AbstractThis paper involves current signal frequency

analysis for airgap eccentricity and bearing damage in

induction motors. Magnetic flux density in the airgap is
calculated to get effect of faults on current signal.
Equation of characteristic frequencies for eccentricity is
revised. In addition, a sudden eccentricity is simulated in
short intervals to get effect of bearing damages. Therefore,
new frequencies are detected for bearing faults that is
related with eccentricity. This knowledge impresses most
of the experimental condition monitoring systems and
diagnosis systems.


KeywordsAirgap Eccentricity, Bearing Damage, Fault
Detection, Induction Motor.
I. INTRODUCTION
nduction motors take important place in factories. Failure of
electrical machines causes loss of money so increment of


Manuscript received December 13, 2008: Revised version received ( ),
2009.
The Author is R&D engineer in Ford-Otosan, Kocaeli, Turkey. Phone:
+90-262 3156764; fax: +90-262 3156004; e-mail: ozelgin@yahoo.com,
iozelgin@ford.com.tr ; web page: http://ozelgin.googlepages.com.

motor reliability has been studied. The main reason of failures
in induction motors are faults such as,

1. Unbalanced rotor
2. Eccentricity
3. Bearing damage
4. Broken bar
5. Winding faults
6. Looseness
7. Oil effect.

These faults can be detected by online monitoring of vibration
and current signals to increase the reliability of machines. A
review of fault detection is given in [1]-[3]. The main reasons
of faults can be arranged as follow,

1. Over temperature
2. Unbalanced temperature


3. Unbalanced pull-force
4. Electromagnetic force
5. Improper lubrication of bearings
Analysis of magnetic flux density for airgap
eccentricity and bearing faults
Ilker Ozelgin
I


Fig. 1 fault detection

INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 4, Volume 2, 2008
162


6. Weak shaft and bearing
7. Over speed and load
8. Improper and weak assembling of rotor and stator.

Experimental setups or condition monitoring systems have
been established for fault detection of induction machines.
Current, vibration, noise, speed, temperature, and torque
signals can be used to detect faults in Fig. 1. Generally, current
signature analysis has been preferred as [4]-[8]. After data
collection, a diagnostic method is necessary to recognize the
fault. Different methods have been used to analyze the data
e.g. fast fourier transform, wavelet transform, fuzzy logic,
neural networks, hilbert transform and new developed methods
as in [8]. Recently, software tools have been developed to
satisfy the requirements of fault diagnosis and repair of electric
machines as in [9]. All experimental studies need correct
technical references or reviews to get best result in fault
diagnosis. Especially, software based decision systems tend to
make wrong decisions.
Reference [1] shows that eccentricity and bearing damage
related faults constitute the huge amount of faults. A
theoretical method is verified by Thomson [10] to observe
eccentricity related frequencies in current and vibration signal.
Results of the method are summarized by Vas [11]. Both of
the studies have a small wrong at the last formula that
calculates eccentricity related frequencies in current signal.
Eccentricity is again researched by Guldemir [12].
In this study, magnetic flux density is analyzed for
eccentricity related frequencies in current signal. To calculate
magnetic flux density, magnetic motor force, and permeance
of air-gap are taken into account.
Bearing damage related frequencies in vibration signal are
evaluated by kinematical equations. Eccentricity is created
instantly in computer program to get effect of bearing damage
on current signal. Study of Schoen is confirmed. Also,
additional frequencies are expressed.
II. AIRGAP ECCENTRICITY
Nonuniform air-gap is called eccentricity in motors. The
main reasons of eccentricity are bad manufacturing tolerances,
unbalanced rotor shaft and weak bearing. Eccentricity causes
unbalanced magnetic pull, vibration and harmonics in current
signal. Eccentricity may be classified in two groups, static
eccentricity and dynamic eccentricity.
A. Static Eccentricity
If rotor rotation center is removed a faraway place which is
parallel to stator center in Fig. 2. This situation is called static
eccentricity. Position of minimum air-gap can not rotate. A
magnetic pull force occurs to the side of minimum air-gap. An
example for static eccentricity is oval stator as in Fig. 3.
B. Dynamic Eccentricity
Rotor rotates around stator center, but it do not rotate
around its center. Hence, minimum airgap and magnetic pull
force rotate with rotor. An example for dynamic eccentricity is
oval rotor as in Fig. 4.
C. Magnetic Flux Density
Magnetic flux density variation at one point of airgap, which
is shown in Fig. 5, gives information about current variation at
the nearest coil. This means magnetic flux density calculation
at one point behaves as a current sensor.
In reference [13], amperes circuital law proves the relation
between current and magnetic flux density.



Fig. 3 oval stator


Fig. 4 oval rotor

Fig. 2 static eccentricity
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 4, Volume 2, 2008
163


B dx I =

(1)

x is any closed path around coil. Formula can be applied for an
coil. Current can be calculated as the formula

I
B

(2)

is length of airgap,

is permeability of air. Therefore,
Magnetic flux density variation is proportional to current
signal variation.
Characteristic frequencies of eccentricity on current signal
are indicated by calculation of magnetic flux density. Magnetic
flux density is the amount of magnetic flux ( ) per unit area
of a section, perpendicular to the direction of flux. Equation
(3) is the mathematical representation of magnetic flux
density.

B
A

= (3)

Magnetic flux is related with magnetic motor force (mmf), and
reluctance (R) of the magnetic flux path.

mmf
R
= (4)

For induction motors, reluctance is the function of airgap
length.

R
A

(5)

Formula of magnetic flux density can be arranged by using
(3)-(5).

mmf
B

= (6)


B mmf = (7)

is permeance. Permeance is the function of airgap length.
Permeability of air gap is worse than permeabilities of stator
and rotor core. Permeabilities of stator and rotor core are taken
infinite. Only, airgap permeability is used. mmf is constructed
by stator and rotor mmf. In references [14] and [15], stator and
rotor mmf are calculated as following equations.

1
st
st
c 0
3 2 N I 1
mmf K sin( w t)
p

=

=

(8)

2
rt rt
r r
rt
c 0
2 Z I
p
1
K sin( (w- w w ) t)
mmf

=

=

+ +

(9)

where

N number of turns at each phase;
st
I stator current;
p number of pole pairs;
degree of stator harmonics;
K

coil factor;
mechanical angle of rotor position;
w angular supply frequency;
t time;
rt
Z number of rotor slot;
rt
I rotor current;
K

coil factor of rotor (for squirelcage motor is 1);

degree of rotor harmonics;
r
w angular speed of rotor.
Degree of stator harmonics is formulated by

1
p(6c 1) = (10)

Degree of rotor harmonics is formulated by

rt
2
2
Z
c
p
0,1, 2, 3, ... c
= +
=
(11)

In calculation of airgap flux density we only use mmf of stator.


Fig. 5 flux density effect on stator coil
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 4, Volume 2, 2008
164


Rotor mmf can be used to improve this study.
Permeance is calculated as following equation,

slot saturation d_eccentricity s_eccentricity
= (12)

Slot and saturation effects are not included in calculation of
permeance. After assumptions, permeance can be written as
air-gap variation due to eccentricity.

1
()

(13)

This formula is rearranged according to static eccentricity.
Permeance of static eccentricity can be defined as

1 1 1
() a cos() (1 cos())

= = =

(14)

a is displacement of rotor. is relative displacement of rotor
that is given by

a

= (15)

Fourier series is applied to permeance of static eccentricity.
Result of fourier analysis is

0 1 2 se
cos() cos(2) ...
1
( )

+ + + = (16)

where
0
2
1
1

(17)

and

i
2 i
i 2
(1 1 )
2.
1

=


i=1,2,3,... (18)

By using (8),(10),(13)-(18) in (7), magnetic flux density is
calculated. Fourier analysis of magnetic flux density shows
frequencies of static eccentricity on current signal.
Permeance for dynamic eccentricity is same with static
eccentricity, but only it rotates with rotor. (16) can be
rearranged for dynamic eccentricity.

r r 0 1 2 de
cos(-w t) cos(2 (-w t)) ...
1
( )

+ + + = (19)

According to appendix table, magnetic flux density in
dynamic eccentricity is given in Fig. 6 and frequency analysis
of magnetic flux density in Fig. 7.
D. Formula of Thomson
In reference [10], eccentricity related frequencies are
predicted by

st_current de st_mmf
(1 s)
f f
p

=

`
)
(20)


where




Fig. 6 variation of magnetic flux density in dynamic eccentricity situation.
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 4, Volume 2, 2008
165


s slip;
f electrical supply frequency;
de
= 1,2,3,;
st_mmf
= 1 effect of stator mmf.

Value of
st_mmf
takes only 1, because the number in front of
w in (4) equals to 1. Value of
st_mmf
ranges from to
+ in reference [10]. In reference [11], it is represented by
1, 3, 5, 7... values. In both of them, value of
st_mmf
is
wrong.
In reference [6]-[8], this value is used as 1 to monitor the
behavior of the sidebands around the fundamental frequency
safely. Nonetheless,
st_mmf
can not take other values.
III. BEARING DAMAGE
A. Bearing Faults and Effect on Vibration Signal
There are four parts in bearing, cage, balls, inner and outer
rings. These parts can be damaged due to over load, corrosion,
improper lubrication and installation. Over load or weak
lubrication causes friction. As a result of friction, temperature
increases, so that oil lost its properties. Also, water and acid is
dangerous for oil. Another point for damage, current flows on
bearings, because of voltage difference between stator and
rotor. Oil behaves as a dielectric material in condenser. These
faults produce small particles in bearing.
Each damaged part on bearing leads to vibration at specific
frequencies. In the case of a bearing damage on inner and
outer ring, characteristic frequencies are respectively identified
as follows

b w
inner r
1 2
Z d cos( )
f f 1 2.
2 d d

= +
+
| | | |
| |
\ \
(21)




b w
outer r
1 2
Z d cos( )
f f 1 2.
2 d d

=
+
| | | |
| |
\ \
(22)

where

b
Z number of balls;
r
f frequency of rotor rotation;
w
d diameter of ball;
1
d diameter of inner ring;
2
d diameter of outer ring;
contact angle of balls.

In reference [16], a real time measurement setup compares
different methods to detect the bearing fault.
B. Bearing Damage Effect on Current Signal
Reference [17] defines formula to calculate the frequencies
on current signal because of bearing faults.

current k
f f k f = (23)

f
k
is one of the characteristic vibration frequency of bearing
faults, k =1,2,3,.


Fig. 7 frequency spectrum of magnetic flux density in dynamic eccentricity situation.

INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 4, Volume 2, 2008
166


Bearing faults should be considered as a sudden eccentricity
problem in order to accurately predict the results of bearing
faults. It is known that bearing faults induce vibration at
characteristic frequencies. As a result of vibration,
displacement of rotor shaft appears. Thus, air-gap changes
according to vibration.
If outer ring damage occurs, it would be described as a
sudden static eccentricity. Since, outer ring do not rotate with
rotor, the static eccentricity equals to outer ring damage as in
Fig. 8.
Inner ring damage is different than outer ring damage,
because inner ring rotates with rotor. Therefore, it is assumed
as a sudden dynamic eccentricity as in Fig. 9.
Refer to (21) and (22), characteristic frequencies are
evaluated for inner and outer rings, respectively 243 Hz and
162 Hz. According to (23), inner ring damage related current
frequencies are 293 Hz and 193 Hz for k=1. Outer ring
damage current frequencies are 212 Hz and 112 Hz for k=1,
374 Hz and 274 Hz for k=2, and 536 Hz and 436 Hz for k=3.
Results of computer program are shown in Fig. 10 and Fig.
11. Formula of Schoen is verified by computer program. Also,
at Fig. 10 for inner ring damage, additional frequencies
appear. These frequencies are based on formula of Thomson.
Refer to (20), some of the values are 220 Hz, 265 Hz and 275
Hz. This means, every frequency in Thomsons formula for
current signal do not only describe eccentricity, but also
describe bearing faults. It should be answered that what is the
difference between eccentricity and bearing fault.
We can not recognize the fault exactly by using current
signature analysis, but reference [4] emphasizes that current
signature analysis is sufficient to detect eccentricity and
bearing faults.
Another point, magnitude of dynamic eccentricity related
frequencies in Fig. 7 are bigger than inner ring damaged
related frequencies in Fig. 10. The reason is that one point
damaged bearing is simulated. Magnitude of the frequencies
shows the severity of defect on bearing.







Fig. 8 animation of outer ring damage


Fig. 9 animation of inner ring damage
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 4, Volume 2, 2008
167


IV. CONCLUSION
st_mmf
equals to 1. Thomson and Vas have used different
values. Use of other
st_mmf
values result with wrong fault
decision.
If eccentricity related frequencies are detected in current
signal. Bearing damage effect should be searched. Vibration
measurement is enough for bearing damage detection.
Both of the results are critical to find faults. The reliability
of Neural Networks for fault detection can be improved by
using current and vibration signals to differentiate the bearing
damage and eccentricity.
APPENDIX
w
d =12 mm,
1
d =80 mm,
2
d =40 mm,
r
f =45 Hz,
b
Z =9,
=0 rad, f =50 Hz, s =0.1, p =1, =0.9, =2 mm,
st
I =5A (rms), N=200, K

=1,
7
4 .10

= H/m.
REFERENCES
[1] S. Nandi, H. A. Toliyat, and X. Li, Condition monitoring and fault
diagnosis of electrical motors-a review, IEEE Trans. Energy Convers.,
vol. 20, no. 5, pp. 719728, Dec. 2005.
[2] M. E. H. Benbouzid, A review of induction motors signature analysis
as a medium for faults detection, IEEE Trans. Indus. Elec., vol. 47, no.
5, pp. 984993, Oct. 2000.
[3] A. Siddique, G. S. Yadava, and B. Singh, A review of stator fault
monitoring techniques of induction motors, IEEE Trans. Energy
Convers., vol. 20, no. 1, pp. 106113, March 2005.
[4] N. Mehla, and R. Dahiya, An approach of condition monitoring of
induction motor using mcsa, International Journal of Systems


Fig. 11 frequency spectrum of magnetic flux density when outer ring is damaged


Fig. 10 frequency spectrum of magnetic flux density when inner ring is damaged
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 4, Volume 2, 2008
168


Applications, Engineering & Development, vol. 1, issue 1, pp. 1317,
July 2007.
[5] N. Mehla, and R. Dahiya, Motor current signature analysis and its
applications in induction motor fault diagnosis International Journal
of Systems Applications, Engineering & Development, vol. 2, issue 1,
pp. 2935, August 2007.
[6] A. Nasiri, and M. Hosseini, Fuzzy-wavelet for detection of eccentricity
in induction motors, 4th WSEAS Int. Con. on Power Engineering
Systems,ICOPES05, Rio de Jaeiro, Brazil, April 2005.
[7] E. L. Bonaldii, L. E. Borges, G. Lambert, and L. E. L. Oliveira A rough
sets based classifier for induction motors fault diagnosis, WSEAS
Trans. on Systems, vol. 2, issue 2, pp. 2271-2278, April 2003.
[8] N. Hu, Y. Zhang, F. Gu, G. Qin, and A. Ball, Early fault detection
using a novel spectrum enhancement method for motor current signature
analysis, 7th WSEAS Int. Con. on Artificial Intelligence, Knowledge
Engineering and Data Bases,AIKED08, University of Cambridge, UK,
February 2008.
[9] V. Kontargyri, C. Tsirekis, S. Kabanaros, A. Sakarellos, A. Moronis,
and N. Kolliopoulos, An expert system for fault diagnosis, repairing
and maintenance of electric machines, 6th WSEAS Int. Con. on
Electrical Engineering, Istanbul, Turkey, May 2007.
[10] J. R. Cameron, W. T. Thomson, and A. B. Dow, Vibration and current
monitoring for detecting airgap eccentricity in large induction motors,
IEE Proc. Inst. Elect. Eng. B, vol. 133, no. 3, pp. 155163, May 1986.
[11] P. Vas, Parameter Estimation,Condition Monitoring, and Diagnosis of
Electrical Machines. Oxford, U.K.: Clarendon, 1993, pp. 307308.
[12] H. Guldemir, Detection of airgap eccentricity using line current
spectrum of induction motors, Elsevier Science Electric Power
Systems Research., 64, no. 2, pp. 109117, Dec. 2003.
[13] D. K. Cheng, Field and Wave Electromagnetics. New York: Addison-
Wesley Company, 1989.
[14] P. L. Alger, Induction Machines Their Behavior and Uses. Switzerland:
Gordon and Breach Science Publishers SA, 1970.
[15] B. Heller, and V. Hamata, Harmonic Field Effects in Induction
Machines. New York: Elsevier, 1977.
[16] A. Fernandez, J. Bilbao, I. Bediaga, A. Gaston, and J. Hernandez,
Feasibility study on diagnostic methods for detection of bearing faults
at an early stage, WSEAS Int. Con. on Dynamic Systems and Control,
pp. 113-118, Venice, Italy, November 2005.
[17] R. R. Schoen, T. G. Habetler, F. Kamran and R. G. Bartheld, Motor
bearing damage detection using stator current monitoring, IEEE Trans.
Indus. Appl., vol. 31, no. 6, pp. 12741279, Nov./Dec. 1995.
[18] S. J. Yang, Low-noise electrical motors. Oxford, New York: Clarendon
Press, 1981, pp. 1536.
[19] I. Boldea and S. A. Nasar, The Induction Machine Handbook. New
York: CRC Press, 2002, ch. 10.
[20] Eschmann, Hasbargen and Weigand, Ball and Roller Bearings Theory,
Design, and Application. New York: John Wiley & Sons Ltd., 1985, pp.
7895.





Ilker Ozelgin was born in Bursa, Turkey in 1981. He received the B.S.
degrees in mechanical engineering and electrical engineering in 2004 and
2005 respectively, and the M.S. degree in control and automation engineering
program in 2006 from Istanbul Technical University, Istanbul, Turkey.
His research interests include fault detection of electrical machines,
modeling of vehicles and electric machines, microcontrollers and hybrid
electric vehicles. He works about hybrid electric vehicles in Ford-Otosan,
Kocaeli, Turkey.
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 4, Volume 2, 2008
169