www.ietdl.

org

Published in IET Electric Power Applications

Received on 10th June 2009
Revised on 6th August 2009
doi: 10.1049/iet-epa.2009.0139

ISSN 1751-8660

Computation of additional losses due to rotor

eccentricity in electrical machines
A. Belahcen A. Arkkio
Department of Electrical Engineering, Helsinki University of Technology, P.O. Box 3000, 02015 TKK, Finland
E-mail: anouar.belahcen@tkk.fi

Abstract: This study investigates the effect of eccentric rotor on the power losses in electrical machines. The
investigations are carried out for an induction machine with PWM-voltage supply and in the cases of dynamic
and static eccentricity. The time-stepping two-dimensional finite element method, with coupled field and
circuit equations, is used for the simulations. The iron losses are computed through a dynamic loss model. It
is shown that the operation of the machine under eccentricity fault results in an increase of both resistive
and iron losses. Further the eccentricity causes additional torque modulation and unbalance magnetic pull
that result in additional bearing friction losses and wearing.

1 Introduction study on the effect of eccentricity on the power losses in

electrical machines is obviously missing from the literature.
With the ever increasing need for energy saving, designers of The goal of this paper is to fill this gap by presenting the
electrical machines have been exploring many possibilities methods required to perform such a study and the results
to manufacture energy-efficient machines. Nowadays, this of applying it to a small-size induction machine. More
is being achieved by optimising the machine design with similar investigations on other machine types and sizes are
respect to power losses while keeping the manufacturing still needed to form a complete picture of the problem.
and material costs at a reasonable level. On the other hand,
the role of the users of these machines is being emphasised The problems related to eccentric-rotor electrical machines
too. The efficient use of the machine is being achieved, have been studied for more than a century [1– 8]. All these
among others, by increasing the use of frequency converters studies were concerned with the unbalance magnetic pull
and other energy conditioning devices. One of the (UMP) and were mainly based on analytical methods
unexplored fields in this respect is the effect of extended (rotating field theory). The analytical analysis is a fruitful
use of faulty machines on the losses, especially eccentric- method as it gives an insight into the causes and effects and
rotor machines. allows the designer to act on different design parameters to
achieve a given goal. Recently, the same problems were
Owing to manufacturing tolerances and aging of the studied once again but the investigation tools changed from
electrical machines, they always present a certain amount of analytical to numerical ones as reluctance networks and
eccentricity. The rotor eccentricity of induction machines finite element method (FEM) [9– 16]. The use of FEM
has been studied from many points of view, although the allowed more accurate analysis of the eccentricity and
emphasis was on the methods to detect the eccentricity at broadened the area of study to the investigation of conical
the early stage and in relation with condition monitoring eccentricity, winding design, iron saturation and control
of electrical machines. However, in some cases more than considerations. In previous works [17, 18], we also studied
50% eccentric machines are allowed to operate in many the possibility of detecting static and dynamic eccentricity
fields of industry as a means to reduce maintenance costs. from the vibrations of induction motors with an emphasis
However, this may lead to an increase in the operation cost on its use in condition monitoring. However, none of the
through increased energy losses and decreased efficiency of above authors paid any attention to the problem of increased
the machine and the frequency converter. An extensive power losses because of the eccentricity.

IET Electr. Power Appl., 2010, Vol. 4, Iss. 4, pp. 259– 266 259
doi: 10.1049/iet-epa.2009.0139 & The Institution of Engineering and Technology 2010
www.ietdl.org

In this paper, we investigate the effect of static and dynamic the total current through it i, and its dc-resistance R
eccentricity on the losses in an induction motor fed from ð
sinusoidal and PWM voltage sources. The investigation @A
u ¼ Ri þ R s dS (3)
method is described and the results from extensive S @t
simulations are presented. The core losses are computed
with a novel method [19] that allows for time tracking of The windings and rotor cage circuit equations are formed by
these losses. This is very important if we want to access the applying Kirchhoff’s laws and (3). The field in the end
phenomena behind the increasing losses in the iron core of windings is modelled by constant impedances in the circuit
the machine. The losses are separated into core and resistive equations. The circuit equations are simultaneously solved
losses in both the stator and rotor of the machine. Different with the field equations. A simplifying assumption is made
levels of eccentricity are simulated. Furthermore, the total on the current density in the stator winding that is assumed
forces acting on the eccentric rotor are presented for the to be uniform on each coil side. This is justified by the use of
reader to obtain an estimate of the increased friction losses. parallel and series connected thin wires to manufacture such a
winding. If the stator winding is form-wound, the method
Although a better estimate of the iron losses can be proposed in [21] is preferable. This method has not been
achieved with a dynamic hysteresis model of the magnetic applied here for the above-mentioned reasons. A complete
material incorporated into finite element (FE) routines description of the FE model used in this work is given in [22].
[20], this option has not been used in this paper for two
reasons. It requires a huge computation time and it is not The above equations are discretised in space and time.
possible to apply it in other commercial software. This last The obtained non-linear system of equations is then solved
option is an essential condition for the methods presented iteratively using Newton – Raphson iteration method.
here to be applied by other designers using commercial FE
software and not developing their own as is the case in
most manufacturing industry.
2.2 Motion and eccentricity
The field and winding circuit equations are written for the stator
and rotor separately and each in its own reference frame. The
2 Computational model field equations are then forced to match in the air gap through
additional air gap equations that are solved simultaneously
2.1 Time stepping FE with the other cross-sectional equations. The motion of the
In this study, the magnetic field in the cross-section of the rotor is modelled by re-creating the mesh in the air gap after
electrical machine is calculated using the two-dimensional rotating the rotor with an angle proportional to the time step
(2D) FEM with the time-stepping scheme. The FEM is and speed. This procedure makes it possible to keep the same
based on the A  F formulation of the magnetic field (A is mesh in both the stator and rotor, whereas only the air gap
the magnetic vector potential and F the reduced scalar mesh is changed from step to step. The algorithm for
potential) and the Crank– Nicholson method is used for meshing the air gap after each time step is rather simple
time integration. The electrical machine is treated as a because of the regular geometry of the machine and the fact
quasi-static magnetic system. The equation to be solved in that the nodes on the boundary of the air gap (stator and rotor
the cross-section geometry of the machine is sides) are prescribed. The eccentricity of the rotor (Fig. 1) is
simulated by either moving the rotor or the stator mesh with
@A a fixed displacement from its original position to obtain,
r  (nr  A) ¼ s  srF (1) respectively, dynamic and static eccentricity. Thus the only
@t
affected part of the models mesh is still the air gap.
In the simulations of an electrical machine, the stator and
rotor cores are made of laminations and their conductivities 2.3 Forces and torque
are set to zero so that the right-hand term of (1) is null.
The reluctivity of iron, n is a non-linear function of the The total force acting on the eccentric rotor at each time step
magnetic vector potential, it is given in the program as a is computed in x- and y-component form. The computation
cubic spline of the square of the amplitude of the magnetic is based on the method of local Jacobian derivative [23],
flux density vector. In conductors such as the rotor bars where a given component of the force is
and stator windings, the reluctivity is that of vacuum and X ð  T 1 @G ð kH k
T 1 @jGj

the gradient of the scalar potential can be defined as Fu ¼ B G Hþ B dH jGj dV
airgap Ve @u 0 @u
elements
u
rF ¼  ez (2) (4)
l
with G is the Jacobian matrix for the transformation from the
where u is the voltage over the conductor and l the conductor reference element to the actual one, jGj its determinant, and u
length. Integrating the current density over the cross-section stands for either x or y. H and B are, respectively, the magnetic
of the conductors leads to a relation between the voltage u, field strength and the magnetic flux density vectors. Detailed

260 IET Electr. Power Appl., 2010, Vol. 4, Iss. 4, pp. 259– 266
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-epa.2009.0139
 www.ietdl.org
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
In (7) – (9), kBk ¼ Bx2 þ By2 is the amplitude of the flux
density vector and u ¼ arctg(By =Bx ) is its angular position.
Bs is the saturation flux density. @ stands for an
infinitesimal variation of the quantities and kc , ke , kh , kr
and b are constants of the material that describe different
loss phenomena.

The power losses are estimated from the energy losses by

dividing by @t and integrating over the volume of the iron
core. In the 2D model, the integration is made over the
Figure 1 Illustration of the static and dynamic eccentricity cross-sectional area and multiplied by the axial length of
When the rotor rotates around its geometrical axis (left), the the machine.
eccentricity is static, whereas the eccentricity is dynamic if the
rotor rotates around any other axis (right) while displaced from The methods for field solution presented above have been
the centre axis of the stator. In electrical machines, both
previously validated through extensive measurements on a
eccentricities coexist
37 kW cage induction machine fed from sinusoidal and
voltage source frequency converter [22]. The method
implementation of (4) is given in [23]. A similar equation based
for loss computation has also been validated through
on the method of local Jacobian derivative from [23] is also used
measurements on the same machine under different
for the computations of the electromagnetic torque.
supply and load conditions [19]. As an example of these
validations, the comparison between the computed and
2.4 A posteriori loss model measured total losses of the machine under no-load (NL)
The electromagnetic loss in the windings and cores of the and full-load (FL) and at different supplies and frequencies
electrical machine are computed step by step from the field (sinusoidal and frequency converter) is given in Fig. 2.
solution and the current densities at each time step. The Some differences between measurements and computations
resistive winding losses are naturally taken into account in are naturally present and can be attributed to many reasons
the field solution because of the coupling between the field (inaccuracy of the field solution because of the 2D
and circuit equations, whereas the core losses are not. This approach, limited accuracy of the measurements mainly
is an a posteriori core loss computation. because of the difficulties related to friction losses etc.).
However, regarding all the assumptions, the agreement
The winding losses are calculated from the computed between measurements and computations is rather
current density J as encouraging.
ð 2
J
Pcu ¼ dw (5)
w s 3 Computational results
The core losses are separated into three components, namely
A 35 kW induction machine with the parameters shown in
the classical eddy-current, the alternating excess and the
Table 1 has been simulated using the methods described
hysteresis losses. The hysteresis losses include a part that is
above. The goal of the simulations was to assess the effect
related to rotational losses.
of the eccentricity on different working quantities such as
The core-loss computation model is presented in [19].
The model is based on energy considerations; thus the
energy lost in the iron core of the machine when the flux
density vector undergoes an infinitesimal variation @B is

@w ¼ @wc þ @we þ @wh (6)

where the classical eddy current, the alternating excess and

the hysteresis energy losses are, respectively
 
@B
@wc ¼ kc  
 @t k@Bk (7)

 
@kBk 1=2

@we ¼ ke   j@kBkj (8) Figure 2 Comparison between the computed total losses
@t  and measured ones at different voltage sources,
loads and frequencies (NL ¼ no-load, FL ¼ full-load,
1  (kBk=Bs )
@wh ¼ kh kBkj@kBkj þ kr kBkj@uj (9) sin ¼ sinusoidal voltage supply and fc ¼ frequency
1þ b(1  (kBk=Bs ))2 converter voltage supply)

IET Electr. Power Appl., 2010, Vol. 4, Iss. 4, pp. 259– 266 261
doi: 10.1049/iet-epa.2009.0139 & The Institution of Engineering and Technology 2010
www.ietdl.org

Table 1 Parameters of the simulated induction machine

Parameter Value Unit

rated power 35 kW
rated voltage 400 V
rated current 64 A
rated slip 3 %
rated frequency 100 Hz
number of poles 4
number of stator slots 48
number of rotor bars 40
Figure 4 Total losses against static and dynamic
connection star
eccentricity levels
Similar trends are visible. The dynamic case shows less loss than
currents, voltages, torque, efficiency and losses both in the the corresponding static one
windings and iron cores. All the simulations are made with
the machine at FL (3% slip). computation, (4) was applied in a one-layer air gap mesh.
The computation of the torque was also carried out on a
In the simulations the machine was fed from a PWM-like single layer air gap mesh. The stator, rotor and air gap
voltage supply generated by the conventional comparison of meshes are shown in Fig. 3. Inserts are added with
the sinusoidal 100 Hz fundamental wave and the triangular zooming at the location of the shortest and longest air gap
carrier wave of 6 kHz. Before the comparison, the third in the case of maximum eccentricity (70%). The regularity
harmonic (300 Hz) with amplitude equal to the sixth of of the air gap mesh even at large eccentricities justifies our
the fundamental harmonic was injected into the use of single layer for the force and torque computations.
fundamental wave to cancel some of the high harmonics
from the final supply wave (this is a feature used by many The computed total losses as function of the eccentricity
manufacturers of PWM frequency converter). To obtain level are shown in Fig. 4 and the efficiency in Fig. 5. The
a steady state as quickly as possible, the time harmonic time-varying eddy-current losses in the iron core of the
approach was used to compute the initial field in the stator are shown in Fig. 6. The peaks in the eddy-current
machine. Further, during the first period of the losses are due to the abrupt change in the amplitude of the
computations, the PWM supply voltage was modulated by voltage supply. This fact is illustrated in Fig. 7, where
a sinusoidal wave of the same frequency as the fundamental the amplitude of the voltage space vector is correlated with
one to minimise the computational transients. A total of the eddy-current losses of the stator. A low value of the
four periods corresponding to 40 ms was computed from voltage results in low values of the field in the machine and
which the results of the two last periods are presented here. thus low eddy-current losses. The increase of the voltage
The simulations solved for the currents and torque among over the rated one results in high values of the field that
other quantities of the machine. For the purpose of forces causes saturation and high values of the eddy-current

Figure 3 Stator, rotor and air gap meshes

A zoom in of the air gap mesh at the smallest and largest locations is shown to justify the use of single layer for the force and torque
computation as well as to exclude the possibility of mesh induced artefacts in the loss computation

262 IET Electr. Power Appl., 2010, Vol. 4, Iss. 4, pp. 259– 266
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-epa.2009.0139
 www.ietdl.org

losses. The separation of losses has been performed in [24].

The losses have been separated to stator resistive and iron
as well as rotor resistive and iron losses. Furthermore, the
iron losses have been separated into their different
components, namely eddy-current, hysteresis and excess
losses. The effects of eccentricity on torque and current
modulation have been presented in [24] too. Fig. 8 shows
different components of the UMP as a function of time.
For these results, the computations were continued up to
(11) and (16) periods of the supply, respectively, for static
and dynamic eccentricity. Indeed, owing to the slip, the
relative position of the minimum air gap and the rotating
Figure 5 Efficiency of the simulated machine under field is varying continuously; thus a better picture of the
different eccentricities UMP, especially in the case of dynamic eccentricity, is
Similar trends can be observed here also with the static case being
reached after the simulation of a large number of periods.
more severe Fig. 9 shows the loci of the total force acting on the
eccentric rotor and its trajectory from the beginning to the
end of the computations for both dynamic and static
eccentricity at a level of 10%. The increase of the rotor
force with eccentricity is what causes bearing friction losses
and wearing. At the beginning of the computations for
Figs. 8 and 9, the rotor was displaced gradually from its
centre position to its final position or orbit during the first
period of the supply. This procedure reduces the
computational transient and enhanced the steady-state
accuracy, but it also means that the results of the first
computation period are meaningless. Fig. 10 shows the
difference in the computed torque between the case of
non-eccentric machine and the case when the machine was
operating at 60% dynamic eccentricity. These results have
been obtained by subtracting the computed torque at the
Figure 6 Eddy-current losses in the stator iron core for non- two cases from each other. A modulation of the torque is
eccentric rotor and different eccentricity cases as function observed, although the average change in the torque is
of time almost zero. This means that although the performance of
Although the trends are the same, a severe increase at some the machine in terms of torque is not affected by the
instants of time can be observed. The presence of the peaks is eccentricity, this latter causes undesirable torque oscillations
explained below in Fig. 7 that might produce vibrations and acoustic noise.

Figure 7 Correlation of the eddy-current losses in the stator Figure 8 Force components acting on the rotor of the
core with the amplitude of the space vector of the supply machine under 10% static and dynamic eccentricity
voltage operation at FL
The machine was fed by PWM at full load in the presence of In the case of static eccentricity the force keeps in the same
no eccentricity. Low-voltage level results in decreased losses, direction but in the case of dynamic eccentricity the direction of
whereas high-voltage levels result in saturation and increased the force depends on the rotor position (this is better seen in
losses Fig. 9)

IET Electr. Power Appl., 2010, Vol. 4, Iss. 4, pp. 259– 266 263
doi: 10.1049/iet-epa.2009.0139 & The Institution of Engineering and Technology 2010
www.ietdl.org

Figure 9 Loci of the force acting on the rotor (left) and trajectory of the rotor centre (right) for different eccentricities
The simulations were started with the rotor at its centre position then gradually displaced to its final position or orbit. Although the rotor
was displaced along the x-axis, the static eccentricity produced forces on the y-axis too because of the slip

will produce additional losses and cause UMP that

contribute to the aging of the machine and reduce its
lifetime. Moreover, the efficiency of the machine will be
reduced resulting in increased energy costs. The computed
efficiency in Fig. 5 represents only the electromagnetic
one. The actual reduction in the efficiency is much higher
because of the increase in the bearing losses with increasing
UMP. The friction losses are difficult to estimate even in
the case of healthy machine and present one of the sources
of errors in the measurements of NL losses [25].

The torque and current fluctuations are small additional

Figure 10 Difference in the electromagnetic torque stresses for the machine windings and shaft, but they
between the cases of non-eccentric rotor and 60% static may have severe effect on the operation of the frequency
eccentricity (the rated torque is 115 Nm) converter. On the one hand, the current modulations
may need to be taken into account by oversizing the
Strong modulation of the torque is seen but the average change in
the torque is almost zero. In the case of dynamic eccentricity, the frequency converter, which is an expensive solution. On the
modulation is less pronounced but a net average torque is seen other hand, the control software of the frequency converter
[24] may work in an erroneous way when estimating the
operation quantities of the machine. This last issue is very
important in sensorless control of induction machine,
4 Discussion where different estimators and algorithms have been
The same machine simulated in this work has been used in designed under the assumption of symmetrical (non-
the previous work related to the condition monitoring of eccentric) machine.
the machine through vibration analysis [17, 18]. In
that work, measurements were made to validate the
The UMP because of 10% static eccentricity is about half
computation results. However, the measurement of the
of the rotor weight (Fig. 9). One can speculate on whether
losses under healthy and faulty machine were not possible
this could be used intentionally to magnetically support the
to carry out because of the sensitivity of the accuracy of
rotor of the machine and thus reduce the mechanical stress
such measurements to different factors and because of the
on the bearing because of the rotor weight (this method
fact that these measurements need extensive use of the
has been used for a while by some manufacturers but it has
machine under faulty condition that would result in
been discarded later). The manufacturing tolerance cannot
destroying it. The validation of the methods presented here
guarantee the accurate positioning of the rotor and control
is done on another healthy machine and extended to the
of the static eccentricity. In machines with small ratio
faulty cases.
between the rotor radius and air gap length, less than 10%
eccentricity is achievable but in other machines with
The computation results presented above show that relatively small air gap and large rotor radius this is not
the operation of the machine under eccentric condition possible because of manufacturing tolerances.

264 IET Electr. Power Appl., 2010, Vol. 4, Iss. 4, pp. 259– 266
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-epa.2009.0139
 www.ietdl.org

5 Conclusion [9] SALON S.J., DEBORTOLI M.J., BUROW D.W., SLAVIK C.J.: ‘Calculation
of circulating current between parallel windings in
The paper presented methods for simulating eccentric rotor induction motors with eccentric rotors by finite element
electrical machines and estimating additional losses and method’. Proc. ICEM, Manchester, UK, September 1992,
other harmful quantities because of static and dynamic pp. 371– 375
eccentricity. The methods have been applied to the
simulation of an eccentric induction machine and related [10] DEBORTOLI M.J., SALON S.J., BUROW D.W., SLAVIK C.J.: ‘Effects of
results have been presented. Although these results are rotor eccentricity and parallel windings on induction
logical and in accordance with the present knowledge and machine behaviour: a study using finite element analysis’,
state-of-art, they cannot be generalised without extensive IEEE Trans. Magn., 1993, 29, (2), pp. 1676 – 1682
analysis of machines of different types and sizes. We
believe that such an analysis and a statistical handling of [11] ARKKIO A., LINDGREN O.: ‘Unbalanced magnetic pull in a
its results will generate new knowledge in the fields of high-speed induction motor with an eccentric rotor’. Proc.
engineering related to electrical machines. ICEM, Paris, France, September 1994, pp. 53– 58

[12] ARKKIO A.: ‘Unbalanced magnetic pull in cage induction

6 Acknowledgment motors – dynamic and static eccentricity’. Proc. ICEM 1996,
This study was accomplished as a part of the research project Vigo, Spain, September, pp. 192– 197
KOMASI financed by the Finnish Funding Agency for
Technology and Innovation (TEKES) diary no. 852/31/05. [13] GARRIGAN N.R., SOONG W.L., STEPHENS C.M., STORACE A., LIPO T.A.:
The author gratefully thanks the Agency for the financial ‘Radial force characteristics of a switched reluctance
support. machine’. Proc. Industry Applications Conf., 34th IAS
Annual Meeting, Phoenix, AZ, USA, October 1999,
pp. 2250 – 2258
7 References
[14] LANTTO E., ARKKIO A., ANTILA M. , POKKI K. , SIMON A.:
[1] FISHER-HINNEN J.: ‘Dynamo design’ (Van Nostrand, 1899) ‘Electromagnetic forces caused by cage induction motor’.
Proc. Seventh Int. Symp. Magnetic Bearings, Zurich,
[2] ROSENBERG E.: ‘Magnetic pull in electric machines’, AIEE Switzerland, August 2000, pp. 589– 594
Trans., 1918, 37, (2), pp. 1425 – 1469
[15] TENHUNEN A., HOLOPAINEN T.P., ARKKIO A.: ‘Impulse method to
[3] ROBINSON R.C.: ‘The calculation of unbalanced magnetic calculate the frequency response of the electromagnetic
pull in synchronous and induction motors’, AIEE Trans., forces on whirling cage rotors’, IEE Proc. Electr. Power
1943, 62, (1), pp. 620– 624 Appl., 2003, 150, (6), pp. 752 – 756

[4] KRONDL M.: ‘Self excited radial vibrations of the rotor of [16] PERERS P., LUNDIN U. , LEIJON M.: ‘Saturation effects on
induction machines with parallel paths in the winding’, Bull. unbalanced magnetic pull in a hydroelectric generator
Assoc. Suisse Electr., 1956, 47, (1), pp. 581– 588 with an eccentric rotor’, IEEE Trans. Magn., 2007, 43,
(10), pp. 3884 – 3890
[5] SCHUISKY V.W.: ‘Magnetic pull in electrical machines due
to eccentricity of the rotor’, ERA Transl. Electrotech. Masch. [17] JOVER P., BELAHCEN A., ARKKIO A.: ‘Signatures of electrical
Ball, 1971, 88, (1), pp. 391– 399 faults in force distribution and vibration pattern of
induction motors’, IEE Proc. Electr. Power Appl., 2006,
[6] JORDAN H., RODER G., WEIS M.: ‘Under what circumstances 153, (4), pp. 523– 529
may mechanical vibrations of the stator core be
expected at supply frequency in four-pole three-phase [18] JOVER P.V., BELAHCEN A., ARKKIO A., LAIHO A., ANTONINO-DAVIU J.A.:
asynchronous machines?’, ERA Transl., Elecktr., 1967, 21, ‘Air-gap force distribution and vibration pattern of
(3), pp. 91– 95 induction motors under dynamic eccentricity’, J. Electr.
Eng., 2008, 90, (3), pp. 209– 218
[7] ZHU Z.Q. , HOWE D.: ‘Effect of rotor eccentricity and
magnetic circuit saturation on acoustic noise and [19] BELAHCEN A., ARKKIO A.: ‘Comprehensive dynamic loss-
vibration of single-phase induction motors’, Electr. Mach. model of electrical steel applied to FE simulation of
Power Syst., 1997, 25, (1), pp. 443 – 457 electrical machines’, IEEE Trans. Magn., 2008, 44, (6),
pp. 886– 889
[8] DORRELL D.G.: ‘Experimental behaviour of unbalanced
magnetic pull in 3-phase induction motors with eccentric [20] DLALA E.: ‘Magnetodynamic vector hysteresis models
rotors and the relationship with tooth saturation’, IEEE for steel laminations of rotating electrical machines’. PhD
Trans. Energy Convers., 1999, 14, (3), pp. 304– 309 thesis, Helsinki University of Technology, 2008, p. 165,

IET Electr. Power Appl., 2010, Vol. 4, Iss. 4, pp. 259– 266 265
doi: 10.1049/iet-epa.2009.0139 & The Institution of Engineering and Technology 2010
www.ietdl.org

http://lib.tkk.fi/Diss/2008/isbn9789512292776/, accessed [23] COULOMB J.L.: ‘A methodology for the determination of

June 2009 global electromechanical quantities from a finite element
analysis and its application to the evaluation of magnetic
[21] ISLAM M.J., ARKKIO A.: ‘Effects of pulse-width-modulated forces, torques and stiffness’, IEEE Trans. Magn., 1983,
supply voltage on eddy currents in the form-wound stator 19, (6), pp. 2514– 2519
winding of a cage induction motor’, IET Electr. Power
Appl., 2009, 3, (1), pp. 50– 58 [24] BELAHCEN A.: ‘Losses in an eccentric rotor induction
machine fed from frequency converter’. Proc. 18th ICEM,
[22] ARKKIO A.: ‘Analysis of induction motors based on the Villamoura, Portugal, September 2008, paper id 883, p. 5
numerical solution of the magnetic field and circuit
equations’, Acta Polytech. Scand., 1987, EI 59, p. 97, [25] PIPPURI J. , ARKKIO A. : ‘Challenges in the segregation
http://lib.hut.fi/Diss/198X/isbn951226076X/, accessed of losses in cage induction machines’. Proc. 18th ICEM,
June 2009 Villamoura, Portugal, September 2008, paper id 1445, p. 5

266 IET Electr. Power Appl., 2010, Vol. 4, Iss. 4, pp. 259– 266
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-epa.2009.0139