas* The AD resistance %alue can be decreased by special
winding materials such as multi<threaded #itG wires or form<wound Roebel bars* Special winding
techniAues such as coil mixing and twisting are also useful (96m6l6inen& ()),)*
Dopper and aluminum are the most common conductor materials in electrical machine windings*
7ecause of the heating of the material& the resistance %alue must be treated with a temperature<
related coefficient o
T
& which depends on the winding material& because the winding temperatures
easily reach the %alue of =) MD and abo%e* Temperature coefficients of resisti%ity for different
metals are presented in Table (*>*


=) )) =) ()) (=) !)) !=) +)) +=) =))
!
+
=
>
-
"
,
)

'reAuency J9GL
S
k
i
n

d
e
p
t
h

J
m
m
L
>)


Table* (*>* Temperature coefficients of resisti%ity for metals typically used in electrical
machines*
Medium JKmL o
T
JPEMDL
Dopper )*)>"R)
<>
!*,R)
<!

Aluminium )*)("R)
<>
+R)
<!


The corrected formula for the DD resistance is thus

ouIc
=
,20
(1 +o

(I
2
-I
1
))
2
(*"
Since the temperature differences in electrical machines are moderate& a couple of hundred degrees
at maximum& the use of a constant temperature coefficient is 3ustified* 2n reality& the temperature
coefficient is not linear& but the accuracy is ne%ertheless usually more than enough for engineering
applications* 'or example& the resisti%ity temperature coefficient of copper follows the eAuation

2u
=
1
2SS.S4 +I
2

1u
(*,
Thus& the temperature coefficients for (-!*=& (,!*= and !,!*= C are )*))+("& )*))!,+ and
)*))!))& respecti%ely* 1A* (*, is %alid for temperatures abo%e -!*= C*
Since the /oule losses are proportional to the second power of the current& the eAuation forms the
basis for understanding how high<efficiency machines should be built* A good principle in
electrical machine design& at least for moderate power ratings& is to keep the /oule losses less than
or eAual to the iron losses* 2t usually helps in keeping the efficiency high& e%en at low operating
speeds*
/oule losses are also generated in permanent magnets& because of induced current harmonics*
There are numerous approaches to analytically estimate /oule losses in a .M materialU howe%er&
complex pole structures reAuire a finite<element<based approach* 1mbedding of the magnets
reduces the effects of current harmonics on the magnets& because rotor laminations damp the flux
density %ariations before they reach the magnets* 2n the surface magnet design there is no such
protection& and different techniAues ha%e to be applied if necessary* Magnet slicing has pro%ed to
be a useful techniAue (.olinder et al*& ,,-)& and it can be implemented either in the axial or
tangential direction*


>)


Table* (*>* Temperature coefficients of resisti%ity for metals typically used in electrical
machines*
Medium JKmL o
T
JPEMDL
Dopper )*)>"R)
<>
!*,R)
<!

Aluminium )*)("R)
<>
+R)
<!


The corrected formula for the DD resistance is thus

ouIc
=
,20
(1 +o

(I
2
-I
1
))
2
(*"
Since the temperature differences in electrical machines are moderate& a couple of hundred degrees
at maximum& the use of a constant temperature coefficient is 3ustified* 2n reality& the temperature
coefficient is not linear& but the accuracy is ne%ertheless usually more than enough for engineering
applications* 'or example& the resisti%ity temperature coefficient of copper follows the eAuation

2u
=
1
2SS.S4 +I
2

1u
(*,
Thus& the temperature coefficients for (-!*=& (,!*= and !,!*= C are )*))+("& )*))!,+ and
)*))!))& respecti%ely* 1A* (*, is %alid for temperatures abo%e -!*= C*
Since the /oule losses are proportional to the second power of the current& the eAuation forms the
basis for understanding how high<efficiency machines should be built* A good principle in
electrical machine design& at least for moderate power ratings& is to keep the /oule losses less than
or eAual to the iron losses* 2t usually helps in keeping the efficiency high& e%en at low operating
speeds*
/oule losses are also generated in permanent magnets& because of induced current harmonics*
There are numerous approaches to analytically estimate /oule losses in a .M materialU howe%er&
complex pole structures reAuire a finite<element<based approach* 1mbedding of the magnets
reduces the effects of current harmonics on the magnets& because rotor laminations damp the flux
density %ariations before they reach the magnets* 2n the surface magnet design there is no such
protection& and different techniAues ha%e to be applied if necessary* Magnet slicing has pro%ed to
be a useful techniAue (.olinder et al*& ,,-)& and it can be implemented either in the axial or
tangential direction*


>


.. Iron losses
2ron losses are probably the most complex loss component of all* The foundations of the loss
e%aluation in an iron core were laid originally by D*. SteinmetG in On the law of Hvsteresis in
",(* According to (SteinmetG& ,"+)& iron losses can be e%aluated by an analytical eAuation
w
Fc
=
h
B
`
1.6
+
c
B
`
2
, (*()
where
< k
h
is the hysteresis coefficient&
< k
e
is the eddy current coefficient&
< f is the affecting freAuency and
< B
`
is the peak %alue of flux density in the material*
The eAuation describes how much energy is lost per cycle in the material* The power loss can be
obtained by multiplying the 1A* (*() by freAuency* The eAuation implies that the losses are
proportional to the area inside the hysteresis loop* The hysteresis loop defines the maximum flux
density limits for gi%en field strengths ending to a certain %alue in both of the odd Auadrants&
which are the saturation limits for the magnetic flux density* Although this eAuation is Auite a
robust approach& it gi%es a generaliGed idea of the phenomenon in the material* According to the
SteinmetG eAuation& the hysteresis losses are proportional to a material<specific hysteresis constant&
and the applied flux density to a certain power& which %aries between *> and ( in the literature*
The latter part presents the eddy current losses in relation to a certain eddy current factor and to the
freAuency of the phenomenon* Many other attempts to determine the magnitude of core losses
ha%e been presented e%er since* A moderniGed %ersion of iron loss hand calculation relies on the
material data pro%ided by electric sheet manufacturers* 2n (5ogt& ,"!)& the eAuation for iron
losses is gi%en in the form

orc
= m
Fc

u
u
u1.5
[
B
`
1
1.5

2
,
(*(
where
< c
u*=
is the specific total loss of the material at *= T and the freAuency under
obser%ation JHEkgL&
< m
'e
is the core mass and
< k
u
is the geometry<related coefficient gi%en in Table (*-*

The loss formulation
u
u
u1.5
can be replaced by a more extensi%e formula& which describes the
harmonic hysteresis losses and the eddy current losses separately

u
:
u1.5
= :
h1,5

h
_

s
Su
] +:
c1,5

c
_

s
Su
]
2

(*((
where
>


.. Iron losses
2ron losses are probably the most complex loss component of all* The foundations of the loss
e%aluation in an iron core were laid originally by D*. SteinmetG in On the law of Hvsteresis in
",(* According to (SteinmetG& ,"+)& iron losses can be e%aluated by an analytical eAuation
w
Fc
=
h
B
`
1.6
+
c
B
`
2
, (*()
where
< k
h
is the hysteresis coefficient&
< k
e
is the eddy current coefficient&
< f is the affecting freAuency and
< B
`
is the peak %alue of flux density in the material*
The eAuation describes how much energy is lost per cycle in the material* The power loss can be
obtained by multiplying the 1A* (*() by freAuency* The eAuation implies that the losses are
proportional to the area inside the hysteresis loop* The hysteresis loop defines the maximum flux
density limits for gi%en field strengths ending to a certain %alue in both of the odd Auadrants&
which are the saturation limits for the magnetic flux density* Although this eAuation is Auite a
robust approach& it gi%es a generaliGed idea of the phenomenon in the material* According to the
SteinmetG eAuation& the hysteresis losses are proportional to a material<specific hysteresis constant&
and the applied flux density to a certain power& which %aries between *> and ( in the literature*
The latter part presents the eddy current losses in relation to a certain eddy current factor and to the
freAuency of the phenomenon* Many other attempts to determine the magnitude of core losses
ha%e been presented e%er since* A moderniGed %ersion of iron loss hand calculation relies on the
material data pro%ided by electric sheet manufacturers* 2n (5ogt& ,"!)& the eAuation for iron
losses is gi%en in the form

orc
= m
Fc

u
u
u1.5
[
B
`
1
1.5

2
,
(*(
where
< c
u*=
is the specific total loss of the material at *= T and the freAuency under
obser%ation JHEkgL&
< m
'e
is the core mass and
< k
u
is the geometry<related coefficient gi%en in Table (*-*

The loss formulation
u
u
u1.5
can be replaced by a more extensi%e formula& which describes the
harmonic hysteresis losses and the eddy current losses separately

u
:
u1.5
= :
h1,5

h
_

s
Su
] +:
c1,5

c
_

s
Su
]
2

(*((
where
>(


< k
h
is the hysteresis

coefficient&
< k
e
is the eddy current

coefficient&
< v
h&=
is the hysteresis core loss component JHEkgL and
< v
e&=
is the eddy current core loss component JHEkgL*
Table* (*-* 8eometry<related core loss coefficients for 1As* (*( and (*(( (5ogt& ,"!)*
Machine type
Teeth ?oke
k
u
k
h
k
e
k
e
k
u

Synchronous machine (*)
*=
*"
(pS)
(<(*(
(p_)
*==<*-
Asynchronous
machine
*"
DD machine (*= *= (*= <

2n 1As* (*( and (*((& the specific total loss of the material is usually pro%ided by the sheet
manufacturer* 9owe%er& the specific total loss can be fitted to the existing manufacturer data* 'or
example& M(-)<!=A by Surahammars 7ruks has specific losses according to Table (*"*
Table* (*"* M(-)<!=A specific loss data according to the Surahammars data sheet*
B
Specific loss JHEkgL
=) 9G )) 9G ()) 9G +)) 9G
)*) )*)!) )*)+) )*),) )*()
)*() )*)-) )*>) )*!-) )*,()
)*!) )*!) )*!+) )*-,) *,,)
)*+) )*(() )*==) *!) !*!!)
)*=) )*!) )*")) *,) +*,+)
)*>) )*+!) *)>) (*>) >*"+)
)*-) )*=+) *!") !*!,) ,*)))
)*") )*>") *-!) +*(>) *+)
)*,) )*"!) (*)) =*(!) +*()
*)) *)) (*=) >*!)) -*!)
*) *()) (*,") -*=) ()*,)
*() *+() !*=) "*"") (+*,)
*!) *-)) +*=) )*=) (,*=)
*+) (*() +*,-) (*=) !=*+)
*=) (*+-) =*,() +*,) +*")
*>) (*")) < < <
*-) !*)=) < < <
*" !*(= < < <

>(


< k
h
is the hysteresis

coefficient&
< k
e
is the eddy current

coefficient&
< v
h&=
is the hysteresis core loss component JHEkgL and
< v
e&=
is the eddy current core loss component JHEkgL*
Table* (*-* 8eometry<related core loss coefficients for 1As* (*( and (*(( (5ogt& ,"!)*
Machine type
Teeth ?oke
k
u
k
h
k
e
k
e
k
u

Synchronous machine (*)
*=
*"
(pS)
(<(*(
(p_)
*==<*-
Asynchronous
machine
*"
DD machine (*= *= (*= <

2n 1As* (*( and (*((& the specific total loss of the material is usually pro%ided by the sheet
manufacturer* 9owe%er& the specific total loss can be fitted to the existing manufacturer data* 'or
example& M(-)<!=A by Surahammars 7ruks has specific losses according to Table (*"*
Table* (*"* M(-)<!=A specific loss data according to the Surahammars data sheet*
B
Specific loss JHEkgL
=) 9G )) 9G ()) 9G +)) 9G
)*) )*)!) )*)+) )*),) )*()
)*() )*)-) )*>) )*!-) )*,()
)*!) )*!) )*!+) )*-,) *,,)
)*+) )*(() )*==) *!) !*!!)
)*=) )*!) )*")) *,) +*,+)
)*>) )*+!) *)>) (*>) >*"+)
)*-) )*=+) *!") !*!,) ,*)))
)*") )*>") *-!) +*(>) *+)
)*,) )*"!) (*)) =*(!) +*()
*)) *)) (*=) >*!)) -*!)
*) *()) (*,") -*=) ()*,)
*() *+() !*=) "*"") (+*,)
*!) *-)) +*=) )*=) (,*=)
*+) (*() +*,-) (*=) !=*+)
*=) (*+-) =*,() +*,) +*")
*>) (*")) < < <
*-) !*)=) < < <
*" !*(= < < <

>!


1Auation (*(, can be fitted to the existing data in Table (*" by selecting the hysteresis and eddy<
current components v
h*=
and v
e*=
as *=) and )*,- HEkg& respecti%ely* The additional eddy current
coefficient needs a %alue of its own for each freAuency* The best fitting is achie%ed with the
coefficient %alues *)& )*-& )*>& and )*= for the freAuencies =)& ))& ()) and +)) 9G& respecti%ely*
The fitting results are presented in 'ig* (*=*

'ig* (*=* Specific losses of M(-)<=)A according to Table (*" data fitted with the core loss
coefficients of Table (*-*
2n the case of known harmonic content in the flux density& the material<specific constants k
e
can be
calculated more accurately by the eAuation

c
=
_ nB
`
n
2
1
B
`

2
(*(!
1Auation (*(! can be used together with the finite<element<based design programs* A similar
approach has also been presented in (.yrh0nen et al*& ())") for the total core losses by eAuation

Fc
=
h,c
B
`
2
m
Fc
(*(+
where
h&e
is the material<specific total loss per kg& including hysteresis and eddy currents& and m
'e

is the mass of the iron circuit* The eAuation can be used to determine losses in each indi%idual part
>!


1Auation (*(, can be fitted to the existing data in Table (*" by selecting the hysteresis and eddy<
current components v
h*=
and v
e*=
as *=) and )*,- HEkg& respecti%ely* The additional eddy current
coefficient needs a %alue of its own for each freAuency* The best fitting is achie%ed with the
coefficient %alues *)& )*-& )*>& and )*= for the freAuencies =)& ))& ()) and +)) 9G& respecti%ely*
The fitting results are presented in 'ig* (*=*

'ig* (*=* Specific losses of M(-)<=)A according to Table (*" data fitted with the core loss
coefficients of Table (*-*
2n the case of known harmonic content in the flux density& the material<specific constants k
e
can be
calculated more accurately by the eAuation

c
=
_ nB
`
n
2
1
B
`

2
(*(!
1Auation (*(! can be used together with the finite<element<based design programs* A similar
approach has also been presented in (.yrh0nen et al*& ())") for the total core losses by eAuation

Fc
=
h,c
B
`
2
m
Fc
(*(+
where
h&e
is the material<specific total loss per kg& including hysteresis and eddy currents& and m
'e

is the mass of the iron circuit* The eAuation can be used to determine losses in each indi%idual part
>+


of the machine& but the problem is that the %alue
h&e
determined by the material manufacturer is
%alid only for sinusoidally %arying flux densities* This problem is usually sol%ed by using
empirically determined loss coefficients presented in Table (*,*
Table* (*,* Dore loss coefficient
h&e
according to the machine type for 1A* (*!*
Machine type ?oke Teeth
Synchronous machine *-@(*= *=@*"
Asynchronous machine *-@(*= *=@*"

Approaches to analytically sol%e eddy current and hysteresis losses are numerous* Dommon to all
these is that the eAuations are more or less based on the classical definitions for eddy current and
hysteresis losses presented abo%e* The problem is that with the modern freAuency<con%erter<fed
systems& the loss e%aluation and material properties are not accurate for classical eAuations*
1xperimental studies ha%e shown a significant increase in core losses with the .HM excitation
(Tutkun& ())()& (7oglietti et al*& ,,)& (7oglietti et al*& ())!) * The actual power dissipation in
freAuency con%erter dri%es is considerably higher than for the pure sinusoidal wa%eform case
owing to the harmonic wa%es in the supply current wa%eform* The current causes the same
distortion to the flux produced in the coils& and thus& increases the amount of induced iron losses*
To minimiGe the .HM effect& the modulation index should be close to unity and the switching
freAuency abo%e = k9G (7oglietti et al*& ,,!)& (7oglietti et al*& ,,=)*

.., 2ontact friction losses in bearings
The bearing losses can usually be determined& depending on the application and bearing type& by
the information pro%ided by the manufacturer* Some manufacturers& such as SC'& pro%ide
analytical tools for the rotating loss e%aluation of bearings* The following analysis of the bearing
friction losses is based on ball bearing structures& because single<row deep groo%e ball bearings are
the most common bearing type used in electrical machines*
'or bearing loads less than ) P of the allowed dynamical loading& in normal operating conditions
and with sufficient lubrication& the rotating friction losses of a bearing can be estimated by

IrIcton
= u.SI
B
(*(=
where
< I is the eAui%alent bearing load J;L&
< D
7
is the bearing bore diameter JmL&
>+


of the machine& but the problem is that the %alue
h&e
determined by the material manufacturer is
%alid only for sinusoidally %arying flux densities* This problem is usually sol%ed by using
empirically determined loss coefficients presented in Table (*,*
Table* (*,* Dore loss coefficient
h&e
according to the machine type for 1A* (*!*
Machine type ?oke Teeth
Synchronous machine *-@(*= *=@*"
Asynchronous machine *-@(*= *=@*"

Approaches to analytically sol%e eddy current and hysteresis losses are numerous* Dommon to all
these is that the eAuations are more or less based on the classical definitions for eddy current and
hysteresis losses presented abo%e* The problem is that with the modern freAuency<con%erter<fed
systems& the loss e%aluation and material properties are not accurate for classical eAuations*
1xperimental studies ha%e shown a significant increase in core losses with the .HM excitation
(Tutkun& ())()& (7oglietti et al*& ,,)& (7oglietti et al*& ())!) * The actual power dissipation in
freAuency con%erter dri%es is considerably higher than for the pure sinusoidal wa%eform case
owing to the harmonic wa%es in the supply current wa%eform* The current causes the same
distortion to the flux produced in the coils& and thus& increases the amount of induced iron losses*
To minimiGe the .HM effect& the modulation index should be close to unity and the switching
freAuency abo%e = k9G (7oglietti et al*& ,,!)& (7oglietti et al*& ,,=)*

.., 2ontact friction losses in bearings
The bearing losses can usually be determined& depending on the application and bearing type& by
the information pro%ided by the manufacturer* Some manufacturers& such as SC'& pro%ide
analytical tools for the rotating loss e%aluation of bearings* The following analysis of the bearing
friction losses is based on ball bearing structures& because single<row deep groo%e ball bearings are
the most common bearing type used in electrical machines*
'or bearing loads less than ) P of the allowed dynamical loading& in normal operating conditions
and with sufficient lubrication& the rotating friction losses of a bearing can be estimated by

IrIcton
= u.SI
B
(*(=
where
< I is the eAui%alent bearing load J;L&
< D
7
is the bearing bore diameter JmL&
>=


< is the constant friction coefficient for bearing according to Table (*) and
< ` is the angular %elocity of the bearing*

Table* (*)* SC' friction coefficients for non<sealed bearings JSC'& ()(L*
7earing type 'riction coefficient
Deep groo%e ball bearing )*))=
Angular contact ball bearing singlea
double row
)*))() E )*))(+
Self<aligning ball bearings )*)))


..4 Additional losses
Additional losses are electromagnetic losses that are not included in the iron and /oule losses* The
additional loss generation originates from the leakage terms& which cause second<order loss
components in the machine structures* Depending on supply current and its harmonic content&
additional losses can be grouped to
< Additional no<load losses that are caused by flux %ariations in the main flux
caused by the stator and rotor permeance %ariations and
< additional load losses that result from the combined effect of the current
linkage wa%eform and the machine leakage components*
7oth groups generate high<freAuency surface losses in the annulus area* The difference is that the
ma3ority of the no<load surface losses are concentrated on the rotor surface* #oss generation is
strengthened by added skewing* The relati%e motion between the rotor and the stator also
generates additional losses in the teeth iron as a result of the main flux pulsation* #oad<oriented
leakage fluxes react in the same way* #oad<oriented losses are also generated in the stator
windings because of the slot leakage component* 2n addition& the end<winding leakage component
interacts with the stator core ends and the machine end space structural parts thereby generating
losses*
There is no direct and accurate method to determine the additional losses of a machine* 9owe%er&
the 21D standard 1; >))!+<( determines that the additional losses are from )* to )*( P of the
input power for salient pole synchronous machines and from )*)= to )*= P for non<salient pole
synchronous machines*

>=


< is the constant friction coefficient for bearing according to Table (*) and
< ` is the angular %elocity of the bearing*

Table* (*)* SC' friction coefficients for non<sealed bearings JSC'& ()(L*
7earing type 'riction coefficient
Deep groo%e ball bearing )*))=
Angular contact ball bearing singlea
double row
)*))() E )*))(+
Self<aligning ball bearings )*)))


..4 Additional losses
Additional losses are electromagnetic losses that are not included in the iron and /oule losses* The
additional loss generation originates from the leakage terms& which cause second<order loss
components in the machine structures* Depending on supply current and its harmonic content&
additional losses can be grouped to
< Additional no<load losses that are caused by flux %ariations in the main flux
caused by the stator and rotor permeance %ariations and
< additional load losses that result from the combined effect of the current
linkage wa%eform and the machine leakage components*
7oth groups generate high<freAuency surface losses in the annulus area* The difference is that the
ma3ority of the no<load surface losses are concentrated on the rotor surface* #oss generation is
strengthened by added skewing* The relati%e motion between the rotor and the stator also
generates additional losses in the teeth iron as a result of the main flux pulsation* #oad<oriented
leakage fluxes react in the same way* #oad<oriented losses are also generated in the stator
windings because of the slot leakage component* 2n addition& the end<winding leakage component
interacts with the stator core ends and the machine end space structural parts thereby generating
losses*
There is no direct and accurate method to determine the additional losses of a machine* 9owe%er&
the 21D standard 1; >))!+<( determines that the additional losses are from )* to )*( P of the
input power for salient pole synchronous machines and from )*)= to )*= P for non<salient pole
synchronous machines*

>>


., -eat transfer
$nderstanding of the heat transfer and the loss generation mechanisms is essential in electrical
machine design* The increasing demand for high<efficiency electrical machines reAuires a careful
design process to achie%e low material consumption and a high torAue density* The design process
of high end electrical machines in%ol%es an increasing amount of consideration of the mechanical
and thermal stresses in the machine* This de%elopment is emphasiGed by the fact that electrical
dri%e systems are expanding to different applications* :ne ma3or dri%ing force has been the
electrification of power transmission in mo%ing machines& %ehicles and %essels*
The design of the electrical circuit is a small part of the design process& e%en though it reAuires a
complicated iteration process as electrical machines are nowadays designed to be application<
specific rather than general purpose machines*

.,.! *undamentals of heat transfer
The general theory recogniGes three different heat transfer methodsB radiation& con%ection and
conduction* 2n electrical machines& the two latter methods are the most important ones* 9eat
transfer by radiation is usually neglected& but it should be taken into account if significant
temperature differences are expected on the radiating surfaces& because the radiation heat transfer
is proportional to the fourth power of the absolute body temperature* Donduction is probably the
most important heat transfer method due to the fact that electrical machines are built mainly of
metallic materials& which are known for their good heat transfer properties* 9owe%er& the
insulation materials limit the heat transfer from the electrical machine to the surrounding medium*
The importance of con%ection in the total heat transfer process increases as the rotating speed
grows andEor if a forced con%ection method is applied to the system*
The heat transfer is defined comprehensi%ely by the four laws of thermodynamics* 2n the design of
electrical machines& the first and second law are of most importance*
The first law of thermodynamics states that the energy of a system can be changed by work done
on or by the system or by applying a heat flow through the system boundaries* The work done by
the system can be di%ided into different partsB mechanical work and work done by the system
boundaries (e*g* pressure change)*
The second law& which states that entropy increases o%er time& means that the heat always flows
from a hotter body to a colder body* The study of the thermal beha%iour of materials is based on
the heat transfer eAuation& which is known as the 'ourier eAuation*
The usual method in the thermal modelling is to apply a thermal network to a specific geometry
for steady<state operation* The model can be further impro%ed by the introduction of a thermal
capacitance matrix& which enables a dynamic study of the system* The networks are usually simple
and fast to sol%e* This is mainly due to the simplification of the geometric parameters and
composition of larger entities* Another method is to use commercial programs such as 'luxb or
>>


., -eat transfer
$nderstanding of the heat transfer and the loss generation mechanisms is essential in electrical
machine design* The increasing demand for high<efficiency electrical machines reAuires a careful
design process to achie%e low material consumption and a high torAue density* The design process
of high end electrical machines in%ol%es an increasing amount of consideration of the mechanical
and thermal stresses in the machine* This de%elopment is emphasiGed by the fact that electrical
dri%e systems are expanding to different applications* :ne ma3or dri%ing force has been the
electrification of power transmission in mo%ing machines& %ehicles and %essels*
The design of the electrical circuit is a small part of the design process& e%en though it reAuires a
complicated iteration process as electrical machines are nowadays designed to be application<
specific rather than general purpose machines*

.,.! *undamentals of heat transfer
The general theory recogniGes three different heat transfer methodsB radiation& con%ection and
conduction* 2n electrical machines& the two latter methods are the most important ones* 9eat
transfer by radiation is usually neglected& but it should be taken into account if significant
temperature differences are expected on the radiating surfaces& because the radiation heat transfer
is proportional to the fourth power of the absolute body temperature* Donduction is probably the
most important heat transfer method due to the fact that electrical machines are built mainly of
metallic materials& which are known for their good heat transfer properties* 9owe%er& the
insulation materials limit the heat transfer from the electrical machine to the surrounding medium*
The importance of con%ection in the total heat transfer process increases as the rotating speed
grows andEor if a forced con%ection method is applied to the system*
The heat transfer is defined comprehensi%ely by the four laws of thermodynamics* 2n the design of
electrical machines& the first and second law are of most importance*
The first law of thermodynamics states that the energy of a system can be changed by work done
on or by the system or by applying a heat flow through the system boundaries* The work done by
the system can be di%ided into different partsB mechanical work and work done by the system
boundaries (e*g* pressure change)*
The second law& which states that entropy increases o%er time& means that the heat always flows
from a hotter body to a colder body* The study of the thermal beha%iour of materials is based on
the heat transfer eAuation& which is known as the 'ourier eAuation*
The usual method in the thermal modelling is to apply a thermal network to a specific geometry
for steady<state operation* The model can be further impro%ed by the introduction of a thermal
capacitance matrix& which enables a dynamic study of the system* The networks are usually simple
and fast to sol%e* This is mainly due to the simplification of the geometric parameters and
composition of larger entities* Another method is to use commercial programs such as 'luxb or
>-


MotorDadb by Dedrat for the e%aluation of the thermal beha%iour of the machine* The
commercial programs would be suitable for this purpose& but they are not always a%ailable and
include some generaliGations that cannot be changed*
1%en though the commercial programs are more accurate than analytical models& at least to some
extent& there are ma3or simplifications in both of the methods*
< The orientation of the motor is usually in the axial direction in the (D plane&
because the main thermal path is considered to be radial*
< 9eat transfer is considered uniform in the radial direction*
< Radiation is usually neglected*
< .hysical properties of materials are considered isotropic*

Thermal networks are usually built such that the geometry is considered as a two<dimensional
plane with the exception of the windings* The temperatures are thus a%eraged& and the total heat
beha%iour is generaliGed* To increase the resolution of the thermal beha%iour& the node number has
to be increased* 9ence& the implementation of the grid and calculations become arduous* Some
automated routines should thus be performed* Hhen increasing the accuracy of the system& it is not
reasonable to calculate the machine geometry as a whole* Similarly as in magnetic calculation& one
pole pitch is usually enough for the determination of thermal properties* To understand the
mechanism of heat transfer in more detail& it is necessary to know the basic laws of
thermodynamics and the main heat transfer methods*

.,. 2onduction
Donduction is the heat transfer method that transfers energy from a higher potential to a lower
potential in order to reach an eAuilibrium state of an eAual potential le%el* :n a microscopic scale&
conduction is the heat transfer method inside solids& liAuids and gases* The molecular motion
causes a con%ection phenomenon in liAuid and gaseous materials& but the heat transfer between
indi%idual particles takes place in the form of conduction heat transfer*
The conduction heat transfer can be expressed in numerous ways* An engineering approach is
pro%ided by rate eAuations& which are used to determine the amount of energy transferred in a
certain time* 'or heat conduction this is known as 'ourier^s law* 'or one<dimensional heat flow it
yields


q''
x
= -k
JI
Jx

(*(>

where q
x
is the heat flux density JHEm
(
L& which is proportional to the thermal conducti%ity k and
the temperature difference cT and in%ersely proportional to the tra%elled distance cx in the x<
>-


MotorDadb by Dedrat for the e%aluation of the thermal beha%iour of the machine* The
commercial programs would be suitable for this purpose& but they are not always a%ailable and
include some generaliGations that cannot be changed*
1%en though the commercial programs are more accurate than analytical models& at least to some
extent& there are ma3or simplifications in both of the methods*
< The orientation of the motor is usually in the axial direction in the (D plane&
because the main thermal path is considered to be radial*
< 9eat transfer is considered uniform in the radial direction*
< Radiation is usually neglected*
< .hysical properties of materials are considered isotropic*

Thermal networks are usually built such that the geometry is considered as a two<dimensional
plane with the exception of the windings* The temperatures are thus a%eraged& and the total heat
beha%iour is generaliGed* To increase the resolution of the thermal beha%iour& the node number has
to be increased* 9ence& the implementation of the grid and calculations become arduous* Some
automated routines should thus be performed* Hhen increasing the accuracy of the system& it is not
reasonable to calculate the machine geometry as a whole* Similarly as in magnetic calculation& one
pole pitch is usually enough for the determination of thermal properties* To understand the
mechanism of heat transfer in more detail& it is necessary to know the basic laws of
thermodynamics and the main heat transfer methods*

.,. 2onduction
Donduction is the heat transfer method that transfers energy from a higher potential to a lower
potential in order to reach an eAuilibrium state of an eAual potential le%el* :n a microscopic scale&
conduction is the heat transfer method inside solids& liAuids and gases* The molecular motion
causes a con%ection phenomenon in liAuid and gaseous materials& but the heat transfer between
indi%idual particles takes place in the form of conduction heat transfer*
The conduction heat transfer can be expressed in numerous ways* An engineering approach is
pro%ided by rate eAuations& which are used to determine the amount of energy transferred in a
certain time* 'or heat conduction this is known as 'ourier^s law* 'or one<dimensional heat flow it
yields


q''
x
= -k
JI
Jx

(*(>

where q
x
is the heat flux density JHEm
(
L& which is proportional to the thermal conducti%ity k and
the temperature difference cT and in%ersely proportional to the tra%elled distance cx in the x<
>"


direction* 1Auation (*!( is the basis of the heat conduction calculations* The eAuation can be
extended to a more general form& but in engineering calculations& which usually in%ol%e a lumped<
parameter techniAue in thermal conduction solutions& the one<directional plane eAuation is more
useful* Table (* presents thermal conducti%ities of common engineering materials*
Table* (** Thermal characteristics of common engineering materials (2ncropera& ())-)*
Material
k JHEm
(
CL
(at !))C)
k JHEm
(
CL
(at +))C)
Darbon steels
Mn d *) P
Si d )* P
>)*= =>*-
*)P d Mn d *> P
)* PW Si d )*> P
=*, +,*"
Mn d *) P
)* PW Si d )*> P
+ +(*(
Stainless steels
A2S2 !)( =* -*!
A2S2 !> !*+ =*(
Aluminium ()(+<T> -- ">
Dold<rolled electric
sheet
()<!) <
Dopper 2ndustrial grade !>)<!"> <
Hater )*>!
)*>"" ((*++=
bars)
Air ((*!R)
<!
!!*"R)
<!

Silicon paste <
Sil%er paste _ (() <
.olyester resin )*=@)*() <
1poxy resin )*!@)*> <
8lass fibre )*( <
Darbon fibre )*= <

.,., &adiation
Radiation is the only heat transfer method that does not reAuire any medium for heat transfer* The
radiation heat transfer is electromagnetic radiation& the wa%elengths of which %ary between )* and
)) em* The wa%elength range includes %isible light and the infrared region& and additionally& the
low<freAuency ultra%iolet region* The absorbed radiation heat depends on the material and the
material surface properties* 2n electrical machines& the reflecti%e surface can be assumed opaAue&
which means that part of the radiation is absorbed and part is reflected* The reason for this is that
no real material can completely absorb radiation heat& but part of the heat is always reflected*
Thus& there is an absorption factor & which is a reference of the medium absorption to an ideal
black body& which completely absorbs the heat exposed* The sum of radiation heat absorbed and
reflected by an opaAue medium is unity*
[ +p = 1 (*(-
>"


direction* 1Auation (*!( is the basis of the heat conduction calculations* The eAuation can be
extended to a more general form& but in engineering calculations& which usually in%ol%e a lumped<
parameter techniAue in thermal conduction solutions& the one<directional plane eAuation is more
useful* Table (* presents thermal conducti%ities of common engineering materials*
Table* (** Thermal characteristics of common engineering materials (2ncropera& ())-)*
Material
k JHEm
(
CL
(at !))C)
k JHEm
(
CL
(at +))C)
Darbon steels
Mn d *) P
Si d )* P
>)*= =>*-
*)P d Mn d *> P
)* PW Si d )*> P
=*, +,*"
Mn d *) P
)* PW Si d )*> P
+ +(*(
Stainless steels
A2S2 !)( =* -*!
A2S2 !> !*+ =*(
Aluminium ()(+<T> -- ">
Dold<rolled electric
sheet
()<!) <
Dopper 2ndustrial grade !>)<!"> <
Hater )*>!
)*>"" ((*++=
bars)
Air ((*!R)
<!
!!*"R)
<!

Silicon paste <
Sil%er paste _ (() <
.olyester resin )*=@)*() <
1poxy resin )*!@)*> <
8lass fibre )*( <
Darbon fibre )*= <

.,., &adiation
Radiation is the only heat transfer method that does not reAuire any medium for heat transfer* The
radiation heat transfer is electromagnetic radiation& the wa%elengths of which %ary between )* and
)) em* The wa%elength range includes %isible light and the infrared region& and additionally& the
low<freAuency ultra%iolet region* The absorbed radiation heat depends on the material and the
material surface properties* 2n electrical machines& the reflecti%e surface can be assumed opaAue&
which means that part of the radiation is absorbed and part is reflected* The reason for this is that
no real material can completely absorb radiation heat& but part of the heat is always reflected*
Thus& there is an absorption factor & which is a reference of the medium absorption to an ideal
black body& which completely absorbs the heat exposed* The sum of radiation heat absorbed and
reflected by an opaAue medium is unity*
[ +p = 1 (*(-
>,


where
< is the absorption factor and
< is the reflection factor
Since absorpti%ity depends on the specific wa%elength& it is not applicable to solid surfaces of
electrical machines& which radiate in a wide freAuency range* Thus& a wa%elength independent
coefficient called relati%e emissi%ity& denoted by c
r
& is introduced* This is a general engineering
assumption and it is known as Vgrey body assumptionF* 2n addition to this& the surfaces are usually
considered diffuse& which means that the absorpti%ity and emissi%ity do not depend on the angle of
radiation*
Hith these assumptions& the radiation heat transfer between solid surfaces is gi%en by the Stefan<
7oltGmann eAuation
q = e
r
o(I
cmIssIvc
4
-I
absorptIvc
4
) (*("
where
< c
r
is relati%e emissi%ity and
< Y S =*>-R)
<"
Hm
<(
C
<+
is the Stefan<7olGman constant*

The relati%e emissi%ity is a combination of the absorpti%e surface and emissi%e surface emissi%ities
and the surface areas according to

r
=
1
1
e
cmIssIvc
+
S
cmIssIvc
S
absorptIvc
_
1
e
absorptIvc
-1]

(*(,
where
< c
ansorpti%e
is the emissi%ity of the absorpti%e surface&
<
c
emissi%e
is the emissi%ity of the emissi%e surface&

<
S
absorpti%e
is the area of the absorpti%e surface and

<
S
emissi%e
is the area of the emissi%e surface*


2n the case of electrical machines& radiation can usually be neglected if the temperature differences
between surfaces are within less than a couple of hundred degrees*

.,.4 2onvection
Don%ection is a special case of conduction heat transfer& since the heat transfer between particles
takes place by conduction* Don%ection always needs a fluid medium* Hithout a fluid& con%ection
cannot take place* The fluid can be either a gas or a mix of gases or liAuids* The difference in
>,


where
< is the absorption factor and
< is the reflection factor
Since absorpti%ity depends on the specific wa%elength& it is not applicable to solid surfaces of
electrical machines& which radiate in a wide freAuency range* Thus& a wa%elength independent
coefficient called relati%e emissi%ity& denoted by c
r
& is introduced* This is a general engineering
assumption and it is known as Vgrey body assumptionF* 2n addition to this& the surfaces are usually
considered diffuse& which means that the absorpti%ity and emissi%ity do not depend on the angle of
radiation*
Hith these assumptions& the radiation heat transfer between solid surfaces is gi%en by the Stefan<
7oltGmann eAuation
q = e
r
o(I
cmIssIvc
4
-I
absorptIvc
4
) (*("
where
< c
r
is relati%e emissi%ity and
< Y S =*>-R)
<"
Hm
<(
C
<+
is the Stefan<7olGman constant*

The relati%e emissi%ity is a combination of the absorpti%e surface and emissi%e surface emissi%ities
and the surface areas according to

r
=
1
1
e
cmIssIvc
+
S
cmIssIvc
S
absorptIvc
_
1
e
absorptIvc
-1]

(*(,
where
< c
ansorpti%e
is the emissi%ity of the absorpti%e surface&
<
c
emissi%e
is the emissi%ity of the emissi%e surface&

<
S
absorpti%e
is the area of the absorpti%e surface and

<
S
emissi%e
is the area of the emissi%e surface*


2n the case of electrical machines& radiation can usually be neglected if the temperature differences
between surfaces are within less than a couple of hundred degrees*

.,.4 2onvection
Don%ection is a special case of conduction heat transfer& since the heat transfer between particles
takes place by conduction* Don%ection always needs a fluid medium* Hithout a fluid& con%ection
cannot take place* The fluid can be either a gas or a mix of gases or liAuids* The difference in
-)


conduction is that the particles are mo%ing* Don%ection can be di%ided into two subcategories&
which are forced and natural con%ection* 2n forced con%ection& extra work is applied to mo%e the
fluid*
The mathematical description of fluid mo%ement is challenging owing to the complex nature of
fluid mechanics* DonseAuently& the eAuation used for con%ection heat transfer is usually a result of
experimental studies* 1xperimental eAuations ha%e to be used with care& because the eAuations
only apply to a certain degree& and the boundary %alues are usually Auite narrow* 2naccuracies in
the system parameter definitions also ha%e a significant effect on the results* The efficiency of
con%ection is described by the ;usselt number Nu& the proportion of the con%ection heat transfer
to the conduction heat transfer in a fluid& by eAuation
Nu =
ol
k

(*!)
< where k is the thermal conducti%ity of fluid&
< o is the con%ection coefficient and
< l is the characteristic length of the surface affected by the con%ection*
The con%ection coefficient o describes the heat dissipation of a material for a unit area at a certain
temperature difference* A higher ;usselt number indicates that heat is remo%ed rather by
con%ection than by conduction* Don%ection coefficients for different fluids and con%ection
methods are presented in Table (*(*
Table* (*(* Thermal conducti%ities k and con%ection coefficients of typical engineering
fluids* The lower %alues are for a )*( mEs %elocity and the higher %alues for a () mEs
%elocity* The %alues are for flat plate con%ection and a )* m long surface*
*luid ;@=5< k

79:m 58 79:m

58
Hater )*>") +))@(")))
Air )*)!)! (@--
1thylene 8lycol )*(> (@()
1ngine oil )*!> >-@-))

As described abo%e& con%ection can be di%ided into two categories* 2ndependent of the method
used& the flow type can also be di%ided into laminar and turbulent flow* These depend on the
geometric properties of the surface and the speed of the fluid& but in any case& the distinction
between laminar and turbulent flow is expressed by the Reynolds number& which describes the
proportion of the fluid density to the %iscosity of the fluid* 'luids with low %iscosity and high
density& such as water and other low<%iscosity liAuids& are susceptible to turbulent flow
characteristics& and are therefore suitable for cooling applications*
.randtl number is the third ma3or dimensionless number in the con%ection heat transfer* .randtl
number describes the proportion of the speed boundary layer to the thermal boundary layer& which
means that as the fluid ad%ances along the surface& the temperature of the fluid rises at the same
-)


conduction is that the particles are mo%ing* Don%ection can be di%ided into two subcategories&
which are forced and natural con%ection* 2n forced con%ection& extra work is applied to mo%e the
fluid*
The mathematical description of fluid mo%ement is challenging owing to the complex nature of
fluid mechanics* DonseAuently& the eAuation used for con%ection heat transfer is usually a result of
experimental studies* 1xperimental eAuations ha%e to be used with care& because the eAuations
only apply to a certain degree& and the boundary %alues are usually Auite narrow* 2naccuracies in
the system parameter definitions also ha%e a significant effect on the results* The efficiency of
con%ection is described by the ;usselt number Nu& the proportion of the con%ection heat transfer
to the conduction heat transfer in a fluid& by eAuation
Nu =
ol
k

(*!)
< where k is the thermal conducti%ity of fluid&
< o is the con%ection coefficient and
< l is the characteristic length of the surface affected by the con%ection*
The con%ection coefficient o describes the heat dissipation of a material for a unit area at a certain
temperature difference* A higher ;usselt number indicates that heat is remo%ed rather by
con%ection than by conduction* Don%ection coefficients for different fluids and con%ection
methods are presented in Table (*(*
Table* (*(* Thermal conducti%ities k and con%ection coefficients of typical engineering
fluids* The lower %alues are for a )*( mEs %elocity and the higher %alues for a () mEs
%elocity* The %alues are for flat plate con%ection and a )* m long surface*
*luid ;@=5< k

79:m 58 79:m

58
Hater )*>") +))@(")))
Air )*)!)! (@--
1thylene 8lycol )*(> (@()
1ngine oil )*!> >-@-))

As described abo%e& con%ection can be di%ided into two categories* 2ndependent of the method
used& the flow type can also be di%ided into laminar and turbulent flow* These depend on the
geometric properties of the surface and the speed of the fluid& but in any case& the distinction
between laminar and turbulent flow is expressed by the Reynolds number& which describes the
proportion of the fluid density to the %iscosity of the fluid* 'luids with low %iscosity and high
density& such as water and other low<%iscosity liAuids& are susceptible to turbulent flow
characteristics& and are therefore suitable for cooling applications*
.randtl number is the third ma3or dimensionless number in the con%ection heat transfer* .randtl
number describes the proportion of the speed boundary layer to the thermal boundary layer& which
means that as the fluid ad%ances along the surface& the temperature of the fluid rises at the same
-


height of examination* The thermal boundary layer stops in a point where no heat transfer occurs
in the normal direction of the surface* The thickness h
t
of this surface is determined by the
eAuation

t
=
(I

-I)
(I

-I

)
(*!
where T
f
is the temperature of the terminal boundary layer that is considered to be the a%erage
temperature between the surface temperature and the ambient temperature in practical calculations*
T
s
is the temperature of the body surface& and T is the temperature of the boundary layer at the
height h
t
* The thickness at which the thermal gradients occur is usually considered h
t
S )*,,*

.,.1 4umped#parameter model
Dompared with the electrical and mechanical design methods& the thermal modelling is a rather
recent approach in the machine design process* Some general guidelines for the thermal
dimensioning of electrical machines ha%e been gi%en in the ,,)s in (Saari& ,,=)& (Cylander&
,,=)& (Mellor& ,,) and (#indstr0m& ,,,) with lumped<parameter models* A more detailed
analysis of permanent magnet machines has also been gi%en by (#iu& ,,!)& (8rauers& ())+) and
(Mellor& ,,)* The general method for the thermal e%aluation of electrical machines is the
lumped<parameter method& which can be described by the matrix eAuation
AT = C
-1
P, (*!(
according to which the temperature difference c) of the machine is directly proportional to the
in%erse of the thermal conductance sAuare matrix % and the heat dissipation single<row matrix P*
The matrix can be presented as a thermal network as shown in 'ig (*>*
-


height of examination* The thermal boundary layer stops in a point where no heat transfer occurs
in the normal direction of the surface* The thickness h
t
of this surface is determined by the
eAuation

t
=
(I

-I)
(I

-I

)
(*!
where T
f
is the temperature of the terminal boundary layer that is considered to be the a%erage
temperature between the surface temperature and the ambient temperature in practical calculations*
T
s
is the temperature of the body surface& and T is the temperature of the boundary layer at the
height h
t
* The thickness at which the thermal gradients occur is usually considered h
t
S )*,,*

.,.1 4umped#parameter model
Dompared with the electrical and mechanical design methods& the thermal modelling is a rather
recent approach in the machine design process* Some general guidelines for the thermal
dimensioning of electrical machines ha%e been gi%en in the ,,)s in (Saari& ,,=)& (Cylander&
,,=)& (Mellor& ,,) and (#indstr0m& ,,,) with lumped<parameter models* A more detailed
analysis of permanent magnet machines has also been gi%en by (#iu& ,,!)& (8rauers& ())+) and
(Mellor& ,,)* The general method for the thermal e%aluation of electrical machines is the
lumped<parameter method& which can be described by the matrix eAuation
AT = C
-1
P, (*!(
according to which the temperature difference c) of the machine is directly proportional to the
in%erse of the thermal conductance sAuare matrix % and the heat dissipation single<row matrix P*
The matrix can be presented as a thermal network as shown in 'ig (*>*
-(



'ig* (*>* Typical thermal network layout fitted to the machine geometry*

7y inserting the sum of related thermal conductances& that is& the incoming branches& of a nodal
point into a corresponding element in the conductance matrix and by inserting the thermal
conductances of indi%idual lea%ing branches into a corresponding element of the next and pre%ious
nodal point& a lumped<parameter network is produced* The pattern can be gi%en as a series
function
0 =
l
l
l
l
l
l
l

1,I
-1
n
=1
-
1,n
-1
. .
-
n,1
-1

n,I
-1
n
=1
1
1
1
1
1
1
1
(*!!
The easiest way to carry out the modelling is to apply simplified geometries* Dylindrical elements
can be e%aluated by the eAuation

cIIndcr
=
lo [
r
out
r
In

2kl
cIIndcr

(*!+
-(



'ig* (*>* Typical thermal network layout fitted to the machine geometry*

7y inserting the sum of related thermal conductances& that is& the incoming branches& of a nodal
point into a corresponding element in the conductance matrix and by inserting the thermal
conductances of indi%idual lea%ing branches into a corresponding element of the next and pre%ious
nodal point& a lumped<parameter network is produced* The pattern can be gi%en as a series
function
0 =
l
l
l
l
l
l
l

1,I
-1
n
=1
-
1,n
-1
. .
-
n,1
-1

n,I
-1
n
=1
1
1
1
1
1
1
1
(*!!
The easiest way to carry out the modelling is to apply simplified geometries* Dylindrical elements
can be e%aluated by the eAuation

cIIndcr
=
lo [
r
out
r
In

2kl
cIIndcr

(*!+
-!


The cylindrical element can be used to model the radial thermal resistances of the rotor and stator
yokes* The cylindrical elements can also be used to model the tooth areaU howe%er& trapeGoid and
sAuare blocks are useful in the modelling of non<cylindrical parts& such as slots& teeth and magnets*
1Auations for trapeGoidal and sAuare element types are

trapczoId
=
h
k[
w
1
+w
2
2
I
taezod
and
(*!=

rcctangIc
=
h
kw
ectangIe
I
ectangIe
,
(*!>
respecti%ely*
More complex eAuations can be used to describe the medium in more detail& but the accuracy
achie%ed this way is usually lost in the con%ection modelling& because the accuracies of the
correlation eAuations are Auite low*

.4 "ummary of the key design areas
1lectrical machine design is a challenging task& which reAuires control o%er multiple parameters
and understanding of the relations between the design areas* Traditional analytical methods are
useful in narrowing down the options of the base design& e%en though the traction machine design
differs from traditional industrial machine designs*
As a traction application does not necessarily ha%e a specific operating point& the traditional
guidelines for machine design cannot be directly applied to a traction machine design* Thus& the
balance between the electromagnetic design& heat transfer and loss generation becomes more
important*
-!


The cylindrical element can be used to model the radial thermal resistances of the rotor and stator
yokes* The cylindrical elements can also be used to model the tooth areaU howe%er& trapeGoid and
sAuare blocks are useful in the modelling of non<cylindrical parts& such as slots& teeth and magnets*
1Auations for trapeGoidal and sAuare element types are

trapczoId
=
h
k[
w
1
+w
2
2
I
taezod
and
(*!=

rcctangIc
=
h
kw
ectangIe
I
ectangIe
,
(*!>
respecti%ely*
More complex eAuations can be used to describe the medium in more detail& but the accuracy
achie%ed this way is usually lost in the con%ection modelling& because the accuracies of the
correlation eAuations are Auite low*

.4 "ummary of the key design areas
1lectrical machine design is a challenging task& which reAuires control o%er multiple parameters
and understanding of the relations between the design areas* Traditional analytical methods are
useful in narrowing down the options of the base design& e%en though the traction machine design
differs from traditional industrial machine designs*
As a traction application does not necessarily ha%e a specific operating point& the traditional
guidelines for machine design cannot be directly applied to a traction machine design* Thus& the
balance between the electromagnetic design& heat transfer and loss generation becomes more
important*
-+



-+



-=


, Implementation of the design methods

The principles of electrical machine design presented in the pre%ious chapters ha%e been
implemented to two prototype machine structures within the application<specific limits* This
chapter presents two salient pole structures& one with a special magnet module (MM) structure and
one with an embedded magnet (1M) structure designed for a %ehicle propulsion system* The
electromagnetic design process is presented in detail along with the transient and steady<state
analysis of the machine performance*

,.! 6ackground of the machine development
.M technology was selected& because .M machines can be built with a high pole number to
achie%e a light and %olume<efficient design* The machine was intended to be connected to the tyre
with a dri%e shaft as presented in 'ig* !**

'ig* !** 1lectrical machine connection to a wheel in the direct dri%e propulsion system*

The allowable machine %oltage and current ratings lead to about !)) 5 line<to<line %oltage ((+*=@
++*- 5) and !)) A line current limits as a result of the battery package limitations set by the
Metropolia design* 2n the machine design process& the field weakening region was limited to about
=) P with a flat battery* The rotating speed corresponding to the maximum a%ailable %oltage was
selected by the desired maximum speed of ()) kmEh* Hith the selected tyres& the rotating speed
reAuired for the maximum speed is about ())) min
<
gi%ing a )*= 5Emin %oltage to rotating speed
ratio*
-=


, Implementation of the design methods

The principles of electrical machine design presented in the pre%ious chapters ha%e been
implemented to two prototype machine structures within the application<specific limits* This
chapter presents two salient pole structures& one with a special magnet module (MM) structure and
one with an embedded magnet (1M) structure designed for a %ehicle propulsion system* The
electromagnetic design process is presented in detail along with the transient and steady<state
analysis of the machine performance*

,.! 6ackground of the machine development
.M technology was selected& because .M machines can be built with a high pole number to
achie%e a light and %olume<efficient design* The machine was intended to be connected to the tyre
with a dri%e shaft as presented in 'ig* !**

'ig* !** 1lectrical machine connection to a wheel in the direct dri%e propulsion system*

The allowable machine %oltage and current ratings lead to about !)) 5 line<to<line %oltage ((+*=@
++*- 5) and !)) A line current limits as a result of the battery package limitations set by the
Metropolia design* 2n the machine design process& the field weakening region was limited to about
=) P with a flat battery* The rotating speed corresponding to the maximum a%ailable %oltage was
selected by the desired maximum speed of ()) kmEh* Hith the selected tyres& the rotating speed
reAuired for the maximum speed is about ())) min
<
gi%ing a )*= 5Emin %oltage to rotating speed
ratio*
->


The a%erage torAue and rotating speed were taken from a dri%e cycle of a specific sports car track
data gi%en by Metropolia& which yielded an a%erage power of (= kH at ))) min
<
* This was set as
the basis of the machine dimensioning* 'or the rated power of (= kH& the %oltage rating results in
a O)) A line current at the ))) min
<
rotating speed*
The high pull<out torAue was a matter of concern from the beginning of the design process*
7ecause of the limiting %alues of the energy source& the maximum output power from the battery is
about +)) kH* As the %ehicle was intended to be eAuipped with four machines& one for each tyre&
the maximum input power for one machine was about )) kH* Hith these initial data& the design
%alues for the electrical machine were gi%en as followsB
< nominal torAue (+) ;m&
< nominal speed ))) min
<
&
< maximum torAue ))) ;m&
< maximum speed ())) min
<
&
< light in weight& the mass limit !" kg for the acti%e parts&
< water cooling in the stator frame&
< length limit for the acti%e parts ()) mm and
< diameter limit for the acti%e parts !") mm*

The dimension limits are based on the actual space a%ailable in the car chassis* The actual space is
somewhat larger& but the dimensions of the acti%e parts are scaled down to accommodate the
water<cooled machine frame in the a%ailable chassis space*
The design process started by analytical calculations& and with the axial space limitation& it was
ob%ious that the magnetic structure of the machine is going to be short compared with the physical
diameter* 'or a water<cooled frame& the current density may %ary between +@- AEmm
(
(5ogt&
,"!)* An air gap flux density %alue of )*" T and a linear current density of += kAEm were decided
to be used as the ground %alues for the electrical design process& resulting in a !> k.a tangential
stress component* The low %alues of the linear current density allow high torAue operation* Thus&
loading of the machine in nominal operation is Auite easy* The magnetic circuit design was made
loose enough to allow space for flux boost in order to get the highest benefit from the reluctance
torAue component at the lowest speed*
The length of the machine acti%e part was set to >= mm* The end space reAuired for traditional
three<phase integral slot windings in the axial direction is close to the pole pitch of the machine*
Hith the rotor outer diameter of ("+ mm& the pole pitch is about => mm& which yields a total acti%e
part length of -- mm* These dimensions pro%ide enough rotor surface area to produce (+) ;m of
constant torAue at the !> k.a tangential stress and a peak torAue of about ) k.a* The axial length
of the rotor is short& because the rest of the space is needed for the end windings* The rotor area
reAuired for the torAue output is produced by increasing the rotor diameter*
To make the stator and the related stator structure light& the machine was decided to be built with a
high pole number* A high pole number allows a light iron core& because the number of flux paths
for the main flux increase linearly as the pole number increases* As the core dimensions are scaled
->


The a%erage torAue and rotating speed were taken from a dri%e cycle of a specific sports car track
data gi%en by Metropolia& which yielded an a%erage power of (= kH at ))) min
<
* This was set as
the basis of the machine dimensioning* 'or the rated power of (= kH& the %oltage rating results in
a O)) A line current at the ))) min
<
rotating speed*
The high pull<out torAue was a matter of concern from the beginning of the design process*
7ecause of the limiting %alues of the energy source& the maximum output power from the battery is
about +)) kH* As the %ehicle was intended to be eAuipped with four machines& one for each tyre&
the maximum input power for one machine was about )) kH* Hith these initial data& the design
%alues for the electrical machine were gi%en as followsB
< nominal torAue (+) ;m&
< nominal speed ))) min
<
&
< maximum torAue ))) ;m&
< maximum speed ())) min
<
&
< light in weight& the mass limit !" kg for the acti%e parts&
< water cooling in the stator frame&
< length limit for the acti%e parts ()) mm and
< diameter limit for the acti%e parts !") mm*

The dimension limits are based on the actual space a%ailable in the car chassis* The actual space is
somewhat larger& but the dimensions of the acti%e parts are scaled down to accommodate the
water<cooled machine frame in the a%ailable chassis space*
The design process started by analytical calculations& and with the axial space limitation& it was
ob%ious that the magnetic structure of the machine is going to be short compared with the physical
diameter* 'or a water<cooled frame& the current density may %ary between +@- AEmm
(
(5ogt&
,"!)* An air gap flux density %alue of )*" T and a linear current density of += kAEm were decided
to be used as the ground %alues for the electrical design process& resulting in a !> k.a tangential
stress component* The low %alues of the linear current density allow high torAue operation* Thus&
loading of the machine in nominal operation is Auite easy* The magnetic circuit design was made
loose enough to allow space for flux boost in order to get the highest benefit from the reluctance
torAue component at the lowest speed*
The length of the machine acti%e part was set to >= mm* The end space reAuired for traditional
three<phase integral slot windings in the axial direction is close to the pole pitch of the machine*
Hith the rotor outer diameter of ("+ mm& the pole pitch is about => mm& which yields a total acti%e
part length of -- mm* These dimensions pro%ide enough rotor surface area to produce (+) ;m of
constant torAue at the !> k.a tangential stress and a peak torAue of about ) k.a* The axial length
of the rotor is short& because the rest of the space is needed for the end windings* The rotor area
reAuired for the torAue output is produced by increasing the rotor diameter*
To make the stator and the related stator structure light& the machine was decided to be built with a
high pole number* A high pole number allows a light iron core& because the number of flux paths
for the main flux increase linearly as the pole number increases* As the core dimensions are scaled
--


down& a larger rotor can be fitted into the same space than in the case of small pole numbers&
which means a larger torAue capability as the rotor area increases* The increase in the pole number
also promotes pull<out torAue production as the pull<out torAue is directly proportional to the
second power of the pole pair number& as can be concluded from 1As* *( and *=@*-*
The rotor diameter and the !") mm limit of the stator stack result in a +" mm space for the stator
stack and the air gap in the radial direction* The lowest pole number to fit the stator geometry
inside the specified dimensions and to allow sufficient slot dimensions is (p S >* This yields a
slot pitch of "*" mm for q S which results in +" slots in the stator* The guide %alues for a
suitable slot pitch of a synchronous machine are between " and (= mm (5ogt& ,"!)* To ensure
safe operation in the maximum speed conditions& it was decided to use low<loss M(-)<=)A grade
sheet metal for the iron cores of the rotor and the stator*
The machine winding was chosen to be a two<layer winding to achie%e short end windings* 2t was
noted that this can lead to problems with the torAue ripple caused by the current linkage harmonics
especially at high current ratings* The one slot per pole per phase configuration results in a slightly
coarse back 1M' wa%eform& but because the machine is intended for automoti%e usage& the torAue
%ibrations are acceptable& at least to some degree* The winding configuration is illustrated in 'ig*
!*(*

'ig* !*(* 7asic winding arrangement of the designed prototype machine*

Thermal conduction fa%ours small slots as the heat dissipation surface is larger compared with the
%olume* The tooth width is optimiGed to keep the flux density of the tooth body at *= T* After this
point& the current linkage reAuired for the magnetiGation increases rapidly* Some points of the AD
magnetiGation cur%e according to the data for the M(-)<=)A sheet by Surahammars 7ruks are
presented in 'ig !*!*
gA <D
gA <D
g7
g7
<7
<7 gD
gD <A
<A
--


down& a larger rotor can be fitted into the same space than in the case of small pole numbers&
which means a larger torAue capability as the rotor area increases* The increase in the pole number
also promotes pull<out torAue production as the pull<out torAue is directly proportional to the
second power of the pole pair number& as can be concluded from 1As* *( and *=@*-*
The rotor diameter and the !") mm limit of the stator stack result in a +" mm space for the stator
stack and the air gap in the radial direction* The lowest pole number to fit the stator geometry
inside the specified dimensions and to allow sufficient slot dimensions is (p S >* This yields a
slot pitch of "*" mm for q S which results in +" slots in the stator* The guide %alues for a
suitable slot pitch of a synchronous machine are between " and (= mm (5ogt& ,"!)* To ensure
safe operation in the maximum speed conditions& it was decided to use low<loss M(-)<=)A grade
sheet metal for the iron cores of the rotor and the stator*
The machine winding was chosen to be a two<layer winding to achie%e short end windings* 2t was
noted that this can lead to problems with the torAue ripple caused by the current linkage harmonics
especially at high current ratings* The one slot per pole per phase configuration results in a slightly
coarse back 1M' wa%eform& but because the machine is intended for automoti%e usage& the torAue
%ibrations are acceptable& at least to some degree* The winding configuration is illustrated in 'ig*
!*(*

'ig* !*(* 7asic winding arrangement of the designed prototype machine*

Thermal conduction fa%ours small slots as the heat dissipation surface is larger compared with the
%olume* The tooth width is optimiGed to keep the flux density of the tooth body at *= T* After this
point& the current linkage reAuired for the magnetiGation increases rapidly* Some points of the AD
magnetiGation cur%e according to the data for the M(-)<=)A sheet by Surahammars 7ruks are
presented in 'ig !*!*
gA <D
gA <D
g7
g7
<7
<7 gD
gD <A
<A
-"



'ig* !*!* Magnetic polariGation with a =) 9G supply as a function of field strength for the
M(-)<=) A sheet (Surahammars 7ruks)*

The largest polariGation at the =) 9G excitation gi%en by the manufacturer for the selected M(-)<
=)A sheet is around *- T& which can be considered a practical flux density limit for the gi%en
sheet grade* The )*( T space for the flux density limit is selected to allow space for flux boost to
promote reluctance torAue production*
#ower flux density %alues can be used& but this causes additional weight and less slot %olume* The
priority in the slot height optimiGation is to allow enough current linkage in the slot to produce
rated torAue and to maintain as small stator diameter as possible because the space limitation is
Auite strict* The optimal slot height depends on the thermal balance* The height of the slot is
limited by the increasing thermal resistance in the radial direction* The main heat flux from the slot
is to the tooth* The slot bottom becomes second* 2t can be characteriGed that e%ery slot has one
tooth in parallel* Thus& the a%erage heat conduction length in the tangential and radial directions is
half of the slot width and slot height& respecti%ely* The assumed parallel heat paths in the slot
region are shown in 'ig* !*+*
-"



'ig* !*!* Magnetic polariGation with a =) 9G supply as a function of field strength for the
M(-)<=) A sheet (Surahammars 7ruks)*

The largest polariGation at the =) 9G excitation gi%en by the manufacturer for the selected M(-)<
=)A sheet is around *- T& which can be considered a practical flux density limit for the gi%en
sheet grade* The )*( T space for the flux density limit is selected to allow space for flux boost to
promote reluctance torAue production*
#ower flux density %alues can be used& but this causes additional weight and less slot %olume* The
priority in the slot height optimiGation is to allow enough current linkage in the slot to produce
rated torAue and to maintain as small stator diameter as possible because the space limitation is
Auite strict* The optimal slot height depends on the thermal balance* The height of the slot is
limited by the increasing thermal resistance in the radial direction* The main heat flux from the slot
is to the tooth* The slot bottom becomes second* 2t can be characteriGed that e%ery slot has one
tooth in parallel* Thus& the a%erage heat conduction length in the tangential and radial directions is
half of the slot width and slot height& respecti%ely* The assumed parallel heat paths in the slot
region are shown in 'ig* !*+*
-,



'ig* !*+* .arallel heat path consisting of one tooth and one slot* .resented heat resistances
are used in slot eAui%alent thermal resistance calculation in 1A* !**

The optimum is found at dimensions for which the parallel connection results in the smallest
thermal resistance* The tooth width is set to constant *> mm because of the *= T saturation
limit* The slot and tooth are considered rectangular elements* 1Auation (*!> is applied to the
thermal resistance calculations* The eAui%alent thermal resistance is calculated as

th,cq
=
(
th1
+
th2
)
th3

th1
+
th2
+
th3

!*

R
R R
th( th!
th
-,



'ig* !*+* .arallel heat path consisting of one tooth and one slot* .resented heat resistances
are used in slot eAui%alent thermal resistance calculation in 1A* !**

The optimum is found at dimensions for which the parallel connection results in the smallest
thermal resistance* The tooth width is set to constant *> mm because of the *= T saturation
limit* The slot and tooth are considered rectangular elements* 1Auation (*!> is applied to the
thermal resistance calculations* The eAui%alent thermal resistance is calculated as

th,cq
=
(
th1
+
th2
)
th3

th1
+
th2
+
th3

!*

R
R R
th( th!
th
")



'ig* !*=* Slot and tooth dimensions*

As a simplification& the slot is modelled as a rectangular element* Dimensions for the rectangular
eAui%alent slot& according to the dimensions in 'ig !*=& are calculated according to

cq
=
y
4
+2(y
1
+y
2
+y
3
)
2

!*(
for the eAui%alent slot height and as
w
cq
=
x
1
+x
2
2

!*!
for the eAui%alent slot width* The de%eloped simple model has shown relati%ely good results in the
slot area estimation* 'or e%aluation& the thermal conducti%ities are selected to be !) HECm for the
core& )*( HECm for the slot impregnation and the slot insulation& !"= HECm for the copper and
)*)(> HECm for air in the contact region* The selected %alues are typical for materials used in the
thermal modelling of an electrical machine (Miller& ,,+)* The resistances R
th
and R
th!
also
include the air contact resistance ()* mm) and )*(+ mm of the slot insulation paper* 7y %arying
the tooth slot height from ) mm to )) mm& the thermal resistance of the parallel connection
%aries as presented in 'ig !*>*
x
=
>
x
+
x s
x

x
(
v
+
v
!
v
(
v

")



'ig* !*=* Slot and tooth dimensions*

As a simplification& the slot is modelled as a rectangular element* Dimensions for the rectangular
eAui%alent slot& according to the dimensions in 'ig !*=& are calculated according to

cq
=
y
4
+2(y
1
+y
2
+y
3
)
2

!*(
for the eAui%alent slot height and as
w
cq
=
x
1
+x
2
2

!*!
for the eAui%alent slot width* The de%eloped simple model has shown relati%ely good results in the
slot area estimation* 'or e%aluation& the thermal conducti%ities are selected to be !) HECm for the
core& )*( HECm for the slot impregnation and the slot insulation& !"= HECm for the copper and
)*)(> HECm for air in the contact region* The selected %alues are typical for materials used in the
thermal modelling of an electrical machine (Miller& ,,+)* The resistances R
th
and R
th!
also
include the air contact resistance ()* mm) and )*(+ mm of the slot insulation paper* 7y %arying
the tooth slot height from ) mm to )) mm& the thermal resistance of the parallel connection
%aries as presented in 'ig !*>*
x
=
>
x
+
x s
x

x
(
v
+
v
!
v
(
v

"



'ig* !*>* Thermal resistance as a function of slot height for the parallel connection of 'ig
!*+ is presented with solid lines* The black line is for a nominal *> mm tooth
width* The red and blue lines are for * and *( times the nominal tooth width&
respecti%ely* The dotted lines present the slot cross<sectional area de%elopment
as a function of slot height*

The local minimum for the thermal resistance of the parallel connection is found at +-*=( mm at
the *> mm tooth width* As the tooth gets wider& the optimal tooth height gets lower %alues
because the thermal conduction path has to be shortened as the heat exchange surface of the slot
bottom decreases* ;e%ertheless& a longer tooth can be used as far as the combined thermal
resistance is lower than that of the narrower teeth* The slot leakage should also be taken into
account to a%oid excess leakage* The slot leakage beha%iour according to slot dimensions
eAui%alent to 'ig* !*> is presented in 'ig* !*-*
) )*) )*)( )*)! )*)+ )*)= )*)> )*)- )*)" )*), )*
)*
)*(
)*!
T
h
e
r
m
a
l

r
e
s
i
s
t
a
n
c
e

J
C
E
H
L
Slot height JmL
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
0
1000
2000
S
l
o
t

a
r
e
a

J
m
m
(
L
"



'ig* !*>* Thermal resistance as a function of slot height for the parallel connection of 'ig
!*+ is presented with solid lines* The black line is for a nominal *> mm tooth
width* The red and blue lines are for * and *( times the nominal tooth width&
respecti%ely* The dotted lines present the slot cross<sectional area de%elopment
as a function of slot height*

The local minimum for the thermal resistance of the parallel connection is found at +-*=( mm at
the *> mm tooth width* As the tooth gets wider& the optimal tooth height gets lower %alues
because the thermal conduction path has to be shortened as the heat exchange surface of the slot
bottom decreases* ;e%ertheless& a longer tooth can be used as far as the combined thermal
resistance is lower than that of the narrower teeth* The slot leakage should also be taken into
account to a%oid excess leakage* The slot leakage beha%iour according to slot dimensions
eAui%alent to 'ig* !*> is presented in 'ig* !*-*
) )*) )*)( )*)! )*)+ )*)= )*)> )*)- )*)" )*), )*
)*
)*(
)*!
T
h
e
r
m
a
l

r
e
s
i
s
t
a
n
c
e

J
C
E
H
L
Slot height JmL
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
0
1000
2000
S
l
o
t

a
r
e
a

J
m
m
(
L
"(



'ig* !*-* Slot leakage as a function of slot height for a double<layer winding* .er unit
%alues are calculated for the nominal operating speed of ))) min
<
& a =) 5
line<to<line %oltage and a )) A line current& which yields an inductance base
%alue L
b
of )*)) 9* The red sAuare indicates the leakage %alue for the selected
slot height* The white rectangle represents the optimal slot height according to
(1A* !*) at the *> mm tooth width*
:wing to the diameter limit of !") mm and the flux density limit of *= T& the slot height has to be
set as !(*= mm& which is considerably less than the model suggests* The situation could be
corrected by decreasing the rotor diameter& but this would endanger the torAue output capability of
the machine*
The other issue related to the heat path optimiGation is the slot siGe* 2t has to be ensured that the
slot is large enough to carry the current reAuired for the nominal torAue* Hinding turn number was
estimated with eAuations presented in Section (**(* The total one<phase winding turn number is
>+& which yields :
N
S" coil sides in one slot to match the back 1M' to the specified supply %oltage
of !)) 5 at a ())) min
<
rotating speed* The selected slot dimensions result in a slot area of (">
mm
(
& which still produces a
=

s
u.SS
sIot
=
8 1uu
u.S 286
= S.6 Amm
2



) )*) )*)( )*)! )*)+ )*)= )*)> )*)- )*)" )*), )*
)*))=
)*)
)*)=
)*)(
)*)(=
)*)!
Slot height JmL
S
l
o
t

l
e
a
k
a
g
e

J
p
u
L
"(



'ig* !*-* Slot leakage as a function of slot height for a double<layer winding* .er unit
%alues are calculated for the nominal operating speed of ))) min
<
& a =) 5
line<to<line %oltage and a )) A line current& which yields an inductance base
%alue L
b
of )*)) 9* The red sAuare indicates the leakage %alue for the selected
slot height* The white rectangle represents the optimal slot height according to
(1A* !*) at the *> mm tooth width*
:wing to the diameter limit of !") mm and the flux density limit of *= T& the slot height has to be
set as !(*= mm& which is considerably less than the model suggests* The situation could be
corrected by decreasing the rotor diameter& but this would endanger the torAue output capability of
the machine*
The other issue related to the heat path optimiGation is the slot siGe* 2t has to be ensured that the
slot is large enough to carry the current reAuired for the nominal torAue* Hinding turn number was
estimated with eAuations presented in Section (**(* The total one<phase winding turn number is
>+& which yields :
N
S" coil sides in one slot to match the back 1M' to the specified supply %oltage
of !)) 5 at a ())) min
<
rotating speed* The selected slot dimensions result in a slot area of (">
mm
(
& which still produces a
=

s
u.SS
sIot
=
8 1uu
u.S 286
= S.6 Amm
2



) )*) )*)( )*)! )*)+ )*)= )*)> )*)- )*)" )*), )*
)*))=
)*)
)*)=
)*)(
)*)(=
)*)!
Slot height JmL
S
l
o
t

l
e
a
k
a
g
e

J
p
u
L
"!


current density %alue for the =) P space factor at the rated current of )) A* The current density is
within the limits for water<cooled synchronous machines as presented in Table (*!* Thus& the
cooling of the machine should not gi%e any cause for concern*
The current<carrying capacity can be referred to the slot siGe as it is directly proportional to the
electrical resistance and thereby directly proportional to the heat dissipation of the slot* The slot
area and the slot heat dissipation capability are presented in 'ig* !*"* The heat dissipation of the
slot is corrected by multiplying the slot copper area by a =*> AEmm
(
current density and by
assuming a =) P filling factor for the slot copper* The slot siGe and the thermal %oltage along the
eAui%alent thermal resistance are presented in 'ig* !*-*


'ig* !*"* Slot area and indicati%e temperature difference in the eAui%alent thermal
resistance* The black contour lines represent the slot cross< sectional area* The
red lines indicate temperature o%er eAui%alent resistance* The red rectangle
shows the %alue for the selected slot height* The white rectangle represents the
optimal slot height according to (1A* !*) at the *> mm tooth width*

'igure !*" shows that with the optimal slot height for the machine geometry& the slot copper heats
more with the same current density %alue* 2n order to achie%e the same heat flux density with a
smaller slot& the current density of the optimal slot geometry would ha%e to be reduced to the %alue
of O+ AEmm
(
*
1
0
0
1
0
0
1
0
0
2
0
0
2
0
0
2
0
0
3
0
0
3
0
0
3
0
0
4
0
0
4
0
0
4
0
0
5
0
0
5
0
0
5
0
0
6
0
0
6
0
0
6
0
0
7
0
0
7
0
0
7
0
0
8
0
0
8
0
0
8
0
0
9
0
0
9
0
0
9
0
0
1
0
0
0
1
0
0
0
1
1
0
0
1
0
1
0
1
0
2
0
2
0
2
0
3
0
3
0
3
0
4
0
4
0
4
0
5
0
5
0
5
0
6
0
6
0
7
0
Slot height JmL
T
o
o
t
h

w
i
d
t
h

J
m
L
)*) )*)( )*)! )*)+ )*)= )*)> )*)- )*)" )*),
)*))>
)*))"
)*)
)*)(
)*)+
)*)>
)*)"
)*)(
)*)((
)*)(+
)*)(>
"!


current density %alue for the =) P space factor at the rated current of )) A* The current density is
within the limits for water<cooled synchronous machines as presented in Table (*!* Thus& the
cooling of the machine should not gi%e any cause for concern*
The current<carrying capacity can be referred to the slot siGe as it is directly proportional to the
electrical resistance and thereby directly proportional to the heat dissipation of the slot* The slot
area and the slot heat dissipation capability are presented in 'ig* !*"* The heat dissipation of the
slot is corrected by multiplying the slot copper area by a =*> AEmm
(
current density and by
assuming a =) P filling factor for the slot copper* The slot siGe and the thermal %oltage along the
eAui%alent thermal resistance are presented in 'ig* !*-*


'ig* !*"* Slot area and indicati%e temperature difference in the eAui%alent thermal
resistance* The black contour lines represent the slot cross< sectional area* The
red lines indicate temperature o%er eAui%alent resistance* The red rectangle
shows the %alue for the selected slot height* The white rectangle represents the
optimal slot height according to (1A* !*) at the *> mm tooth width*

'igure !*" shows that with the optimal slot height for the machine geometry& the slot copper heats
more with the same current density %alue* 2n order to achie%e the same heat flux density with a
smaller slot& the current density of the optimal slot geometry would ha%e to be reduced to the %alue
of O+ AEmm
(
*
1
0
0
1
0
0
1
0
0
2
0
0
2
0
0
2
0
0
3
0
0
3
0
0
3
0
0
4
0
0
4
0
0
4
0
0
5
0
0
5
0
0
5
0
0
6
0
0
6
0
0
6
0
0
7
0
0
7
0
0
7
0
0
8
0
0
8
0
0
8
0
0
9
0
0
9
0
0
9
0
0
1
0
0
0
1
0
0
0
1
1
0
0
1
0
1
0
1
0
2
0
2
0
2
0
3
0
3
0
3
0
4
0
4
0
4
0
5
0
5
0
5
0
6
0
6
0
7
0
Slot height JmL
T
o
o
t
h

w
i
d
t
h

J
m
L
)*) )*)( )*)! )*)+ )*)= )*)> )*)- )*)" )*),
)*))>
)*))"
)*)
)*)(
)*)+
)*)>
)*)"
)*)(
)*)((
)*)(+
)*)(>
"+


The magnet module (MM) topology was selected to minimiGe the consumption of magnet material
e%en though an embedded magnet (1M) arrangement would be mechanically more stable* The
fixing of the magnets in the MM design is essential for the machine endurance*
The further de%elopment of the machine structure was performed with finite element methods by
inspecting different machine geometries for enhanced torAue production within the limits of the
existing dimensions and power supply*
2t was disco%ered that a magnetic pole construction with sheet metal side bars would be beneficial
from the perspecti%e of the torAue output as discussed in Section *!*(* The only concern was the
fixing of the magnet unit on the rotor surface* Thus& from the %ery beginning& the machine design
was a challenging task& because adhesi%e fixing has generally been a%oided in these machine
designs& e%en though the manufacturer specifications for the adhesi%e holding capability would
seem con%incing* The resulting pole configuration of the electrical machine is presented in 'ig*
!*,*

'ig* !*,* MM machine pole geometry with a glued magnet pole structure*

The pole structure produces similar beha%iour of the direct< and Auadrature<axis magnetic
conducti%ity as in the embedded magnet design* The difference is that the iron cams are separated
from the magnets to pre%ent short<circuiting of the magnets* This ensures better utiliGation of the
magnet material in the flux production as the leakage flux caused by the iron passage short<
.M .M
"+


The magnet module (MM) topology was selected to minimiGe the consumption of magnet material
e%en though an embedded magnet (1M) arrangement would be mechanically more stable* The
fixing of the magnets in the MM design is essential for the machine endurance*
The further de%elopment of the machine structure was performed with finite element methods by
inspecting different machine geometries for enhanced torAue production within the limits of the
existing dimensions and power supply*
2t was disco%ered that a magnetic pole construction with sheet metal side bars would be beneficial
from the perspecti%e of the torAue output as discussed in Section *!*(* The only concern was the
fixing of the magnet unit on the rotor surface* Thus& from the %ery beginning& the machine design
was a challenging task& because adhesi%e fixing has generally been a%oided in these machine
designs& e%en though the manufacturer specifications for the adhesi%e holding capability would
seem con%incing* The resulting pole configuration of the electrical machine is presented in 'ig*
!*,*

'ig* !*,* MM machine pole geometry with a glued magnet pole structure*

The pole structure produces similar beha%iour of the direct< and Auadrature<axis magnetic
conducti%ity as in the embedded magnet design* The difference is that the iron cams are separated
from the magnets to pre%ent short<circuiting of the magnets* This ensures better utiliGation of the
magnet material in the flux production as the leakage flux caused by the iron passage short<
.M .M
"=


circuiting is pre%ented* The magnet itself consists of three parts as presented in 'ig !*)* The top
and bottom layers of the magnets are separated with a high<permeability S:MA#:?b soft
magnetic composite layer* The purpose of the SMD layer is to guide the flux in the A<axis direction
for high reluctance torAue operation* The soft magnetic composite piece between the magnets is
also easier from the manufacturing point of %iew& and compared with the solid block solution& the
core losses are much lower* 9owe%er& the mechanical structure is weaker compared with
embedded designs& but the lower magnet %olume and the option to use a magnet material with a
low remanence %alue can make difference in an application that does not reAuire high operating
speeds* The direct<axis inductance of the surface magnet machine is also slightly smaller
compared with the embedded magnet design& which usually results in higher pull<out torAue
characteristics*
The results of the finite element calculations were promising& but because of the current limitation&
the machine pull<out torAue maximum was lowered to the %alue of -)) ;m* The mechanical
strength of the machine ga%e some cause for concern& but e%entually& the construction of the
prototype machine was accomplished*
,.!.! )esting and mechanical aspects
2n the final structure& the topmost layer of the magnet was slit towards the air gap in the axial
direction to reduce the eddy current effects* The magnet stack was bundled up with a two<
compound composite adhesi%e& and the whole stack was tied to the rotor surface with a two<
component Aralditeb epoxy resin* The magnet module is shown in 'ig* !*)*

'ig* !*)* Three<layer magnet module of the MM machine* The top and bottom layers are
made of a ;e+,!a grade magnet material* The material of the middle layer is a
Somaloy ==) composite* The whole assembly is epoxy coated* The slitting of
the topmost magnet is not shown because of the epoxy coating*

The fixing of the magnets was a challenging task* 'irst& the machine was tested with the adhesi%e
fixing of the magnets* This resulted in a failure in the magnet fixing during the initial testing*
Apparently& the siGing could not resist the forces acting inside the machine& e%en though the
"=


circuiting is pre%ented* The magnet itself consists of three parts as presented in 'ig !*)* The top
and bottom layers of the magnets are separated with a high<permeability S:MA#:?b soft
magnetic composite layer* The purpose of the SMD layer is to guide the flux in the A<axis direction
for high reluctance torAue operation* The soft magnetic composite piece between the magnets is
also easier from the manufacturing point of %iew& and compared with the solid block solution& the
core losses are much lower* 9owe%er& the mechanical structure is weaker compared with
embedded designs& but the lower magnet %olume and the option to use a magnet material with a
low remanence %alue can make difference in an application that does not reAuire high operating
speeds* The direct<axis inductance of the surface magnet machine is also slightly smaller
compared with the embedded magnet design& which usually results in higher pull<out torAue
characteristics*
The results of the finite element calculations were promising& but because of the current limitation&
the machine pull<out torAue maximum was lowered to the %alue of -)) ;m* The mechanical
strength of the machine ga%e some cause for concern& but e%entually& the construction of the
prototype machine was accomplished*
,.!.! )esting and mechanical aspects
2n the final structure& the topmost layer of the magnet was slit towards the air gap in the axial
direction to reduce the eddy current effects* The magnet stack was bundled up with a two<
compound composite adhesi%e& and the whole stack was tied to the rotor surface with a two<
component Aralditeb epoxy resin* The magnet module is shown in 'ig* !*)*

'ig* !*)* Three<layer magnet module of the MM machine* The top and bottom layers are
made of a ;e+,!a grade magnet material* The material of the middle layer is a
Somaloy ==) composite* The whole assembly is epoxy coated* The slitting of
the topmost magnet is not shown because of the epoxy coating*

The fixing of the magnets was a challenging task* 'irst& the machine was tested with the adhesi%e
fixing of the magnets* This resulted in a failure in the magnet fixing during the initial testing*
Apparently& the siGing could not resist the forces acting inside the machine& e%en though the
">


calculated safety factor according to the tensile strength was high* The machine could be stopped
fast enough to pre%ent any se%ere damage*
A carbon fibre banding was also tested& but the fibring process failed in the curing mold
installation& and the band was damaged* This resulted in a band failure at the =)) rpm speed and
led to a total destruction of the test rotor and the windings* The rotor was next secured with a
fibreglass band and an elastic polybutadiene resin* This arrangement showed good endurance& but
the band failed in the final tests at ())) rpmU howe%er& the machine did not suffer any additional
damage*
Although the test results were adeAuate and feasible& at least to some degree& an additional series
of tests were performed with a welded stainless steel band* 2t was known beforehand that stainless
steel has a high coefficient of temperature enlargement& and a possible failure was therefore
expected* 9owe%er& the stainless band performed extremely well up to the =)) rpm operating
speed& and no weakening was noticed e%en in the maximum torAue tests* The ())) rpm operating
speed was again too much* The band yielded and the tension caused the band to touch the stator
surface as a result of the excess heating of the band* The operation was normaliGed again when the
rotating speed was lowered to =)) min
<
* The finite element method %erifies the excess heating of
the band caused by the conducting losses* Dalculations at the rated current show eddy current loss
generation of !> H already at the ))) min
<
rotating speed* The eddy current losses increase to
the le%el of !)) H at the ())) min
<
rotating speed* The band causes about a *( P unit decrease
in the efficiency ratings at the rated (= kH power* Thus& it is not ad%isable to use conducting
banding at these freAuency le%els*
This magnet module machine type still needs further de%elopment to o%ercome the mechanical
problems& but thus far& the results are promising* 9owe%er& it was decided to design a new rotor
structure instead to ensure safe machine operation* The new rotor structure was designed with
buried magnets to the same frame geometry as the pre%ious one*
,.!. "econd generation
To ensure safe operation o%er =)) min
<
operation& an embedded magnet structure was
introduced* To achie%e similar operation characteristics& the back electromoti%e force wa%eform
was eAualiGed* The embedded magnet topology is presented in 'ig* !**
">


calculated safety factor according to the tensile strength was high* The machine could be stopped
fast enough to pre%ent any se%ere damage*
A carbon fibre banding was also tested& but the fibring process failed in the curing mold
installation& and the band was damaged* This resulted in a band failure at the =)) rpm speed and
led to a total destruction of the test rotor and the windings* The rotor was next secured with a
fibreglass band and an elastic polybutadiene resin* This arrangement showed good endurance& but
the band failed in the final tests at ())) rpmU howe%er& the machine did not suffer any additional
damage*
Although the test results were adeAuate and feasible& at least to some degree& an additional series
of tests were performed with a welded stainless steel band* 2t was known beforehand that stainless
steel has a high coefficient of temperature enlargement& and a possible failure was therefore
expected* 9owe%er& the stainless band performed extremely well up to the =)) rpm operating
speed& and no weakening was noticed e%en in the maximum torAue tests* The ())) rpm operating
speed was again too much* The band yielded and the tension caused the band to touch the stator
surface as a result of the excess heating of the band* The operation was normaliGed again when the
rotating speed was lowered to =)) min
<
* The finite element method %erifies the excess heating of
the band caused by the conducting losses* Dalculations at the rated current show eddy current loss
generation of !> H already at the ))) min
<
rotating speed* The eddy current losses increase to
the le%el of !)) H at the ())) min
<
rotating speed* The band causes about a *( P unit decrease
in the efficiency ratings at the rated (= kH power* Thus& it is not ad%isable to use conducting
banding at these freAuency le%els*
This magnet module machine type still needs further de%elopment to o%ercome the mechanical
problems& but thus far& the results are promising* 9owe%er& it was decided to design a new rotor
structure instead to ensure safe machine operation* The new rotor structure was designed with
buried magnets to the same frame geometry as the pre%ious one*
,.!. "econd generation
To ensure safe operation o%er =)) min
<
operation& an embedded magnet structure was
introduced* To achie%e similar operation characteristics& the back electromoti%e force wa%eform
was eAualiGed* The embedded magnet topology is presented in 'ig* !**
"-



'ig* !** .ole arrangement of the embedded magnet machine*

The two<magnet arrangement has been maintained for the sake of impro%ed inductance ratio& and
additionally& the pole is shaped in the A<axis to achie%e a similar back 1M' wa%eform as that of
the surface magnet structure* The magnet remanence has to be high because of the high stray
fluxes in the magnet sides* These two structures are compared in the following sections*

,. "teady#state analysis
This chapter presents the results of the steady<state finite element analysis on the designed
machine types* The results of the analytical approach in the pre%ious chapters are used as the basis
of the finite element modelling* The analytical approach and the finite<element<based methods
should go hand in hand in the design process* The finite<element<based computer programs allow a
more thorough analysis of the geometry<related effects on the machine performance compared
with the analytical methodsU howe%er& a careful study of the initial parameters helps in a%oiding
unnecessary computations*
.M .M
"-



'ig* !** .ole arrangement of the embedded magnet machine*

The two<magnet arrangement has been maintained for the sake of impro%ed inductance ratio& and
additionally& the pole is shaped in the A<axis to achie%e a similar back 1M' wa%eform as that of
the surface magnet structure* The magnet remanence has to be high because of the high stray
fluxes in the magnet sides* These two structures are compared in the following sections*

,. "teady#state analysis
This chapter presents the results of the steady<state finite element analysis on the designed
machine types* The results of the analytical approach in the pre%ious chapters are used as the basis
of the finite element modelling* The analytical approach and the finite<element<based methods
should go hand in hand in the design process* The finite<element<based computer programs allow a
more thorough analysis of the geometry<related effects on the machine performance compared
with the analytical methodsU howe%er& a careful study of the initial parameters helps in a%oiding
unnecessary computations*
.M .M
""


,..! 6ack electromotive force
7oth of the machines are designed to produce similar back electromoti%e force characteristics in
order to achie%e a similar magnetic beha%iour* The magnet material for the MM structure is
;eorem ;e +,!a* The magnet grade data are presented in 'ig* !*(*

'ig* !*(* ;e+,!a grade data by ;eorem Magnets*
The embedded magnet structure reAuires a magnet grade with a larger remanence& because the A<
axis pro%ides a short circuit path for the magnet flux* The magnet grade of the embedded magnet
machine is ;e-=! a of ;eorem magnets* The magnet grade data are presented in 'ig* !*!*
""


,..! 6ack electromotive force
7oth of the machines are designed to produce similar back electromoti%e force characteristics in
order to achie%e a similar magnetic beha%iour* The magnet material for the MM structure is
;eorem ;e +,!a* The magnet grade data are presented in 'ig* !*(*

'ig* !*(* ;e+,!a grade data by ;eorem Magnets*
The embedded magnet structure reAuires a magnet grade with a larger remanence& because the A<
axis pro%ides a short circuit path for the magnet flux* The magnet grade of the embedded magnet
machine is ;e-=! a of ;eorem magnets* The magnet grade data are presented in 'ig* !*!*
",



'ig* !*!* ;e-=!a grade data by ;eorem Magnets*
'or finite element calculations& the magnet characteristics are considered for the )) MD
temperature* This yields a *)- T remanence for the MM machine and a *( T remanent flux
density for the 1M machine* Hhen comparing the magnet grade data& the difference in the
temperature tolerance becomes e%ident* The demagnetiGation cur%e at the )) MD temperature of
the ;e-=! grade has a knee below the )*! T flux density* At this temperature& the rotating speed
limit would be about !*, p*u* This can be estimated by comparing the magnet remanence knee
point %alue with the Gero field strength %alue at a gi%en temperature* Dompared with the ;e +,!a
grade& there is a considerably higher risk of permanent demagnetiGation in o%erspeed operation or
in short<circuit situations& if the temperature is high enough* The rotating speed limit estimates in
per unit %alues according to the magnet grade data are gi%en in Table !**
Table* !** 1stimated rotating speed limits for different temperatures according to the
magnet data presented in 'igs* !*( and !*! for the magnets ;e+,!a and -=!a&
respecti%ely*
Temperature JMDL ;e-=!a ;e+,!a
") ( (no limit)
)) !*, (no limit)
() (*! (no limit)
=) *=( +*"=
") (no data) (*),

",



'ig* !*!* ;e-=!a grade data by ;eorem Magnets*
'or finite element calculations& the magnet characteristics are considered for the )) MD
temperature* This yields a *)- T remanence for the MM machine and a *( T remanent flux
density for the 1M machine* Hhen comparing the magnet grade data& the difference in the
temperature tolerance becomes e%ident* The demagnetiGation cur%e at the )) MD temperature of
the ;e-=! grade has a knee below the )*! T flux density* At this temperature& the rotating speed
limit would be about !*, p*u* This can be estimated by comparing the magnet remanence knee
point %alue with the Gero field strength %alue at a gi%en temperature* Dompared with the ;e +,!a
grade& there is a considerably higher risk of permanent demagnetiGation in o%erspeed operation or
in short<circuit situations& if the temperature is high enough* The rotating speed limit estimates in
per unit %alues according to the magnet grade data are gi%en in Table !**
Table* !** 1stimated rotating speed limits for different temperatures according to the
magnet data presented in 'igs* !*( and !*! for the magnets ;e+,!a and -=!a&
respecti%ely*
Temperature JMDL ;e-=!a ;e+,!a
") ( (no limit)
)) !*, (no limit)
() (*! (no limit)
=) *=( +*"=
") (no data) (*),

,)


2n order to keep the machine siGe small& the permanent magnet material should be utiliGed as well
as possible* This reAuirement results in a sAuared air gap flux wa%eform* This can be obser%ed
from the back electromoti%e force calculation in 'lux (D* The results are shown in 'ig* !*+*

'ig* !*+* Simulated line<to<line back electromoti%e wa%eforms of the prototype machines*
The dashed line indicates the 1M designs while the solid line is for the MM
structure* The one slot pitch skewing of the actual machine results in a more
sinusoidal wa%eform*
The non<skewed back 1M' wa%eform is not sinusoidal& because the pole shapes are not formed to
produce a sinusoidal wa%eform* ;e%ertheless& the skewing of the actual machine corrects this
situation* The pole shapes of the machines are designed to maximiGe the flux generation with the
gi%en magnet material* The pole<generated air gap flux wa%eforms are presented in 'ig* !*=*
) ( ! + = > - "
x )
<!
<(=)
<())
<=)
<))
<=)
)
=)
))
=)
())
(=)
Times JsL
7
a
c
k

e
l
e
c
t
r
o
m
o
t
i
%
e

f
o
r
c
e

J
5
L


1M
MM
1M skewed
MM skewed
,)


2n order to keep the machine siGe small& the permanent magnet material should be utiliGed as well
as possible* This reAuirement results in a sAuared air gap flux wa%eform* This can be obser%ed
from the back electromoti%e force calculation in 'lux (D* The results are shown in 'ig* !*+*

'ig* !*+* Simulated line<to<line back electromoti%e wa%eforms of the prototype machines*
The dashed line indicates the 1M designs while the solid line is for the MM
structure* The one slot pitch skewing of the actual machine results in a more
sinusoidal wa%eform*
The non<skewed back 1M' wa%eform is not sinusoidal& because the pole shapes are not formed to
produce a sinusoidal wa%eform* ;e%ertheless& the skewing of the actual machine corrects this
situation* The pole shapes of the machines are designed to maximiGe the flux generation with the
gi%en magnet material* The pole<generated air gap flux wa%eforms are presented in 'ig* !*=*
) ( ! + = > - "
x )
<!
<(=)
<())
<=)
<))
<=)
)
=)
))
=)
())
(=)
Times JsL
7
a
c
k

e
l
e
c
t
r
o
m
o
t
i
%
e

f
o
r
c
e

J
5
L


1M
MM
1M skewed
MM skewed
,



'ig* !*=* :ne<pole air gap flux density wa%eforms normal to the surface for the MM and
1M structures in no<load conditions without skewing* The estimate is from the
middle of the annulus with a pole position similar to 'igs* !*, and !** The
notching of the wa%eform is caused by the slotting effect*
7ecause of the stronger magnets& the flux wa%eform of the 1M structure has higher flux density
%alues* The flux maximiGation results in a sAuare wa%eform of the air gap flux density* The
wa%eform can be made sinusoidal by pole shaping& but this results in a poor utiliGation of the
magnet material*

,.. "ynchronous inductances
The static finite element model illustrates the torAue output capability of the machine* There are
two ways to analyse the machine performance* A direct method is to obtain steady<state
performance points by a 'lux (D simulation* 1%en though the current<fed calculations are
relati%ely fast& the calculations still take time* Another method is to use a 'lux (D solution for the
two<axis model synchronous inductances as a basis for an analytical performance analysis*
As presented in Section *!& the two<axis model synchronous inductances can be used to determine
the machine performance characteristics* The magnitude of the synchronous inductance depends
) ) () !) +) =)
)
)*
)*(
)*!
)*+
)*=
)*>
)*-
)*"
)*,

.ole width JmmL

A
i
r
g
a
p

f
l
u
x

d
e
n
s
i
t
y

J
5
s
E
m
(
L


1M
MM
,



'ig* !*=* :ne<pole air gap flux density wa%eforms normal to the surface for the MM and
1M structures in no<load conditions without skewing* The estimate is from the
middle of the annulus with a pole position similar to 'igs* !*, and !** The
notching of the wa%eform is caused by the slotting effect*
7ecause of the stronger magnets& the flux wa%eform of the 1M structure has higher flux density
%alues* The flux maximiGation results in a sAuare wa%eform of the air gap flux density* The
wa%eform can be made sinusoidal by pole shaping& but this results in a poor utiliGation of the
magnet material*

,.. "ynchronous inductances
The static finite element model illustrates the torAue output capability of the machine* There are
two ways to analyse the machine performance* A direct method is to obtain steady<state
performance points by a 'lux (D simulation* 1%en though the current<fed calculations are
relati%ely fast& the calculations still take time* Another method is to use a 'lux (D solution for the
two<axis model synchronous inductances as a basis for an analytical performance analysis*
As presented in Section *!& the two<axis model synchronous inductances can be used to determine
the machine performance characteristics* The magnitude of the synchronous inductance depends
) ) () !) +) =)
)
)*
)*(
)*!
)*+
)*=
)*>
)*-
)*"
)*,

.ole width JmmL

A
i
r
g
a
p

f
l
u
x

d
e
n
s
i
t
y

J
5
s
E
m
(
L


1M
MM
,(


on the magnetiGation state of the machine* 8enerally& the direct<axis inductance is Auite stable& but
the Auadrature<axis inductance is sensiti%e to the magnetiGation state of the machine*
Traditional two<axis model synchronous inductances can be e%aluated by the flux (D software*
The inductances are calculated by applying magnetiGation to an indi%idual flux axis* The d<axis is
placed to point to the coil centre in order to determine the d<axis synchronous inductance* The A<
axis is e%aluated in a similar way by rotating the pole by ,) electrical degrees to obtain the A<axis
in the magnetiGing direction*
The steady<state solution of the 'lux (D sol%er does not take into account the end<winding leakage
term* Thus& the leakage in the end<winding region is determined analytically* #ater on& the end<
winding leakage term is also applied to the transient calculations of the %oltage<supplied models*
The end<winding leakage term is calculated by eAuation
I

=
4m

qN
2

0
(2l
cw

cw
+w
w

w
) !*+
where
< l
ew
is the axial height of the end winding&
< W
w
is the width of the coil span and
<
ew
and
w
are the permeance coefficients of the end winding according to to
(5ogt& ,"!)*
The terms
ew
and
w
are permeance coefficients of the winding geometry* The cross<section of the
end<winding region of the winding scheme presented in 'ig* !*" is in accordance with 'ig* !*>*
Thus& the corresponding coefficients
ew
and
w
ha%e %alues )*!(+ and )*(+! (5ogt &,"!)&
respecti%ely*

'ig* !*>* 8eometrical arrangement for the end winding of a half slot winding results in
the %alues of )*!(+ and )*(+! for
l
and
w
& respecti%ely (5ogt& ,"!) for the
end<winding leakage term calculation*
,(


on the magnetiGation state of the machine* 8enerally& the direct<axis inductance is Auite stable& but
the Auadrature<axis inductance is sensiti%e to the magnetiGation state of the machine*
Traditional two<axis model synchronous inductances can be e%aluated by the flux (D software*
The inductances are calculated by applying magnetiGation to an indi%idual flux axis* The d<axis is
placed to point to the coil centre in order to determine the d<axis synchronous inductance* The A<
axis is e%aluated in a similar way by rotating the pole by ,) electrical degrees to obtain the A<axis
in the magnetiGing direction*
The steady<state solution of the 'lux (D sol%er does not take into account the end<winding leakage
term* Thus& the leakage in the end<winding region is determined analytically* #ater on& the end<
winding leakage term is also applied to the transient calculations of the %oltage<supplied models*
The end<winding leakage term is calculated by eAuation
I

=
4m

qN
2

0
(2l
cw

cw
+w
w

w
) !*+
where
< l
ew
is the axial height of the end winding&
< W
w
is the width of the coil span and
<
ew
and
w
are the permeance coefficients of the end winding according to to
(5ogt& ,"!)*
The terms
ew
and
w
are permeance coefficients of the winding geometry* The cross<section of the
end<winding region of the winding scheme presented in 'ig* !*" is in accordance with 'ig* !*>*
Thus& the corresponding coefficients
ew
and
w
ha%e %alues )*!(+ and )*(+! (5ogt &,"!)&
respecti%ely*

'ig* !*>* 8eometrical arrangement for the end winding of a half slot winding results in
the %alues of )*!(+ and )*(+! for
l
and
w
& respecti%ely (5ogt& ,"!) for the
end<winding leakage term calculation*
,!


According to the winding dimensions& the leakage term according to 1A* (*!) is )*)=!- m9&
which yields )*)=( in per unit %alues* The inductance base %alue for the per unit con%ersion is
calculated for a ))) min
<
rotating speed& a =) 5 line<to<line %oltage and a )) A line current&
yielding *)! m9* The synchronous inductances of both machine structures& including the end
leakage term& are presented in 'ig* !*-*

'ig* !*-* Direct< and Auadrature<axis inductances as a per unit %alue presentation as a
function of supply current* The current is aligned with one axis at a time to
obtain the inductance %alues for the gi%en axis* The method is applied separately
for the d< and A<axes*
'ig !*- shows that the 1M machine has higher inductance ratings than the MM machine& which is
in good agreement with the theory of inductance calculation& because the 1M configuration has
more iron in the flux path compared with the MM configuration* The direct<axis inductance is )*+>
for the 1M structure and )*+ for the MM structure* 7ecause the direct<axis synchronous
inductance is in%ersely proportional to the breako%er torAue& the breako%er torAue %alues for the
MM and 1M machine according to Table !*( inductances would be (*++ and (*- times the
nominal torAue of (+) ;m if the saliency and saturation effects are neglected*
The Auadrature<axis inductances are higher compared with the direct<axis inductances& as
expected* The Auadrature<axis inductances seem to saturate hea%ily as the A<axis current increases&
especially in the 1M structure* 7ecause of the armature reaction& the flux path starts to shift in the
A<axis direction& which causes the A<axis inductance to saturate*
<=)) <+)) <!)) <()) <)) ) )) ()) !)) +)) =))
)*(
)*!
)*+
)*=
)*>
)*-
)*"
)*,
RMS axis current JAL
.
e
r

u
n
i
t

i
n
d
u
c
t
a
n
c
e


MM #
d
MM #
A
1M #
d
1M #
A
,!


According to the winding dimensions& the leakage term according to 1A* (*!) is )*)=!- m9&
which yields )*)=( in per unit %alues* The inductance base %alue for the per unit con%ersion is
calculated for a ))) min
<
rotating speed& a =) 5 line<to<line %oltage and a )) A line current&
yielding *)! m9* The synchronous inductances of both machine structures& including the end
leakage term& are presented in 'ig* !*-*

'ig* !*-* Direct< and Auadrature<axis inductances as a per unit %alue presentation as a
function of supply current* The current is aligned with one axis at a time to
obtain the inductance %alues for the gi%en axis* The method is applied separately
for the d< and A<axes*
'ig !*- shows that the 1M machine has higher inductance ratings than the MM machine& which is
in good agreement with the theory of inductance calculation& because the 1M configuration has
more iron in the flux path compared with the MM configuration* The direct<axis inductance is )*+>
for the 1M structure and )*+ for the MM structure* 7ecause the direct<axis synchronous
inductance is in%ersely proportional to the breako%er torAue& the breako%er torAue %alues for the
MM and 1M machine according to Table !*( inductances would be (*++ and (*- times the
nominal torAue of (+) ;m if the saliency and saturation effects are neglected*
The Auadrature<axis inductances are higher compared with the direct<axis inductances& as
expected* The Auadrature<axis inductances seem to saturate hea%ily as the A<axis current increases&
especially in the 1M structure* 7ecause of the armature reaction& the flux path starts to shift in the
A<axis direction& which causes the A<axis inductance to saturate*
<=)) <+)) <!)) <()) <)) ) )) ()) !)) +)) =))
)*(
)*!
)*+
)*=
)*>
)*-
)*"
)*,
RMS axis current JAL
.
e
r

u
n
i
t

i
n
d
u
c
t
a
n
c
e


MM #
d
MM #
A
1M #
d
1M #
A
,+


Since the two<axis model %alues are calculated separately for the d< and A<axes& the estimate does
not take into account the cross<saturation effect on the inductances* 2n this type of traction
machines& an indi%idual axis analysis in not sufficient for accurate performance e%aluation as
presented in the following sections* The direct< and Auadrature<axis inductances depend on both
current components* The cross<magnetiGing effect can be e%aluated by calculating the inductances
of indi%idual axes in interference with the intersecting current component as presented in
(Stumberger et al*& ())!)* 2nductances are calculated according to the eAuation

I
d,q
=
o
d,q
(i
d,q
, i
q,d
)
oi
d,q
=
A
d,q
Ai
d,q

!*=

The direct< and Auadrature<axis inductance graphs for the MM and 1M arrangements are presented
in 'igs* !*"@!*(*

'ig* !*"* .er unit direct<axis synchronous inductance including the cross<saturation effect
of the Auadrature<axis current on the MM machine*
,+


Since the two<axis model %alues are calculated separately for the d< and A<axes& the estimate does
not take into account the cross<saturation effect on the inductances* 2n this type of traction
machines& an indi%idual axis analysis in not sufficient for accurate performance e%aluation as
presented in the following sections* The direct< and Auadrature<axis inductances depend on both
current components* The cross<magnetiGing effect can be e%aluated by calculating the inductances
of indi%idual axes in interference with the intersecting current component as presented in
(Stumberger et al*& ())!)* 2nductances are calculated according to the eAuation

I
d,q
=
o
d,q
(i
d,q
, i
q,d
)
oi
d,q
=
A
d,q
Ai
d,q

!*=

The direct< and Auadrature<axis inductance graphs for the MM and 1M arrangements are presented
in 'igs* !*"@!*(*

'ig* !*"* .er unit direct<axis synchronous inductance including the cross<saturation effect
of the Auadrature<axis current on the MM machine*
,=



'ig* !*,* .er unit Auadrature<axis inductance including the cross<saturation effect of the
direct<axis current on the MM machine*

'ig* !*()* .er unit direct<axis synchronous inductance including the cross<saturation effect
of the Auadrature<axis current on the 1M machine*
,=



'ig* !*,* .er unit Auadrature<axis inductance including the cross<saturation effect of the
direct<axis current on the MM machine*

'ig* !*()* .er unit direct<axis synchronous inductance including the cross<saturation effect
of the Auadrature<axis current on the 1M machine*
,>



'ig* !*(* .er unit Auadrature<axis inductance including the cross<saturation effect of the
direct<axis current on the 1M machine*
The inductances are symmetrical with the positi%e and negati%e %alues of the Auadrature<axis
current in both L
d
and L
A
* The main difference lies in the direct<axis current component* Hith the
demagnetiGing direct<axis current& the inductance le%els usually remain steady as presented abo%e
in 'ig* !*-* An increasing torAue causes the direct<axis inductance to saturate earlier& which
narrows down the field weakening region of the machine* Dompared with the 1M design& the
magnet module inductances respond slowly to current changes* The laminate material in the rotor
surface of the 1M machine and the iron paths to the Auadrature axis saturate more easily under
increased magnetiGation& which causes large %ariations in the inductance %alues*

,.., )or.ue production
2n the normal motoring operation of the non<salient pole machine& the i
d
S ) control usually
de%elops the best torAue to current ratio& and the torAue is controlled by the amount of positi%e I
A

current* 'or non<salient pole designs& the best combination is found by using a suitable amount of
negati%e direct<axis current along with the positi%e Auadrature<axis current& as presented in Section
*!*!*
Traditionally& indi%idual axis inductances ha%e been used for salient pole machine performance
estimation* The per unit %alues of 'ig* !*- are gi%en in Table !*(*
,>



'ig* !*(* .er unit Auadrature<axis inductance including the cross<saturation effect of the
direct<axis current on the 1M machine*
The inductances are symmetrical with the positi%e and negati%e %alues of the Auadrature<axis
current in both L
d
and L
A
* The main difference lies in the direct<axis current component* Hith the
demagnetiGing direct<axis current& the inductance le%els usually remain steady as presented abo%e
in 'ig* !*-* An increasing torAue causes the direct<axis inductance to saturate earlier& which
narrows down the field weakening region of the machine* Dompared with the 1M design& the
magnet module inductances respond slowly to current changes* The laminate material in the rotor
surface of the 1M machine and the iron paths to the Auadrature axis saturate more easily under
increased magnetiGation& which causes large %ariations in the inductance %alues*

,.., )or.ue production
2n the normal motoring operation of the non<salient pole machine& the i
d
S ) control usually
de%elops the best torAue to current ratio& and the torAue is controlled by the amount of positi%e I
A

current* 'or non<salient pole designs& the best combination is found by using a suitable amount of
negati%e direct<axis current along with the positi%e Auadrature<axis current& as presented in Section
*!*!*
Traditionally& indi%idual axis inductances ha%e been used for salient pole machine performance
estimation* The per unit %alues of 'ig* !*- are gi%en in Table !*(*
,-


Table* !*(* .er unit inductances of the MM and 1M structure* The currents are aligned
according to the corresponding axes as presented in 'ig* !*-*
Durrent JAL L
d&1M
L
A&1M
#
d&MM
#
A&MM

<=))*) )*!, )*!! )*!> )*!
<+=)*) )*!, )*!+ )*!- )*!!
<+))*) )*+) )*!> )*!" )*!=
<!=)*) )*+ )*!, )*!, )*!-
<!))*) )*+! )*+( )*+) )*+)
<(=)*) )*++ )*+= )*+) )*+!
<())*) )*++ )*+, )*+ )*+=
<=)*) )*+= )*=! )*+ )*+-
<))*) )*+> )*=- )*+ )*+"
<=)*) )*+> )*>+ )*+ )*+,
<)*) )*+> )*>> )*+ )*+,
)*) )*+> )*>, )*+ )*+,
=)*) )*+> )*-( )*+ )*+,
))*) )*+> )*-= )*+ )*+,
=)*) )*+= )*-- )*+ )*+,
())*) )*+( )*>- )*+) )*+,
(=)*) )*!, )*=, )*!" )*+"
!))*) )*!= )*== )*!> )*+-
!=)*) )*!( )*= )*! )*+=
+))*) )*(, )*+- )*(- )*+!
+=)*) )*(> )*+! )*(= )*!,
=))*) )*(= )*+) )*(! )*!>

According to the estimated two<axis model inductances& the torAue output of the machine can be
estimated by 1A* *,* The torAue production as a function of current angle is presented in 'ig* !*((*
,-


Table* !*(* .er unit inductances of the MM and 1M structure* The currents are aligned
according to the corresponding axes as presented in 'ig* !*-*
Durrent JAL L
d&1M
L
A&1M
#
d&MM
#
A&MM

<=))*) )*!, )*!! )*!> )*!
<+=)*) )*!, )*!+ )*!- )*!!
<+))*) )*+) )*!> )*!" )*!=
<!=)*) )*+ )*!, )*!, )*!-
<!))*) )*+! )*+( )*+) )*+)
<(=)*) )*++ )*+= )*+) )*+!
<())*) )*++ )*+, )*+ )*+=
<=)*) )*+= )*=! )*+ )*+-
<))*) )*+> )*=- )*+ )*+"
<=)*) )*+> )*>+ )*+ )*+,
<)*) )*+> )*>> )*+ )*+,
)*) )*+> )*>, )*+ )*+,
=)*) )*+> )*-( )*+ )*+,
))*) )*+> )*-= )*+ )*+,
=)*) )*+= )*-- )*+ )*+,
())*) )*+( )*>- )*+) )*+,
(=)*) )*!, )*=, )*!" )*+"
!))*) )*!= )*== )*!> )*+-
!=)*) )*!( )*= )*! )*+=
+))*) )*(, )*+- )*(- )*+!
+=)*) )*(> )*+! )*(= )*!,
=))*) )*(= )*+) )*(! )*!>

According to the estimated two<axis model inductances& the torAue output of the machine can be
estimated by 1A* *,* The torAue production as a function of current angle is presented in 'ig* !*((*
,"



'ig* !*((* 1stimated torAue output as a function of current angle calculated according to
the synchronous inductance data presented in 'ig* !*- and Table !*(* The solid
line is for the MM machine while the dashed line is for the 1M machine* The
currents range from =) A
RMS
to +)) A
RMS
at =) A
RMS
steps* Stator /oule losses
are neglected* .ermanent magnet flux and supply %oltage are p*u*
The estimation of indi%idual component torAue production according to the inductances of Table
!*( shows peak torAues below current angles of ,) degrees at the current ratings abo%e ()) A&
which is caused by the re%erse of the inductance ratio as the saturation reduces the Auadrature<axis
inductance below the direct<axis inductance %alues& as presented in 'ig !*((* The negati%e torAue
at low current angle %alues is caused by the reluctance torAue being larger than the synchronous
torAue component*
As a comparison with 'ig* !*((& the electromagnetic torAue estimation according to the magneto<
static model of 'lux(Db at )) A of excitation is presented in 'ig* !*(!*
) () +) >) ") )) () +) >) ")
<())
)
())
+))
>))
"))
)))
Durrent angle
T
o
r
A
u
e

J
;
m
L
,"



'ig* !*((* 1stimated torAue output as a function of current angle calculated according to
the synchronous inductance data presented in 'ig* !*- and Table !*(* The solid
line is for the MM machine while the dashed line is for the 1M machine* The
currents range from =) A
RMS
to +)) A
RMS
at =) A
RMS
steps* Stator /oule losses
are neglected* .ermanent magnet flux and supply %oltage are p*u*
The estimation of indi%idual component torAue production according to the inductances of Table
!*( shows peak torAues below current angles of ,) degrees at the current ratings abo%e ()) A&
which is caused by the re%erse of the inductance ratio as the saturation reduces the Auadrature<axis
inductance below the direct<axis inductance %alues& as presented in 'ig !*((* The negati%e torAue
at low current angle %alues is caused by the reluctance torAue being larger than the synchronous
torAue component*
As a comparison with 'ig* !*((& the electromagnetic torAue estimation according to the magneto<
static model of 'lux(Db at )) A of excitation is presented in 'ig* !*(!*
) () +) >) ") )) () +) >) ")
<())
)
())
+))
>))
"))
)))
Durrent angle
T
o
r
A
u
e

J
;
m
L
,,



'ig* !*(!* TorAue production as a function of current angle according to the 'lux (D model
at the )) A current*
The torAue cur%e is distorted& and the peak torAue point is not as distinct as the theory suggests*
This is mainly due to the absence of skewing in the modelling* The permeance differences
resulting from the rotation of the rotor in the static analysis cause fluctuation in the torAue cur%e&
and thus& do not gi%e a reliable estimate on the torAue beha%iour* ;e%ertheless& the reluctance peak
is higher for the embedded design& and the earlier occurrence of the reluctance torAue maximum
seems to fa%our the magnet module design* The finite element e%aluation also predicts a torAue
peak around a ,)<degree current angle at the nominal current& which is in line with the analytical
e%aluation*
At higher current le%els& the finite element approach suggests current angles abo%e %alues of ,)
degrees& which is in line with the theory presented in Section *! but differs from the analytical
two<axis model results presented in 'ig !*((* At higher current %alues& the torAue cur%es start to
reach a uniform shape for both of the machines* The peak torAue production takes place around
()<degree current angle for both machines* The 'lux (D estimates for torAue at the !)) A and
+)) A currents are presented in 'igs* !*(+ and !*(=*
) () +) >) ") )) () +) >) ")
<=)
)
=)
))
=)
())
(=)
!))
Durrent angle J
o
L
T
o
r
A
u
e

J
;
m
L


MM T
tot
MM T
rel
1M T
t ot
1M T
rel
,,



'ig* !*(!* TorAue production as a function of current angle according to the 'lux (D model
at the )) A current*
The torAue cur%e is distorted& and the peak torAue point is not as distinct as the theory suggests*
This is mainly due to the absence of skewing in the modelling* The permeance differences
resulting from the rotation of the rotor in the static analysis cause fluctuation in the torAue cur%e&
and thus& do not gi%e a reliable estimate on the torAue beha%iour* ;e%ertheless& the reluctance peak
is higher for the embedded design& and the earlier occurrence of the reluctance torAue maximum
seems to fa%our the magnet module design* The finite element e%aluation also predicts a torAue
peak around a ,)<degree current angle at the nominal current& which is in line with the analytical
e%aluation*
At higher current le%els& the finite element approach suggests current angles abo%e %alues of ,)
degrees& which is in line with the theory presented in Section *! but differs from the analytical
two<axis model results presented in 'ig !*((* At higher current %alues& the torAue cur%es start to
reach a uniform shape for both of the machines* The peak torAue production takes place around
()<degree current angle for both machines* The 'lux (D estimates for torAue at the !)) A and
+)) A currents are presented in 'igs* !*(+ and !*(=*
) () +) >) ") )) () +) >) ")
<=)
)
=)
))
=)
())
(=)
!))
Durrent angle J
o
L
T
o
r
A
u
e

J
;
m
L


MM T
tot
MM T
rel
1M T
t ot
1M T
rel
))



'ig* !*(+* TorAue production as a function of current angle according to the 'lux (D model
at the !)) A current*

) () +) >) ") )) () +) >) ")
<())
<))
)
))
())
!))
+))
=))
>))
-))
"))
Durrent angle J
o
L
T
o
r
A
u
e

J
;
m
L


MM T
tot
MM T
rel
1M T
t ot
1M T
rel
))



'ig* !*(+* TorAue production as a function of current angle according to the 'lux (D model
at the !)) A current*

) () +) >) ") )) () +) >) ")
<())
<))
)
))
())
!))
+))
=))
>))
-))
"))
Durrent angle J
o
L
T
o
r
A
u
e

J
;
m
L


MM T
tot
MM T
rel
1M T
t ot
1M T
rel
)



'ig* !*(=* TorAue production as a function of current angle according to the 'lux (D model
at the +)) A current*
The reluctance torAue production can be seen clearly as a flux path shift at the higher supply
current %alues compared for example with the nominal point current %alue of )) A* 9owe%er& the
cur%e of the reluctance component according to the current angle is completely uniAue for both of
the machines* The reluctance torAue production reAuires large currents and demagnetiGing current
in the direct axis*
The two<axis inductance estimation alone does not gi%e an exact estimate of the machine
inductance beha%iour& because the air gap flux density wa%eform is not purely sinusoidal (see 'ig*
!*=)& (Miller& ,,+)* The peak %alues of torAue at different excitation currents seem to match
closely& but the difference in the occurring current angle is significant* The armature reaction&
which consists of the d<axis and the A<axis field should be taken into account in more detail to
achie%e a better correspondence of the analytical and finite<element<based estimates* Hith the
cross<saturation taken into account& the torAue cur%e of the analytical model is corrected close to
the finite element estimates as presented in 'ig* !*(>*
) () +) >) ") )) () +) >) ")
<())
)
())
+))
>))
"))
)))
Durrent angle J
o
L
T
o
r
A
u
e

J
;
m
L


MM T
tot
MM T
rel
1M T
t ot
1M T
rel
)



'ig* !*(=* TorAue production as a function of current angle according to the 'lux (D model
at the +)) A current*
The reluctance torAue production can be seen clearly as a flux path shift at the higher supply
current %alues compared for example with the nominal point current %alue of )) A* 9owe%er& the
cur%e of the reluctance component according to the current angle is completely uniAue for both of
the machines* The reluctance torAue production reAuires large currents and demagnetiGing current
in the direct axis*
The two<axis inductance estimation alone does not gi%e an exact estimate of the machine
inductance beha%iour& because the air gap flux density wa%eform is not purely sinusoidal (see 'ig*
!*=)& (Miller& ,,+)* The peak %alues of torAue at different excitation currents seem to match
closely& but the difference in the occurring current angle is significant* The armature reaction&
which consists of the d<axis and the A<axis field should be taken into account in more detail to
achie%e a better correspondence of the analytical and finite<element<based estimates* Hith the
cross<saturation taken into account& the torAue cur%e of the analytical model is corrected close to
the finite element estimates as presented in 'ig* !*(>*
) () +) >) ") )) () +) >) ")
<())
)
())
+))
>))
"))
)))
Durrent angle J
o
L
T
o
r
A
u
e

J
;
m
L


MM T
tot
MM T
rel
1M T
t ot
1M T
rel
)(



'ig* !*(>* 1stimated torAue output as a function of current angle calculated according to
the cross<saturated inductances* The solid line is for the MM machine& while the
dashed line is for the 1M machine* The currents range from =) A
RMS
to +))
A
RMS
at =) A
RMS
steps* Stator /oule losses are neglected* .ermanent magnet flux
and supply %oltage are p*u*
The torAue production cur%e is now in line with the theory* The steady<state current analysis
suggests a similar beha%iour of the torAue production in the finite element modelling* 1%en though
the inductance charts seem to differ substantially& as shown in 'igs* !*"@!*(& the torAue
production beha%iour is surprisingly similar*
7oth of the designs show increased torAue production capability because of the rotor saliency*
The reluctance torAue component cur%es are presented in 'ig !*(- and the proportion of the
reluctance torAue production in the total torAue production is presented as a function of current
angle in 'ig* !*("*
) () +) >) ") )) () +) >) ")
)
())
+))
>))
"))
)))
())
Durrent angle JML
T
o
r
A
u
e

J
;
m
L
)(



'ig* !*(>* 1stimated torAue output as a function of current angle calculated according to
the cross<saturated inductances* The solid line is for the MM machine& while the
dashed line is for the 1M machine* The currents range from =) A
RMS
to +))
A
RMS
at =) A
RMS
steps* Stator /oule losses are neglected* .ermanent magnet flux
and supply %oltage are p*u*
The torAue production cur%e is now in line with the theory* The steady<state current analysis
suggests a similar beha%iour of the torAue production in the finite element modelling* 1%en though
the inductance charts seem to differ substantially& as shown in 'igs* !*"@!*(& the torAue
production beha%iour is surprisingly similar*
7oth of the designs show increased torAue production capability because of the rotor saliency*
The reluctance torAue component cur%es are presented in 'ig !*(- and the proportion of the
reluctance torAue production in the total torAue production is presented as a function of current
angle in 'ig* !*("*
) () +) >) ") )) () +) >) ")
)
())
+))
>))
"))
)))
())
Durrent angle JML
T
o
r
A
u
e

J
;
m
L
)!



'ig* !*(-* Reluctance torAue component as a function of current angle* The currents range
from =) A
RMS
to +)) A
RMS
at =) A
RMS
steps* The dashed line indicates the 1M
machine and the solid line for the MM machine*

) () +) >) ") )) () +) >) ")
<())
<=)
<))
<=)
)
=)
))
=)
())
(=)
!))
Durrent angle JML
R
e
l
u
c
t
a
n
c
e

t
o
r
A
u
e

J
;
m
L
*
)!



'ig* !*(-* Reluctance torAue component as a function of current angle* The currents range
from =) A
RMS
to +)) A
RMS
at =) A
RMS
steps* The dashed line indicates the 1M
machine and the solid line for the MM machine*

) () +) >) ") )) () +) >) ")
<())
<=)
<))
<=)
)
=)
))
=)
())
(=)
!))
Durrent angle JML
R
e
l
u
c
t
a
n
c
e

t
o
r
A
u
e

J
;
m
L
*
)+



'ig* !*("* .roportion of reluctance torAue in the total torAue production as a function of
current angle* The currents range from =) A
RMS
to +)) A
RMS
at =) A
RMS
steps*
The dashed line indicates the 1M machine and the solid line for the MM
machine*
The torAue cur%es in 'ig* !*(- and !*(" demonstrate a clear benefit of the reluctance torAue
production in the total torAue output* The reluctance component is around () P of the total torAue
output at high current %alues* The analytical and finite element results show that the machines
under study ha%e potential& and the reluctance torAue component can be increased significantly
with a suitable amount of current*
The design process shows that special attention has to be paid to the modelling process* The
e%aluation suggests that the estimation of the cross<saturation effect is essential in the torAue
production analysis of a traction machine* The results are also important from the perspecti%e of
the dri%e parameter selection* Domparing with 'igs* !*(( and !*(>& the incorrect selection of dri%e
parameters can lead to a O() P loss in peak torAue production*
To achie%e the best performance& the electrical machine should be dri%en with a minimum current
control& which results in the highest torAue to current ratio for the gi%en operation point as
presented in 'ig* !*(,*
) () +) >) ") )) () +) >) ")
<
<)*"
<)*>
<)*+
<)*(
)
)*(
)*+
)*>
Durrent angle JML
R
e
l
u
c
t
a
n
c
e

t
o
r
A
u
e

t
o

t
o
t
a
l

t
o
r
A
u
e

p
*
u
*
)+



'ig* !*("* .roportion of reluctance torAue in the total torAue production as a function of
current angle* The currents range from =) A
RMS
to +)) A
RMS
at =) A
RMS
steps*
The dashed line indicates the 1M machine and the solid line for the MM
machine*
The torAue cur%es in 'ig* !*(- and !*(" demonstrate a clear benefit of the reluctance torAue
production in the total torAue output* The reluctance component is around () P of the total torAue
output at high current %alues* The analytical and finite element results show that the machines
under study ha%e potential& and the reluctance torAue component can be increased significantly
with a suitable amount of current*
The design process shows that special attention has to be paid to the modelling process* The
e%aluation suggests that the estimation of the cross<saturation effect is essential in the torAue
production analysis of a traction machine* The results are also important from the perspecti%e of
the dri%e parameter selection* Domparing with 'igs* !*(( and !*(>& the incorrect selection of dri%e
parameters can lead to a O() P loss in peak torAue production*
To achie%e the best performance& the electrical machine should be dri%en with a minimum current
control& which results in the highest torAue to current ratio for the gi%en operation point as
presented in 'ig* !*(,*
) () +) >) ") )) () +) >) ")
<
<)*"
<)*>
<)*+
<)*(
)
)*(
)*+
)*>
Durrent angle JML
R
e
l
u
c
t
a
n
c
e

t
o
r
A
u
e

t
o

t
o
t
a
l

t
o
r
A
u
e

p
*
u
*
)=



'ig* !*(,* TorAue per ampere production as a function of current angle* The currents range
from =) A
RMS
to +)) A
RMS
at =) A
RMS
steps* The dashed line is for the 1M
machine while the solid line is for the MM machine*
The highest torAue to ampere ratios are achie%ed at current angles at which the reluctance torAue
component reaches the highest %alues*

,..4 Performance over speed range
The machine performance o%er the operating speed range is important in traction dri%es* 5arying
supply conditions reAuire good understanding of the machine state at different operating points*
The torAue production characteristic cur%es as a function of rotating speed for p*u supply %oltage
are presented in 'ig* !*!)* At speeds below p*u*& the constant %oltage beha%iour is shown for
curiosity* Ceeping the %oltage constant results in strongly limited torAue production below the
rated speed as the current ()*=@+ p*u*) is needed to increase the stator flux linkage& which must in
all conditions match with the %oltage supply* Domparison with 'ig* !*!! shows that when )*(= p*u*
%oltage is used& the torAue peak at + p*u* current is found at a lower speed& accordingly* These
figures are not intended to suggest any control method for the machines but only illustrate the
torAue producing capabilities at different speeds and constant %oltages*

) () +) >) ") )) () +) >) ")
)
)*=

*=
(
(*=
!
Durrent angle JML
T
o
r
A
u
e

p
e
r

A
m
p
e
r
e
)=



'ig* !*(,* TorAue per ampere production as a function of current angle* The currents range
from =) A
RMS
to +)) A
RMS
at =) A
RMS
steps* The dashed line is for the 1M
machine while the solid line is for the MM machine*
The highest torAue to ampere ratios are achie%ed at current angles at which the reluctance torAue
component reaches the highest %alues*

,..4 Performance over speed range
The machine performance o%er the operating speed range is important in traction dri%es* 5arying
supply conditions reAuire good understanding of the machine state at different operating points*
The torAue production characteristic cur%es as a function of rotating speed for p*u supply %oltage
are presented in 'ig* !*!)* At speeds below p*u*& the constant %oltage beha%iour is shown for
curiosity* Ceeping the %oltage constant results in strongly limited torAue production below the
rated speed as the current ()*=@+ p*u*) is needed to increase the stator flux linkage& which must in
all conditions match with the %oltage supply* Domparison with 'ig* !*!! shows that when )*(= p*u*
%oltage is used& the torAue peak at + p*u* current is found at a lower speed& accordingly* These
figures are not intended to suggest any control method for the machines but only illustrate the
torAue producing capabilities at different speeds and constant %oltages*

) () +) >) ") )) () +) >) ")
)
)*=

*=
(
(*=
!
Durrent angle JML
T
o
r
A
u
e

p
e
r

A
m
p
e
r
e
)>



'ig* !*!)* Domparison of torAue production at currents )*=@+ per unit at )*= per unit steps*
The solid line is for the MM machine and the dashed line for the 1M machine*
'igure shows the a%ailable maximum torAue when permanent magnet flux and
supply %oltage are p*u*
Domparison of the torAue cur%es in 'ig* !*!) shows the better field weakening characteristics of
the MM machine* The torAue<producing capability of the 1M machine is slightly better at higher
current %alues* 9igh pull<out torAue %alues are achie%ed at subnominal speeds where the excess
%oltage is used to boost the flux linkage* This point represents the maximum torAue output of the
machine* Depending on the supply %oltage le%el& the peak torAue point shifts along the rotating
speed axis& which is in this case around +*= times the nominal torAue %alue for both of the
machines* 9igher peak torAue le%els can be achie%ed& but it reAuires more current to achie%e a
higher %oltage to speed ratio to increase the flux linkage le%el*
The field weakening beha%iour of both machines shows the saturation effect of inductances at high
current %alues& which narrows down the field weakening region* 2n order to achie%e better torAue
output& the current le%el has to be decreased as the speed increases* The torAue production cur%es
of the machines are presented separately in 'igs* !*! and !*!( in addition to the constant power
cur%es*
) )*= *= (
)
)*=

*=
(
(*=
!
!*=
+
+*=
.er unit rotating speed
.
e
r

u
n
i
t

t
o
r
A
u
e
)>



'ig* !*!)* Domparison of torAue production at currents )*=@+ per unit at )*= per unit steps*
The solid line is for the MM machine and the dashed line for the 1M machine*
'igure shows the a%ailable maximum torAue when permanent magnet flux and
supply %oltage are p*u*
Domparison of the torAue cur%es in 'ig* !*!) shows the better field weakening characteristics of
the MM machine* The torAue<producing capability of the 1M machine is slightly better at higher
current %alues* 9igh pull<out torAue %alues are achie%ed at subnominal speeds where the excess
%oltage is used to boost the flux linkage* This point represents the maximum torAue output of the
machine* Depending on the supply %oltage le%el& the peak torAue point shifts along the rotating
speed axis& which is in this case around +*= times the nominal torAue %alue for both of the
machines* 9igher peak torAue le%els can be achie%ed& but it reAuires more current to achie%e a
higher %oltage to speed ratio to increase the flux linkage le%el*
The field weakening beha%iour of both machines shows the saturation effect of inductances at high
current %alues& which narrows down the field weakening region* 2n order to achie%e better torAue
output& the current le%el has to be decreased as the speed increases* The torAue production cur%es
of the machines are presented separately in 'igs* !*! and !*!( in addition to the constant power
cur%es*
) )*= *= (
)
)*=

*=
(
(*=
!
!*=
+
+*=
.er unit rotating speed
.
e
r

u
n
i
t

t
o
r
A
u
e
)-



'ig* !*!* TorAue production cur%es of the MM structure based on the inductance %alues of
the saturation model* The red cur%es are torAue cur%es for currents )*=& & (& !
and + p*u* The black labelled contour lines are torAue cur%es for constant power
in per unit %alues* The back electromoti%e force and supply %oltage are p*u*

) )*= *= (
)
)*=

*=
(
(*=
!
!*=
+
+*=
.er unit rotating speed
.
e
r

u
n
i
t

t
o
r
A
u
e
L
)*=
)*=
)
*=

*=
*=

*=

*
=
(
(
(
(
)-



'ig* !*!* TorAue production cur%es of the MM structure based on the inductance %alues of
the saturation model* The red cur%es are torAue cur%es for currents )*=& & (& !
and + p*u* The black labelled contour lines are torAue cur%es for constant power
in per unit %alues* The back electromoti%e force and supply %oltage are p*u*

) )*= *= (
)
)*=

*=
(
(*=
!
!*=
+
+*=
.er unit rotating speed
.
e
r

u
n
i
t

t
o
r
A
u
e
L
)*=
)*=
)
*=

*=
*=

*=

*
=
(
(
(
(
)"



'ig* !*!(* TorAue production cur%es of the 1M structure based on the inductance %alues of
the saturation model* The red cur%es are torAue cur%es for currents )*=@+ per
unit at )*= per unit steps* The black labelled contour lines are constant power
cur%es in per unit %alues* The back electromoti%e force and supply %oltage are
p*u*
According to the analytical e%aluation& the field weakening characteristics of both machine types
are sufficient for the gi%en application when considering the nominal point operation* The 1M
machine field weakening characteristics are slightly better around nominal current %alues as a
result of the larger direct<axis inductance %alue when compared with the MM design*
;e%ertheless& the field weakening characteristics of the machines seem to be sufficient for the
gi%en application*
;ow& our only concern is the low battery %oltage* The torAue production cur%es as function of
rotating speed for )*(= p* u* supply %oltages are presented in 'ig* !*!!*
) )*= *= (
)
)*=

*=
(
(*=
!
!*=
+
+*=
.er unit rotating speed
.
e
r

u
n
i
t

t
o
r
A
u
e
)*(=
)*(=
)*(=
)*(=
)*-=
)*-=
)*-=
)
*
-
=

)"



'ig* !*!(* TorAue production cur%es of the 1M structure based on the inductance %alues of
the saturation model* The red cur%es are torAue cur%es for currents )*=@+ per
unit at )*= per unit steps* The black labelled contour lines are constant power
cur%es in per unit %alues* The back electromoti%e force and supply %oltage are
p*u*
According to the analytical e%aluation& the field weakening characteristics of both machine types
are sufficient for the gi%en application when considering the nominal point operation* The 1M
machine field weakening characteristics are slightly better around nominal current %alues as a
result of the larger direct<axis inductance %alue when compared with the MM design*
;e%ertheless& the field weakening characteristics of the machines seem to be sufficient for the
gi%en application*
;ow& our only concern is the low battery %oltage* The torAue production cur%es as function of
rotating speed for )*(= p* u* supply %oltages are presented in 'ig* !*!!*
) )*= *= (
)
)*=

*=
(
(*=
!
!*=
+
+*=
.er unit rotating speed
.
e
r

u
n
i
t

t
o
r
A
u
e
)*(=
)*(=
)*(=
)*(=
)*-=
)*-=
)*-=
)
*
-
=

),



'ig* !*!!* Domparison of torAue production at currents )*=@+ per unit at )*= per unit steps*
The solid line is for the MM machine and the dashed line for the 1M machine*
The figure shows the a%ailable maximum torAue when the permanent magnet
flux is per unit and the supply %oltage )*(= p*u*
Hith the (= P supply %oltage le%el& the machines are is still capable of producing peak torAue of
+*= times the nominal torAue& but the nominal speed reAuires o%er =) P of excess current* This
supply %oltage le%el is too low compared with the cut of le%el of the Altairnano =) Ah cell& the
cut<off %oltage of which is about >) P of the normal operating %oltage of !-+ 5 line<to<line*
9owe%er& a low %oltage le%el can lead to problems with the thermal endurance if this is not taken
into account in the design process* Thus& in the traction machine design process& it is necessary
also to take into account the thermal characteristics of the machine*

,., )ime stepping analysis
Transient magnetic analysis allows a more detailed e%aluation of the machine performance o%er
the operating speed range* 2t allows the determination of the iron and /oule losses and currents to
define the efficiency and power factor* Time stepping analyses are conducted with %oltage<fed
models and sinusoidal %oltage wa%eforms* The drawback of a sinusoidal supply wa%eform is the
optimistic estimate of the loss beha%iour under different load conditions* .HM<fed machines tend
) )*( )*+ )*> )*" *(
)
)*=

*=
(
(*=
!
!*=
+
+*=
.er unit rotating speed
.
e
r

u
n
i
t

t
o
r
A
u
e
),



'ig* !*!!* Domparison of torAue production at currents )*=@+ per unit at )*= per unit steps*
The solid line is for the MM machine and the dashed line for the 1M machine*
The figure shows the a%ailable maximum torAue when the permanent magnet
flux is per unit and the supply %oltage )*(= p*u*
Hith the (= P supply %oltage le%el& the machines are is still capable of producing peak torAue of
+*= times the nominal torAue& but the nominal speed reAuires o%er =) P of excess current* This
supply %oltage le%el is too low compared with the cut of le%el of the Altairnano =) Ah cell& the
cut<off %oltage of which is about >) P of the normal operating %oltage of !-+ 5 line<to<line*
9owe%er& a low %oltage le%el can lead to problems with the thermal endurance if this is not taken
into account in the design process* Thus& in the traction machine design process& it is necessary
also to take into account the thermal characteristics of the machine*

,., )ime stepping analysis
Transient magnetic analysis allows a more detailed e%aluation of the machine performance o%er
the operating speed range* 2t allows the determination of the iron and /oule losses and currents to
define the efficiency and power factor* Time stepping analyses are conducted with %oltage<fed
models and sinusoidal %oltage wa%eforms* The drawback of a sinusoidal supply wa%eform is the
optimistic estimate of the loss beha%iour under different load conditions* .HM<fed machines tend
) )*( )*+ )*> )*" *(
)
)*=

*=
(
(*=
!
!*=
+
+*=
.er unit rotating speed
.
e
r

u
n
i
t

t
o
r
A
u
e
)


to ha%e a higher loss generation because of the increased harmonic content of the supply* A finite<
element<based loss analysis is discussed in detail in Dhapter +*
,.,.! Aoltage#fed dynamic model
2n the finite element analysis& the %oltage<fed model gi%es more accurate %alues for the breako%er
torAue* This calculation includes the analytically determined end<winding leakage component
calculated in Section !*(*(* 1%en though the end<winding leakage component is small& the leakage
component becomes relati%ely significant in the torAue output analysis because the .M machine
synchronous inductances are low* The transient calculations are carried out by assuming a () MD
temperature for both of the machines* The transient magnetic torAue production including the end<
winding leakage component for the =) 5 line<to<line supply at a ))) min
<
rotating speed is
presented in 'ig* !*!+*

'ig* !*!+* TorAue production as a function of current with a transient<%oltage<dri%en
simulation at a ))) min
<
rotating speed and a =) 5 line<to<line supply
%oltage*
Dompared with the static current model& the torAue cur%e starts to bend after the ()) A supply
current* The peak %alue of torAue is achie%ed around !))A of supply current* An increase in the
torAue output reAuires a higher supply %oltage as 1A* *( suggests* 2n the MM configuration& the
flux starts to shift in the A<axis direction as the static model predicts* 2n the 1M structure& the .M
) =) )) =) ()) (=) !)) !=)
)
))
())
!))
+))
=))
>))
RMS Durrent JAL
T
o
r
A
u
e

J
;
m
L


1M
MM
)


to ha%e a higher loss generation because of the increased harmonic content of the supply* A finite<
element<based loss analysis is discussed in detail in Dhapter +*
,.,.! Aoltage#fed dynamic model
2n the finite element analysis& the %oltage<fed model gi%es more accurate %alues for the breako%er
torAue* This calculation includes the analytically determined end<winding leakage component
calculated in Section !*(*(* 1%en though the end<winding leakage component is small& the leakage
component becomes relati%ely significant in the torAue output analysis because the .M machine
synchronous inductances are low* The transient calculations are carried out by assuming a () MD
temperature for both of the machines* The transient magnetic torAue production including the end<
winding leakage component for the =) 5 line<to<line supply at a ))) min
<
rotating speed is
presented in 'ig* !*!+*

'ig* !*!+* TorAue production as a function of current with a transient<%oltage<dri%en
simulation at a ))) min
<
rotating speed and a =) 5 line<to<line supply
%oltage*
Dompared with the static current model& the torAue cur%e starts to bend after the ()) A supply
current* The peak %alue of torAue is achie%ed around !))A of supply current* An increase in the
torAue output reAuires a higher supply %oltage as 1A* *( suggests* 2n the MM configuration& the
flux starts to shift in the A<axis direction as the static model predicts* 2n the 1M structure& the .M
) =) )) =) ()) (=) !)) !=)
)
))
())
!))
+))
=))
>))
RMS Durrent JAL
T
o
r
A
u
e

J
;
m
L


1M
MM



flux tends to bend against the A<axis and short circuit with the ad3acent pole* This effect is less
intense in the MM structure because of the air gap path* To achie%e higher torAue ratings& the
supply %oltage has to be increased*

'ig* !*!=* 'lux paths of the MM (left) and 1M (right) configurations at the =)5
##
and
()) A supply*
The breako%er point of torAue production at the =) 5 supply is reached 3ust below a current of
!)) A with both of the machines at ))) min
<
* This means that the machines can only produce a
torAue maximum around =)) ;m with a !))A excitation current at the =) 5 line<to<line supply
%oltage* An optimal current angle to produce a higher torAue cannot be achie%ed without
increasing the supply %oltage* To achie%e higher pull<out torAue le%els& there has to be %oltage
reser%e designed in the system to reach the benefits from the reluctance torAue component as
presented in Section *!*
,.,. "hort#circuit endurance
7ecause of the low direct<axis synchronous inductance& the sustained short<circuit current of a .M
machine tends to be large* The short<circuiting current can cause a large enough demagnetiGing
flux in the d<axis direction to permanently demagnetiGe the magnets* The MM configuration is
especially %ulnerable as there is no iron in the d<axis direction to saturate and protect the magnet
closest to the air gap* The 1M configuration has thus better opportunities to withstand short
circuits*
Magnets can be protected by selecting the magnet grade carefully& which means choosing grades
without the !
rd
Auadrant remanence knee& but this usually leads to larger machine dimensions&
because the magnets ha%ing some counter<magnetiGation resistance ha%e lower energy products
compared with the high B
r
grades* A suitably high stator leakage also helps to protect the magnets
as it decreases the demagnetiGing flux of the machine*
To obser%e the short<circuit endurance of the machine structure& a three<phase short circuit
e%aluation was carried out in a 'lux(D en%ironment* Short circuit is applied to all three phases in
no<load conditions of a generator* The short<circuiting is timed to take place in a point where one
line<to<line %oltage sign is changing from negati%e to positi%e* This time period results in the
highest peak in the short<circuiting current in that particular phase* Short circuit currents are
presented in 'ig !*!>*



flux tends to bend against the A<axis and short circuit with the ad3acent pole* This effect is less
intense in the MM structure because of the air gap path* To achie%e higher torAue ratings& the
supply %oltage has to be increased*

'ig* !*!=* 'lux paths of the MM (left) and 1M (right) configurations at the =)5
##
and
()) A supply*
The breako%er point of torAue production at the =) 5 supply is reached 3ust below a current of
!)) A with both of the machines at ))) min
<
* This means that the machines can only produce a
torAue maximum around =)) ;m with a !))A excitation current at the =) 5 line<to<line supply
%oltage* An optimal current angle to produce a higher torAue cannot be achie%ed without
increasing the supply %oltage* To achie%e higher pull<out torAue le%els& there has to be %oltage
reser%e designed in the system to reach the benefits from the reluctance torAue component as
presented in Section *!*
,.,. "hort#circuit endurance
7ecause of the low direct<axis synchronous inductance& the sustained short<circuit current of a .M
machine tends to be large* The short<circuiting current can cause a large enough demagnetiGing
flux in the d<axis direction to permanently demagnetiGe the magnets* The MM configuration is
especially %ulnerable as there is no iron in the d<axis direction to saturate and protect the magnet
closest to the air gap* The 1M configuration has thus better opportunities to withstand short
circuits*
Magnets can be protected by selecting the magnet grade carefully& which means choosing grades
without the !
rd
Auadrant remanence knee& but this usually leads to larger machine dimensions&
because the magnets ha%ing some counter<magnetiGation resistance ha%e lower energy products
compared with the high B
r
grades* A suitably high stator leakage also helps to protect the magnets
as it decreases the demagnetiGing flux of the machine*
To obser%e the short<circuit endurance of the machine structure& a three<phase short circuit
e%aluation was carried out in a 'lux(D en%ironment* Short circuit is applied to all three phases in
no<load conditions of a generator* The short<circuiting is timed to take place in a point where one
line<to<line %oltage sign is changing from negati%e to positi%e* This time period results in the
highest peak in the short<circuiting current in that particular phase* Short circuit currents are
presented in 'ig !*!>*
(



'ig* !*!>* Three<phase short<circuit currents at ))) min
<
speed for the MM and 1M
structures* Ha%eforms include DD offset since the short circuit starts at peak
flux linkage*
The peak currents are >)-*= A for the MM machine and =!-*) for the 1M machine* The short
circuits current peaks occur at +*) ms and !*>- ms after the initialiGation of the short circuit* The
flux density graphs of the short circuit for the MM and 1M configurations for the corresponding
time periods are presented in 'ig* !*!-*

)*))( )*))+ )*))> )*))" )*) )*)( )*)+ )*)> )*)" )*)(
<())
<))
)
))
())
!))
+))
=))
>))
-))
Time JsL
D
u
r
r
e
n
t

J
A
L


1M
MM
(



'ig* !*!>* Three<phase short<circuit currents at ))) min
<
speed for the MM and 1M
structures* Ha%eforms include DD offset since the short circuit starts at peak
flux linkage*
The peak currents are >)-*= A for the MM machine and =!-*) for the 1M machine* The short
circuits current peaks occur at +*) ms and !*>- ms after the initialiGation of the short circuit* The
flux density graphs of the short circuit for the MM and 1M configurations for the corresponding
time periods are presented in 'ig* !*!-*

)*))( )*))+ )*))> )*))" )*) )*)( )*)+ )*)> )*)" )*)(
<())
<))
)
))
())
!))
+))
=))
>))
-))
Time JsL
D
u
r
r
e
n
t

J
A
L


1M
MM
!



'ig* !*!-* Three<phase short<circuit flux density graph of the magnet surfaces of the MM
and 1M configurations according to the 'lux (D simulation* The flux densities
are gi%en in normal direction on the surface of the magnet closest to the air gap*
Hith the selected ;e+,!a grade for the MM configuration& at least partial demagnetiGation is
ine%itable if the magnet temperature exceeds =) MD as shown in the data sheet in 'ig !*(* The
;e-=!a grade of the 1M configuration is more sensiti%e to the temperature because of the higher
B
r
as shown in 'ig !*!* Thus& the 1M configuration can only handle three<phase short circuits
below magnet temperatures of )) MD* The leading edge of the 1M magnet suffers most*
2nterestingly& the highest stress occurs at the lagging edge of the MM structure* According to the
short<circuit analysis& it would be beneficial to replace the magnet grade of the 1M configuration
to a more temperature tolerant grade* ;e%ertheless& in both cases& a long<term short circuit will
e%entually demagnetiGe the magnets as the magnets heat up*

,.4 )hermal network
An analytical thermal model was de%eloped to study the thermal properties of the machine in
steady<state operation* The model was extended to allow dynamic obser%ation o%er time*
Temperature modelling is essential in .M structures owing to the temperature sensiti%ity of the
.M material in general* A dynamic model is needed to %alidate the machine performance e%en for
) = ) = () (= !) != +)
)
)*
)*(
)*!
)*+
)*=
)*>
)*-
)*"
)*,

Magnet width JmmL

'
l
u
x

d
e
n
s
i
t
y

J
5
s
E
m
(
L


1M
MM
!



'ig* !*!-* Three<phase short<circuit flux density graph of the magnet surfaces of the MM
and 1M configurations according to the 'lux (D simulation* The flux densities
are gi%en in normal direction on the surface of the magnet closest to the air gap*
Hith the selected ;e+,!a grade for the MM configuration& at least partial demagnetiGation is
ine%itable if the magnet temperature exceeds =) MD as shown in the data sheet in 'ig !*(* The
;e-=!a grade of the 1M configuration is more sensiti%e to the temperature because of the higher
B
r
as shown in 'ig !*!* Thus& the 1M configuration can only handle three<phase short circuits
below magnet temperatures of )) MD* The leading edge of the 1M magnet suffers most*
2nterestingly& the highest stress occurs at the lagging edge of the MM structure* According to the
short<circuit analysis& it would be beneficial to replace the magnet grade of the 1M configuration
to a more temperature tolerant grade* ;e%ertheless& in both cases& a long<term short circuit will
e%entually demagnetiGe the magnets as the magnets heat up*

,.4 )hermal network
An analytical thermal model was de%eloped to study the thermal properties of the machine in
steady<state operation* The model was extended to allow dynamic obser%ation o%er time*
Temperature modelling is essential in .M structures owing to the temperature sensiti%ity of the
.M material in general* A dynamic model is needed to %alidate the machine performance e%en for
) = ) = () (= !) != +)
)
)*
)*(
)*!
)*+
)*=
)*>
)*-
)*"
)*,

Magnet width JmmL

'
l
u
x

d
e
n
s
i
t
y

J
5
s
E
m
(
L


1M
MM
+


highly arbitrary load cycles& in which constant operating %alues are repeatedly exceeded* This
applies especially to mobile solutions*
The de%eloped analytical model includes !" nodal points* The distribution of nodal points across
the geometry is presented in Appendix 22*
The nodal points are distributed as followsB
< = in the stator core&
< = in the windings&
< in the annulus&
< > in the magnet pole&
< ! in the rotor core&
< ! in the axle&
< + in the bearings&
< > in the end shields&
< ! in the frame and
< ( in the end space*

The model is basically in the (D form in addition to the axial heat flow paths in the frame& shaft
and windings* The main heat flow path is in the radial direction through the laminate stacks
because the insulation layer of the laminate sheets pre%ents the heat flow in the axial direction* The
stator frame shell area is di%ided into two parts in the radial direction* The axial conducti%ity in the
laminate stacks is neglected because the magnitude of axial heat conduction is negligible
compared with the heat conduction in the radial direction*
2n order to match the thermal resistances of the teeth and rotor pole area with the circular
resistance elements& the radial thermal resistances of the teeth area are di%ided by the slot number
because of the parallel heat flow paths in the tooth and slot region* The same applies to the slot
axial resistances owing to the parallel heat flow paths in the axial direction* The radial thermal
resistances of the rotor pole area are di%ided by the pole number for the same reason*

,.4.! *rame region and end#winding space
9alf of the frame is treated as a fluid cooling surface area and half as a natural<con%ection<affected
surface in parallel with the pre%ious according to (Richter& ,>() as
o
naturaI
= 6.S +u.uS(I
s
-I

)
!*>


+


highly arbitrary load cycles& in which constant operating %alues are repeatedly exceeded* This
applies especially to mobile solutions*
The de%eloped analytical model includes !" nodal points* The distribution of nodal points across
the geometry is presented in Appendix 22*
The nodal points are distributed as followsB
< = in the stator core&
< = in the windings&
< in the annulus&
< > in the magnet pole&
< ! in the rotor core&
< ! in the axle&
< + in the bearings&
< > in the end shields&
< ! in the frame and
< ( in the end space*

The model is basically in the (D form in addition to the axial heat flow paths in the frame& shaft
and windings* The main heat flow path is in the radial direction through the laminate stacks
because the insulation layer of the laminate sheets pre%ents the heat flow in the axial direction* The
stator frame shell area is di%ided into two parts in the radial direction* The axial conducti%ity in the
laminate stacks is neglected because the magnitude of axial heat conduction is negligible
compared with the heat conduction in the radial direction*
2n order to match the thermal resistances of the teeth and rotor pole area with the circular
resistance elements& the radial thermal resistances of the teeth area are di%ided by the slot number
because of the parallel heat flow paths in the tooth and slot region* The same applies to the slot
axial resistances owing to the parallel heat flow paths in the axial direction* The radial thermal
resistances of the rotor pole area are di%ided by the pole number for the same reason*

,.4.! *rame region and end#winding space
9alf of the frame is treated as a fluid cooling surface area and half as a natural<con%ection<affected
surface in parallel with the pre%ious according to (Richter& ,>() as
o
naturaI
= 6.S +u.uS(I
s
-I

)
!*>


=


where
< T
s
is the ob3ect surface temperature and
< T
f
is the temperature of the surrounding medium*

The eAuation of natural con%ection is also applied to both end<shield surfaces* The thermal
resistances of the frame& stator and rotor yoke and rotor shaft in the radial direction are modelled
by 1A* (*!+* The placement of temperature nodes in the frame region is presented in 'ig* !*!"*

'ig* !*!"* Temperature node placement in the frame region* The duplicate numbering of
the nodal points refers to the corresponding numbering in each end of the
machine* The whole thermal network is presented in A..1;D2I 22*
1nd shields are also modelled as cylindrical plates with one node at each end of the machine* The
axial thermal resistance is also taken into account* The bearings are modelled in the radial
direction only* 7ecause of the complicated structure of the bearing& the radial thermal conducti%ity
is taken into account as an eAui%alent air gap of )*! mm as suggested by (Staton et al*& ())=)* The
nodal point arrangement for bearings is presented in 'ig !*!,* The contact resistance to the end
shield and shaft are also modelled as air on the inner and outer race surfaces*
=


where
< T
s
is the ob3ect surface temperature and
< T
f
is the temperature of the surrounding medium*

The eAuation of natural con%ection is also applied to both end<shield surfaces* The thermal
resistances of the frame& stator and rotor yoke and rotor shaft in the radial direction are modelled
by 1A* (*!+* The placement of temperature nodes in the frame region is presented in 'ig* !*!"*

'ig* !*!"* Temperature node placement in the frame region* The duplicate numbering of
the nodal points refers to the corresponding numbering in each end of the
machine* The whole thermal network is presented in A..1;D2I 22*
1nd shields are also modelled as cylindrical plates with one node at each end of the machine* The
axial thermal resistance is also taken into account* The bearings are modelled in the radial
direction only* 7ecause of the complicated structure of the bearing& the radial thermal conducti%ity
is taken into account as an eAui%alent air gap of )*! mm as suggested by (Staton et al*& ())=)* The
nodal point arrangement for bearings is presented in 'ig !*!,* The contact resistance to the end
shield and shaft are also modelled as air on the inner and outer race surfaces*
>



'ig* !*!,* Thermal resistance displacement of a bearing*

,.4. )eeth and slots
The tooth area is di%ided into four parts& rectangular in the tooth tip and body and trapeGoidal in
the tooth bottom and neck as presented in 'ig* !*+)* The calculation of these thermal resistances
applies 1As* (*!= and (*!>*

'ig* !*+)* Thermal resistances of the slot and tooth areas* The whole thermal network is
presented in A..1;D2I 22*
>



'ig* !*!,* Thermal resistance displacement of a bearing*

,.4. )eeth and slots
The tooth area is di%ided into four parts& rectangular in the tooth tip and body and trapeGoidal in
the tooth bottom and neck as presented in 'ig* !*+)* The calculation of these thermal resistances
applies 1As* (*!= and (*!>*

'ig* !*+)* Thermal resistances of the slot and tooth areas* The whole thermal network is
presented in A..1;D2I 22*
-



This information can be used to determine the eAui%alent numbers of parallel strings in slots in the
corresponding directions* 1Aui%alent slot dimensions are calculated by applying the dimensions of
'ig* !*= to 1As* !*( and !*!* The number of conductors in the slot height is thus
n
h
=

cq
n
strIngs
w
cq

!*-
and
n
w
=
n
strIngs
n
h

!*"
for the number of conductors in the length of slot width* The eAui%alent resistances in the slot
height and slot width (in other words& radial and tangential directions& respecti%ely) can be
calculated as

sIot,radIaI
=
n
h
k
u
n
w
l
Fc
+

cq
-n
h
J
strIng
z
Impr
n
w
l
Fc
+
2
sIot,Ins
z
sIot,Ins
w
cq
l
Fc

!*,
and

sIot,tangcntIaI
=
n
w
k
u
n
h
l
Fc
+
w
cq
-n
w
J
strIng
z
Impr
n
h
l
Fc
+
2
sIot,Ins
z
sIot,Ins

cq
l
Fc

!*)

respecti%ely* 7ecause the thermal resistances represent the thermal resistance in the tangential
direction for the whole width of the slot& the tangential thermal resistance has to be di%ided by two
as e%ery slot has one tooth in parallel with each other* Thus& the heat passage length is half of the
slot width in a%erage* 2n addition to the slot radial and tangential direction& there has to be a
thermal contact resistance added to the heat path because the impregnation lea%es air pockets in
the coil sides (Cylander& ,,=) and (Saari& ,,=)* These can be modelled as small air gaps* The
lengths of typical contact region air gaps are presented in Table !*!*
Table* !*!* Suggested air gaps in the thermal contact area (Saari& ,,=)&
(Cylander& ,,=)*
Material 1Aui%alent air gap length JemL
Hinding to core ))@!))
Dast iron frame to stator& or more than )) kH =)@")
Aluminum frame to stator& or less than ))kH !)@+)

-



This information can be used to determine the eAui%alent numbers of parallel strings in slots in the
corresponding directions* 1Aui%alent slot dimensions are calculated by applying the dimensions of
'ig* !*= to 1As* !*( and !*!* The number of conductors in the slot height is thus
n
h
=

cq
n
strIngs
w
cq

!*-
and
n
w
=
n
strIngs
n
h

!*"
for the number of conductors in the length of slot width* The eAui%alent resistances in the slot
height and slot width (in other words& radial and tangential directions& respecti%ely) can be
calculated as

sIot,radIaI
=
n
h
k
u
n
w
l
Fc
+

cq
-n
h
J
strIng
z
Impr
n
w
l
Fc
+
2
sIot,Ins
z
sIot,Ins
w
cq
l
Fc

!*,
and

sIot,tangcntIaI
=
n
w
k
u
n
h
l
Fc
+
w
cq
-n
w
J
strIng
z
Impr
n
h
l
Fc
+
2
sIot,Ins
z
sIot,Ins

cq
l
Fc

!*)

respecti%ely* 7ecause the thermal resistances represent the thermal resistance in the tangential
direction for the whole width of the slot& the tangential thermal resistance has to be di%ided by two
as e%ery slot has one tooth in parallel with each other* Thus& the heat passage length is half of the
slot width in a%erage* 2n addition to the slot radial and tangential direction& there has to be a
thermal contact resistance added to the heat path because the impregnation lea%es air pockets in
the coil sides (Cylander& ,,=) and (Saari& ,,=)* These can be modelled as small air gaps* The
lengths of typical contact region air gaps are presented in Table !*!*
Table* !*!* Suggested air gaps in the thermal contact area (Saari& ,,=)&
(Cylander& ,,=)*
Material 1Aui%alent air gap length JemL
Hinding to core ))@!))
Dast iron frame to stator& or more than )) kH =)@")
Aluminum frame to stator& or less than ))kH !)@+)

"


Hindings are di%ided into fi%e different nodal points* Three nodes are placed in the stator coreU
one in the centre and one in each end of the stator stack* The last two nodes are for the end
windings as presented in 'igs* !*!" and !*+)*
Axial heat flow is e%aluated for the copper only because the heat<carrying capability of the
impregnation material is considerably lower compared with the copper* The copper area is
calculated with the actual copper turn number according to
S
u
= u.2S

n
paraIIcI
J
strIng
2

!*
where
< :
N
is the number of coil turn sides in one slot&
< n
parallel
is the number of parallel strings in the coil turn and
< d
string
is the diameter of one string*

Thus& the axial thermal resistance of a single coil is

u,axIaI
=
l
Fc
k
u
S
u

!*(
The end<winding axial resistance is calculated similarly& but the end<winding length is calculated
from the eAuation
l
cnd
= 1.2
p
+u.uS ,
!*!
which is an experimental formula for the end<winding length (5ogt& ,"!)* The result has to be
multiplied by )*= to obtain the thermal resistance in the middle of the end winding loop*

,.4., Air gap region
The e%aluation of the air region thermal resistances is based on the Taylor number Ta& which
describes the heat exchange (con%ection and conduction) between two concentric cylinders& in
which the inner cylinder rotates* 2f no axial flow is present& the annulus flow is usually Auite
laminar* The transition to turbulent flow begins at higher rotating speeds* 1Auation for the
calculation of the Taylor number according to (7ecker et al*& ,>() is
Io =
0
2
r
8
o
3
v
2
F
g

!*+

"


Hindings are di%ided into fi%e different nodal points* Three nodes are placed in the stator coreU
one in the centre and one in each end of the stator stack* The last two nodes are for the end
windings as presented in 'igs* !*!" and !*+)*
Axial heat flow is e%aluated for the copper only because the heat<carrying capability of the
impregnation material is considerably lower compared with the copper* The copper area is
calculated with the actual copper turn number according to
S
u
= u.2S

n
paraIIcI
J
strIng
2

!*
where
< :
N
is the number of coil turn sides in one slot&
< n
parallel
is the number of parallel strings in the coil turn and
< d
string
is the diameter of one string*

Thus& the axial thermal resistance of a single coil is

u,axIaI
=
l
Fc
k
u
S
u

!*(
The end<winding axial resistance is calculated similarly& but the end<winding length is calculated
from the eAuation
l
cnd
= 1.2
p
+u.uS ,
!*!
which is an experimental formula for the end<winding length (5ogt& ,"!)* The result has to be
multiplied by )*= to obtain the thermal resistance in the middle of the end winding loop*

,.4., Air gap region
The e%aluation of the air region thermal resistances is based on the Taylor number Ta& which
describes the heat exchange (con%ection and conduction) between two concentric cylinders& in
which the inner cylinder rotates* 2f no axial flow is present& the annulus flow is usually Auite
laminar* The transition to turbulent flow begins at higher rotating speeds* 1Auation for the
calculation of the Taylor number according to (7ecker et al*& ,>() is
Io =
0
2
r
8
o
3
v
2
F
g

!*+

,



where
< is the mechanical angular speed&
< r
]
is the a%erage air gap radius&
< o is the air gap length&
< v is the kinematic %iscosity of the air gap fluid and
< F
g
is a correcting factor depending on the cylinder dimensions*
F
g
can be neglected in practical cases where the air gap length is considerably smaller compared
with the rotating radius (oED
r
W)* )* Taylor numbers beyond -+) are considered as a turbulent
region (7ecker et al*& ,>()* 2n this region& the ;usselt number is described as
Nu = u.4u9Io
0.241
-1S7Io
-0.75

!*=
At lower Taylor numbers& the %alue of the ;usselt number is considered to be Nu S( (7ecker et
al*& ,>()* The con%ection coefficient from the stator to the rotor o%er the characteristic length of
air gap is
o
8
=
k
I
Nu
2o
.
!*>
Radiation is not taken into account because the effect is negligible as a result of the small
temperature difference between the rotor and stator surfaces*

,.4.4 &otor region
A rotor model allows modelling of the embedded and module magnets* The Vpole shoeF area is
neglected in the magnet module design* The magnets are modelled with rectangular elements
ha%ing nodal points at the top and bottom of the magnets* The arrangement is presented in 'ig*
!*+*

,



where
< is the mechanical angular speed&
< r
]
is the a%erage air gap radius&
< o is the air gap length&
< v is the kinematic %iscosity of the air gap fluid and
< F
g
is a correcting factor depending on the cylinder dimensions*
F
g
can be neglected in practical cases where the air gap length is considerably smaller compared
with the rotating radius (oED
r
W)* )* Taylor numbers beyond -+) are considered as a turbulent
region (7ecker et al*& ,>()* 2n this region& the ;usselt number is described as
Nu = u.4u9Io
0.241
-1S7Io
-0.75

!*=
At lower Taylor numbers& the %alue of the ;usselt number is considered to be Nu S( (7ecker et
al*& ,>()* The con%ection coefficient from the stator to the rotor o%er the characteristic length of
air gap is
o
8
=
k
I
Nu
2o
.
!*>
Radiation is not taken into account because the effect is negligible as a result of the small
temperature difference between the rotor and stator surfaces*

,.4.4 &otor region
A rotor model allows modelling of the embedded and module magnets* The Vpole shoeF area is
neglected in the magnet module design* The magnets are modelled with rectangular elements
ha%ing nodal points at the top and bottom of the magnets* The arrangement is presented in 'ig*
!*+*

()



'ig* !*+* Thermal resistances of a pole region* 2n the 1M structure there is a thermal
resistance of the pole shoe* 2n the MM structure instead& the thermal resistance is
considered Gero* The thermal contact areas between the node pairs (@!& +@=
and >@- are taken into account as adhesi%e 3oints in the MM structure and as
air gaps in the 1M structure* The iron path acts as a short<circuit path of heat
flux for each pole*
There are also contact resistance areas for magnet surfaces* The thermal contact area in the magnet
module machine is modelled as an insulation layer of ()) em in thickness* This area is replaced by
air in the embedded magnet structure* The magnet pole also has a side path for heat flux in the
Auadrature<axis direction& modelled with a rectangular element in both structures* The tangential
thermal conducti%ity of the rotor is neglected in the model*

,.1 &esults of the thermal modelling
The de%eloped lumped thermal model is used to predict the thermal beha%iour of the machine in
different load conditions* The model analysis is di%ided into time<dependent and steady<state
analyses* The steady<state analysis pro%ides fast e%aluation of temperatures in static load
conditions but does not gi%e information about the transient characteristics* Hith a transient
model& it is possible to e%aluate the machine endurance for specific load cycles& in which the rated
%alues are repeatedly exceeded* The losses are e%aluated by the 'lux (D program and applied to
specific nodes of the thermal model*

,.1.! "teady#state analysis
Cnowledge on material parameters is important in thermal modelling* 2n general& the list of
materials is Auite short* The thermal conducti%ities of metals %ary between () and !) HECm
depending on their silicon content (Miller& ,,+)* #ower thermal conducti%ity %alues are related to
a higher silicon content* 2n this analysis& the thermal conducti%ity of the sheet metal is set to !)
()



'ig* !*+* Thermal resistances of a pole region* 2n the 1M structure there is a thermal
resistance of the pole shoe* 2n the MM structure instead& the thermal resistance is
considered Gero* The thermal contact areas between the node pairs (@!& +@=
and >@- are taken into account as adhesi%e 3oints in the MM structure and as
air gaps in the 1M structure* The iron path acts as a short<circuit path of heat
flux for each pole*
There are also contact resistance areas for magnet surfaces* The thermal contact area in the magnet
module machine is modelled as an insulation layer of ()) em in thickness* This area is replaced by
air in the embedded magnet structure* The magnet pole also has a side path for heat flux in the
Auadrature<axis direction& modelled with a rectangular element in both structures* The tangential
thermal conducti%ity of the rotor is neglected in the model*

,.1 &esults of the thermal modelling
The de%eloped lumped thermal model is used to predict the thermal beha%iour of the machine in
different load conditions* The model analysis is di%ided into time<dependent and steady<state
analyses* The steady<state analysis pro%ides fast e%aluation of temperatures in static load
conditions but does not gi%e information about the transient characteristics* Hith a transient
model& it is possible to e%aluate the machine endurance for specific load cycles& in which the rated
%alues are repeatedly exceeded* The losses are e%aluated by the 'lux (D program and applied to
specific nodes of the thermal model*

,.1.! "teady#state analysis
Cnowledge on material parameters is important in thermal modelling* 2n general& the list of
materials is Auite short* The thermal conducti%ities of metals %ary between () and !) HECm
depending on their silicon content (Miller& ,,+)* #ower thermal conducti%ity %alues are related to
a higher silicon content* 2n this analysis& the thermal conducti%ity of the sheet metal is set to !)
(


HECm as there were no accurate data a%ailable of the specific grade used in this study* The
thermal conducti%ity of electric steel is highly dependent on the silicon content of the steel along
with the electrical conducti%ity* 1lectric steels of low specific loss %alues ha%e a higher silicon
content to reduce the steel resisti%ity in order to decrease the eddy current losses* A high silicon
content has the same effect on thermal conducti%ity*
The thermal characteristics of aluminium are difficult to model since physical properties %ary
according to the material grade* The thermal conducti%ity of cast aluminium is around +) HECm&
but certain alloys are close to the ()) HECm rating of pure aluminium (Miller& ,,+)& (2ncropera&
())-)* The alloy for the frame and rotor shaft is tool aluminium& which has a typical thermal
conducti%ity of ") HECm*
2nsulating sheets and adhesi%es usually ha%e thermal conducti%ity %alues around )*( to )*! HECm*
'or slot impregnation resins& thermal conducti%ities around )*( ha%e been reported for
polyethylene compounds* More expensi%e epoxy resins can reach HECm %alues and abo%e*
Material<specific data for thermal conductance calculations are presented in Tables !*+ and !*=*
Table* !*+* Material<specific data for thermal conductance calculations*
Material Thermal conducti%ity JHECmL
Sheet metal !)
Dopper !"=
Aluminium +)
Slot impregnation )*(
Slot insulation )*(

The thermal conducti%ities of solid materials are not updated according to the temperature because
the accurate temperature beha%iour is not known for the industrial grade materials* .re%ious
studies on the thermal performance of electrical machines ha%e shown good agreement of the
models with the selected material parameters (Rilla& ())>)*
The air properties (thermal conducti%ity& density& kinematic and dynamic %iscosity) are updated
according to the temperature because air properties are sensiti%e to the pre%ailing temperature* The
effect is significant in the thermal conducti%ity 3oints and especially in the air gap region* The
thermal conducti%ity of air as a function of temperature is presented in 'ig* !*+(*
(


HECm as there were no accurate data a%ailable of the specific grade used in this study* The
thermal conducti%ity of electric steel is highly dependent on the silicon content of the steel along
with the electrical conducti%ity* 1lectric steels of low specific loss %alues ha%e a higher silicon
content to reduce the steel resisti%ity in order to decrease the eddy current losses* A high silicon
content has the same effect on thermal conducti%ity*
The thermal characteristics of aluminium are difficult to model since physical properties %ary
according to the material grade* The thermal conducti%ity of cast aluminium is around +) HECm&
but certain alloys are close to the ()) HECm rating of pure aluminium (Miller& ,,+)& (2ncropera&
())-)* The alloy for the frame and rotor shaft is tool aluminium& which has a typical thermal
conducti%ity of ") HECm*
2nsulating sheets and adhesi%es usually ha%e thermal conducti%ity %alues around )*( to )*! HECm*
'or slot impregnation resins& thermal conducti%ities around )*( ha%e been reported for
polyethylene compounds* More expensi%e epoxy resins can reach HECm %alues and abo%e*
Material<specific data for thermal conductance calculations are presented in Tables !*+ and !*=*
Table* !*+* Material<specific data for thermal conductance calculations*
Material Thermal conducti%ity JHECmL
Sheet metal !)
Dopper !"=
Aluminium +)
Slot impregnation )*(
Slot insulation )*(

The thermal conducti%ities of solid materials are not updated according to the temperature because
the accurate temperature beha%iour is not known for the industrial grade materials* .re%ious
studies on the thermal performance of electrical machines ha%e shown good agreement of the
models with the selected material parameters (Rilla& ())>)*
The air properties (thermal conducti%ity& density& kinematic and dynamic %iscosity) are updated
according to the temperature because air properties are sensiti%e to the pre%ailing temperature* The
effect is significant in the thermal conducti%ity 3oints and especially in the air gap region* The
thermal conducti%ity of air as a function of temperature is presented in 'ig* !*+(*
((



'ig* !*+(* Thermal conducti%ity of air as a function of temperature*
Table* !*=* .hysical properties of air at the )) MD temperature*
Nuantity Thermal conducti%ity $nit
Thermal conducti%ity )*)!( Hm
<
C
<

Density )*,= kg m
<!

Cinematic %iscosity (!*)>R)
<>
m
(
s
<

Dynamic %iscosity (*,R)
<>
kg m
<
s
<


The model is tested for a rated speed of ))) min
<
and a (= kH load* The loss distribution is
based on the 'lux (D analysis* Dopper losses are estimated analytically* The loss data for the
thermal model are gi%en in Table !*>*




) )) ()) !)) +)) =)) >)) -)) "))
)*)(
)*)(=
)*)!
)*)!=
)*)+
)*)+=
)*)=
)*)==
)*)>
)*)>=
)*)-
T
h
e
r
m
a
l

c
o
n
d
u
c
t
i
%
i
t
y

J
H
E
m
C
L
Temperature J
o
DL
((



'ig* !*+(* Thermal conducti%ity of air as a function of temperature*
Table* !*=* .hysical properties of air at the )) MD temperature*
Nuantity Thermal conducti%ity $nit
Thermal conducti%ity )*)!( Hm
<
C
<

Density )*,= kg m
<!

Cinematic %iscosity (!*)>R)
<>
m
(
s
<

Dynamic %iscosity (*,R)
<>
kg m
<
s
<


The model is tested for a rated speed of ))) min
<
and a (= kH load* The loss distribution is
based on the 'lux (D analysis* Dopper losses are estimated analytically* The loss data for the
thermal model are gi%en in Table !*>*




) )) ()) !)) +)) =)) >)) -)) "))
)*)(
)*)(=
)*)!
)*)!=
)*)+
)*)+=
)*)=
)*)==
)*)>
)*)>=
)*)-
T
h
e
r
m
a
l

c
o
n
d
u
c
t
i
%
i
t
y

J
H
E
m
C
L
Temperature J
o
DL
(!


Table* !*>* 1stimated loss data in the rated point for the MM and 1M structures at the
()MD temperature*
#oss data MM JHL 1MJHL
Dopper loss (! (!
Stator core loss =+ =!
Tooth losses ,) )+
Top magnet loss != !>
#ower magnet loss >
Middle piece ( +
Rotor yoke loss - (+
7earing losses (SC') (!- (!-
Additional losses (> (>
Total -"> "+

According to the 'lux (D and analytical calculations& the loss beha%iour seems similar in both
machines* The main difference seems to lie in the magnet /oule losses and the rotor core losses*
Surprisingly& the top<mounted magnets seem to ha%e almost an eAual amount of losses* 2n the
actual machine& the magnet close to the air gap in the magnet module is slitted to decrease the
losses* The absence of core material in the d<axis direction causes additional stress to the lower
magnet of the MM structure compared with the 1M structure* According to the 21D standard& the
additional losses for the output power of (= kH are (=*- H* The bearing losses are estimated
according to the SC' calculators for a gi%en bearing type* 2t was noted that lubrication %iscosity
has a slight effect on the bearing friction losses in the sealed bearing structure* The embedded
magnet machine has through<%entilation channels in the end<shield region* This is taken into
account by applying con%ecti%e cooling to the end<winding space node with the air fluid data
%alues of the a%erage temperature between the ambient and machine inner air space* The ambient
air temperature is considered to be (= MD and cooling fluid temperature != MD*
The results of the steady<state model according to Table !*> loss data are presented in Table !*-*
Table* !*-* Temperature rise of the steady<state thermal model at rated %alues*
Temperature node MMJMDL 1M JMDL
Slot middle "=* =!*>
Doil end ,)*> =>*,
Stator yoke +=*) (>*(
Tooth middle ==*= !>*"
Top magnet >)*( =,*)
Rotor yoke >* =,*!
2nner air =+*= (,*,

(!


Table* !*>* 1stimated loss data in the rated point for the MM and 1M structures at the
()MD temperature*
#oss data MM JHL 1MJHL
Dopper loss (! (!
Stator core loss =+ =!
Tooth losses ,) )+
Top magnet loss != !>
#ower magnet loss >
Middle piece ( +
Rotor yoke loss - (+
7earing losses (SC') (!- (!-
Additional losses (> (>
Total -"> "+

According to the 'lux (D and analytical calculations& the loss beha%iour seems similar in both
machines* The main difference seems to lie in the magnet /oule losses and the rotor core losses*
Surprisingly& the top<mounted magnets seem to ha%e almost an eAual amount of losses* 2n the
actual machine& the magnet close to the air gap in the magnet module is slitted to decrease the
losses* The absence of core material in the d<axis direction causes additional stress to the lower
magnet of the MM structure compared with the 1M structure* According to the 21D standard& the
additional losses for the output power of (= kH are (=*- H* The bearing losses are estimated
according to the SC' calculators for a gi%en bearing type* 2t was noted that lubrication %iscosity
has a slight effect on the bearing friction losses in the sealed bearing structure* The embedded
magnet machine has through<%entilation channels in the end<shield region* This is taken into
account by applying con%ecti%e cooling to the end<winding space node with the air fluid data
%alues of the a%erage temperature between the ambient and machine inner air space* The ambient
air temperature is considered to be (= MD and cooling fluid temperature != MD*
The results of the steady<state model according to Table !*> loss data are presented in Table !*-*
Table* !*-* Temperature rise of the steady<state thermal model at rated %alues*
Temperature node MMJMDL 1M JMDL
Slot middle "=* =!*>
Doil end ,)*> =>*,
Stator yoke +=*) (>*(
Tooth middle ==*= !>*"
Top magnet >)*( =,*)
Rotor yoke >* =,*!
2nner air =+*= (,*,

(+


The estimated machine temperatures show clearly the benefit of the through<%entilated structure*
According to the e%aluation& the machine operates within safe temperature limits at the rated speed
and torAue* The model was used to determine the maximum load %alue for the machines within the
temperature limits* The windings of the machine are designed for a == MD limit* At a +) A
current& the 1M structure reaches the winding temperature limit* 2n this point& the magnets are
approximately in a (= MD temperature* The MM structure& on the other hand& meets the winding
temperature limit at a ( A current* 9owe%er& the magnets are at a much lower temperature of
= MD in this point* The analytical prediction of thermal characteristics shows that the machine
dimensioning should be at the safe side*

,.1. )ransient analysis
2n addition to the thermal conductance matrix of the static model& a thermal capacitance matrix
was generated* The capacitance matrix is built as a diagonal matrix according to Section (*!*=*
:nly solid materials are taken into account because the thermal capacitances in the air region are
considerably smaller* The material data needed to construct the matrix are presented in Table !*"*
Table* !*"* Material parameters for thermal capacitance calculations*
Material
Specific heat capacity at
))MD J/kg
<
C
<
L
Density Jkg m
<!
L
Sheet metal +=) ->))
Dopper !"= ",>)
Aluminium ,)) (-))
Slot impregnation ()) ,))
2nsulation ()) ,))

2teration is carried out by a (
nd
<order Adams<Moulton method& as it is suggested to be fast and
stable tool for the analysis of thermal gradient (Cylander& ,,=)* The results of the dynamic model
according to the MM and 1M structure are presented in 'ig* !*+!* The loss data are taken from
Table !*>*
(+


The estimated machine temperatures show clearly the benefit of the through<%entilated structure*
According to the e%aluation& the machine operates within safe temperature limits at the rated speed
and torAue* The model was used to determine the maximum load %alue for the machines within the
temperature limits* The windings of the machine are designed for a == MD limit* At a +) A
current& the 1M structure reaches the winding temperature limit* 2n this point& the magnets are
approximately in a (= MD temperature* The MM structure& on the other hand& meets the winding
temperature limit at a ( A current* 9owe%er& the magnets are at a much lower temperature of
= MD in this point* The analytical prediction of thermal characteristics shows that the machine
dimensioning should be at the safe side*

,.1. )ransient analysis
2n addition to the thermal conductance matrix of the static model& a thermal capacitance matrix
was generated* The capacitance matrix is built as a diagonal matrix according to Section (*!*=*
:nly solid materials are taken into account because the thermal capacitances in the air region are
considerably smaller* The material data needed to construct the matrix are presented in Table !*"*
Table* !*"* Material parameters for thermal capacitance calculations*
Material
Specific heat capacity at
))MD J/kg
<
C
<
L
Density Jkg m
<!
L
Sheet metal +=) ->))
Dopper !"= ",>)
Aluminium ,)) (-))
Slot impregnation ()) ,))
2nsulation ()) ,))

2teration is carried out by a (
nd
<order Adams<Moulton method& as it is suggested to be fast and
stable tool for the analysis of thermal gradient (Cylander& ,,=)* The results of the dynamic model
according to the MM and 1M structure are presented in 'ig* !*+!* The loss data are taken from
Table !*>*
(=



'ig* !*+!* Temperature rise in C of different machine parts according to the transient
analysis for the MM and 1M structures at rated %alues* The loss %alues are based
on Table !*-* The numbers in brackets are the corresponding node numbers of
the applied thermal model*

The cooling arrangement of the 1M structure shows a clear benefit related to the temperature
change* The time constant of the 1M structure for the temperature stabiliGation is also
considerably lower compared with the totally enclosed structure of the MM machine& as expected*
The corresponding time constants for the coil end temperature are O== minutes for the MM
machine and O+> minutes for the 1M machine*
,.1., "ensitivity of the developed model
The sensiti%ity of the analytical thermal model is inspected by %arying the thermal conducti%ities
of materials and referring them to the selected %alues* The selected thermal conducti%ities
pro%ided in Table !*= ha%e shown good agreement in pre%ious machine studies (Rilla& ())>)* The
material parameters should ne%ertheless be selected carefully& and the sensiti%ity of the model
should always be checked by %arying the material parameters* Additional care has to be taken in
the selection of the insulating material parameters* This is clearly seen in Table !*,& where the
effects of thermal conducti%ities on the slot temperature node are gi%en*
Table* !*,* Table for winding temperature estimation depending on the selected thermal
conducti%ity of the material* The resulting temperature difference is presented for each
material separately*
(=



'ig* !*+!* Temperature rise in C of different machine parts according to the transient
analysis for the MM and 1M structures at rated %alues* The loss %alues are based
on Table !*-* The numbers in brackets are the corresponding node numbers of
the applied thermal model*

The cooling arrangement of the 1M structure shows a clear benefit related to the temperature
change* The time constant of the 1M structure for the temperature stabiliGation is also
considerably lower compared with the totally enclosed structure of the MM machine& as expected*
The corresponding time constants for the coil end temperature are O== minutes for the MM
machine and O+> minutes for the 1M machine*
,.1., "ensitivity of the developed model
The sensiti%ity of the analytical thermal model is inspected by %arying the thermal conducti%ities
of materials and referring them to the selected %alues* The selected thermal conducti%ities
pro%ided in Table !*= ha%e shown good agreement in pre%ious machine studies (Rilla& ())>)* The
material parameters should ne%ertheless be selected carefully& and the sensiti%ity of the model
should always be checked by %arying the material parameters* Additional care has to be taken in
the selection of the insulating material parameters* This is clearly seen in Table !*,& where the
effects of thermal conducti%ities on the slot temperature node are gi%en*
Table* !*,* Table for winding temperature estimation depending on the selected thermal
conducti%ity of the material* The resulting temperature difference is presented for each
material separately*
(>


Material
)hermal
conductivity
range
79:m58
+M#slot
temperature
7B28
C)
7B28
"M#slot
temperature
7B28
C)
7B28
Dopper !>)@+)) "+*-@"+*> )* ,* <
1lectric sheet ()@+) "-*=@"!*) +*= (=*)@=*, +*
Slothinsulation )*(@)*! "+*> < ,* <
Slothimpregnation )*(@)*> "+*>@->*, -*- ,*@))*" "*!
'rame +)@()) "+*>@"+*= )* ,*@"*! )*")
All high < All low < "-*=@-+*" (*- (=*)@,>*, ("*

According to the results of Table !*,& the selection of the thermal conducti%ity of the slot
impregnation has the highest influence on the model results* The low conducti%ity %alues refer to
typical polyester resins* 9igh )*> HEmC %alues are achie%able with more expensi%e epoxy resins*
The thermal conducti%ity of the slot insulation sheet seems to ha%e a negligible effect on the
temperature distribution* The 1M structure has lower temperature changes as a result of the
through<%entilated structure* This can be seen clearly in the changes in the slot impregnation
conducti%ity* The second highest influence on the temperatures is the thermal conducti%ity of the
sheet metal*


,.3 2onclusions on the machine design
According to the thermal analyses made on the structures& the proposed design methods resulted in
two functional machine structures*
The performance analysis by two<axis model inductances is only indicati%e for these machine
types* A comparison between the finite element method and the analytical approach shows the
weakness of the traditional analytical model& which does not take into account the armature
reaction accurately because the direct< and Auadrature<axis inductances obtained from the 'lux (D
calculation for the d< and A<axes indi%idually do not describe the magnetic state of the machine
correctly*
The enhanced method with the cross<saturation gi%es better results* Thus& finite<element<based
methods are needed for the more accurate machine performance estimation of these machine
types* Dare must be taken because an accurate control of parameters in the %oltage< and current<
fed models is not possible* Therefore& an iterati%e approach is mandatory in the design process*
The %oltage le%els and load angles ha%e to be %aried to find an optimum current angle*
The magnet module design seems to suffer from mechanical problems in the magnet fixing* Thus&
the embedded machine structure is more fa%ourable because of its better mechanical stability* The
through<%entilated structure also shows benefits in cooling& and this should be considered in
mobile applications as the traditional 2D1 solutions also ha%e to be protected against en%ironment*
(>


Material
)hermal
conductivity
range
79:m58
+M#slot
temperature
7B28
C)
7B28
"M#slot
temperature
7B28
C)
7B28
Dopper !>)@+)) "+*-@"+*> )* ,* <
1lectric sheet ()@+) "-*=@"!*) +*= (=*)@=*, +*
Slothinsulation )*(@)*! "+*> < ,* <
Slothimpregnation )*(@)*> "+*>@->*, -*- ,*@))*" "*!
'rame +)@()) "+*>@"+*= )* ,*@"*! )*")
All high < All low < "-*=@-+*" (*- (=*)@,>*, ("*

According to the results of Table !*,& the selection of the thermal conducti%ity of the slot
impregnation has the highest influence on the model results* The low conducti%ity %alues refer to
typical polyester resins* 9igh )*> HEmC %alues are achie%able with more expensi%e epoxy resins*
The thermal conducti%ity of the slot insulation sheet seems to ha%e a negligible effect on the
temperature distribution* The 1M structure has lower temperature changes as a result of the
through<%entilated structure* This can be seen clearly in the changes in the slot impregnation
conducti%ity* The second highest influence on the temperatures is the thermal conducti%ity of the
sheet metal*


,.3 2onclusions on the machine design
According to the thermal analyses made on the structures& the proposed design methods resulted in
two functional machine structures*
The performance analysis by two<axis model inductances is only indicati%e for these machine
types* A comparison between the finite element method and the analytical approach shows the
weakness of the traditional analytical model& which does not take into account the armature
reaction accurately because the direct< and Auadrature<axis inductances obtained from the 'lux (D
calculation for the d< and A<axes indi%idually do not describe the magnetic state of the machine
correctly*
The enhanced method with the cross<saturation gi%es better results* Thus& finite<element<based
methods are needed for the more accurate machine performance estimation of these machine
types* Dare must be taken because an accurate control of parameters in the %oltage< and current<
fed models is not possible* Therefore& an iterati%e approach is mandatory in the design process*
The %oltage le%els and load angles ha%e to be %aried to find an optimum current angle*
The magnet module design seems to suffer from mechanical problems in the magnet fixing* Thus&
the embedded machine structure is more fa%ourable because of its better mechanical stability* The
through<%entilated structure also shows benefits in cooling& and this should be considered in
mobile applications as the traditional 2D1 solutions also ha%e to be protected against en%ironment*
(-


The through<%entilated structure pro%ides a higher steady<state load capability& and the lower
temperature rise time constants allow a better dynamic performance*

(-


The through<%entilated structure pro%ides a higher steady<state load capability& and the lower
temperature rise time constants allow a better dynamic performance*

("



("



(,


4 Analysis of the e/perimental studies
A set of tests were performed for both machine types* The power output capability and thermal
stress tests were performed to compare the performance of the machine structures* The test bench
and the cooling arrangement were the same for both of the machines*
To simulate the cooling arrangement of a mo%ing machine& the cooling of the machine was
arranged with an aluminium heat exchanger and a ( 5
DD
electric water pump with a measured
flow of )*( kgEs* The heat exchanger was exposed to forced cooling with an auxiliary axial fan*
The embedded magnet machine has additional cooling %ents on both end shields for enhanced
con%ection heat transfer* 'or comparison& the surface magnet module machine is a totally enclosed
design* The surface magnet module and the embedded magnet machines are referred to from here
onwards as an MM and 1M& respecti%ely*

4.! )est set#up
The test set<up consists of two test machines& a three<phase +)) 5 and !== kH A77 induction
machine& a Magtroll ())) ;m TorAue meter& a ?okogawa .T<+))) power analyser and a 'luke
thermometer unit for the .T<)) temperature elements* Details of the test set<up are presented in
A..1;D2I 2* The test set<up is shown in 'ig* +**

'ig* +** Test set<up with the MM machine in the front& load machine in the back and heat
exchanger on the left*

(,


4 Analysis of the e/perimental studies
A set of tests were performed for both machine types* The power output capability and thermal
stress tests were performed to compare the performance of the machine structures* The test bench
and the cooling arrangement were the same for both of the machines*
To simulate the cooling arrangement of a mo%ing machine& the cooling of the machine was
arranged with an aluminium heat exchanger and a ( 5
DD
electric water pump with a measured
flow of )*( kgEs* The heat exchanger was exposed to forced cooling with an auxiliary axial fan*
The embedded magnet machine has additional cooling %ents on both end shields for enhanced
con%ection heat transfer* 'or comparison& the surface magnet module machine is a totally enclosed
design* The surface magnet module and the embedded magnet machines are referred to from here
onwards as an MM and 1M& respecti%ely*

4.! )est set#up
The test set<up consists of two test machines& a three<phase +)) 5 and !== kH A77 induction
machine& a Magtroll ())) ;m TorAue meter& a ?okogawa .T<+))) power analyser and a 'luke
thermometer unit for the .T<)) temperature elements* Details of the test set<up are presented in
A..1;D2I 2* The test set<up is shown in 'ig* +**

'ig* +** Test set<up with the MM machine in the front& load machine in the back and heat
exchanger on the left*

!)


The results obtained are based on the freAuency con%erter unit and the de%eloped control method
by 5acon :y* The electrical %alues are obtained with the power analyser by applying a s
sampling time in the data capture*
4.!.! 4oad analysis
The measured root mean sAuare line<to<line back 1M' %alues for the MM and 1M machines& at
the ambient temperature of (= MD and ))) min
<
rated speed& are !=*+ 5 and =+*! 5&
respecti%ely* The measurement was carried out by rotating the measured machine with a load
machine and measuring the line<to<line %oltage with a power analyser* The estimated %alue for the
back 1M' according to the finite element calculations at the rated speed of ))) min
<
is ==*) 5
for the MM machine and =,*, 5 for the 1M machine*
The difference in the 1M' ratings is mostly due to the dimensioning error in the magnet module
design caused by the end effect in the rotors& which is explained by the fact that the permanent
magnet ends tend to short<circuit part of the main flux* Thus& the rotor length in the MM machine
is too short* According to (/ussila& ()))& the physical length of a permanent magnet rotor should
be the length of the stator stack plus two times the air gap length at each end of the machine* Thus&
the machine rotor length should be
l
rotor
= l
Fc
+4o 4.1
to match the effecti%e stator core length* DonseAuently& the rotor flux is smaller than the optimum
=)*) 5 because of the length of >= mm instead of the optimal >" mm* This error was corrected in
the 1M machine dimensioning* An additional flux decrease is caused by the slitting of the surface
magnet in the axial direction* The adhesi%e layers between the magnet pieces decrease the space
factor of the magnet pole*
The difference shows directly in the load characteristics* The MM machine torAue output is around
) P smaller with currents of eAual siGe when compared with the 1M machine* 7oth machines are
measured in motoring operation for !))& -)) and =)) min
<
rotating speeds* 2n addition& the ()))
min
<
rotating speed is measured only for the 1M machine because of the mechanical restrictions
of the MM structure* The output power and torAue of the machines are presented in Tables +* and
+*(* The current %alues refer to the nominal current I
n
S)) A*






!)


The results obtained are based on the freAuency con%erter unit and the de%eloped control method
by 5acon :y* The electrical %alues are obtained with the power analyser by applying a s
sampling time in the data capture*
4.!.! 4oad analysis
The measured root mean sAuare line<to<line back 1M' %alues for the MM and 1M machines& at
the ambient temperature of (= MD and ))) min
<
rated speed& are !=*+ 5 and =+*! 5&
respecti%ely* The measurement was carried out by rotating the measured machine with a load
machine and measuring the line<to<line %oltage with a power analyser* The estimated %alue for the
back 1M' according to the finite element calculations at the rated speed of ))) min
<
is ==*) 5
for the MM machine and =,*, 5 for the 1M machine*
The difference in the 1M' ratings is mostly due to the dimensioning error in the magnet module
design caused by the end effect in the rotors& which is explained by the fact that the permanent
magnet ends tend to short<circuit part of the main flux* Thus& the rotor length in the MM machine
is too short* According to (/ussila& ()))& the physical length of a permanent magnet rotor should
be the length of the stator stack plus two times the air gap length at each end of the machine* Thus&
the machine rotor length should be
l
rotor
= l
Fc
+4o 4.1
to match the effecti%e stator core length* DonseAuently& the rotor flux is smaller than the optimum
=)*) 5 because of the length of >= mm instead of the optimal >" mm* This error was corrected in
the 1M machine dimensioning* An additional flux decrease is caused by the slitting of the surface
magnet in the axial direction* The adhesi%e layers between the magnet pieces decrease the space
factor of the magnet pole*
The difference shows directly in the load characteristics* The MM machine torAue output is around
) P smaller with currents of eAual siGe when compared with the 1M machine* 7oth machines are
measured in motoring operation for !))& -)) and =)) min
<
rotating speeds* 2n addition& the ()))
min
<
rotating speed is measured only for the 1M machine because of the mechanical restrictions
of the MM structure* The output power and torAue of the machines are presented in Tables +* and
+*(* The current %alues refer to the nominal current I
n
S)) A*






!



Table* +** MM machine power output and torAue production gi%en according to the
rotating speed and the load current* The %alues highlighted by grey indicate the nominal
point* The %alues are in JkH ii ;mL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) (* ii --*) "*"= ii ()*) (=*(,ii >*) I
))) "*(- ii -,*) (*>) ii (*) -*)> ii >(*, (*(! ii ()(*-
-)) =*"- ii ")*) ,*)() ii (!*) (*), ii >+*, =*) ii ()=*,
!)) (*>" ii "=*! !*,,) ii (-*) =*!+) ii -)*) >*>>) ii ((*)


Table* +*(* 1M machine power output and torAue production gi%en according to the
rotating speed and the load current* The %alues highlighted by grey indicate the nominal
point in JkH ii ;mL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) ()* ii ,>*) !)*> ii +!*) +)*+( ii ,!*) +,*+( ii (!>*)
=)) =* ii ,>*) ((*-- ii +=*) !)*!!ii ,!* !"*) ii (+(*)
))) )*) ii ,=*) =*" ii +=*) ()*+( ii ,=*) (=*== ii (++*)
-)) -* ii ,-*) )*"= ii +"*) +*=, ii ,,*) "*" ii (+"*)
!)) !*+) ii ))*) +*-") ii =(*) >*!") ii ()!*) -*,") ii (=+*)

7ecause of the loss of the .M flux& the MM machine does not reach the rated (+) ;m torAue at the
nominal current rating as shown in Table +** The 1M structure& on the other hand& meets the
reAuired torAue rating* The importance of the nominal point selection is also shown in the torAue
output characteristics* The torAue production of the machines decreases Auite significantly at the
rated current when the operating speed increases* This is due to the decreasing back 1M' as the
magnets deteriorate thermally as a result of increasing losses as discussed later in Section +*(*!*
The steady<state torAue production at the rated current for both of the machines is presented in 'ig*
+*( as a function of rotating speed*
!



Table* +** MM machine power output and torAue production gi%en according to the
rotating speed and the load current* The %alues highlighted by grey indicate the nominal
point* The %alues are in JkH ii ;mL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) (* ii --*) "*"= ii ()*) (=*(,ii >*) I
))) "*(- ii -,*) (*>) ii (*) -*)> ii >(*, (*(! ii ()(*-
-)) =*"- ii ")*) ,*)() ii (!*) (*), ii >+*, =*) ii ()=*,
!)) (*>" ii "=*! !*,,) ii (-*) =*!+) ii -)*) >*>>) ii ((*)


Table* +*(* 1M machine power output and torAue production gi%en according to the
rotating speed and the load current* The %alues highlighted by grey indicate the nominal
point in JkH ii ;mL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) ()* ii ,>*) !)*> ii +!*) +)*+( ii ,!*) +,*+( ii (!>*)
=)) =* ii ,>*) ((*-- ii +=*) !)*!!ii ,!* !"*) ii (+(*)
))) )*) ii ,=*) =*" ii +=*) ()*+( ii ,=*) (=*== ii (++*)
-)) -* ii ,-*) )*"= ii +"*) +*=, ii ,,*) "*" ii (+"*)
!)) !*+) ii ))*) +*-") ii =(*) >*!") ii ()!*) -*,") ii (=+*)

7ecause of the loss of the .M flux& the MM machine does not reach the rated (+) ;m torAue at the
nominal current rating as shown in Table +** The 1M structure& on the other hand& meets the
reAuired torAue rating* The importance of the nominal point selection is also shown in the torAue
output characteristics* The torAue production of the machines decreases Auite significantly at the
rated current when the operating speed increases* This is due to the decreasing back 1M' as the
magnets deteriorate thermally as a result of increasing losses as discussed later in Section +*(*!*
The steady<state torAue production at the rated current for both of the machines is presented in 'ig*
+*( as a function of rotating speed*
!(



'ig* +*(* Donstant torAue production of the machine& with the nominal %oltage le%el and
steady<state temperature& as a function of rotating speed at the nominal current
according to the %alues in Tables +* and +*(*
The torAue decay in the 1M structure is around - P in the torAue output when comparing with the
!)) rpm and ())) rpm %alues at the nominal current* The drop is caused by a decrease in the back
1M' %oltage caused by the increasing magnet temperature* According to 'ig* +*& the torAue
decay rate of the MM configuration is about the same as with the 1M structure*
The beha%iour in 'ig* +*( suggests similar loss distribution characteristics for both of the machine
configurations* Additional peak torAue test resulted in a ="= ;m maximum torAue %alue for the
MM machine and >>) ;m maximum %alue for the 1M machine at the !() A current* The
percentage difference in the peak torAue production is about the same as in the back electromoti%e
measurement* The peak torAues are low compared with the finite element results& which yield
nearly ")) ;m %alues for both of the machines at the O!)) A current* According to the finite
element calculations& the supply %oltage should be about *= times the back 1M' %oltage*
Apparently& the freAuency con%erter control method is not capable of producing an optimal current
angle control in order to achie%e the desired operation of the machine* According to the measured
data& the machine operating state is presented as phasor diagrams in 'igs* +*! and +*+*
) =)) ))) =)) ())) (=))
,)
())
()
(()
(!)
(+)
(=)
(>)
Rotating speed Jmin
<
L
T
o
r
A
u
e

J
;
m
L


1M
MM
!(



'ig* +*(* Donstant torAue production of the machine& with the nominal %oltage le%el and
steady<state temperature& as a function of rotating speed at the nominal current
according to the %alues in Tables +* and +*(*
The torAue decay in the 1M structure is around - P in the torAue output when comparing with the
!)) rpm and ())) rpm %alues at the nominal current* The drop is caused by a decrease in the back
1M' %oltage caused by the increasing magnet temperature* According to 'ig* +*& the torAue
decay rate of the MM configuration is about the same as with the 1M structure*
The beha%iour in 'ig* +*( suggests similar loss distribution characteristics for both of the machine
configurations* Additional peak torAue test resulted in a ="= ;m maximum torAue %alue for the
MM machine and >>) ;m maximum %alue for the 1M machine at the !() A current* The
percentage difference in the peak torAue production is about the same as in the back electromoti%e
measurement* The peak torAues are low compared with the finite element results& which yield
nearly ")) ;m %alues for both of the machines at the O!)) A current* According to the finite
element calculations& the supply %oltage should be about *= times the back 1M' %oltage*
Apparently& the freAuency con%erter control method is not capable of producing an optimal current
angle control in order to achie%e the desired operation of the machine* According to the measured
data& the machine operating state is presented as phasor diagrams in 'igs* +*! and +*+*
) =)) ))) =)) ())) (=))
,)
())
()
(()
(!)
(+)
(=)
(>)
Rotating speed Jmin
<
L
T
o
r
A
u
e

J
;
m
L


1M
MM
!!



'ig* +*!* .er unit phasor presentation of the MM machine at the ))) min
<
rotating speed
with nominal current and load* The power factor is )*"=*


'ig* +*+* .er unit phasor presentation of the 1M machine at the ))) min
<
rotating speed
with nominal current and load* The power factor is cos S )*""*
According to the measured data& the freAuency con%erter dri%es the MM machine with the positi%e
I
d
current at the nominal operating %alues because of the low 1M' %oltage le%el* :n the other
hand& the 1M machine is fed with a O- P excess %oltage at the same torAue reference*

4.!. +fficiency and loss analysis
Machine efficiencies are calculated from the measured input and output powers of the machines*
The low %oltage le%el causes the dominating effect of /oule losses in the machine efficiency
ratings* The efficiency %alues are presented in Tables +*! and +*+ according to the rotating speed
!!



'ig* +*!* .er unit phasor presentation of the MM machine at the ))) min
<
rotating speed
with nominal current and load* The power factor is )*"=*


'ig* +*+* .er unit phasor presentation of the 1M machine at the ))) min
<
rotating speed
with nominal current and load* The power factor is cos S )*""*
According to the measured data& the freAuency con%erter dri%es the MM machine with the positi%e
I
d
current at the nominal operating %alues because of the low 1M' %oltage le%el* :n the other
hand& the 1M machine is fed with a O- P excess %oltage at the same torAue reference*

4.!. +fficiency and loss analysis
Machine efficiencies are calculated from the measured input and output powers of the machines*
The low %oltage le%el causes the dominating effect of /oule losses in the machine efficiency
ratings* The efficiency %alues are presented in Tables +*! and +*+ according to the rotating speed
!+


and current* The =)) min
<
speed full current %alues for the MM machine were not measured
because of a motor failure*
Table* +*!* Measured MM machine efficiency at different rotating speeds and currents*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) )*, )*,! )*,! <
))) )*, )*,( )*,( )*,
-)) )*,) )*, )*,) )*",
!)) )*", )*"- )*"= )*"(

Table* +*+* Measured 1M machine efficiency at different rotating speeds and currents*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) )*," )*,> )*,> )*,+
=)) )*,- )*,> )*,= )*,+
))) )*,> )*,= )*,= )*,!
-)) )*,= )*,= )*,+ )*,(
!)) )*,! )*,( )*"- )*">

The measured efficiencies of the MM machine in different load and speed conditions are Auite
poor for a permanent magnet machine* 1%en though the efficiencies of the MM structure are lower
compared with the 1M configuration& the absolute losses are %ery close to each other* The total
heat dissipations of the machines are presented in Tables +*= and +*>*

Table* +*=* Total loss dissipation of the MM machine in JHL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) (() !>) -") <
))) "!) +) +-) ,")
-)) >") ,) (,) "!)
!)) !!) >) ,=) +")

!+


and current* The =)) min
<
speed full current %alues for the MM machine were not measured
because of a motor failure*
Table* +*!* Measured MM machine efficiency at different rotating speeds and currents*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) )*, )*,! )*,! <
))) )*, )*,( )*,( )*,
-)) )*,) )*, )*,) )*",
!)) )*", )*"- )*"= )*"(

Table* +*+* Measured 1M machine efficiency at different rotating speeds and currents*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) )*," )*,> )*,> )*,+
=)) )*,- )*,> )*,= )*,+
))) )*,> )*,= )*,= )*,!
-)) )*,= )*,= )*,+ )*,(
!)) )*,! )*,( )*"- )*">

The measured efficiencies of the MM machine in different load and speed conditions are Auite
poor for a permanent magnet machine* 1%en though the efficiencies of the MM structure are lower
compared with the 1M configuration& the absolute losses are %ery close to each other* The total
heat dissipations of the machines are presented in Tables +*= and +*>*

Table* +*=* Total loss dissipation of the MM machine in JHL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) (() !>) -") <
))) "!) +) +-) ,")
-)) >") ,) (,) "!)
!)) !!) >) ,=) +")

!=


Table* +*>* Total loss dissipation of the 1M machine in JHL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) =)) =) -)) !+()
=)) +") ,+) >) (!)
))) !,) -+) =) "")
-)) !=) =!) "-) +")
!)) (() +( ,) !!)

There is inconsistency in the measurement results at low current %alues* The measured torAue
%alues are so low that the accuracy of the ())) ;m torAue transducer causes problems* This can
also be obser%ed from the cur%e* The machine efficiencies according to Tables +*! and +*+ are
presented in 'igs* +*= and +*>& respecti%ely*

'ig* +*=* MM machine efficiency according to Table +*! with the estimated efficiencies at
the ))) min
<
rotating speed*
!=


Table* +*>* Total loss dissipation of the 1M machine in JHL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) =)) =) -)) !+()
=)) +") ,+) >) (!)
))) !,) -+) =) "")
-)) !=) =!) "-) +")
!)) (() +( ,) !!)

There is inconsistency in the measurement results at low current %alues* The measured torAue
%alues are so low that the accuracy of the ())) ;m torAue transducer causes problems* This can
also be obser%ed from the cur%e* The machine efficiencies according to Tables +*! and +*+ are
presented in 'igs* +*= and +*>& respecti%ely*

'ig* +*=* MM machine efficiency according to Table +*! with the estimated efficiencies at
the ))) min
<
rotating speed*
!>



'ig* +*>* 1M machine efficiencies according to Table +*+ with the estimated efficiency at
the ))) min
<
rotating speed*
The 1M machine efficiency is higher at low current %alues as expected* The closed design on the
other hand has more arc in the efficiency cur%es because of the restricted heat remo%al* The heat
remo%al to the cooling fluid decreases at lower loss %alues as a result of the decreasing
temperature difference* The measured losses at ))) min
<
and the )) A current are close to the
estimated loss %alues presented in Table !*>*
4. )hermal characteristics
Machine temperatures were measured according to the load points presented abo%e* The results are
compared with the de%eloped thermal model in order to %erify the %alidity of the model* The
steady<state properties are important for the model accuracy& but a more important aspect is the
dynamic performance in %arying load conditions* The physical parameters for the modelling are
the same as presented in Table !*= in Section !*=** The specific heat capacities of the dynamic
model are the same as in Table !*"*

4..! "teady#state temperatures
As expected& the temperature beha%iour of the machines seems to be similar& but the benefit of the
%entilated structure of the 1M machine starts to show at the -)) min
<
speed* The temperature rises
of the MM and 1M machines are presented in Tables +*- and +*"& respecti%ely*

!>



'ig* +*>* 1M machine efficiencies according to Table +*+ with the estimated efficiency at
the ))) min
<
rotating speed*
The 1M machine efficiency is higher at low current %alues as expected* The closed design on the
other hand has more arc in the efficiency cur%es because of the restricted heat remo%al* The heat
remo%al to the cooling fluid decreases at lower loss %alues as a result of the decreasing
temperature difference* The measured losses at ))) min
<
and the )) A current are close to the
estimated loss %alues presented in Table !*>*
4. )hermal characteristics
Machine temperatures were measured according to the load points presented abo%e* The results are
compared with the de%eloped thermal model in order to %erify the %alidity of the model* The
steady<state properties are important for the model accuracy& but a more important aspect is the
dynamic performance in %arying load conditions* The physical parameters for the modelling are
the same as presented in Table !*= in Section !*=** The specific heat capacities of the dynamic
model are the same as in Table !*"*

4..! "teady#state temperatures
As expected& the temperature beha%iour of the machines seems to be similar& but the benefit of the
%entilated structure of the 1M machine starts to show at the -)) min
<
speed* The temperature rises
of the MM and 1M machines are presented in Tables +*- and +*"& respecti%ely*

!-



Table* +*-* Measured end<winding temperatures of the MM machine in MD*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) >+*= -,*( )*- <
))) =,*, -*" ,+*> ("*(
-)) =*> >>*+ ",*) ()*"
!)) +(*, ="*) "(*( +*,

Table* +*"* Measured end<winding temperatures of the 1M machine in MD*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) =>*+ >=*> ")*) ))*-
=)) =*+ >* ->*! ,,*+
))) +"*) =,*) ->*+ )*+
-)) ++*> =>*- ->*! )=*+
!)) +)* =+*) ->*( )"*-

Table +*- shows that the temperature rise occurs as expected* The load point at =)) min
<
and ))
P torAue was not measured to protect the magnets from o%erheating because no information about
the magnet surface temperature was a%ailable during the measurement& and the -)) min
<

temperature at the )) P torAue was already o%er hundred degrees*
At the rated %alues of ))) min
<
and the )) A current& the steady<state thermal model results in
(> MD for the end<winding temperature of the MM machine* The corresponding temperature for
the 1M structure is )+*" MD* The de%eloped model seems to agree well with the measured data*
The corresponding load points for the ))) min
<
rotating speed are shown in 'ig* +*-*
!-



Table* +*-* Measured end<winding temperatures of the MM machine in MD*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) >+*= -,*( )*- <
))) =,*, -*" ,+*> ("*(
-)) =*> >>*+ ",*) ()*"
!)) +(*, ="*) "(*( +*,

Table* +*"* Measured end<winding temperatures of the 1M machine in MD*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) =>*+ >=*> ")*) ))*-
=)) =*+ >* ->*! ,,*+
))) +"*) =,*) ->*+ )*+
-)) ++*> =>*- ->*! )=*+
!)) +)* =+*) ->*( )"*-

Table +*- shows that the temperature rise occurs as expected* The load point at =)) min
<
and ))
P torAue was not measured to protect the magnets from o%erheating because no information about
the magnet surface temperature was a%ailable during the measurement& and the -)) min
<

temperature at the )) P torAue was already o%er hundred degrees*
At the rated %alues of ))) min
<
and the )) A current& the steady<state thermal model results in
(> MD for the end<winding temperature of the MM machine* The corresponding temperature for
the 1M structure is )+*" MD* The de%eloped model seems to agree well with the measured data*
The corresponding load points for the ))) min
<
rotating speed are shown in 'ig* +*-*
!"



'ig* +*-* Domparison of the measured coil temperatures and model estimates for a )))
min
<
rotating speed* The secondary axis presents the temperature difference of
the estimated model to the measured %alues*

The temperature rise characteristics of the 1M machine for each constant speed in Table +*"
follow the theory for indi%idual rotating speeds& but the ad%antage of %enting is e%ident as the
rotating speed increases* The winding temperatures decrease at eAui%alent loading when the
rotating speed increases* 2ncreasing the rotating speed enhances the coil end con%ection and
thereby decreases the coil end temperature* The effect is opposite in the enclosed structure because
the coil end does not ha%e direct contact to the ambient air*




)*+ )*= )*> )*- )*" )*,
<"
<>
<+
<(
)
(
T
e
m
p
e
r
a
t
u
r
e

J
M
D
L
.er unit current


0.4 0.5 0.6 0.7 0.8 0.9 1
40
60
80
100
120
140
T
e
m
p
e
r
a
t
u
r
e

d
i
f
f
e
r
e
n
c
e

J
C
L
1M calculated
1M measured
MM calculated
MM measured
1M difference
MM difference
!"



'ig* +*-* Domparison of the measured coil temperatures and model estimates for a )))
min
<
rotating speed* The secondary axis presents the temperature difference of
the estimated model to the measured %alues*

The temperature rise characteristics of the 1M machine for each constant speed in Table +*"
follow the theory for indi%idual rotating speeds& but the ad%antage of %enting is e%ident as the
rotating speed increases* The winding temperatures decrease at eAui%alent loading when the
rotating speed increases* 2ncreasing the rotating speed enhances the coil end con%ection and
thereby decreases the coil end temperature* The effect is opposite in the enclosed structure because
the coil end does not ha%e direct contact to the ambient air*




)*+ )*= )*> )*- )*" )*,
<"
<>
<+
<(
)
(
T
e
m
p
e
r
a
t
u
r
e

J
M
D
L
.er unit current


0.4 0.5 0.6 0.7 0.8 0.9 1
40
60
80
100
120
140
T
e
m
p
e
r
a
t
u
r
e

d
i
f
f
e
r
e
n
c
e

J
C
L
1M calculated
1M measured
MM calculated
MM measured
1M difference
MM difference
!,


4.. 2omparison with the dynamic model
The rated point %alues of Table !*- were selected to e%aluate the dynamic model performance* The
temperature rise cur%es are presented in 'ig* +*"*

'ig* +*"* Doil end temperature rise of the MM and 1M machines as a function of time at
))) min
<
and a )) A rated current*

The de%eloped dynamic model shows good performance for the MM machine& as shown in 'ig*
+*"* According to the slope& the modelled heat capacitance seems to be large for both machines
because the model temperature rise is slower compared with the measured data* The difference is
more pronounced with the 1M machine with the %entilated frame* The result is an indicator of the
analytical model inaccuracy resulting from the selected boundary conditions and simplifications in
the machine geometry* The results can be corrected to some extent by %erification tests and
correction coefficients& but in new machine designs& this is not possible* At best& the model is only
a good estimate of the phenomenon itself*
The model was also tested to obtain actual measured race track data* The torAue and rotating speed
profiles from tests on the ;jrnburgring ;ordschleife race track are presented in 'ig* +*,*
!,


4.. 2omparison with the dynamic model
The rated point %alues of Table !*- were selected to e%aluate the dynamic model performance* The
temperature rise cur%es are presented in 'ig* +*"*

'ig* +*"* Doil end temperature rise of the MM and 1M machines as a function of time at
))) min
<
and a )) A rated current*

The de%eloped dynamic model shows good performance for the MM machine& as shown in 'ig*
+*"* According to the slope& the modelled heat capacitance seems to be large for both machines
because the model temperature rise is slower compared with the measured data* The difference is
more pronounced with the 1M machine with the %entilated frame* The result is an indicator of the
analytical model inaccuracy resulting from the selected boundary conditions and simplifications in
the machine geometry* The results can be corrected to some extent by %erification tests and
correction coefficients& but in new machine designs& this is not possible* At best& the model is only
a good estimate of the phenomenon itself*
The model was also tested to obtain actual measured race track data* The torAue and rotating speed
profiles from tests on the ;jrnburgring ;ordschleife race track are presented in 'ig* +*,*
+)



'ig* +*,* Measured track data from the test lap on ;jrbrugring ;ordschleife* The rotating
speed is presented in black and the torAue in dark grey* The measured data are
from the right rear machine with an approximately >)) ;m peak torAue %alue*
The measured track data were used as a feed signal for the dynamic thermal model to estimate the
temperature beha%iour of the 1M machine in track conditions* The rotating speed& supply current
and cooling fluid temperatures were taken into account in the simulation* The estimated coil
temperature profile of a single lap on the ;jrburgring ;ordschleife track is presented in 'ig* +*)
together with the measured coil temperature*

'ig* +*)* Measured (black) and estimated (gray) coil temperature in a single lap
measurement on ;jrburgring ;ordschleife*
+)



'ig* +*,* Measured track data from the test lap on ;jrbrugring ;ordschleife* The rotating
speed is presented in black and the torAue in dark grey* The measured data are
from the right rear machine with an approximately >)) ;m peak torAue %alue*
The measured track data were used as a feed signal for the dynamic thermal model to estimate the
temperature beha%iour of the 1M machine in track conditions* The rotating speed& supply current
and cooling fluid temperatures were taken into account in the simulation* The estimated coil
temperature profile of a single lap on the ;jrburgring ;ordschleife track is presented in 'ig* +*)
together with the measured coil temperature*

'ig* +*)* Measured (black) and estimated (gray) coil temperature in a single lap
measurement on ;jrburgring ;ordschleife*
+


The simulation model shows good agreement with the measured coil temperature e%en though the
ambient temperature and con%ecti%e cooling effects of the mo%ing %ehicle could not be replicated
in the simulation procedure* The model starts to de%iate from the track data around !=) second
point* 2t is possible that the modelled air gap con%ection coefficient is o%erestimated at high
rotating speed %alues& which causes the model to cool more efficiently*

4.., +stimation of the magnet temperature
The increasing rotating speed causes increasing core losses and additional stray load losses in the
machine& especially in the rotor& which cause degradation of the magnet polariGation& that is&
re%ersible demagnetiGation of the magnet* The measured back 1M' of the 1M machine is =+*! 5
and respecti%ely& !=*+ 5 for the MM machine at the (= D temperature* The corresponding %alues
for the back 1M' in different load conditions were measured between the changes in the machine
load* According to these data& the magnet temperature can be estimated& because the magnet
remanent flux density decreases by about )* P for e%ery Cel%in degree as the magnet
temperature rises* The estimated magnet temperatures are presented in Tables +*, and +*)& and
correspondingly& in 'igs* +* and +*(*
Table* +*,* 1stimated magnet temperatures JMDL of the MM machine according to the
back 1M' measurement*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) >- -" ", <
))) =, >> -" ,+
-)) = =- >> --
!)) !- +) +, >>

Table* +*)* 1stimated magnet temperatures JMDL of the 1M machine according to the
back 1M' measurement*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) >) >+ - ")
=)) =! =- >( -(
))) +, =( =, >,
-)) +( +> =+ >=
!)) !!*, +(*)+ +=*>> =+*)(

+


The simulation model shows good agreement with the measured coil temperature e%en though the
ambient temperature and con%ecti%e cooling effects of the mo%ing %ehicle could not be replicated
in the simulation procedure* The model starts to de%iate from the track data around !=) second
point* 2t is possible that the modelled air gap con%ection coefficient is o%erestimated at high
rotating speed %alues& which causes the model to cool more efficiently*

4.., +stimation of the magnet temperature
The increasing rotating speed causes increasing core losses and additional stray load losses in the
machine& especially in the rotor& which cause degradation of the magnet polariGation& that is&
re%ersible demagnetiGation of the magnet* The measured back 1M' of the 1M machine is =+*! 5
and respecti%ely& !=*+ 5 for the MM machine at the (= D temperature* The corresponding %alues
for the back 1M' in different load conditions were measured between the changes in the machine
load* According to these data& the magnet temperature can be estimated& because the magnet
remanent flux density decreases by about )* P for e%ery Cel%in degree as the magnet
temperature rises* The estimated magnet temperatures are presented in Tables +*, and +*)& and
correspondingly& in 'igs* +* and +*(*
Table* +*,* 1stimated magnet temperatures JMDL of the MM machine according to the
back 1M' measurement*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) >- -" ", <
))) =, >> -" ,+
-)) = =- >> --
!)) !- +) +, >>

Table* +*)* 1stimated magnet temperatures JMDL of the 1M machine according to the
back 1M' measurement*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) >) >+ - ")
=)) =! =- >( -(
))) +, =( =, >,
-)) +( +> =+ >=
!)) !!*, +(*)+ +=*>> =+*)(

+(



'ig* +** Magnet temperature %alues based on Table +*, for the MM machine*

'ig* +*(* Magnet temperature %alues based on Table +*) for the 1M machine*
) ) () !) +) =) >) -) ") ,) ))
!))
-))
)))
=))
Temperature JMDL
R
o
t
a
t
i
n
g

s
p
e
e
d

J
m
i
n
<

L


)*(2n
)*+2n
)*>2n
)*"2n
) ) () !) +) =) >) -) ")
!))
-))
)))
=))
()))
Temperature JMDL
R
o
t
a
t
i
n
g

s
p
e
e
d

J
m
i
n
<

L


)*(2n
)*+2n
)*>2n
)*"2n
+(



'ig* +** Magnet temperature %alues based on Table +*, for the MM machine*

'ig* +*(* Magnet temperature %alues based on Table +*) for the 1M machine*
) ) () !) +) =) >) -) ") ,) ))
!))
-))
)))
=))
Temperature JMDL
R
o
t
a
t
i
n
g

s
p
e
e
d

J
m
i
n
<

L


)*(2n
)*+2n
)*>2n
)*"2n
) ) () !) +) =) >) -) ")
!))
-))
)))
=))
()))
Temperature JMDL
R
o
t
a
t
i
n
g

s
p
e
e
d

J
m
i
n
<

L


)*(2n
)*+2n
)*>2n
)*"2n
+!


Hhen comparing with the rated<%alue operation at ))) min
<
and the )) A current& the static
thermal model shows good agreement with the estimated magnet temperatures* The temperature
model gi%es ,*( MD for the magnet temperature of the MM machine* :n the other hand& the
magnet temperature of the 1M machine is "> MD& which is slightly too high when compared with
the >, MD %alue in Table +*)* 2t seems that the con%ection from the rotor shaft slee%e should be
taken into account because the structure in Auestion has a large con%ection surface area to the
machine inner air* The enclosed structure does not gain any significant benefit from the axle slee%e
modelling& because the machine inner air is considerably higher compared with the through<
%entilated structure* The shaft of the 1M machine is shown in 'ig* +*!*

'ig* +*!* Rotor shaft of the 1M machine*

4..4 2ooling efficiency
The radiator inlet and outlet temperatures were also measured with two separate .T<)) elements&
and the heat remo%ed by the cooling system was calculated for the measured water flow of () g
per second* 9eat remo%al in the eAui%alent load points and rotating speeds is presented in Tables
+* and +*(*





+!


Hhen comparing with the rated<%alue operation at ))) min
<
and the )) A current& the static
thermal model shows good agreement with the estimated magnet temperatures* The temperature
model gi%es ,*( MD for the magnet temperature of the MM machine* :n the other hand& the
magnet temperature of the 1M machine is "> MD& which is slightly too high when compared with
the >, MD %alue in Table +*)* 2t seems that the con%ection from the rotor shaft slee%e should be
taken into account because the structure in Auestion has a large con%ection surface area to the
machine inner air* The enclosed structure does not gain any significant benefit from the axle slee%e
modelling& because the machine inner air is considerably higher compared with the through<
%entilated structure* The shaft of the 1M machine is shown in 'ig* +*!*

'ig* +*!* Rotor shaft of the 1M machine*

4..4 2ooling efficiency
The radiator inlet and outlet temperatures were also measured with two separate .T<)) elements&
and the heat remo%ed by the cooling system was calculated for the measured water flow of () g
per second* 9eat remo%al in the eAui%alent load points and rotating speeds is presented in Tables
+* and +*(*





++


Table* +** Radiator heat remo%al rate for the MM machine in JHL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) =+( >( -(> I
))) +(+ +," =," -=)
-)) !-( +(+ =!> >""
!)) (-> !>) +"+ >=)

Table* +*(* Radiator heat remo%al rate for the 1M machine in JHL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) (,( !) !=) !-"
=)) (=" (>> (," !="
))) (++ (+" (-+ (">
-)) (!> (+( (-> !>+
!)) "> (+ (++ (,"

Dompared with the totally enclosed structure of the MM machine& the heat remo%al through water
in the 1M machine cooling is considerably lower* This could be anticipated because the heat
exchange rate is dependent on the temperature difference* 7ecause of the %ented structure& the
stator stack temperature is lower than in the totally enclosed MM machine* The ma3ority of the
heat exits the machine structure through the %enting* The heat remo%al is still considerable
compared with the total heat production of the machines* The water heat remo%al rate compared
with the total losses is presented in Tables +*! and +*+*
Table* +*!* Ratio of water cooling heat remo%al to the total losses of the MM machine*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) )*++ )*+= )*+ <
))) )*= )*++ )*+ )*!"
-)) )*== )*+> )*+ )*!"
!)) )*"+ )*=, )*= )*++


++


Table* +** Radiator heat remo%al rate for the MM machine in JHL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) =+( >( -(> I
))) +(+ +," =," -=)
-)) !-( +(+ =!> >""
!)) (-> !>) +"+ >=)

Table* +*(* Radiator heat remo%al rate for the 1M machine in JHL*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) (,( !) !=) !-"
=)) (=" (>> (," !="
))) (++ (+" (-+ (">
-)) (!> (+( (-> !>+
!)) "> (+ (++ (,"

Dompared with the totally enclosed structure of the MM machine& the heat remo%al through water
in the 1M machine cooling is considerably lower* This could be anticipated because the heat
exchange rate is dependent on the temperature difference* 7ecause of the %ented structure& the
stator stack temperature is lower than in the totally enclosed MM machine* The ma3ority of the
heat exits the machine structure through the %enting* The heat remo%al is still considerable
compared with the total heat production of the machines* The water heat remo%al rate compared
with the total losses is presented in Tables +*! and +*+*
Table* +*!* Ratio of water cooling heat remo%al to the total losses of the MM machine*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

=)) )*++ )*+= )*+ <
))) )*= )*++ )*+ )*!"
-)) )*== )*+> )*+ )*!"
!)) )*"+ )*=, )*= )*++


+=


Table* +*+* Ratio of water cooling heat remo%al to the total losses of the 1M machine*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) )*=" )*(- )*( )*
=)) )*=+ )*(" )*, )*>
))) )*>( )*!+ )*(+ )*=
-)) )*>" )*+> )*!( )*(=
!)) )*"! )*= )*(- )*((

The measuring accuracy of the torAue meter is )* P of the measuring range& which means that the
error margin is ( ;m* The offset is high& and there is thus a certain error in the measured %alue& but
still& the %alues are comparable with other measured %alues because the other point also includes
the same error* The magnitude of the measured %alues can be kept correct* Ceeping this in mind&
the effect of the water cooling is undisputed in the MM machine construction* 2n the low<speed
area and moderate power ratings& the ma3ority of losses are remo%ed by water& but the cooling
efficiency of water seems to decrease Auite dramatically as the loading increases* The situation is
better in the 1M machine where the ma3ority of losses are remo%ed by con%ection through the
%enting holes of the machine* According to the measurements& the decrease in the water cooling
efficiency seems to be Auite linear if the !)) min
<
speed is ignored*

4., 2onclusions
According to the tests& there are no significant differences in the machine performance of the two
machine types studied* The analytical and finite<element<based modelling seems to agree with the
measured performance data when considering operation at rated torAue %alues* The measured peak
torAues could not be achie%ed with the test set<up* :ne possible reason is related to the control of
the machine* The control method is not capable of producing a correct current angle for high
torAue operation* 9owe%er& the test results show that the lack of back 1M' %oltage in the MM
machine results in a lower performance compared with the 1M machine* The difference is highest
in the peak torAue production*
The temperature measurements show good agreement with the de%eloped lumped<parameter
thermal model* 'urther de%elopment of the model is needed for the through<%entilated structure as
the magnet temperature estimates are high compared with the estimates for the measured decrease
in the back 1M'*
The dynamic model also agrees well with the measured data& but the thermal capacitance %alues
are too high* A more accurate model of the geometry could sol%e this problem*
+=


Table* +*+* Ratio of water cooling heat remo%al to the total losses of the 1M machine*
min
#!
=.4I
n
=.3I
n
=.@I
n
I
n

())) )*=" )*(- )*( )*
=)) )*=+ )*(" )*, )*>
))) )*>( )*!+ )*(+ )*=
-)) )*>" )*+> )*!( )*(=
!)) )*"! )*= )*(- )*((

The measuring accuracy of the torAue meter is )* P of the measuring range& which means that the
error margin is ( ;m* The offset is high& and there is thus a certain error in the measured %alue& but
still& the %alues are comparable with other measured %alues because the other point also includes
the same error* The magnitude of the measured %alues can be kept correct* Ceeping this in mind&
the effect of the water cooling is undisputed in the MM machine construction* 2n the low<speed
area and moderate power ratings& the ma3ority of losses are remo%ed by water& but the cooling
efficiency of water seems to decrease Auite dramatically as the loading increases* The situation is
better in the 1M machine where the ma3ority of losses are remo%ed by con%ection through the
%enting holes of the machine* According to the measurements& the decrease in the water cooling
efficiency seems to be Auite linear if the !)) min
<
speed is ignored*

4., 2onclusions
According to the tests& there are no significant differences in the machine performance of the two
machine types studied* The analytical and finite<element<based modelling seems to agree with the
measured performance data when considering operation at rated torAue %alues* The measured peak
torAues could not be achie%ed with the test set<up* :ne possible reason is related to the control of
the machine* The control method is not capable of producing a correct current angle for high
torAue operation* 9owe%er& the test results show that the lack of back 1M' %oltage in the MM
machine results in a lower performance compared with the 1M machine* The difference is highest
in the peak torAue production*
The temperature measurements show good agreement with the de%eloped lumped<parameter
thermal model* 'urther de%elopment of the model is needed for the through<%entilated structure as
the magnet temperature estimates are high compared with the estimates for the measured decrease
in the back 1M'*
The dynamic model also agrees well with the measured data& but the thermal capacitance %alues
are too high* A more accurate model of the geometry could sol%e this problem*
+>


To sum up& the machine performance is at the estimated le%el with the embedded magnet structure*
This would also apply to the magnet module design with corrected rotor dimensions that are due to
the permanent magnet end effect* 9owe%er& the mechanics of the magnet module has to be
de%eloped further* The comparison still shows the benefit of the through<%entilated structure o%er
the totally enclosed machine* 2t would be ad%isable to consider through<%entilated structures in
mobile applications for a higher performance if suitable protection against ambient conditions
could be arranged*
'urther testing by Metropolia has shown good results of the 1M machine performance in the field
weakening region resulting in se%eral speed records on track and ice* Among the most significant
achie%ements is the -*)!*()( electric %ehicle speed record of (=(*), kmEh on ice& which means
o%er =) P field weakening of permanent magnet polariGation on fully charged batteries*

+>


To sum up& the machine performance is at the estimated le%el with the embedded magnet structure*
This would also apply to the magnet module design with corrected rotor dimensions that are due to
the permanent magnet end effect* 9owe%er& the mechanics of the magnet module has to be
de%eloped further* The comparison still shows the benefit of the through<%entilated structure o%er
the totally enclosed machine* 2t would be ad%isable to consider through<%entilated structures in
mobile applications for a higher performance if suitable protection against ambient conditions
could be arranged*
'urther testing by Metropolia has shown good results of the 1M machine performance in the field
weakening region resulting in se%eral speed records on track and ice* Among the most significant
achie%ements is the -*)!*()( electric %ehicle speed record of (=(*), kmEh on ice& which means
o%er =) P field weakening of permanent magnet polariGation on fully charged batteries*

+-


1 2onclusions
This thesis presents a study on the permanent magnet traction machine design for a performance<
oriented %ehicle application with a four<wheel dri%e* The study pro%ides two different approaches
to produce a light<weight machine structure with increased saliency for enhanced torAue
production and field weakening characteristicsB A traditional embedded magnet machine design
with inherent saliency and a special magnet module design to increase the machine saliency are
introduced*

1.! 2ontributions of the doctoral thesis
According to the thermal analysis carried out for the structures& the implemented design methods
resulted in two functional machine configurations* 7oth of the introduced concepts show an
increased torAue output capability as a result of the increased saliency*
The magnet module design seems to suffer from mechanical problems in the magnet fixing* Thus&
the embedded machine structure is a more fa%ourable solution because of its better mechanical
stability* 'urthermore& the through<%entilated structure shows benefits in cooling& and this should
be considered in mobile applications as traditional 2D1 solutions ha%e also to be protected against
the en%ironment* A through<%entilated structure pro%ides higher steady state load capability and
the lower temperature rise time constants allow a better dynamic performance*
7ecause a traction application does not necessarily ha%e a specific operating point& the traditional
guidelines for machine design cannot be directly applied to a traction machine design* Thus& the
balance between the electromagnetic design (including design limitations related to energy source
selection)& heat transfer and loss generation becomes more important& especially in /oule<loss<
dominated machine structures as presented in the doctoral thesis*
According to the simulations& a performance analysis by indi%idual two<axis model inductances is
only indicati%e for these machine types* A comparison between the finite element method and the
analytical approach shows the weakness of a traditional analytical model& which does not take into
account the armature reaction accurately enough because the direct< and Auadrature<axis
inductances obtained from the 'lux (D calculation for the d< and A<axes indi%idually do not
correctly describe the magnetic state of the machine*
An extended finite<element<based approach is needed to analyse the cross magnetiGation effect on
the machine performance in order to fully exploit the torAue output potential* ;e%ertheless& an
analytical approach is efficient in the basic dimensioning and prediction of the physical beha%iour
as shown in the simple tool de%elopment carried out for the slot and tooth dimensioning to
e%aluate the heat transfer efficiency in the stator stack*
According to the tests& there are no significant differences in the performances between these two
machines* The analytical and finite<element<based modelling seems to agree with the measured
performance data when considering operation at rated torAue %alues* 9owe%er& the test results
+-


1 2onclusions
This thesis presents a study on the permanent magnet traction machine design for a performance<
oriented %ehicle application with a four<wheel dri%e* The study pro%ides two different approaches
to produce a light<weight machine structure with increased saliency for enhanced torAue
production and field weakening characteristicsB A traditional embedded magnet machine design
with inherent saliency and a special magnet module design to increase the machine saliency are
introduced*

1.! 2ontributions of the doctoral thesis
According to the thermal analysis carried out for the structures& the implemented design methods
resulted in two functional machine configurations* 7oth of the introduced concepts show an
increased torAue output capability as a result of the increased saliency*
The magnet module design seems to suffer from mechanical problems in the magnet fixing* Thus&
the embedded machine structure is a more fa%ourable solution because of its better mechanical
stability* 'urthermore& the through<%entilated structure shows benefits in cooling& and this should
be considered in mobile applications as traditional 2D1 solutions ha%e also to be protected against
the en%ironment* A through<%entilated structure pro%ides higher steady state load capability and
the lower temperature rise time constants allow a better dynamic performance*
7ecause a traction application does not necessarily ha%e a specific operating point& the traditional
guidelines for machine design cannot be directly applied to a traction machine design* Thus& the
balance between the electromagnetic design (including design limitations related to energy source
selection)& heat transfer and loss generation becomes more important& especially in /oule<loss<
dominated machine structures as presented in the doctoral thesis*
According to the simulations& a performance analysis by indi%idual two<axis model inductances is
only indicati%e for these machine types* A comparison between the finite element method and the
analytical approach shows the weakness of a traditional analytical model& which does not take into
account the armature reaction accurately enough because the direct< and Auadrature<axis
inductances obtained from the 'lux (D calculation for the d< and A<axes indi%idually do not
correctly describe the magnetic state of the machine*
An extended finite<element<based approach is needed to analyse the cross magnetiGation effect on
the machine performance in order to fully exploit the torAue output potential* ;e%ertheless& an
analytical approach is efficient in the basic dimensioning and prediction of the physical beha%iour
as shown in the simple tool de%elopment carried out for the slot and tooth dimensioning to
e%aluate the heat transfer efficiency in the stator stack*
According to the tests& there are no significant differences in the performances between these two
machines* The analytical and finite<element<based modelling seems to agree with the measured
performance data when considering operation at rated torAue %alues* 9owe%er& the test results
+"


show that the lack of back 1M' %oltage in the magnet module (MM) machine results in a lower
performance compared with the 1M machine* 2n addition& the welded steel band generates
significant losses& which clearly reduces the o%erall efficiency of the MM structure*
The temperature measurements also show good agreement with the de%eloped lumped<parameter
thermal model* 'urther de%elopment of the model is needed for the through<%entilated structure&
because the magnet temperature estimates are high compared with the modelled %alues including
the decrease in the back 1M'* The dynamic model also agrees well with the measured data& but
the thermal capacitance %alues are too high resulting in a de%iation in low transients*
The estimated peak torAues could not be achie%ed with the test set<up* This is possibly related to
the control of the machine* The control method is not capable of producing a correct current angle
for high torAue operation* Additional tests by Metropolia ha%e shown good results of the field
weakening capability resulting in se%eral speed records on %arious tracks*

1. Prospects of the future work
At least the following topics of further study ha%e been identified in the course of the doctoral
workB

< Determination of optimum magnetic circuit dimensions considering the utiliGation of
%oltage reser%e in torAue production*
< 1ffect of %ariation in the supply %oltage le%el on the selection of the nominal point*
< De%elopment of an analytical lumped<parameter model for the optimiGation of
machine heat remo%al*
< De%elopment of an analytical cross<magnetiGation model for the fast e%aluation of
machine parameters*
< De%elopment of magnet pole fixing methods in the magnet module design*




+"


show that the lack of back 1M' %oltage in the magnet module (MM) machine results in a lower
performance compared with the 1M machine* 2n addition& the welded steel band generates
significant losses& which clearly reduces the o%erall efficiency of the MM structure*
The temperature measurements also show good agreement with the de%eloped lumped<parameter
thermal model* 'urther de%elopment of the model is needed for the through<%entilated structure&
because the magnet temperature estimates are high compared with the modelled %alues including
the decrease in the back 1M'* The dynamic model also agrees well with the measured data& but
the thermal capacitance %alues are too high resulting in a de%iation in low transients*
The estimated peak torAues could not be achie%ed with the test set<up* This is possibly related to
the control of the machine* The control method is not capable of producing a correct current angle
for high torAue operation* Additional tests by Metropolia ha%e shown good results of the field
weakening capability resulting in se%eral speed records on %arious tracks*

1. Prospects of the future work
At least the following topics of further study ha%e been identified in the course of the doctoral
workB

< Determination of optimum magnetic circuit dimensions considering the utiliGation of
%oltage reser%e in torAue production*
< 1ffect of %ariation in the supply %oltage le%el on the selection of the nominal point*
< De%elopment of an analytical lumped<parameter model for the optimiGation of
machine heat remo%al*
< De%elopment of an analytical cross<magnetiGation model for the fast e%aluation of
machine parameters*
< De%elopment of magnet pole fixing methods in the magnet module design*




+,


&eferences


(Altairnano& ()),) Altairnano& ()),& Altairnano 50Ah batterv cell technical paper&
J:nlineL& JAccessed ) /une ()(L& A%ailable from
www*altairnano*com*
(Arkadan et al*& ())-) Arkadan& A*A*& Al<Aawar& ;* and 9anbali& A*A*& \Design
:ptimiGation of SynRM Dri%es for 915 .ower Train
Applications&\ 2n IEEE international Conference on Electric
Machines and Drives& 5ol* *& ())-& pp* ")@"+*
(7arcaro et al*& ())") 7arcaro M*& 7ianchi& ;* and Magnussen '*& [.M motors for
hybrid electric %ehicles&\ 2n Universities Power Engineering
Conference& ())"& pp* @=*
(7ecker et al*& ,>() 7ecker& C* M* and Caye& /*& ,>(& [Measurements of Diabatic
flow in annulus with inner rotating cylinder&\ Journal of Heat
Transfer& no "+& pp* ,-@)=*
(7oglietti et al*& ,,) 7oglietti A*& 'erraris .*& #aGGari M* and .rofumo '*& [2ron losses
in magnetic materials with six<step and .HM in%erter supply&\
IEEE Transactions on Magnetics& 5ol* (-& 2ss* >& ,,& pp* =!!+@
=!!>*
(7oglietti et al*& ,,!) 7oglietti A*& 'erraris .*& #aGGari M* and .rofumo '*& [1ffects of
different modulation index on the iron losses in soft magnetic
materials supplied by .HM in%erter&\ IEEE Transactions on
Magnetics& 5ol* (,& 2ss* >& ,,!& pp* !(!+@!(!>*
(7oglietti et al*& ,,=) 7oglietti A*& 'erraris .*& #aGGari M* and .astorelli M*& [Dhange of
the iron losses with the switching supply freAuency in soft
magnetic materials supplied by .HM in%erter&\ IEEE
Transactions on Magnetics& 5ol* !& 2ss* >& ,,=& pp* +(=)@+(=(*
(7oglietti et al*& ())!) 7oglietti A*& Da%agnino*A& #aGGari M* and .astorelli M*&
\.redicting iron losses in soft magnetic materials with arbitrary
%oltage supplyB An engineering approach&\ IEEE Transactions on
Magnetics& 5ol* !,& 2ss* (& ())!& pp* ,"@,",*
(Daricchi et al*& ,,+) Daricchi& '*& Drescimbini& '*& 'edeli& 1* and ;oioa& 8*& [Design
and construction of a wheel<directly<coupled axial<flux .M motor
prototype for 15s&\ 2n IEEE Industrv Applications Societv Annual
Meeting& 5ol* & ,,+& pp* (=+@(>*
+,


&eferences


(Altairnano& ()),) Altairnano& ()),& Altairnano 50Ah batterv cell technical paper&
J:nlineL& JAccessed ) /une ()(L& A%ailable from
www*altairnano*com*
(Arkadan et al*& ())-) Arkadan& A*A*& Al<Aawar& ;* and 9anbali& A*A*& \Design
:ptimiGation of SynRM Dri%es for 915 .ower Train
Applications&\ 2n IEEE international Conference on Electric
Machines and Drives& 5ol* *& ())-& pp* ")@"+*
(7arcaro et al*& ())") 7arcaro M*& 7ianchi& ;* and Magnussen '*& [.M motors for
hybrid electric %ehicles&\ 2n Universities Power Engineering
Conference& ())"& pp* @=*
(7ecker et al*& ,>() 7ecker& C* M* and Caye& /*& ,>(& [Measurements of Diabatic
flow in annulus with inner rotating cylinder&\ Journal of Heat
Transfer& no "+& pp* ,-@)=*
(7oglietti et al*& ,,) 7oglietti A*& 'erraris .*& #aGGari M* and .rofumo '*& [2ron losses
in magnetic materials with six<step and .HM in%erter supply&\
IEEE Transactions on Magnetics& 5ol* (-& 2ss* >& ,,& pp* =!!+@
=!!>*
(7oglietti et al*& ,,!) 7oglietti A*& 'erraris .*& #aGGari M* and .rofumo '*& [1ffects of
different modulation index on the iron losses in soft magnetic
materials supplied by .HM in%erter&\ IEEE Transactions on
Magnetics& 5ol* (,& 2ss* >& ,,!& pp* !(!+@!(!>*
(7oglietti et al*& ,,=) 7oglietti A*& 'erraris .*& #aGGari M* and .astorelli M*& [Dhange of
the iron losses with the switching supply freAuency in soft
magnetic materials supplied by .HM in%erter&\ IEEE
Transactions on Magnetics& 5ol* !& 2ss* >& ,,=& pp* +(=)@+(=(*
(7oglietti et al*& ())!) 7oglietti A*& Da%agnino*A& #aGGari M* and .astorelli M*&
\.redicting iron losses in soft magnetic materials with arbitrary
%oltage supplyB An engineering approach&\ IEEE Transactions on
Magnetics& 5ol* !,& 2ss* (& ())!& pp* ,"@,",*
(Daricchi et al*& ,,+) Daricchi& '*& Drescimbini& '*& 'edeli& 1* and ;oioa& 8*& [Design
and construction of a wheel<directly<coupled axial<flux .M motor
prototype for 15s&\ 2n IEEE Industrv Applications Societv Annual
Meeting& 5ol* & ,,+& pp* (=+@(>*
=)


(Dhan& ,,>) Dhan& D*D* and Dhau& C*T*& [An ad%anced permanent magnet
motor dri%e system for battery<powered electric %ehicles&\
Transactions on Jehicular Technologv& 5ol* += & 2ss* & ,,>& pp*
")@""*
(Dhau et al*& ())") Dhau& C*T* Dhan& D*D* and #iu D*& [:%er%iew of .ermanent<
Magnet 7rushless Dri%es for 1lectric and 9ybrid 1lectric
5ehicles&\ Transactions on Industrial Electronics, 5ol* ==*& 2ss* >&
())"& pp* ((+>@((=-*
(D%etko%ski et al*& ())>) D%etko%ski& 8*& .etko%ska& #* and 8air& S*& [TorAue analysis of
axial field .M synchronous motor for 15&\ 2n International
Svmposium on Power Electronics, Electric Drives, Automation
and Motion& ())>& pp* (+@(-*
(1ckermann& ())) 1ckermann 1*& ())& World Historv of the Automobile& Society of
Automoti%e 1ngineers 2nc* Harrendale*
(1lliot& ,,!) 1lliot R*& ,,!& Electromagnetics. Historv, Theorv, and
Applications&
st
ed*& ;ew /erseyB Hiley<2111 .ress*
('aiG& ())>) 'aiG& /* and Cey%ani<7orou3eni 7*& \:ptimal Design of 2nternal
.ermanent Magnet Motor for StarterE8enerator of 9ybrid 1lectric
5ehicle&\ 2n Power Electronics and Motion Control Conference&
())>& pp* ,"+@,,)*
(8rauers& ())+) 8rauers A* and Casinathan .*& ['orce Density #imits in low@
speed .M machines due to temperature and reactance&\ IEEE
Transactions on Energv Conversion& 5ol* +& 2ss* !& September
())+& pp* ="@=(=*
(9eikkil6& ())() 9eikkil6& T& Permanent magnet svnchronous motor for industrial
inverter appilications - analvsis and design& Acta uni%ersitatis
#appeenrantaensis ;o* !+, Doctoral thesis& ())(& #appeenranta
$ni%ersity of Technology*
(96m6l6inen& ()),) 96m6l6inen 9*& ()),& [Skin effect and %oltage stresses in .HM
freAuency con%erter dri%en motors&\ MasterFs thesis&
#appeenranta $ni%ersity of Technology*
(21A& ()) 2nternational 1nergy Agency& ()& Kev World Energv statistics&
Technical report& J:nlineL& JAccessed ) /une ()(L& A%ailable
from www*iea*org*
(2111 Doll*& ,,!) 2111 DolloAuium on Motors and Drives for batterv powered
Propulsion& Digest ;o* ")& ,,!*
=)


(Dhan& ,,>) Dhan& D*D* and Dhau& C*T*& [An ad%anced permanent magnet
motor dri%e system for battery<powered electric %ehicles&\
Transactions on Jehicular Technologv& 5ol* += & 2ss* & ,,>& pp*
")@""*
(Dhau et al*& ())") Dhau& C*T* Dhan& D*D* and #iu D*& [:%er%iew of .ermanent<
Magnet 7rushless Dri%es for 1lectric and 9ybrid 1lectric
5ehicles&\ Transactions on Industrial Electronics, 5ol* ==*& 2ss* >&
())"& pp* ((+>@((=-*
(D%etko%ski et al*& ())>) D%etko%ski& 8*& .etko%ska& #* and 8air& S*& [TorAue analysis of
axial field .M synchronous motor for 15&\ 2n International
Svmposium on Power Electronics, Electric Drives, Automation
and Motion& ())>& pp* (+@(-*
(1ckermann& ())) 1ckermann 1*& ())& World Historv of the Automobile& Society of
Automoti%e 1ngineers 2nc* Harrendale*
(1lliot& ,,!) 1lliot R*& ,,!& Electromagnetics. Historv, Theorv, and
Applications&
st
ed*& ;ew /erseyB Hiley<2111 .ress*
('aiG& ())>) 'aiG& /* and Cey%ani<7orou3eni 7*& \:ptimal Design of 2nternal
.ermanent Magnet Motor for StarterE8enerator of 9ybrid 1lectric
5ehicle&\ 2n Power Electronics and Motion Control Conference&
())>& pp* ,"+@,,)*
(8rauers& ())+) 8rauers A* and Casinathan .*& ['orce Density #imits in low@
speed .M machines due to temperature and reactance&\ IEEE
Transactions on Energv Conversion& 5ol* +& 2ss* !& September
())+& pp* ="@=(=*
(9eikkil6& ())() 9eikkil6& T& Permanent magnet svnchronous motor for industrial
inverter appilications - analvsis and design& Acta uni%ersitatis
#appeenrantaensis ;o* !+, Doctoral thesis& ())(& #appeenranta
$ni%ersity of Technology*
(96m6l6inen& ()),) 96m6l6inen 9*& ()),& [Skin effect and %oltage stresses in .HM
freAuency con%erter dri%en motors&\ MasterFs thesis&
#appeenranta $ni%ersity of Technology*
(21A& ()) 2nternational 1nergy Agency& ()& Kev World Energv statistics&
Technical report& J:nlineL& JAccessed ) /une ()(L& A%ailable
from www*iea*org*
(2111 Doll*& ,,!) 2111 DolloAuium on Motors and Drives for batterv powered
Propulsion& Digest ;o* ")& ,,!*
=


(2ncropera& ())-) 2ncropera '*.* and DeHitt D* .*& ())-& Fundamentals of Heat and
Mass Transfer& >
th
ed*& ;ew ?orkB /ohn Hiley k Sons& 2nc*
(/ung et al*& ())-) /ung /<H*& 9ong /<.* and Cim ?<C*& [Dharacteristic Analysis and
Domparison of 2.MSM for 915 According to .ole and Slot
Dombination&\ 2n Jehicle Power and Propulsion Conference,
())-& pp* --"@-"!*
(/ussila& ())) /ussila 9*& ())& Concentrated Winding multiphase permanent
magnet machine design and electromagnetic properties - case
axial flux machine& Acta $ni%ersitatis #appeenrantaensis ;o* !-+&
Doctoral thesis& #appeenranta $ni%ersity of Technology*
(Cylander& ,,=) Cylander& 8*& ,,=& Thermal modeling of small cage induction
motors& Technical report ;o* (>=& Dhalmers $ni%ersity of
Technology& 80teborg& Sweden& 'ebruary ,,=*
(#arminie& ())!) #arminie /* and #owry /*& ())!& Electric Jehicle Technologv
Explained& DhichesterB /ohn Hiley k Sons& #td*
(#indstr0m& ,,,) #indstr0m /*& ,,,& Thermal Model of a Permanent Magnet
Motor for A Hvbrid Electric Jehicle& Technical report& Dhalmers
$ni%ersity of Technology& 80teborg& Sweden*
(#iu et al*& ,,!) #iu* /*& 9owe D*& Mellor .*9* and /enkins M*C*& [Thermal
analysis of permanent magnet machines&\ 2n the Sixth
International Conference on Electrical Machines and Drives& ;o*
!->& September "@)& ,,!& pp* !=,@!>+*
(Mellor et al*& ,,) Mellor .*9*& Roberts D* and Turner D*R*& [#umped parameter
thermal model for electrical machines of T1'D design&\ 2n IEEE
Proceedings B Electric power applications, 5ol* !"& 2ss* =& ,,&
pp* ()=@("*
(Metropolia& ()),) 9elsinki Metropolia $ni%ersity of Applied Sciences& ()),&
Metropolia Electric Race-About media archive, J:nlineL&
JAccessed ) /une ()(L& www*raceabout*fi*
(Miller& ,,+) 9endershot /*R* /r* and Miller T*/*1*& ,,+& Design of Brushless
Permanent-Magnet motors& :xfordB Magna .hysics .ublishing
and Dlaredon .ress*
(;eorem +,!& ())") ;eorem ())"a& Neorem magnet Ne493a data sheet, J:nlineL&
JAccessed ) /une ()(L& A%ailable from www*neorem*fi.
(;eorem -=!& ())") ;eorem ())"b& Neorem magnet Ne753a data sheet, J:nlineL&
JAccessed ) /une ()(L& A%ailable from www*neorem*fi.
=


(2ncropera& ())-) 2ncropera '*.* and DeHitt D* .*& ())-& Fundamentals of Heat and
Mass Transfer& >
th
ed*& ;ew ?orkB /ohn Hiley k Sons& 2nc*
(/ung et al*& ())-) /ung /<H*& 9ong /<.* and Cim ?<C*& [Dharacteristic Analysis and
Domparison of 2.MSM for 915 According to .ole and Slot
Dombination&\ 2n Jehicle Power and Propulsion Conference,
())-& pp* --"@-"!*
(/ussila& ())) /ussila 9*& ())& Concentrated Winding multiphase permanent
magnet machine design and electromagnetic properties - case
axial flux machine& Acta $ni%ersitatis #appeenrantaensis ;o* !-+&
Doctoral thesis& #appeenranta $ni%ersity of Technology*
(Cylander& ,,=) Cylander& 8*& ,,=& Thermal modeling of small cage induction
motors& Technical report ;o* (>=& Dhalmers $ni%ersity of
Technology& 80teborg& Sweden& 'ebruary ,,=*
(#arminie& ())!) #arminie /* and #owry /*& ())!& Electric Jehicle Technologv
Explained& DhichesterB /ohn Hiley k Sons& #td*
(#indstr0m& ,,,) #indstr0m /*& ,,,& Thermal Model of a Permanent Magnet
Motor for A Hvbrid Electric Jehicle& Technical report& Dhalmers
$ni%ersity of Technology& 80teborg& Sweden*
(#iu et al*& ,,!) #iu* /*& 9owe D*& Mellor .*9* and /enkins M*C*& [Thermal
analysis of permanent magnet machines&\ 2n the Sixth
International Conference on Electrical Machines and Drives& ;o*
!->& September "@)& ,,!& pp* !=,@!>+*
(Mellor et al*& ,,) Mellor .*9*& Roberts D* and Turner D*R*& [#umped parameter
thermal model for electrical machines of T1'D design&\ 2n IEEE
Proceedings B Electric power applications, 5ol* !"& 2ss* =& ,,&
pp* ()=@("*
(Metropolia& ()),) 9elsinki Metropolia $ni%ersity of Applied Sciences& ()),&
Metropolia Electric Race-About media archive, J:nlineL&
JAccessed ) /une ()(L& www*raceabout*fi*
(Miller& ,,+) 9endershot /*R* /r* and Miller T*/*1*& ,,+& Design of Brushless
Permanent-Magnet motors& :xfordB Magna .hysics .ublishing
and Dlaredon .ress*
(;eorem +,!& ())") ;eorem ())"a& Neorem magnet Ne493a data sheet, J:nlineL&
JAccessed ) /une ()(L& A%ailable from www*neorem*fi.
(;eorem -=!& ())") ;eorem ())"b& Neorem magnet Ne753a data sheet, J:nlineL&
JAccessed ) /une ()(L& A%ailable from www*neorem*fi.
=(


(;erg& et al*& ())) ;erg /*& Rilla M* and .yrh0nen /*& [Thermal Analysis of Radial<
'lux 1lectrical Machines with a 9igh .ower Density&\ IEEE
Transactions on Industrial Electronics& 5ol* ==& 2ss* )& ())& pp*
!=+!@!==+*
(;eudorfer et al*& ())") ;eudorfer& 9*& Hicker& ;* and 7inder A*& [Domparison of Three
Different 1lectric .owertrains for the $se in 9ybrid 1lectric
5ehicles&\ 2n IET Conference on Power Electronics& Machines
and Drives& ())"& pp* =)@=+*
(:hyama et al*& ())>)* :hyama C*& ;ashed M*;*'*& Aso C*& 'u3ii 9* and $ehara 9*&
[Design using 'inite 1lement Analysis of Switched Reluctance
Motor for 1lectric 5ehicle&\ 2n ICTTA& ())>& pp* -(-@-!(*
(.olinder et al*& ,,-) .olinder 9* and 9oe3imakers M*/*& [1ddy<current losses in the
permanent magnets of a .M machine&\ 2n the Eighth International
Conference on Electrical Machines and Drives& ,,-& pp* !"@
+(*
(.ost& ()))) .ost R*'* and Ryuto% D*D*& [The 2nductrackB a simpler approach
to magnetic le%itation&\ IEEE Transactions on Applied
Superconductivitv& 5ol* )& 2ss* & ()))& pp* ,)@,)+*
(.rofumo et al*& ,,-) .rofumo& '*& Thang T* and Tenconi& A*& [Axial flux machines
dri%esB a new %iable solution for electric cars&\ IEEE Transactions
on Industrial Electronics& 5ol*++& 2ss* & ,,-& pp* !,@+=*
(.yrh0nen et al*& ())") .yrh0nen /*& /okinen T* and 9rabo%ko%a 5*& ())"& Design of
Rotating Electrical Machines& DhichesterB /ohn Hiley k Sons*
(.yrh0nen et al*& ())) .yrh0nen /*& Ruuskanen 5*& ;erg /*& .uranen& /* and /ussila& 9*&
[.ermanent<Magnet #ength 1ffects in AD Machines&\ IEEE
Transactions on Magnetics& 5ol* +>& 2ss* )& ())& pp* !-"!@
!-",*
(Nionghua et al*& ())!) Nionghua T*& Shuanghong H*& Thiyuan M*& Hei 8* and ?ihui N*&
[Design of a =)kH Switched Reluctance Machine for 915
.ropulsion System&\ 2n the 58th IEEE Conference on Jehicular
Technologv& 5ol* =& :ctober >@,& pp* !()-@!(*
(Ramamurthy& ())) Ramamurthy S*S* and 7alda /*D*& [SiGing a Switched Reluctance
Motor for electric %ehicles&\ IEEE Transactions on Industrv
Applications& 5ol* !-& 2ss* =& ())& pp* (=>@(>+*
(Richter& ,>() Richter R*& ,>(& Elektrische Maschinen. Svnchronmaschinen und
Einankerumformer (1lectrical machinesB Synchronous machines
and Rotary Don%erters) 22& 7asle and StuttgartB 7irkh6user 5erlag*
=(


(;erg& et al*& ())) ;erg /*& Rilla M* and .yrh0nen /*& [Thermal Analysis of Radial<
'lux 1lectrical Machines with a 9igh .ower Density&\ IEEE
Transactions on Industrial Electronics& 5ol* ==& 2ss* )& ())& pp*
!=+!@!==+*
(;eudorfer et al*& ())") ;eudorfer& 9*& Hicker& ;* and 7inder A*& [Domparison of Three
Different 1lectric .owertrains for the $se in 9ybrid 1lectric
5ehicles&\ 2n IET Conference on Power Electronics& Machines
and Drives& ())"& pp* =)@=+*
(:hyama et al*& ())>)* :hyama C*& ;ashed M*;*'*& Aso C*& 'u3ii 9* and $ehara 9*&
[Design using 'inite 1lement Analysis of Switched Reluctance
Motor for 1lectric 5ehicle&\ 2n ICTTA& ())>& pp* -(-@-!(*
(.olinder et al*& ,,-) .olinder 9* and 9oe3imakers M*/*& [1ddy<current losses in the
permanent magnets of a .M machine&\ 2n the Eighth International
Conference on Electrical Machines and Drives& ,,-& pp* !"@
+(*
(.ost& ()))) .ost R*'* and Ryuto% D*D*& [The 2nductrackB a simpler approach
to magnetic le%itation&\ IEEE Transactions on Applied
Superconductivitv& 5ol* )& 2ss* & ()))& pp* ,)@,)+*
(.rofumo et al*& ,,-) .rofumo& '*& Thang T* and Tenconi& A*& [Axial flux machines
dri%esB a new %iable solution for electric cars&\ IEEE Transactions
on Industrial Electronics& 5ol*++& 2ss* & ,,-& pp* !,@+=*
(.yrh0nen et al*& ())") .yrh0nen /*& /okinen T* and 9rabo%ko%a 5*& ())"& Design of
Rotating Electrical Machines& DhichesterB /ohn Hiley k Sons*
(.yrh0nen et al*& ())) .yrh0nen /*& Ruuskanen 5*& ;erg /*& .uranen& /* and /ussila& 9*&
[.ermanent<Magnet #ength 1ffects in AD Machines&\ IEEE
Transactions on Magnetics& 5ol* +>& 2ss* )& ())& pp* !-"!@
!-",*
(Nionghua et al*& ())!) Nionghua T*& Shuanghong H*& Thiyuan M*& Hei 8* and ?ihui N*&
[Design of a =)kH Switched Reluctance Machine for 915
.ropulsion System&\ 2n the 58th IEEE Conference on Jehicular
Technologv& 5ol* =& :ctober >@,& pp* !()-@!(*
(Ramamurthy& ())) Ramamurthy S*S* and 7alda /*D*& [SiGing a Switched Reluctance
Motor for electric %ehicles&\ IEEE Transactions on Industrv
Applications& 5ol* !-& 2ss* =& ())& pp* (=>@(>+*
(Richter& ,>() Richter R*& ,>(& Elektrische Maschinen. Svnchronmaschinen und
Einankerumformer (1lectrical machinesB Synchronous machines
and Rotary Don%erters) 22& 7asle and StuttgartB 7irkh6user 5erlag*
=!


(Rilla& ())>) Rilla M&& ())>& [Thermal modelling of permanent magnet
synchronous machine\& MasterFs thesis& #appeenranta uni%ersity
of Technology*
(Rooks& ())+) Rooks /*A* and Hallace A*C* (())+)& [1nergy efficiency of
5SDs&\ IEEE Industrv Applications Maga:ine )& pp* =-@>*
(Ruotsalainen& ()() Ruotsalainen S*& ()(& Afoneuvofen fa tvkoneiden voimansiirron
kehittminen& JDe%elopment of %ehicle and working machine
transmissionL& 'inal report (in 'innish)& 9elsinki Metropolia
$ni%ersity of Applied Sciences*
(Russenschuck& ())) Russenschuck S*& ())& Field computation for Accelerator
Magnets& HeinheimB Hiley<5D9 5erlag 8mb9 k Do* C8aA*
(Saari& ,,=) Saari /*& ,,=& Thermal Modelling of High speed induction
machines& Acta .olytechnica Scandina%ica& 1lectrical 1ngineering
series ;o* "(& 1spoo*
(SC'& ()() SC'& ()(& J:nlineL& JAccessed ) :ctober ()(L& A%ailable
from www*skf*com
(Shidore et al*& ())-) Shidore& ;*& 7ohn& T*& Duoba& M*& #ohse<7usch& 9* and Sharer&
.*& l.915 ^all<electric range^ and fuel economy in charge<
sustaining mode for low S:D operation of the /DS 5#+ M #i<
ion battery using battery 92#&l 2n the 23rd International Electric
Jehicle Svmposium (EJS23)& Anaheim& DA& Dec* ())-*
(Staton et al*& ())=) Staton& D*& 7oglietti& A* and Da%agnino& A* [Sol%ing the More
Difficult Aspects of 1lectric Motor Thermal Analysis in Small
and Medium SiGe 2ndustrial 2nduction Motors&\ IEEE
Transactions on Energv Conversion& 5ol* ()& 2ss* !& ())=& pp*
>()@>("
(SteinmetG& ,"+) SteinmetG D*.*& [:n the #aw of 9ysteresis&\ 2n IEEE
proceedings& 5ol* -(& 2ss* (& ,"+& pp* ,-@((*
(Stumberger et al*& ())!) Stumberger& 7*& Stumberger& 8*& Dolinar& D*& 9amler& A* and
Trlep& M*& [1%aluation of saturation and cross<magnetiGation
effects in interior permanent<magnet synchronous motor&\ IEEE
Transactions on Industrv Applications& 5ol* !,*& 2ss* =*& ())!&
pp*(>+@(-*
(TR7& ())>) Transportation Research 7oard of the ;ational Academies& [Tires
and .assenger 5ehicle 'uel 1conomy\& Special report (">& $SA&
())>*
=!


(Rilla& ())>) Rilla M&& ())>& [Thermal modelling of permanent magnet
synchronous machine\& MasterFs thesis& #appeenranta uni%ersity
of Technology*
(Rooks& ())+) Rooks /*A* and Hallace A*C* (())+)& [1nergy efficiency of
5SDs&\ IEEE Industrv Applications Maga:ine )& pp* =-@>*
(Ruotsalainen& ()() Ruotsalainen S*& ()(& Afoneuvofen fa tvkoneiden voimansiirron
kehittminen& JDe%elopment of %ehicle and working machine
transmissionL& 'inal report (in 'innish)& 9elsinki Metropolia
$ni%ersity of Applied Sciences*
(Russenschuck& ())) Russenschuck S*& ())& Field computation for Accelerator
Magnets& HeinheimB Hiley<5D9 5erlag 8mb9 k Do* C8aA*
(Saari& ,,=) Saari /*& ,,=& Thermal Modelling of High speed induction
machines& Acta .olytechnica Scandina%ica& 1lectrical 1ngineering
series ;o* "(& 1spoo*
(SC'& ()() SC'& ()(& J:nlineL& JAccessed ) :ctober ()(L& A%ailable
from www*skf*com
(Shidore et al*& ())-) Shidore& ;*& 7ohn& T*& Duoba& M*& #ohse<7usch& 9* and Sharer&
.*& l.915 ^all<electric range^ and fuel economy in charge<
sustaining mode for low S:D operation of the /DS 5#+ M #i<
ion battery using battery 92#&l 2n the 23rd International Electric
Jehicle Svmposium (EJS23)& Anaheim& DA& Dec* ())-*
(Staton et al*& ())=) Staton& D*& 7oglietti& A* and Da%agnino& A* [Sol%ing the More
Difficult Aspects of 1lectric Motor Thermal Analysis in Small
and Medium SiGe 2ndustrial 2nduction Motors&\ IEEE
Transactions on Energv Conversion& 5ol* ()& 2ss* !& ())=& pp*
>()@>("
(SteinmetG& ,"+) SteinmetG D*.*& [:n the #aw of 9ysteresis&\ 2n IEEE
proceedings& 5ol* -(& 2ss* (& ,"+& pp* ,-@((*
(Stumberger et al*& ())!) Stumberger& 7*& Stumberger& 8*& Dolinar& D*& 9amler& A* and
Trlep& M*& [1%aluation of saturation and cross<magnetiGation
effects in interior permanent<magnet synchronous motor&\ IEEE
Transactions on Industrv Applications& 5ol* !,*& 2ss* =*& ())!&
pp*(>+@(-*
(TR7& ())>) Transportation Research 7oard of the ;ational Academies& [Tires
and .assenger 5ehicle 'uel 1conomy\& Special report (">& $SA&
())>*
=+


(Tutkun& ())() Tutkun ;* and Moses A*/*& [Measurement of power loss
distribution in a typical stator core under .HM %oltage
excitation&\ Journal of Magnetism and Magnetic Materials& ;o*
(>(& ())(& pp* (!)@(!+*
(5ogt& ,"!) 5ogt C*& ,"!& Elektrische Maschinen. Berechnung rotierender
elektrischer Maschinen (1lectrical machines* Design of rotating
electrical machines)& !
rd
edition& 7erlinB 517 5erlag 8mb9*
(Hatanabe et al*& ())-) Hatanabe& C*& Cawasaki& Aida& S*& ComatsuGaki& A* and Miki& 2*&
[Dri%ing force characteristics of +)kH switched reluctance motor
for electric %ehicle&\ 2n ICEMS& ())-& pp* ",+@","*
(H18& ()),) Technical guide @ 2nduction motors fed by .HM freAuency
in%erters& J:nlineL& JAccessed ) /une ()(L& A%ailable from
www*weg*net*
(Hu et al*& ())() Hu H*& Dunlop /*7* and Dollocott S*/*& [Design optimiGation of
switched reluctance motor by electromagnetic and thermal finite
element analysis&\ 2n INTERMAG, GS-01& ())!*
(Iia& ,,") Iia& H* and Dhin& .*& [A Specially Designed 15 .M Motor
Dri%e&\ 2n Proceedings of Energv management and Power
Deliverv& 5ol* (*& ,,"& pp* -"@-((*
(Iingming et al*& ()),) Iingming T*& Iuhui H*& 'eng T*& Tao '* and Iinhua 8*& [An
optimal method of power parameters^ matching design in DM.M
based 915&\ 2n ICEMS& ()),& pp* @>*
(Iiong et al*& ())") Iiong& H*& Hu& T* ?in& D*& Dhen& #*& [1conomical comparison
of three hybrid electric car solutions&\ 2n IEEE Jehicle Power and
Propulsion Conference& Shanghai& !@= September& ())"& pp*@>*
(Teraouila et al*& ())>) Teraouila& M*& 7enbouGid& M*1*9* and Diallo& D*& [1lectric motor
dri%e selection issues for 915 propulsion systemsB a comparati%e
study&\ IEEE Transactions on Jehicular Technologv& ())>&
pp*-=>@->+*

=+


(Tutkun& ())() Tutkun ;* and Moses A*/*& [Measurement of power loss
distribution in a typical stator core under .HM %oltage
excitation&\ Journal of Magnetism and Magnetic Materials& ;o*
(>(& ())(& pp* (!)@(!+*
(5ogt& ,"!) 5ogt C*& ,"!& Elektrische Maschinen. Berechnung rotierender
elektrischer Maschinen (1lectrical machines* Design of rotating
electrical machines)& !
rd
edition& 7erlinB 517 5erlag 8mb9*
(Hatanabe et al*& ())-) Hatanabe& C*& Cawasaki& Aida& S*& ComatsuGaki& A* and Miki& 2*&
[Dri%ing force characteristics of +)kH switched reluctance motor
for electric %ehicle&\ 2n ICEMS& ())-& pp* ",+@","*
(H18& ()),) Technical guide @ 2nduction motors fed by .HM freAuency
in%erters& J:nlineL& JAccessed ) /une ()(L& A%ailable from
www*weg*net*
(Hu et al*& ())() Hu H*& Dunlop /*7* and Dollocott S*/*& [Design optimiGation of
switched reluctance motor by electromagnetic and thermal finite
element analysis&\ 2n INTERMAG, GS-01& ())!*
(Iia& ,,") Iia& H* and Dhin& .*& [A Specially Designed 15 .M Motor
Dri%e&\ 2n Proceedings of Energv management and Power
Deliverv& 5ol* (*& ,,"& pp* -"@-((*
(Iingming et al*& ()),) Iingming T*& Iuhui H*& 'eng T*& Tao '* and Iinhua 8*& [An
optimal method of power parameters^ matching design in DM.M
based 915&\ 2n ICEMS& ()),& pp* @>*
(Iiong et al*& ())") Iiong& H*& Hu& T* ?in& D*& Dhen& #*& [1conomical comparison
of three hybrid electric car solutions&\ 2n IEEE Jehicle Power and
Propulsion Conference& Shanghai& !@= September& ())"& pp*@>*
(Teraouila et al*& ())>) Teraouila& M*& 7enbouGid& M*1*9* and Diallo& D*& [1lectric motor
dri%e selection issues for 915 propulsion systemsB a comparati%e
study&\ IEEE Transactions on Jehicular Technologv& ())>&
pp*-=>@->+*

==


A..1;D2I 2B Test eAuipment
Magtroll TM !+ ())) ;m inline torAue transducer
< combined error of linearity and hysteresis W )*P of rated torAue
< Temperature influence on the Geroasensiti%ity W )*P of rated torAueE) C
< 2nfluence of speed on the Gero torAue signal W )*)P of rated torAue E))) min
<

< #ong term stability of sensiti%ity W )*)=P of rated torAue E year
Magtroll TM >+)) Transducer display
< Accuracy of speed measurement )*)P of reading
< Accuracy of TorAue )*)P of range ()5)
?okogawa .G<+))) .ower analyGer
< Temperature (!MD !MD
< AccuracyB
Table 2* ?okogawa .T<+))) accuracy table
'reAuencies 5oltageEDurrent .ower
DD m()*(P of rdg g )*P of rng) m()*(P of rdg g )*P of rng)
)* 9G W f W ) 9G m()*(P of rdg g )*P of rng) m()*(P of rdg g )*)=P of rng)
) 9G W f W += 9G m()*(P of rdg g )*)=P of rng) m()*(P of rdg g )*)(=P of rng)
+= 9G W f W k9G m()*P of rdg g )*)=P of rng) m()*P of rdg g )*)(=P of rng)
k9G W f W ) k9G m()*P of rdg g )*)=P of rng) m()*P of rdg g )*)+P of rng)
) k9G W f W n =) k9G m()*(P of rdg g )*P of rng) m()*(P of rdg g )*)=P of rng)
=) k9G W f W )) k9G m()*>P of rdg g )*(P of rng) m()*>P of rdg g )*P of rng)
)) k9G W f W ()) k9G m()*>P of rdg g )*(P of rng) m(*=P of rdg g )*=P of rng)
()) k9G W f W +)) k9G m(P of rdg g )*(P of rng) m(*=P of rdg g )*=P of rng)
+)) k9G W f W =)) k9G
mJ()* g )*))>f)P of rdg g)*(P
of mgL
mJ()* g )*)),f)P of rdg
g)*=Pof rngL
=)) k9G W f W M9G
mJ()* g )*))> Z f)P of rdg g(P
of rngL
mJ()* g )*)),f)P of rdg
g*=Pof rngL
M9G W f W = M9G
mJ()* g )*))> Z f)P of rdg g(P
of rngL
@


==


A..1;D2I 2B Test eAuipment
Magtroll TM !+ ())) ;m inline torAue transducer
< combined error of linearity and hysteresis W )*P of rated torAue
< Temperature influence on the Geroasensiti%ity W )*P of rated torAueE) C
< 2nfluence of speed on the Gero torAue signal W )*)P of rated torAue E))) min
<

< #ong term stability of sensiti%ity W )*)=P of rated torAue E year
Magtroll TM >+)) Transducer display
< Accuracy of speed measurement )*)P of reading
< Accuracy of TorAue )*)P of range ()5)
?okogawa .G<+))) .ower analyGer
< Temperature (!MD !MD
< AccuracyB
Table 2* ?okogawa .T<+))) accuracy table
'reAuencies 5oltageEDurrent .ower
DD m()*(P of rdg g )*P of rng) m()*(P of rdg g )*P of rng)
)* 9G W f W ) 9G m()*(P of rdg g )*P of rng) m()*(P of rdg g )*)=P of rng)
) 9G W f W += 9G m()*(P of rdg g )*)=P of rng) m()*(P of rdg g )*)(=P of rng)
+= 9G W f W k9G m()*P of rdg g )*)=P of rng) m()*P of rdg g )*)(=P of rng)
k9G W f W ) k9G m()*P of rdg g )*)=P of rng) m()*P of rdg g )*)+P of rng)
) k9G W f W n =) k9G m()*(P of rdg g )*P of rng) m()*(P of rdg g )*)=P of rng)
=) k9G W f W )) k9G m()*>P of rdg g )*(P of rng) m()*>P of rdg g )*P of rng)
)) k9G W f W ()) k9G m()*>P of rdg g )*(P of rng) m(*=P of rdg g )*=P of rng)
()) k9G W f W +)) k9G m(P of rdg g )*(P of rng) m(*=P of rdg g )*=P of rng)
+)) k9G W f W =)) k9G
mJ()* g )*))>f)P of rdg g)*(P
of mgL
mJ()* g )*)),f)P of rdg
g)*=Pof rngL
=)) k9G W f W M9G
mJ()* g )*))> Z f)P of rdg g(P
of rngL
mJ()* g )*)),f)P of rdg
g*=Pof rngL
M9G W f W = M9G
mJ()* g )*))> Z f)P of rdg g(P
of rngL
@


=>


A..1;D2I 22B Thermal network



Magnethins
.>
Middle piece
.=
Magnethins
.+
Magnethtop
.!
Magnethins
.(
.olea 7and
.
Air gap
.>
Toothh
.-
'rame
.(
Statorhyoke
Toothh+
.+
Toothh!
.=
Toothh(
.,
Slothstick
.!
Axlehax
.
.()
D+
D!
D(
D
D>
D=
D>
D=
D+
D!
D(
D
D(
Hater con%
Slothtan
.)
Slothrad
Rotorhyoke
."
Magnethins
.-
Magnethbottom
.,
D!
D(
D
Slothax
."
.!!a!+
Axlehcontact
.() .(a((
7earinghcontact
D(
7earinghrad
D(
.(!a(+
7earinghcontact
D(
.(=a(>
1ndhshieldhrad
.(-a("
'ramehax
.(,a!)
.!a!(
D(
D(
1ndhwinding
.!=a!>
'luidhin
.!-a!"
'luidhin
.!-a!"
1ndhcon%
D(
D, D,
=>


A..1;D2I 22B Thermal network



Magnethins
.>
Middle piece
.=
Magnethins
.+
Magnethtop
.!
Magnethins
.(
.olea 7and
.
Air gap
.>
Toothh
.-
'rame
.(
Statorhyoke
Toothh+
.+
Toothh!
.=
Toothh(
.,
Slothstick
.!
Axlehax
.
.()
D+
D!
D(
D
D>
D=
D>
D=
D+
D!
D(
D
D(
Hater con%
Slothtan
.)
Slothrad
Rotorhyoke
."
Magnethins
.-
Magnethbottom
.,
D!
D(
D
Slothax
."
.!!a!+
Axlehcontact
.() .(a((
7earinghcontact
D(
7earinghrad
D(
.(!a(+
7earinghcontact
D(
.(=a(>
1ndhshieldhrad
.(-a("
'ramehax
.(,a!)
.!a!(
D(
D(
1ndhwinding
.!=a!>
'luidhin
.!-a!"
'luidhin
.!-a!"
1ndhcon%
D(
D, D,
=-


A..1;D2I 222B Test machine dimensions

>*)
R+)*=
++*)
>*)
+)*)
*>)
R=*=)
++*!M
((*!"
!*=)
((*=)M
*=(
!*))
(*>+
*-)
(>*-)
*>)
R*=)
R*=)
(*!M
R,)*))
((*=)M
((*=)M
R+(*)
=*))
!*=)
>*)) +)*))
*(=M
(,*=>
R+*==
!*>=))
Magnet module (MM) dimensions
1mbedded magnet (1M) dimensions
Stator diemensions
=-


A..1;D2I 222B Test machine dimensions

>*)
R+)*=
++*)
>*)
+)*)
*>)
R=*=)
++*!M
((*!"
!*=)
((*=)M
*=(
!*))
(*>+
*-)
(>*-)
*>)
R*=)
R*=)
(*!M
R,)*))
((*=)M
((*=)M
R+(*)
=*))
!*=)
>*)) +)*))
*(=M
(,*=>
R+*==
!*>=))
Magnet module (MM) dimensions
1mbedded magnet (1M) dimensions
Stator diemensions
ACTA UNIVERSITATIS LAPPEENRANTAENSIS



455. HUJALA, MAIJA. Structural dynamics in global pulp and paper industry. 2011. Diss.

456. KARVONEN, MATTI. Convergence in industry evolution. 2011. Diss.

457. KINNUNEN, TEEMU .Bag-of-features approach to unsupervised visual object categorisation. 2011.
Diss.

458. RUUSKANEN, VESA. Design aspects of megawatt-range direct-driven permanent magnet wind
generators. 2011. Diss.

459. WINTER, SUSANNA. Network effects: scale development and implications for new product
performance. 2011. Diss.

460. JSKELINEN, ANSSI. Integrating user experience into early phases of software development.
2011. Diss.

461. KRIINEN, TOMMI. Polymer surface modification by atomic layer deposition. 2011. Diss.

462. KOCHURA, ALEKSEY. Growth, magnetic and transport properties of InSb and II-IV-As2
semiconductors doped with manganese. 2011. Diss.

463. PUTKIRANTA, ANTERO. Possibilities and challenges of longitudinal studies in operations
management. 2011. Diss.

464. HAPPONEN, ARI.Muuttuvaan kysyntn sopeutuva varastonohjausmalli. 2011. Diss.

465. VASAVA, PARITOSH. Application of computational fluid dynamics in modelling blood flow in human
thoracic aorta. 2011. Diss.

466. PURO, LIISA.Identification of extractives and polysaccharides as foulants in membrane filtration of
pulp and paper mill effluents. 2011. Diss.

467. LAPPALAINEN, PIA.Socially Competent Leadership predictors, impacts and skilling in
engineering. 2012. Diss.

468. PLAMTHOTTATHIL, ANSHY OONNITTAN.Application of electrokinetic Fenton process for the
remediation of soil contaminated with HCB. 2012. Diss.

469. EBRAHIMI, FATEMEH. Synthesis of percarboxylic acids in microreactor. 2012. Diss.

470. JANTUNEN, SAMI.Making sense of software product requirements. 2012. Diss.

471. VILKO, JYRI.Approaches to supply chain risk management: identification, analysis and control.
2012. Diss.

472. TANSKANEN, VESA.CFD modelling of direct contact condensation in suppression pools by
applying condensation models of separated flow. 2012. Diss.

473. HUHTANEN MIKKO. Software for design of experiments and response modelling of cake filtration
applications. 2012. Diss.

474. PARJANEN, SATU. Creating possibilities for collective creativity
Brokerage functions in practice-based innovation. 2012. Diss.

475. KUKKONEN, SAKU. Generalized differential evolution for global multi-objective optimization with
constraints. 2012. Diss.
ACTA UNIVERSITATIS LAPPEENRANTAENSIS



455. HUJALA, MAIJA. Structural dynamics in global pulp and paper industry. 2011. Diss.

456. KARVONEN, MATTI. Convergence in industry evolution. 2011. Diss.

457. KINNUNEN, TEEMU .Bag-of-features approach to unsupervised visual object categorisation. 2011.
Diss.

458. RUUSKANEN, VESA. Design aspects of megawatt-range direct-driven permanent magnet wind
generators. 2011. Diss.

459. WINTER, SUSANNA. Network effects: scale development and implications for new product
performance. 2011. Diss.

460. JSKELINEN, ANSSI. Integrating user experience into early phases of software development.
2011. Diss.

461. KRIINEN, TOMMI. Polymer surface modification by atomic layer deposition. 2011. Diss.

462. KOCHURA, ALEKSEY. Growth, magnetic and transport properties of InSb and II-IV-As2
semiconductors doped with manganese. 2011. Diss.

463. PUTKIRANTA, ANTERO. Possibilities and challenges of longitudinal studies in operations
management. 2011. Diss.

464. HAPPONEN, ARI.Muuttuvaan kysyntn sopeutuva varastonohjausmalli. 2011. Diss.

465. VASAVA, PARITOSH. Application of computational fluid dynamics in modelling blood flow in human
thoracic aorta. 2011. Diss.

466. PURO, LIISA.Identification of extractives and polysaccharides as foulants in membrane filtration of
pulp and paper mill effluents. 2011. Diss.

467. LAPPALAINEN, PIA.Socially Competent Leadership predictors, impacts and skilling in
engineering. 2012. Diss.

468. PLAMTHOTTATHIL, ANSHY OONNITTAN.Application of electrokinetic Fenton process for the
remediation of soil contaminated with HCB. 2012. Diss.

469. EBRAHIMI, FATEMEH. Synthesis of percarboxylic acids in microreactor. 2012. Diss.

470. JANTUNEN, SAMI.Making sense of software product requirements. 2012. Diss.

471. VILKO, JYRI.Approaches to supply chain risk management: identification, analysis and control.
2012. Diss.

472. TANSKANEN, VESA.CFD modelling of direct contact condensation in suppression pools by
applying condensation models of separated flow. 2012. Diss.

473. HUHTANEN MIKKO. Software for design of experiments and response modelling of cake filtration
applications. 2012. Diss.

474. PARJANEN, SATU. Creating possibilities for collective creativity
Brokerage functions in practice-based innovation. 2012. Diss.

475. KUKKONEN, SAKU. Generalized differential evolution for global multi-objective optimization with
constraints. 2012. Diss.

476. LAAKSONEN, JONNA.Tactile-proprioceptive robotic grasping. 2012. Diss.

477. KALLIO, ANNE. Enhancing absorptive capacity in a non-research and development context
An action research approach to converting individual observations into organizational awareness.
2012. Diss.

478. LTTIL, LAURI. Improving transportation and warehousing efficiency with simulation based
decision support systems. 2012. Diss.

479. OYOMNO, WERE.Usable privacy preservation in mobile electronic personality. 2012. Diss.

480. LINNALA, MIKKO.Simulation and optimization tools in paper machine concept design. 2012. Diss.

481. KORPIJRVI, JUHA.Aging based maintenance and reinvestment scheduling of electric distribution
network. 2012. Diss.

482. KORHONEN, JUHAMATTI.Active inverter output filtering methods. 2012. Diss.

483. KLODOWSKI, ADAM. Flexible multibody approach in bone strain estimation during physical activity:
quantifying osteogenic potential. 2012. Diss.

484. VUORENMAA, MARKKU. Osaamisen johtaminen pk-yrityksen kansainvlisen kasvun elinkaarella.
2012. Diss.

485. RAUTIAINEN, MARITA. Dynamic ownership in family business systems a portfolio business
approach. 2012. Diss.

486. LILIUS, REIJO. THE FINNISH IT INDUSTRIES IN TRANSITION Defining and measuring the
Finnish software product and IT services industries by applying theoretical frameworks . 2012. Diss.

487. TUOMINEN, PASI. The purpose of consumer co-operation: implications for the management and
governance of co-operatives. 2012. Diss.

488. SAARI, ESA. Suurnopeus-turbokoneroottoreiden termodynaaminen ja mekaaninen mallinnus sek
rakenneanalyysi. 2012. Diss.

489. PAANANEN, MIKKO. On innovative search: the use of internal and external sources of innovation
among Finnish innovators. 2012. Diss.

490. BELOVA, POLINA. Quasiclassical approach to the vortex state in iron-based superconductors.
2012. Diss.

491. HIETANEN, IIRO. Design and characterization of large area position sensitive radiation detectors.
2012. Diss.

492. PSSIL, ANNE. A reflexive model of research-based theatre Processing innovation of the cross-
road of theatre, reflection and practice-based innovation activities. 2012. Diss.

493. RIIPINEN, TOMI. Modeling and control of the power conversion unit in a solid oxide fuel cell
environment. 2012. Diss.

494. RANTALAINEN, TUOMAS.Simulation of structural stress history based on dynamic analysis. 2012.
Diss.

495. SALMIMIES, RIINA.Acidic dissolution of iron oxides and regeneration of a ceramic filter medium.
2012. Diss.

496. VAUTERIN, JOHANNA JULIA.The demand for global student talent: Capitalizing on the value of
university-industry collaboration. 2012. Diss.

476. LAAKSONEN, JONNA.Tactile-proprioceptive robotic grasping. 2012. Diss.

477. KALLIO, ANNE. Enhancing absorptive capacity in a non-research and development context
An action research approach to converting individual observations into organizational awareness.
2012. Diss.

478. LTTIL, LAURI. Improving transportation and warehousing efficiency with simulation based
decision support systems. 2012. Diss.

479. OYOMNO, WERE.Usable privacy preservation in mobile electronic personality. 2012. Diss.

480. LINNALA, MIKKO.Simulation and optimization tools in paper machine concept design. 2012. Diss.

481. KORPIJRVI, JUHA.Aging based maintenance and reinvestment scheduling of electric distribution
network. 2012. Diss.

482. KORHONEN, JUHAMATTI.Active inverter output filtering methods. 2012. Diss.

483. KLODOWSKI, ADAM. Flexible multibody approach in bone strain estimation during physical activity:
quantifying osteogenic potential. 2012. Diss.

484. VUORENMAA, MARKKU. Osaamisen johtaminen pk-yrityksen kansainvlisen kasvun elinkaarella.
2012. Diss.

485. RAUTIAINEN, MARITA. Dynamic ownership in family business systems a portfolio business
approach. 2012. Diss.

486. LILIUS, REIJO. THE FINNISH IT INDUSTRIES IN TRANSITION Defining and measuring the
Finnish software product and IT services industries by applying theoretical frameworks . 2012. Diss.

487. TUOMINEN, PASI. The purpose of consumer co-operation: implications for the management and
governance of co-operatives. 2012. Diss.

488. SAARI, ESA. Suurnopeus-turbokoneroottoreiden termodynaaminen ja mekaaninen mallinnus sek
rakenneanalyysi. 2012. Diss.

489. PAANANEN, MIKKO. On innovative search: the use of internal and external sources of innovation
among Finnish innovators. 2012. Diss.

490. BELOVA, POLINA. Quasiclassical approach to the vortex state in iron-based superconductors.
2012. Diss.

491. HIETANEN, IIRO. Design and characterization of large area position sensitive radiation detectors.
2012. Diss.

492. PSSIL, ANNE. A reflexive model of research-based theatre Processing innovation of the cross-
road of theatre, reflection and practice-based innovation activities. 2012. Diss.

493. RIIPINEN, TOMI. Modeling and control of the power conversion unit in a solid oxide fuel cell
environment. 2012. Diss.

494. RANTALAINEN, TUOMAS.Simulation of structural stress history based on dynamic analysis. 2012.
Diss.

495. SALMIMIES, RIINA.Acidic dissolution of iron oxides and regeneration of a ceramic filter medium.
2012. Diss.

496. VAUTERIN, JOHANNA JULIA.The demand for global student talent: Capitalizing on the value of
university-industry collaboration. 2012. Diss.