> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <

Calculation of Unbalanced Magnetic Pull in a

Salient-Pole Synchronous Generator Using
Finite-element Method and Measured Shaft
Orbit
Damir arko, Member, IEEE, Drago Ban, Member, IEEE, Ivan Vazdar, Vladimir Jari
an, n, n

Abstract In this paper the unbalanced magnetic forces which

act upon the rotor of a salient-pole synchronous generator due to
eccentric motion of the rotor shaft in the presence of magnetic
field in no-load and loaded condition are calculated using the
finite-element method. The displacement of the rotor is modeled
using the actual shaft orbit measured on a 5 MVA salient-pole
generator driven by a gas turbine in a cogeneration plant. The
influences of the stator winding parallel paths and the rotor
damper winding on attenuation of unbalanced magnetic pull are
analyzed using the finite-element method. It is shown that under
linear conditions the correlations between separate harmonic
components of the shaft orbit, the resulting harmonic
components of the unbalanced magnetic pull and the induced
stator winding currents can be established in both the no-load
and loaded condition.
Index TermsGenerators, vibrations, electromagnetic forces,
force, damping , finite element methods, rotors, stators, armature

NOMENCLATURE
U
V

U
V

K
ss
sk

s(t)

Magnitude of the vector of forward precession

Magnitude of the vector of backward precession
Phase shift of the vector of forward precession
Phase shift of the vector of backward precession
Angular speed of precession
Distance between shafts center and its center of
gravity
Angular speed of rotation of the shaft around its center
Shaft orbit in the complex form
complex vector of static displacement
complex vector of kinetic displacement
instantaneous displacement of the rotor

Manuscript received January 05, 2011. Accepted for publication June 1,

2011.
Copyright 2011 IEEE. Personal use of this material is permitted. However,
permission to use this material for any other purposes must be obtained from
the IEEE by sending a request to [email protected].
Damir arko and Drago Ban are with the Faculty of Electrical Engineering
and Computing, Department of Electric Machines, Drives and Automation,
Unska 3, 10000 Zagreb, Croatia (phone: +385 1 6129 706; fax: +385 1 6129
705; e-mail: [email protected]; [email protected]).
Ivan Vazdar is with KONAR Generators and Motors Inc., Fallerovo
etalite 22, 10000 Zagreb, Croatia (e-mail: [email protected]).
Vladimir Jari is with MARTING d.o.o., Alberta Fortisa 171, 10000 Zagreb,
Croatia (e-mail: [email protected]).

Amplitude, frequency and phase shift of the nth

harmonic component of the shaft orbit
Fx, Fy Constant average forces in x and y direction
Dr
Rotor outer diameter
D
Stator inner diameter
0
Air gap size along the centerline of the pole shoe
kc
Carter factor
l
Axial length of the stator core
B1
Fundamental component of the air-gap flux density
Ephase Rms value of the induced no-load phase voltage
f
Rated frequency
w
Number of stator winding turns connected in series
p
Number of pole pairs
fd1
Stator winding distribution factor for fundamental
component
fp1
Stator winding pitch factor for fundamental component
q
Number of slots per pole and phase
y
Coil pitch of the stator winding
p
Pole pitch
s
Slot pitch
Ns
Number of stator slots
do
Width of the slot opening
Fradial Total radial force on the rotor
Fx0, Fx1, Fx2, x components of the UMP originating from the
static deflection, the first and the second
harmonic components of the shaft orbit
Fy0, Fy1, Fy2 y components of the UMP originating from the
static deflection, the first and the second
harmonic components of the shaft orbit
Ns
Number of stator slots
Fa, Ff Armature and field winding MMFs
Ff1
Fundamental component of the field winding MMF
Fad1, Fai1
Magnitudes of the direct and inverse
components of the armature winding MMF
ad0, ai0
Initial phase shifts of the direct and inverse
components of the armature winding MMF
s
Angular coordinate along the stator circumference
e
Electrical angular frequency

Total air-gap permeance

Harmonic component of the air-gap permeance without

eccentricity

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
se

Air-gap permeance component due to static

eccentricity
r, Radial and tangential components of the Maxwell
stress tensor
0
Permeability of vacuum

I. INTRODUCTION

HE magnetic field in a synchronous machine

interacting with the armature winding currents or the
salient poles creates tangential electromagnetic forces on the
rotor which produces torque. The radial forces are generated
at the same time, but in the case when the stator and rotor are
perfectly concentric having a symmetrical air gap around the
perimeter, the radial forces are cancelled out. These forces
depend not only on electromagnetic parameters of the
machine, but also on the geometric shape of the stator and
rotor and the presence of an eccentric motion of the rotor.
When the rotor is eccentric the unbalanced magnetic pull
(UMP) occurs in which case the net radial force is developed
and in interaction with the mechanical system may cause
unwanted, harmful vibrations. A detailed model of this
interaction is presented in [1] and [2] for the case of a twopole turbogenerator whose rotor is modeled by means of finite
beam elements, the bearing effects by stiffness and damping
coefficients and the foundation by pedestals. The UMP is
added to other exciting forces to correctly reproduce the
dynamic behavior of a real machine during no-load operation.
The general case of eccentric motion includes dynamic
eccentricity when the rotor center follows an arbitrary
trajectory whose geometric center does not coincide with the
center of the stator. In addition, the stator and rotor shape may
deviate from an ideal symmetric shape. In [3] it has been
shown that shape deviations in a hydrogenerator may give rise
to different whirling motions, including both backward and
forward precession. In both cases the UMP will arise due to
asymmetry in the air gap. The two important parameters of the
unbalanced forces are frequency and amplitude which implies
their periodic nature due to rotating motion of the machine.
In salient-pole synchronous machines the stator winding
very often has parallel paths. The eccentric motion of the rotor
induces circulating currents within parallel paths which have a
damping effect on the UMP. This influence is noticeable in
induction [4] and permanent-magnet motors [5] as well. In
addition, the salient-poles usually contain the squirrel-cage
type rotor damper winding which further attenuates the UMP
[6]. Both influences have been studied in this paper.
The calculation of unbalanced magnetic forces is essential
for the analysis of vibrations and evaluation of the mechanical
stress which arises in various parts of the machine. The two
common approaches are analytical methods and the finiteelement method (FEM). The analytical solutions are usually
based on winding functions, air gap permeance and lumped
parameters or conformal mapping can be used as well. In [7]
the permeance function has been used to calculate analytically
the waveform of the air-gap flux density for the purpose of

detecting eccentricity in a synchronous generator. Simond et

al. [8] used the air-gap permeance approach to derive
measurable modulation functions which carry information
about static and dynamic eccentricities and stator and rotor
bore deformations of a salient-pole synchronous generator. In
[9] the air-gap MMF for a machine with eccentricity has been
formulated in an analytical model in a manner to distinguish
the effect of the homopolar flux which is significant for twopole machines. For the purpose of detecting eccentricity in
induction machines the analytical modeling based on air-gap
permeance and modified winding functions [10]-[12] or the
MMF function [13] has be used to distinguish the frequency
components in the stator current or in the complex apparent
power modulus [14] that appear due to rotor eccentricity. In
[15] the effect of a ball bearing fault has been represented by
permeance variation which is a complex sum of an infinite
number of rotating eccentricities. Those permeance variations
reflect on the harmonic content of the stator current which can
be used for diagnosing the bearing faults. Various diagnostic
techniques for detection of faults in induction machines,
which includes eccentricity as well, have been summarized in
[16]. This paper provides a comprehensive list of useful
references in the field of induction motor diagnostics.
In [17] conformal mapping has been used to calculate the
air-gap permeance function, flux density and UMP in a
permanent-magnet motor with rotor eccentricity.
The common problem of the analytical solutions is their
limitation to linear problems, but their advantage is short
calculation time. The nonlinearity can be taken into account as
in [6] where lumped parameters of the nonlinear analytical
model are found by means of optimization using the results of
finite-element (FE) simulations in which case the parameters
are valid only for a single operating point. The analytical
permeance network models can inherently take into account
saturation and also model the eccentric motion of the rotor. In
[18] this approach is used for dynamic simulation of air-gap
eccentricity in an induction motor.
The FEM includes saturation locally by default, which
makes it inherently more time consuming and in the case of
salient pole synchronous machines with high number of poles
and fractional slot windings may result in the necessity to
model a large portion of the machine to take into
consideration the geometric symmetry. In addition, in the
presence of eccentricity there is no geometric symmetry, so
the whole cross section of the machine must be modeled
regardless of the number of slots and poles. This may result in
an extremely large number of elements in the FE mesh. In the
FE based studies of UMP Perers et al. [19] reports a mesh size
of 85000 elements for a 14 pole generator, while Lin Wang et
al. [20] reports a mesh size of 158760 elements for an 88 pole
generator. The eccentric motion of the rotor in FEM for any
type of electric machine is most effectively modeled using a
time stepping method, which also allows for the inclusion of
the electric circuits into the model to account for the effects of
the stator winding and the rotor damper winding. Timestepping is used in [21] to model eccentricity in a permanent-

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
magnet synchronous motor for the purpose of fault diagnosis
and in [22] to calculate additional losses due to rotor
eccentricity in an induction motor with PWM-voltage supply.
However, the calculation of forces in the FEM is sensitive to
discretization errors. In that regard the Maxwell stress tensor
approach is more sensitive than the virtual work method. In
[23] it has been shown that the value of torque calculated by
the virtual work method is very consistent irrespective of the
mesh refinement and changes marginally with a change in the
number of layers in the air-gap mesh. This can be important
for calculation of UMP because rotor displacement is usually
expressed in micrometers, which can be significantly smaller
than the size of the finite-element mesh in the air gap.
In [24] the dynamic eccentricity in a mechanical run prior
to excitation of the generator is modeled by observing the
whirling motion at a single frequency, which assumes that
shaft orbit has a circular shape. In reality, the shaft orbit
consists of multiple harmonic components which all together
affect the UMP in the machine. In linear models each
harmonic component can be treated separately and the total
UPM is equal to the sum of UMPs for individual harmonic
components. However, this approach is not valid under
nonlinear conditions in which case permeance networks or the
FEM are the most suitable approaches.
The contribution of this paper is a model of the measured
shaft orbit of a salient-pole synchronous generator and the
associated vibration displacements in two directions,
perpendicular to one another, with all their significant
harmonic components which have been used to calculate the
unbalanced magnetic pull by means of the FEM coupled with
circuit equations for the stator winding and the rotor damper
winding. For calculation of forces the virtual work method has
been applied. The no-load operation and the loaded condition
have both been included in the study and the damping effects
of the stator winding parallel paths and the rotor damper
winding have been taken into account.
II. THEORETICAL BACKGROUND
Shaft vibrations are defined as rapid motions of the shaft
which are related to its deflection in rotating motion. The
deflection of the shaft is a deformation of its elastic line which
connects all geometric centers of the shaft cross sections along
the axis of the bearings. This axis is an imaginary line which
connects geometric centers of the bearing bores and is aligned
with the z axis of the absolute reference frame. Since the shaft
bends under the action of static radial forces, its centers of
rotation form a static deflection line. In rotation the kinetic
displacement sk is added to the static displacement ss. (Fig. 1)
The motion of a shaft cross section is composed of a motion
of the shaft center C around the point S of the static deflection
line and also of a rotation of the cross section around the shaft
center C. This motion of the shaft is called precession. The
precession is defined by the motion of the shafts center of
gravity which can be given in the complex plane by equation
z Ue

j t U

Ve

j t V

ce jt

(1)

where U and V are the magnitudes, and U and V are the

phase shifts of the vectors of forward and backward
precession respectively, is the angular speed of precession, c
is the distance between the shafts center and the shafts center
of gravity and is the angular speed of rotation of the shaft
around its center.
The path of the shaft center in the radial plane is called
shaft orbit (marked as K in Fig. 1) and is given by a general
equation in the complex form
K ss sk ss Ue

j t U

Ve

j t V

(2)

where ss is the complex vector of static displacement and sk

is the complex vector of kinetic displacement.
When >0 and V = 0 the precession is given by a vector
Uejt, which represents the pure forward precession. In that
case the shaft orbit has the same orientation as the rotation of
the shaft.
When <0 and U = 0 the precession is given by a vector
Vejt, which represents the pure backward precession. In that
case the orientation of the shaft orbit is opposite to the rotation
of the shaft.
The most general type of motion contains both forward and
backward precession as described by (2). When = the
synchronous forward or synchronous backward precession
occurs.
In electric machines the vibrations of the fundamental
frequency are caused by the unbalance of the masses in
rotation or the unbalanced magnetic pull. Very often higher
order harmonic components appear which can be caused by
the fluid whirl in poorly lubricated fluid bearings [25]. The
equation for the shaft orbit in that case is
K ss Ue

j t U

Ve

j t V

an e

j n t n

(3)

n 1

where an, n and n are the amplitude, the frequency and the
phase shift of the nth harmonic component and k is the
maximum number of higher order harmonic components.
Note that the higher order harmonic components do not affect
the time needed to go over one closed loop of the shaft orbit.
They only affect the shape of the orbit.
The size, shape and location of the shaft orbit vary with the
speed of rotation and load. In general, the kinetic shaft
displacements reach maxima at the critical speed. The most
common type of motion in synchronous machines is a
synchronous forward precession. The backward precession
can also occur with anisotropic bearings [25]. In the forward
precession the deflection line retains the sagged shape which
it has in the nonresonant motion, while in the backward
precession the deflection line alternates, i.e. the shaft flexes
twice for every turn around its center. This alternate flexion
requires high energy which makes the backward precession
less likely to appear.

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <

Fig. 1. Kinetic shaft orbit

III.

MEASURED ECCENTRICITY

The salient-pole synchronous generator for which the

detailed design datasheets and the results of measured rotor
vibrations are available is used in a cogeneration plant with a
gas turbine as a prime mover. The basic data of the generator
are shown in TABLE I.
The measurement and analysis of bearing and shaft
vibrations have been carried out by means of Brel & Kjr
VIBROTEST 60 vibration analyzer and DEWE-50-PCI-16
data acquisition system. The electrodynamic sensors (Brel &
Kjr VS080) for measurement of bearing vibrations have
been fixed on bearings housings in two directions (x, y). The
non contact sensors (Brel & Kjr IN 085) for measurement
of shaft vibrations have been fixed through holes in the shaft
casing near drive end (DE) bearing in two directions (x, y).
Therefore, the shaft vibrations have been measured relatively
to the bearing casing which is mounted on the machine frame.
The tracking signal for determination of the phase angle of
fundamental component of vibration speed vectors has been
obtained from the rotor shaft by means of a photosensitive
transducer. The measured variables are vibration
displacement, vibration velocity and vibration acceleration.
The most useful quantity for our analysis is the vibration
displacement which is given in the form of a shaft orbit graph.
The principle measurement scheme is shown in Fig. 2.
The shaft orbit has been initially recorded for the cold unit
in a mechanical run at 1500 rpm (Fig. 3) with both field and
armature winding currents equal to zero. The rotor shaft
exhibits dynamic eccentricity with forward synchronous
precession. The coordinates of time integrated mean values of
static shaft displacement are (-11.2, 82.6) m. In this case
there is no electromagnetic field in the machine and hence no
UMP that might contribute to eccentric motion of the shaft.
The eccentric motion occurs mainly due to gravity and finite
stiffness of the shaft which cause it to deflect and shift its
center of gravity away from the axis of the bearings. As the
shaft rotates a rotating centrifugal force appears which acts
from the bearing axis towards the center of gravity. This force
causes the eccentric motion of the shaft. In addition, eccentric
motion in a mechanical run without excitation can also appear

due to wear of the bearings or errors during assembly.

The recorded shaft orbit for the case of no-load operation at
synchronous speed of 1500 rpm with an excitation current of
36.5 A and induced rated armature voltage of 10500 V is
shown in Fig. 4a. The orbit graph for the loaded condition
(P = 1.84 MW, Q = 0.18 MVAr) with a thermally stable unit
at 1500 rpm is shown in Fig. 4b. This generator operates in a
cogeneration plant of a natural gas processing facility. The
power demand in the facility at the time when the
measurements were carried out did not allow the rated load of
the generator to be achieved. The coordinates of time
integrated mean values of static shaft displacement for no-load
operation are (-12.57, 78.26) m, and for loaded generator
(9.49, 89.91) m.
From Fig. 3 and Fig. 4a it is apparent that the presence of
magnetic field in no-load operation, and consequently the
unbalanced magnetic pull, affects the shaft orbit to a small
extent. However, when the generator is loaded (Fig. 4b) the
size of the orbit increases by approximately 100 % and the
static displacement is altered as well.
IV. FINITE-ELEMENT MODEL FOR CALCULATION OF
UNBALANCED MAGNETIC PULL
The eccentric motion of the rotor in the case when the
generator is magnetized either in no-load or loaded operation
will result in unbalanced electromagnetic forces which act
upon the rotor and consequently on the bearings of the
machine.
TABLE I.
DATA OF THE SYNCHRONOUS GENERATOR
Rated power
Rated power factor
Rated voltage
Voltage regulation
Rated frequency
Rated speed
Overspeed
Moment of inertia
Stator inner diameter
Air gap size at the centerline of the pole shoe
Air gap size at the edge of the pole shoe
Width of the slot opening
Axial length

5000 kVA
0.8
10500 V
10%
50 Hz
1500 rpm
1800 rpm
304 kgm2
800 mm
5.5 mm
8.8 mm
14.2 mm
820 mm

Fig. 2. Principle scheme for measurement of synchronous generator

vibrations

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <

moves. In each time step the air-gap mesh along the boundary
between the two layers is modified so that nodes along the
boundary are mutual for both layers. Simultaneous rotary and
linear motions of the rotor are thus allowed. The stator
winding coils are objects in the FE model and their mutual
connections are modeled as an electric circuit which is solved
simultaneously with the rest of the model. The induced
voltages due to both rotary and linear motions are calculated
by FEM, and external current or voltage sources, resistors,
capacitors, inductors and other elements can be added to that
circuit as well. The end-winding leakage inductance has also
been added to the model. It has been calculated analytically
using a 3-D method based on closed form solution of
Neumann integrals [26].

Fig. 3. Recorded shaft orbit in mechanical run of the cold unit at 1500 rpm.
Field and armature winding currents are equal to zero.

The goal is to use the finite-element method to simulate the

measured eccentric motion of the shaft and calculate the
resulting unbalanced magnetic pull. Since the shaft orbit for
every revolution of the rotor is slightly different (see Fig. 4),
for the purpose of FE simulation the average orbits are
calculated.
The harmonic content of the average orbits is shown in
TABLE II. Since there is no backward precession, i.e. V in (3)
equals zero, the shapes of the orbits can be reconstructed
using data from TABLE II if (3) is written in a slightly
different form, i.e.

K ss U x cos t U x jU y sin t U y
k

axn cos n t axn ja yn sin n t ayn

n2

(4)

(a)

where ss , Ux, Uy, Ux, Uy, axn, ayn, axn and ayn are given in
TABLE II. Note that Ux, Uy, Ux and Uy refer to the
fundamental component of the shaft orbit. Only the first 10
harmonic components of the Fourier series are given, i.e.
k = 10. According to Fig. 5, the first 10 components are
sufficient for good reconstruction of the shaft orbit.
The eccentric motion for one revolution of the rotor is
modeled using Infolytica MagNet 7.1.1 software by
simultaneously combining the rotation of the rotor about its
axis and the motion of the rotor axis with respect to the stator
center according to Fig. 5a for the no-load condition and
according to Fig. 5b for the loaded generator. The time needed
for the rotor axis to go through one full eccentric orbit is the
same as the time it takes the rotor to make one full revolution
about its axis at the speed of 1500 rpm. The time-stepping
method is used to model the motion of the rotor. The air-gap
is divided in two layers along its centerline. The layer along
the stator bore is stationary, while the other layer is attached to
the rotor and slides along the stationary one as the rotor

(b)
Fig. 4. Recorded shaft orbit in (a) no-load condition (1500 rpm, 10500 V),
(b) loaded condition (P = 1.84 MW, Q = 0.18 MVAr, 1500 rpm, 10500 V)

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
TABLE II.
HARMONIC CONTENT OF THE AVERAGE SHAFT ORBITS

n
2
3
4
5
6
7
8
9
10

225 rad/s

ss

-12.57+j78.26 m

Ux, m
22.415
axn, m
2.483
0.707
0.935
0.263
0.293
0.241
0.108
0.233
0.180

Uy, m
24.242
ayn, m
1.674
1.171
0.883
0.690
0.243
0.112
0.048
0.276
0.139

Ux, rad

Uy, rad

-0.763

-2.059

axn, rad

ayn,rad

2.708
3.087
-2.900
-2.644
-0.012
-3.042
0.286
0.878
-0.398

-0.825
1.420
0.302
-0.730
-2.109
0.288
-2.275
-0.332
0.899

120
Shaft displacement in y direction, m

NO-LOAD CONDITION

225 rad/s
9.49+j89.91 m

Ux, m
52.028
axn, m
2.735
1.683
0.761
1.232
0.447
0.220
0.079
0.209
0.192

Uy, m
43.073
ayn, m
1.515
0.493
0.505
1.813
0.765
0.257
0.145
0.294
0.197

Ux, rad
-1.860
axn, rad
-1.189
0.408
0.742
-0.032
-0.311
2.326
-2.394
-2.152
-1.478

Measured shaft orbit

Reconstructed shaft orbit using
first 10 harmonic components

100
90
80
70
60

-40 -30
-20
-10
0
10
Shaft displacement in x direction, m

20

(a)

Uy, rad
-0.240

160

ayn,rad
1.621
-1.102
0.146
-1.792
-1.098
-0.473
-0.504
0.029
-0.690

In these simulations it is assumed that the generator exhibits

parallel precession, i.e. the rotor axis is always parallel to the
stator axis. In reality the shaft is not infinitely rigid, so it will
always have a slightly arched shape because of which the
orbits measured at the bearings will differ from the orbits in
other cross sections of the shaft. The actual shapes of the
orbits along the shaft affect the UMP and vice versa, so for the
accurate analysis the electromagnetic and the structural
analysis problems must be coupled. Such coupling was not
possible in our case due to limitations of the FE software
designated for solving only electromagnetic problems.
The nonlinear structural analysis models of the rotor shaft
based on finite beam elements which take into account
stiffness, damping, gyroscopic effects, weight and external
forces (UMP at no-load, inherent unbalance and stiffness
asymmetry) to model the static and dynamic deflection lines
along the shaft of a two-pole turbogenerator with cylindrical
rotor have been reported in [1]-[2]. The results indicate that it
is possible to reliably model the shaft motion when comparing
the simulated and measured orbits at the bearings. It is then
justifiable to assume that such a model also works well for the
other nodes along the shaft where it was not possible to
measure the shaft orbits.

Shaft displacement in y direction, m

n
2
3
4
5
6
7
8
9
10

110

50
-50

LOADED CONDITION

ss

150

Measured shaft orbit

Reconstructed shaft orbit using
first 10 harmonic components

140
130
120
110
100
90
80
70
60
50
40
-50 -40 -30 -20 -10 0 10 20 30 40 50 60 70
Shaft displacement in x direction, m
(b)

Fig. 5. Measured (averaged) and reconstructed shaft orbit for (a) no-load and
(b) loaded condition

V. SIMULATION RESULTS
The FE simulations used for calculation of the UMP
resulting from the eccentric motion of the rotor according to
the recorded shaft orbits for no-load and loaded conditions
have been carried out for the following three cases:
1) Force calculation without stator winding parallel paths
and without rotor damper winding,
2) Force calculation with stator winding parallel paths and
without rotor damper winding,
3) Force calculation with stator winding parallel paths and
with rotor damper winding.

A. Case 1
This is the simplest case in which the damping effects of the
stator winding parallel paths and the rotor damper winding
have not been included. Therefore, the resulting UMP will

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
originate entirely from instantaneous displacements of the
rotor shaft center with respect to the center of the stator bore.
The results of simulations for both operating conditions are
unbalanced electromagnetic forces given as a function of time,
which are shown in Fig. 6. The maximum radial force for noload condition occurs at t = 12.4 ms and equals 2359 N. This
time instant corresponds to the maximum average shaft
displacement of 100.2 m. In addition, due to static
displacement of the shaft, the constant average force
F = 1750 N will appear.
Similarly, for the loaded generator the maximum radial
force occurs at t = 37.4 ms and equals 2518 N. This time
instant corresponds to the maximum average shaft
displacement of 133.8 m. The constant average force in the
case of the loaded generator equals F = 1612 N.
The maximum instantaneous force depends on the
maximum instantaneous displacement. On the other hand, the
size of the orbit reflects on the peak-to-peak variation of the
force. The peak-to-peak value of the total radial force in noload condition is Fp-p = 1294 N, while for the loaded generator
Fp-p = 1742 N.

Force, N

5000

3000

In order to establish a link between the harmonic content of

the total unbalanced force on the rotor and the shape of the
shaft orbit, a simple analytical model for calculation of forces
is employed. The instantaneous single sided magnetic pull at
eccentric rotor position can be calculated according to [27]
F t

FEM: Fe=
FEM: with saturation
Analytical

Ephase

kd1
0.01

0.02
Time, s

0.03

0.04

7000

FEM: with saturation

Analytical

Force, N

6000

B 1

5000

Dl
p

(6)

sin q 2
q sin 2

sin 6 100 2
6sin 100 2

0.9561

(7)
(8)

Ephase 2 p
2 fwf d1 f p1 Dl

5747 2 2
2 50 96 0.9561 0.9397 0.8 0.82 (9)

0.9144 T

4000
3000

The Carter factor follows from the well known expression

2000

kc

1000
0
0

2 fwkd1kp1 B 1

where q is the number of slots per pole and phase and y/p is
the coil pitch to pole pitch ratio. The air-gap flux density is
then

9000
8000

kp1 sin y 2 p sin 14 2 18 0.9397

(a)
FEM: Fe=

(5)

where Dr = D-20 is the rotor outer diameter (D is the stator

inner diameter), 0 is the air gap size along the centerline of
the pole shoe, kc is the Carter factor, l is the axial length of the
stator core, s(t) is the instantaneous displacement of the rotor
(Fig. 5a) and B1 is the fundamental component of the air-gap
flux density. The geometric parameters D, l and 0 are given
in TABLE I. The Carter factor is used to take into account the
presence of stator slots and obtain the effective air gap size.
The air gap flux density can be calculated from

2000
1000
0

Dr l s t 2
B 1
4 0 k c 0

where Ephase is the rms value of the induced no-load phase

voltage, f is the rated frequency, w is the number of stator
winding turns connected in series, p is the number of pole
pairs, kd1 and kp1 are the winding distribution and pitch factors
for fundamental component. The no-load FE simulation has in
this particular case been carried out using measured rated noload field current. The calculated rms value of the induced
phase voltage in that case equals 5747 V, which is 5.2 %
smaller than rated voltage of 6062 V measured for that same
field current. The distribution and pitch factors are given by

6000

4000

0.01

0.02
Time, s

0.03

0.04

(b)
Fig. 6. Waveforms of the total unbalanced magnetic pull in (a) no-load and
(b) loaded condition calculated analytically and numerically using FEM
(simulation WITHOUT parallel paths and WITHOUT damper winding)

s
2
2d o
d o 0 d o
s
ln 1
atan


2 0 d o 2 0

1.1597

(10)

where s = (D)/(Ns) is the slot pitch, Ns is the number of

stator slots and do is the width of the slot opening.
Since all parameters in (5), except s(t), are constant, it
appears that the waveform of the unbalanced magnetic pull

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
will be directly proportional to the shape of the shaft orbit,
which is determined by s(t). The calculation of force in (5)
assumes that iron is infinitely permeable, so comparison with
FE simulation is possible if the same assumption is made in
the FE model. This is accomplished by setting the relative
permeability of the stator and rotor core to 109 to simulate an
infinitely permeable material. However, in order to obtain the
same fundamental component of the induced phase voltage
and thus the same fundamental component of the air-gap flux
density as in the case when the actual B-H curves of the stator
and rotor core are taken into account, the field current had to
be reduced. Otherwise, with the same field current as in the
nonlinear simulation, due to increased overall permeance of
the magnetic circuit in the linear model, higher values of the
air-gap flux density and induced voltage would have been
obtained.
The waveforms of the total unbalanced force on the rotor in
no-load condition calculated analytically using (5) and
numerically using FEM with saturation and with infinitely
permeable iron are shown in Fig. 6a. The small oscillations
that are superimposed on the basic waveforms are due to
slotting effect and due to discretization errors which occur
because the shifts of the rotor are small relative to the size of
the FE mesh in the air gap.
The waveforms indicate that fundamental frequency is
25 Hz. The fundamental frequency of the force in the case of
whirling motion is equal to the frequency of precession [6].
Since in our case is equal to the angular speed of rotation of
the shaft around its center, in a four pole machine at 1500 rpm
this corresponds to the frequency of 25 Hz.
There is a very good agreement between analytically and
numerically calculated waveforms with infinitely permeable
iron. This result clearly indicates that the shape of the shaft
orbit is the dominant factor which determines the waveform of
the UMP.
The strong influence of saturation in the stator and rotor
core on the magnitude of force is also apparent. The peak
value of force is reduced by a factor of 2.36 when saturation is
taken into account. In a sense by adjusting the field current
and the value of induced voltage to be the same in nonlinear
and linear models, the saturation has been taken into account
to some extent. However, the equality of induced voltages
does not necessarily lead to the equality of calculated forces
with and without the presence of saturation in the FE model.
The correlation with the shaft orbit is also visible in Fig. 7a
which compares harmonic content of the shaft orbit and of the
total unbalanced force on the rotor (without the average
component). The magnitudes are normalized with respect to
their fundamental components (F1 and s1).
Similar analysis can be done for the loaded condition as
well. However, in this case the fundamental component of the
air-gap flux density depends on the resulting actions of the
field winding and the armature winding MMFs. In the FE
model an iterative procedure is used to determine the field
current and the position of the armature current vector which
yield the required active and reactive power output

(P = 1.84 MW, Q = 0.18 MVAr) in the cases of both saturated

machine and unsaturated machine with infinitely permeable
iron. Since the presence of eccentricity does not affect the
average power output, in order to save time and reduce the
geometry of the FE model to one pole pitch, the field current
and the position of the armature current vector have been
determined by modeling circular motion without eccentricity.
The field and armature currents thus calculated are introduced
into models with eccentric motion defined according to Fig.
5b in order to calculate the UMP.
The flux density B1 used in (5) is calculated from the
fundamental component of the induced phase voltage
resulting from the FE simulation of the loaded machine with
saturation. In that case Ephase = 5986 V, which yields
B1 = 0.9525 T.
The waveforms of the total unbalanced force on the rotor in
the loaded condition calculated analytically using (5) and
numerically using FEM with saturation and with infinitely
permeable iron are shown in Fig. 6b. In this case there is also
a very good agreement between analytically and numerically
calculated waveforms with infinitely permeable iron.
According to Fig. 7b the correlation of the harmonic contents
of the shaft orbit and of the UMP calculated analytically and
numerically is apparent in the loaded generator as well.
B. Case 2
When the eccentric motion occurs in the presence of
parallel paths in the stator winding, it induces currents which
circulate within the branches connected in parallel. For the
loaded condition, the circulating components can be extracted
from the total phase currents by subtracting the currents in the
parallel branches of each phase and dividing their difference
by two. These circulating currents have a damping effect on
the UMP because they try to cancel the variation of magnetic
field that caused them. The attenuation of the UMP is visible
in the waveforms of the total UMP shown in Fig. 8.
C. Case 3
The actual generator has a rotor damper winding with 12
bars on every pole shoe which are shorted by end rings. There
are no connections between damper windings of adjacent
poles. The eccentric motion of the rotor causes the variation of
the flux linkage in the damper winding and induces the
currents which also produce the damping effect and further
attenuate the UMP. This effect is visible in the waveforms of
the total UMP shown in Fig. 8.
The waveforms of the UMP for all three cases in no-load
and loaded condition are compared in Fig. 8 and the resulting
values of the peak radial force are compared in TABLE III.
Similarly, the harmonic contents are compared in Fig. 9. The
attenuating effects of the stator winding parallel paths and
particularly of the rotor damper winding are also apparent
from those figures. Moreover, the damping does not only
affect the magnitude, but also the harmonic content of the
UMP. Note that in Case 2 the induced current in the stator
winding gives rise to a fairly high 100 Hz component. The

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
origin of this particular harmonic component can be
determined by simulating the eccentric motion using one
particular harmonic component of the shaft orbit at a time.
Unfortunately, this approach can only be used in a linear
model, since it employs the principle of superposition which
cannot be used in a nonlinear model. Nevertheless, the linear
approach can also give a valid insight into the origin of the
100 Hz component. For this purpose a linear finite-element
model has been used for the cases of no-load operation with
and without stator winding parallel paths. In both cases the
rotor damper winding has not been included. For each case
three simulations have been performed. The static deflection
ss , the first and the second harmonic components of the shaft
orbit defined in TABLE II for no-load operation have been
simulated respectively. From each simulation the x and y
components of the UMP have been calculated (Fig. 10 and
Fig. 11). In addition, for the case when stator winding parallel
paths are included, the induced stator currents which circulate
within parallel branches of the phase windings have been
calculated (Fig. 12). The total radial force resulting from all
three simulations is equal to
2

Force: FEM with saturation (F1=512.7 N)

F/F1 and s/s1

Force: Analytical (F1=1230.2 N)

Shaft orbit (s1=23.21 m)

0.6
0.4
0.2
0

25

50

75 100 125 150 175 200 225 250

Frequency, Hz

(a)
1
Force: FEM with saturation (F1=704.1 N)
Force: FEM with Fe= (F1=2367.8 N)

F/F1 and s/s1

0.8

Force: Analytical (F1=2398.1 N)

Shaft orbit (s1=41.64 m)

0.6
0.4
0.2
0

(11)

where Fx0, Fx1, Fx2, Fy0, Fy1 and Fy2 are the x and y components
of the UMP originating from the static deflection, the first and
the second harmonic components of the shaft orbit
respectively. The same total radial force can be calculated
from a single FE simulation in which the static deflection and
the first two harmonic components of the shaft orbit are used
simultaneously to define the eccentric motion of the rotor. Of
course, the same principle can be expanded to simulate
independently an arbitrary number of shaft orbit harmonic
components. However, in that case the simulations would
require very dense FE mesh and small time steps in order to
correctly account for individual higher harmonic components
of the shaft orbit due to their small magnitude and high
frequency.
Note that the total radial force in Fig. 10b is comparable to
the force in Fig. 6a. It is basically the same waveform, except
that in Fig. 10b the static deflection and two harmonic
components of the shaft orbit are used, while in Fig. 6a all 10
harmonic components are used to simulate the eccentric
motion of the rotor. Similarly, the total radial force in Fig. 11b
can be compared with the force in Fig. 8a calculated for Case
2. The difference is in the magnitude of the force since the
result in Fig. 11b has been obtained using linear model with
static deflection and two harmonic components of the shaft
orbit, while in Fig. 8a all 10 harmonic components are used in
a nonlinear model.

Force: FEM with Fe= (F1=1230.5 N)

0.8

25

50

75 100 125 150 175 200 225 250

Frequency, Hz

(b)
Fig. 7. Comparison of normalized harmonic content of the shaft orbit and of
the total unbalanced magnetic pull in (a) no-load and (b) loaded condition
calculated numerically and analytically (simulation WITHOUT parallel paths
and WITHOUT damper winding)
3000
2500

WITHOUT stator parallel paths and WITHOUT rotor damper winding

WITH stator parallel paths and WITHOUT rotor damper winding
WITH stator parallel paths and WITH rotor damper winding

2000
Force, N

Fx 0 Fx1 Fx 2 Fy 0 Fy1 Fy 2

1500
1000
500
0
0

0.01

0.02
Time, s

0.03

0.04

(a)
3000

WITHOUT stator parallel paths and WITHOUT rotor damper winding

WITH stator parallel paths and WITHOUT rotor damper winding
WITH stator parallel paths and WITH rotor damper winding

2500
Force, N

Fradial

2000
1500
1000
500
0
0

0.01

0.02
Time, s

0.03

0.04

(b)
Fig. 8. Waveforms of the total unbalanced magnetic pull in (a) no-load and
(b) loaded condition for all three cases

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
5500

600
WITHOUT stator parallel paths and WITHOUT rotor damper winding
WITH stator parallel paths and WITHOUT rotor damper winding
WITH stator parallel paths and WITH rotor damper winding

5000
Total radial force, N

500
400
Force, N

10

300
200
100

4500
4000
3500
3000

25

50

75 100 125 150 175 200 225 250

Frequency, Hz

2500
0

(a)

0.01

0.02
Time, s

900

Fig. 10. Force components in x and y direction in no-load operation

WITHOUT stator parallel paths and WITHOUT rotor damper winding
calculated using linear FEM resulting from the following harmonic
components of the shaft orbit: (a) static deflection (constant term), 1st
harmonic, and 2nd harmonic, (d) static, 1st and 2nd together

WITH stator parallel paths and WITHOUT rotor damper winding

WITH stator parallel paths and WITH rotor damper winding

700
600
Force, N

0.04

(b)

WITHOUT stator parallel paths and WITHOUT rotor damper winding

800

0.03

500
400

4000

300
3000

200
100

2000

25

50

Force, N

75 100 125 150 175 200 225 250

Frequency, Hz
(b)

Fy0

Force, N

3000

Fy1
Fy2

Fx2

1000

0.01

0.015

Fx2
0.02 0.025
Time, s

0.03

0.035

0.04

(a)

5000
4000
3000
2000

0
0

-1000

Fx0
0.005

0.005

1000

-2000
0

-2000
0

Total radial force, N

4000

Fx1

Fx1

Fx0

Loaded
Average
Peak force
force
(% attenuation)
1736 N
2518 N (0 %)
1133 N
2087 N (17.1 %)
654 N
1233 N (51.0 %)

5000

2000

Fy2

1000

-1000

TABLE III.
COMPARISON OF PEAK RADIAL FORCE FOR ALL THREE CASES

Case 1
Case 2
Case 3

Fy0

Fig. 9. Comparison of harmonic contents of the UMP in (a) no-load and (b)
loaded condition

No-load
Average
Peak force
force
(% attenuation)
1790 N
2359 N (0 %)
1220 N
2022 N (14.3 %)
645 N
1005 N (57.4 %)

Fy1

0.01

0.02
Time, s

0.03

0.04

(b)
0.01

0.015

0.02 0.025
Time, s

(a)

0.03

0.035

0.04

Fig. 11. Force components in x and y direction in no-load operation WITH

stator parallel paths and WITHOUT rotor damper winding calculated using
linear FEM resulting from the following harmonic components of the shaft
orbit: (a) static deflection (constant term), 1st harmonic and 2nd harmonic, (b)
static, 1st and 2nd together

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <

Ff Ff 1 cos ps e t

Stator current, A

0
-0.25

0.01

0.02
Time, s

0.03

0.04

(a)
0.4
iU1

0.3

iV1
iW1

0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
0

0.01

0.02
Time, s

0.03

0.04

0.03

0.04

(b)
0.04
iU2

0.03

iV2
iW2

0.02
0.01
0
-0.01
-0.02

(13)

(14)

where is the harmonic component of the air-gap permeance

without eccentricity, se is an additional permeance
component due to static eccentricity and se0 is the angular
position of the maximum air-gap reduction. The radial
component of the air-gap flux density is then equal to
Br Fa Ff

0.25

-1
0

-0.03
0

(15)

The UMP can be calculated from the radial r and

tangential components of the Maxwell stress tensor given
by

0.01

0.02
Time, s

(c)
1.5
iU0+iU1+iU2

1.25

iV0+iV1+iV2

1
Stator current, A

cos s e t se cos s se0
p
0,2 p ,4 p

iW0

-0.5

where Ff1 is the magnitude of the fundamental component

MMF and p is the number of pole pairs. In the presence of
static deflection of the rotor, i.e. static eccentricity, the air-gap
permeance can be approximated using a term [28], [29]

iV0

-0.75

(12)

where s is the angular coordinate along the stator

circumference, e is the electrical angular frequency, Fad1,
Fai1, ad0 and ai0 are the magnitudes and the initial phase
shifts of the direct and inverse components respectively. The
fundamental component of the field winding MMF equals

iU0

0.5

Stator current, A

Fa Fad1 cos s e t ad0 Fai1 cos s e t ai0

1
0.75

Stator current, A

The results in Fig. 11 indicate that the 100 Hz component of

the UMP emerges in the case when there is static deflection of
the rotor and the circulating currents are induced in the stator
winding parallel branches. If one considers the principle
cross-section of a four-pole machine, then the winding axes
and the reference directions of the phase currents for normal
operation are given according to Fig. 13a. However, in the
case when there are only induced circulating currents flowing
in the stator winding, the instantaneous currents in the parallel
branches of one phase have the same value, but they flow in
the opposite directions. Therefore, the referent direction of
current in adjacent parallel branch of every phase winding
changes resulting in the winding structure and the positions of
the winding axes as shown in Fig. 13b. Therefore, the
opposite directions of currents in the parallel branches of the
same phase have the effect of turning the four-pole winding
into a two-pole winding with winding axes of phases U and W
mutually shifted by 60 mechanical degrees. This effect is
illustrated in Fig. 14 which shows the flux lines of the
armature winding field due to induced circulating currents
shown in Fig. 12a at t = 10 ms with field winding current set
to zero. The phase shifts between the winding axes and the
phase shifts of induced circulating currents indicate that this is
not a symmetrical three-phase system, which gives rise to both
direct and inverse components of the armature winding MMF
whose fundamental component in a four-pole generator is
given by

11

iW0+iW1+iW2

0.75
0.5
0.25
0
-0.25
-0.5
-0.75
-1
0

0.01

0.02
Time, s

0.03

0.04

(d)
Fig. 12. Induced stator winding currents in no-load condition calculated using
linear FEM which circulate within the branches connected in parallel
(WITHOUT rotor damper winding taken into account) resulting from the
following harmonic components of the shaft orbit: (a) static deflection
(constant term), (b) 1st harmonic, (c) 2nd harmonic, (d) static, 1st and 2nd
together

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <

12

Fx f1 cos 2 se0 ad0 2 cos ad0

cos 2e t ad0 2 se0 f 2 cos 2se0 ai0
2 cos 2e t ai0 cos 2e t ai0 2 se0
Fy f1 sin 2 se0 ad0 2sin ad0

(18)

sin 2e t ad0 2 se0 f 2 sin 2se0 ai0

2sin 2e t ai0 sin 2e t ai0 2 se0

where
(a)

(b)

Fig. 13. Principle cross-section of the four-pole generator with double layer
short pitched stator winding arrangement, referent current directions and
positions of the phase winding axes for the cases of (a) normal operation, and
(b) operation with open-circuited stator winding and induced circulating
currents in the parallel branches

f1

160

se2 Ff1 Fad1 Dl ,

r Br2 B2 2 0 , Br B 0

(16)

where 0 is the permeability of vacuum. The usual assumption

for salient-pole machines is that B is negligible compared to
Br. The detailed analysis which confirms this assumption is
given in [30]. Hence, the x and y components of the UMP are
given by
2

Fx

Fy

Br2 D
l cos s d s
20 2
Br2 D
l sin s d s
2 0 2

(17)

For simplicity, if only additional permeance due to static

eccentricity is used in (14) combined with (15) and (17), the
resulting expressions for Fx and Fy are

160

se2 Ff 1 Fai1 Dl .

(19)

From these solutions it is apparent that Fx and Fy consist of

constant terms and harmonic components at 2e, which
correspond to the frequency of 100 Hz. The FEM results (Fig.
11) also indicate the presence of this harmonic component.
The more complete solution is obtained by adding also the
dominant air-gap permeance terms 0+2pcos[2(ps-et)] in
(14) and (15), but they are omitted due to very lengthy
solution of (17) thus obtained. For instance, the inclusion of
only 0 yields additional constant terms and harmonic
components at 2e, while 2pcos[2(ps-et)] yields additional
constant terms and harmonic components at 2e and 4e.
VI.

Fig. 14. Flux lines of the armature winding field due to induced circulating
currents in the winding parallel paths with field winding current set to zero
without rotor damper winding and with only static deflection of the shaft orbit
taken into account

f2

CONCLUSION

The calculation of unbalanced magnetic pull in a 5 MVA,

4 pole synchronous generator based on known geometry of
the machine, finite-element method and measured shaft orbits
in no-load operation at 10500 V, 50 Hz, 1500 rpm and loaded
condition has been presented.
The UMP has been analyzed for three different cases with
respect to damping effects of the stator winding parallel paths
and the rotor damper winding. It has been shown that in the
case when the damping effects are not taken into account the
harmonic content of the shaft orbit is closely related to the
harmonic content of the UMP since then the UMP is
proportional to the instantaneous displacement of the rotor.
When damping is included, both the magnitude and the
harmonic content of the UMP are affected. The more exact
correlations between harmonic components of the UMP,
individual harmonic components of the shaft orbit and the
damping effects of the stator winding and the rotor damper
winding can be established by simulating the separate
harmonic components of the shaft orbit in both no-load and
loaded condition. However, this approach in not directly
applicable to nonlinear finite-element simulations because
superposition cannot be used under nonlinear conditions. A
way around this problem would be to simulate the no-load or
loaded condition with the actual shaft orbit and store the
values of permeability related to the nodes of the FE mesh.
After that the UMP can be calculated separately for each
harmonic component of the shaft orbit always using the same
stored nodal values of permeability in order to establish all the

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
correlations correctly. This principle can be applied to all
three cases studied in the paper.
The results shown in the paper indicate that for this
generator the stator winding parallel paths alone attenuate the
UMP by 14.3 % in no-load and 17.1 % in loaded condition.
The rotor damper winding and the stator winding parallel
paths together yield the total attenuation of 57.4 % in no-load
and 51.0 % in loaded condition.

[16]

[17]

[18]

REFERENCES
[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

P. Pennacchi and L. Frosini, Dynamical behaviour of a three-phase

generator due to unbalanced magnetic pull, IEE Proc.-Electr. Power
Appl., Vol. 152, No. 6, pp. 1389-1400, November 2005
P. Pennacchi, Computational model for calculating the dynamical
behaviour of generators caused by unbalanced magnetic pull and
experimental validation, Journal of Sound and Vibration, No.312, pp.
332-353, 2008
N.L.P. Lundstrm and Jan-Olov Aidanp, Whirling frequencies and
amplitudes due to deviations of generator shape, International Journal
of Non-Linear Mechanics, No. 43, pp. 933-940, 2008
A. Tenhunen, Finite-Element Calculation Of Unbalanced Magnetic Pull
And Circulating Current Between Parallel Windings In Induction Motor
With Non-Uniform Eccentric Rotor, Proceedings of Electromotion '01,
pp. 19-24, Bologna, Italy, 19-20 June, 2001
D.M. Ionel and I.P. Brown, On the effect of winding parallel paths in
synchronous permanent magnet electric machine drives, 14th Biennial
IEEE Conference on Electromagnetic Field Computation (CEFC),
Chicago, IL, 9-12 May 2010
A. Burakov and A. Arkkio, Low-Order Parametric Force Model for
Eccentric-Rotor Electrical Machine Equipped With Parallel Stator
Windings and Rotor Cage, IET Electric Power Applications, Vol. 1,
Issue 4, pp. 532-542, 2007
B. A. T. Iamamura, Y. Le Menach, A. Tounzi, N. Sadowski and E.
Guillot, Study of Synchronous Generator Eccentricities Using
Analytical Approach and FEM, ICEM 2010 19th International
Conference on Electrical Machines, 6 pages, Rome, Italy, September 68, 2010
J. Simond, Mai Tu Xuan and R. Wetter, An Innovative Inductive AirGap Monitoring for Large Low Speed Hydro-Generators, ICEM 2008
18th International Conference on Electrical Machines, pp. 1-5,
Vilamoura, Portugal, September 6-9, 2008
A. Di Gerlando, G.M. Foglia and R. Perini, Analytical Modelling of
Unbalanced Magnetic Pull in Isotropic Electrical Machines, ICEM
2008 18th International Conference on Electrical Machines, pp. 1-6,
Vilamoura, Portugal, September 6-9, 2008
M. Sahraoui, A. Ghoggal, S.E. Zouzou and M.E. Benbouzid, Analysis
of Dynamic Eccentricity Related Signatures on Stator Current of
Squirrel Cage Induction Motors, ICEM 2008 18th International
Conference on Electrical Machines, pp. 1-6, Vilamoura, Portugal,
September 6-9, 2008
A. Ghoggal, A. Aboubou, S.E: Zouzou, M. Sahraoui and H. Razik,
Considerations about the modeling and simulation of air-gap
eccentricity in induction motors, Proc. of IECON 06 - 32nd Annual
Conf. of the IEEE Industrial Electronics Society, pp. 4987-4992, Paris,
France, November 7-10, 2006
J. Faiz and M. Ojaghi, Unified winding function approach for dynamic
simulation of different kinds ofeccentricity faults in cage induction
machines, IET Electr. Power Appl., Vol. 3, Iss. 5, pp. 461470,
September 2009
M. Bldt, P. Granjon, B. Raison and G. Rostaing, Models for Bearing
Damage Detection in Induction Motors Using Stator Current
Monitoring, IEEE Trans. Ind. Electronics, Vol. 55, No. 4, pp. 18131822, April 2008
Mhamed Drif and A. J. Marques Cardoso, Airgap-Eccentricity Fault
Diagnosis, in Three-Phase Induction Motors, by the Complex Apparent
Power Signature Analysis, IEEE Trans. Ind. Electronics, Vol. 55, No.
3, pp. 1404-1410, March 2008
B. Akin, U. Orguner, H.A. Toliyat, and M. Rayner, Low Order PWM
Inverter Harmonics Contributions to the Inverter-Fed Induction Machine

[19]

[20]

[21]

[22]

[23]

[24]

[25]
[26]

[27]

[28]

[29]

[30]

13

Fault Diagnosis, IEEE Trans. Ind. Electronics, Vol. 55, No. 2, pp. 610619, February 2008
A. Bellini, F. Filippetti, C. Tassoni and G.A. Capolino, Advances in
Diagnostic Techniques for Induction Machines, IEEE Trans. Ind.
Electronics, Vol.55, No.12, pp. 4109-4126, December 2008
J. T. Li, Z. J. Liu and L. H. A. Nay, Effect of Radial Magnetic Forces in
Permanent Magnet Motors With Rotor Eccentricity, IEEE Trans. Mag,
Vol. 43, No. 6, pp. 2525-2527, June 2007
A., Mahyob, M.Y.O. Elmoctar, P. Reghem, G. Barakat, Induction
Machine Modelling Using Permeance Network Method For Dynamic
Simulation of Air-Gap Eccentricity, 12th European Conference on
Power Electronics and Applications (EPE 2007), pp. 1-9, Aalborg,
Denmark, 2-5 September 2007
R. Perers, U. Lundin and M. Leijon, Saturation Effects on Unbalanced
Magnetic Pull in a Hydroelectric Generator With an Eccentric Rotor,
IEEE Trans. Mag., Vol. 43, No. 10, pp. 3884-3890, October 2007
Lin Wang, R.W. Cheung, Zhiyun Ma, Jiangjun Ruan and Ying Peng,
Finite-Element Analysis of Unbalanced Magnetic Pull in a Large
Hydro-Generator Under Practical Operations, IEEE Trans. Mag, Vol.
44, No. 6, pp. 1558-1561, June 2008
B.M. Ebrahimi, J. Faiz, and M.J. Roshtkhari, Static-, Dynamic-, and
Mixed-Eccentricity Fault Diagnoses in Permanent-Magnet Synchronous
Motors, IEEE Trans. Ind. Electronics, Vol. 56, No. 11, pp. 4727-4739,
November 2009
A. Belahcen, and A. Arkkio, Computation of additional losses due to
rotor eccentricity in electrical machines, IET Electr. Power Appl.,
Vol. 4, Iss. 4, pp. 259266, 2010
S. Salon, S. Bhatia and D. Burow, Some Aspects of Torque
Calculations in Electrical Machines, IEEE Trans. Mag, Vol. 33, No. 2,
pp. 2018-2021, March 1997
D. Guo, F. Chu, D. Chen, The Unbalanced Magnetic Pull And Its
Effects On Vibration In A Three-Phase Generator With Eccentric
Rotor, Journal of Sound and Vibration 254 (2), pp. 297312, 2002
Agnieszka Muszyska, Rotordynamics, Taylor & Francis, 2005
D. Ban, D. arko, I. Mandi, Turbogenerator End-Winding Leakage
Inductance Calculation Using a 3-D Analytical Approach Based on the
Solution of Neumann Integrals, IEEE Trans. Mag, Vol. 20, No. 1,
pp. 98-105, March 2005
T. Schneider and A. Binder, Design and Evaluation of a 60 000 rpm
Permanent Magnet Bearingless High Speed Motor, 7th International
Conference on Power Electronics and Drive Systems (PEDS '07), pp. 18, Bangkok, Thailand, November 27-30, 2007
G. Joksimovic, C. Bruzzese, and E. Santini, Static Eccentricity
Detection in Synchronous Generators by Field Current and Stator
Voltage Signature Analysis Part I: Theory, ICEM 2010 19th
International Conference on Electrical Machines, 6 pages, Rome, Italy,
September 6-8, 2010
G. Joksimovic, C. Bruzzese, and E. Santini, Static Eccentricity
Detection in Synchronous Generators by Field Current and Stator
Voltage Signature Analysis Part II: Measurements, ICEM 2010 19th
International Conference on Electrical Machines, 5 pages, Rome, Italy,
September 6-8, 2010
L. Frosini and P. Pennacchi, The effect of rotor eccentricity on the
radial and tangential electromagnetic stresses in synchronous machines,
Proc. of IECON 06 - 32nd Annual Conf. of the IEEE Industrial
Electronics Society, pp. 1287-1292, Paris, France, November 7-10, 2006

Damir arko (M06) was born in 1972 in

Zagreb, Croatia. He received the B.S. and M.S.
degrees in electrical engineering from the
University of Zagreb in 1995 and 1999
respectively and the Ph.D. degree from the
University of Wisconsin-Madison in 2004.
Currently he is an Assistant Professor at the
Department of Electrical Machines, Drives and
Automation, Faculty of Electrical Engineering
and Computing, University of Zagreb, Croatia.
His research activities are related to design,
modeling and optimization of electrical machines and power transformers and
testing of electrical machines.

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
Drago Ban (M01) was born in 1939 in VrpoljeSibenik, Croatia. He received the B.Sc., M.Sc.
and Ph.D. degrees in electrical engineering from
the Faculty of Electrical Engineering, University
of Zagreb, Zagreb, Croatia in 1965, 1975 and
1987 respectively.
From 1968 till 1989 he worked as a
development and project engineer in KONARElectrical Industry, Zagreb, Croatia. He is
currently a full Professor at the Department of
Electrical Machines Drives and Automation,
Faculty of Electrical Engineering and
Computing, University of Zagreb, Croatia. His
scientific and professional activities have been related to the field of electrical
machinery, electrical drives and technical diagnostics.
Professor Ban has a long experience in organization of international
conferences, seminars and lectures. He was the chairman of several
international conferences on electrical drives and power electronics: EDPE
1996, EDPE 98 and EDPE 2000. In 2002 he was the general chairman of the
10th International Power Electronics and Motion Control Conference, EPEPEMC 2002, 9-11. Sept. Dubrovnik, Croatia. He is a member of Editorial
Board of the Journal of Electrical Engineering, several international
professional associations, and a member of Croatian Academy of Engineering.

14