DEVELOPMENT OF A TOOL TO DETECT FAULTS IN INDUCTION

MOTORS VIA CURRENT SIGNATURE ANALYSIS

Mark Fenger, Mark Susnik and Paul Laderoute

William T. Thomson

Iris Power Engineering

CANADA

University of Abertay
SCOTLAND

ABSTRACT
This paper will demonstrate through industrial case histories, how current signature analysis can reliably
diagnose rotor winding problems in induction motors. Traditional CSA measurements can result in false alarms
and/or misdiagnosis of healthy machines due to the presence of current frequency components in the stator current
resulting from non-rotor related conditions such as mechanical load fluctuations, gearboxes, etc. Theoretical
advancements have now made it possible to predict many of these components, thus making CSA testing a much
more robust and less error prone technology. Based on these theoretical developments, case histories are presented
which demonstrate the ability to separate current components resulting from mechanical gearboxes from those
resulting from broken rotor bars. From this data, a new hand held instrument for reliable detection of broken rotor
bars, air gap eccentricity, shorted turns in LV stator windings and mechanical phenomena/problems in induction
motor drives is being developed. Detection of the inception of these problems prior to failure facilitates remedial
action to be carried out thus avoiding the significant costs associated with unexplained down time due to unexpected
failures.

INTRODUCTION
Operators of electrical motors are under continual pressure to reduce maintenance costs and prevent
unscheduled downtimes that can result in lost production and revenue. Application of condition based maintenance
strategies rely on specialized on-line monitors to reliably provide a measure of the health of the drive system. Thus,
unexpected failures and consequent downtime may be avoided and/or the time between planned shutdowns for
planned maintenance may be increased. In addition, downtime for testing is limited. Maintenance and operational
costs are thus reduced. During the past twenty years, there has been a substantial amount of fundamental research
into the creation of condition monitoring and diagnostic techniques for plant maintenance for induction motors [15].
This paper describes the development of an instrument that will reliably diagnose problems such as broken
rotor bars or abnormal levels of air gap eccentricity in induction motor drives during normal in-service operation.
In-depth theoretical derivations for detecting electrical and mechanical problems via on-line current signature
analysis [9]. The main focus of the paper is on the key features embedded into the intelligent hand-held CSA
instrument and the interactive communication possibilities via networking.

DETECTION OF BROKEN BARS

It is well known that a 3-phase symmetrical stator winding fed from a symmetrical supply with frequency
f1 , will produce a resultant forward rotating magnetic field at synchronous speed and if exact symmetry exists there
will be no resultant backward rotating field. Any asymmetry of the supply or stator winding impedances will cause
a resultant backward rotating field from the stator winding. When applying the same rotating magnetic field
fundamentals to the rotor winding, the first difference comp ared to the stator winding is that the frequency of the
induced electro-magnetic force and current in the rotor winding is at slip frequency, i.e. sf1 , and not at the supply
frequency:
The rotor currents in a cage winding produce an effective 3-phase magnetic field with the same number of
poles as the stator field but rotating at slip frequency f2 = sf1 with respect to the rotating rotor. With a symmetrical
cage winding, only a forward rotating field exists. If rotor asymmetry occurs then there will also be a resultant

Iris Rotating Machine Conference

June 2003, Santa Monica, CA

Page 1 of 8

backward rotating field at slip frequency with respect to the forward rotating rotor. As a result, the backward
rotating field with respect to the rotor induces an e.m.f. and current in the stator winding at:
fsb =f1 (1-2s) Hz

(1)

This is referred to as the lower twice slip frequency sideband due to broken rotor bars. There is therefore a
cyclic variation of current that causes a torque pulsation at twice slip frequency (2sf1) and a corresponding speed
oscillation, which is also a function of the drive inertia. This speed oscillation can reduce the magnitude (amps) of
the f1 (1-2s) sideband but an upper sideband current component at f1 (1+2s) is induced in the stator winding due to
rotor oscillation [6].
The upper sideband is enhanced by the third time harmonic flux. Broken rotor bars therefore result in
current components being induced in the stator winding at frequencies given by:
fsb =f1 (12s) Hz

(2)

These are the classical twice slip frequency sidebands due to broken rotor bars.

DETECTION OF AIR GAP ECCENTRICITY

Air gap eccentricity may be detected by identifying the characteristic current signature pattern being
indicative of abnormal levels of air gap eccentricity and to then trend that signature [7-9]. The specific frequencies
of the current components indicative of air gap eccentricity may be calculated by :

1 s
n ws .
f ec = f 1 (R nd )
p

fec
f1
R
nd
s
p
n ws

=
=
=
=
=
=
=

(3)

frequency components which are a function of air gap eccentricity)

supply frequency)
number of rotor slots
1
slip
pole-pairs
1, 3, 5, 7etc

With nd = 0 in equation (3) gives the classical rotor slot passing frequency components - a series of components
spaced at twice the supply frequency, 2f1 , apart. The signature pattern of specific rotor slot passing frequencies and
the two components from equation (3) with n d = 1 can be used to identify abnormal levels of air gap eccentricity.

DETECTION OF MECHANICAL INFLUENCES

Changes in air gap eccentricity results in changes in the air gap flux waveform. With dynamic eccentricity
the rotor position can vary and any oscillation in the radial air gap length results in variations in the air gap flux.
Consequently this can induce stator current components given by [10,11]:
fe = f1 mfr

(5)

where
f1 = supply frequency
fr = rotational speed frequency of the rotor
m = 1,2,3.............harmonic number
fe = current components due to air gap changes

Iris Rotating Machine Conference

June 2003, Santa Monica, CA

Page 2 of 8

This means that problems such as shaft/coupling misalignment, bearing wear, roller element bearing defects and
mechanical problems that result in dynamic rotor disturbances can be potentially detected due to changes in the
current spectrum.

THE INFLUENCE OF GEARBOXES

As seen from above, mechanical oscillations will give rise to additional current components in the
frequency spectrum. As will be shown in the following, gearboxes may give rise to current components of
frequencies close to or similar to those of broken bar components. Hence, to perform a reliable diagnosis of a rotor
winding for motors connected to a gearbox, the influence of gearbox components in the spectrum need be
considered.
Specifically, slow revolving shafts will give rise to current components around the main supply frequency
components as prescribed by equation (5) where the rotational speed frequency of the shaft, rotating with Nr rpm,
may be calculated as fr = Nr/60.
Consider the case where a 60 Hz motor rated 300 HP, 575 V, 885 rpm is connected to a gearbox. The
output speed from the gearbox is 19.39 rpm. The shaft speed of 19.39 rpm gives rise to two frequency components
symmetrically distributed 0.32 Hz around the main supply frequency, i.e. the specific rotational speed frequencies
are 59.67 Hz and 60.32 Hz. Harmonics of these are distributed 0.64 Hz, 0.97 Hz, 1.29 Hz, 1.62 Hz etc
symmetrically around f1 .
From equation (2) it can be seen that the frequency range in which broken rotor bars can be detected is defined by.
fl.max = (1-2s fl)f1

(6)

where sfl is the full load slip. For instance, for the machine in this example the full load slip is 0.0166 thus yielding
a maximum lower frequency component of 58 Hz and consequently a maximum upper component of 62 Hz. Hence,
up to the 6th harmonic of the rotational frequency of the shaft are present in the full load frequency band.
Furthermore, to accurately measure and trend the number of broken rotor bars, it is imperative that the rotational
speed frequencies of gearbox shafts be detected, identified and omitted from the broken bar analysis.
Field testing have proved identification of mechanical components in the broken bar spectrum a non-trivial task
even when the gearbox name-plate data is known. Firstly, the gearbox may be old and thus subjected to wear and
tear over the years. In this case, the next result is that gearbox components do not always appear at the expected
frequency positions. Secondly, the name-plate data on the housing may not reflect the actual gear ratios. In this
case, the current components due to gearboxes do not appear near the expected frequency positions. A sophisticated
algorithm has been developed which takes in to account these phenomena and allow for rejection of mechanical
current components during the broken bar analysis, thus enhancing the reliability of the analysis it self.

DEVELOPMENT REQUIREMENTS FOR A NEW CSA MONITOR

The keyword for condition based-monitoring is reliability and in the case of the instrument this includes a
number of crucial features:

Unambiguous diagnosis of a fault over a range of motor ratings

Correct estimation of the slip for any given load conditions for a range of motor designs and power ratings
Clear discrimination between the unique current signature patterns caused by a fault and any current
components induced due to normal characteristics of the drive system.
Current components caused by the effect of mechanical load (for example coal crushers, conveyors and
gearboxes in the drive train) must be reliably diagnosed since they can be misinterpreted as components
from broken rotor bars etc [4,12].
Eliminate the need for an expert to interpret the acquired data by applying reliable, advanced diagnostic
algorithms to the spectra.

Iris Rotating Machine Conference

June 2003, Santa Monica, CA

Page 3 of 8

Correct estimation of the slip together with discrimination between

current components due to mechanical phenomena and current components
due to broken rotor bars is, without doubt, the most critical aspect of current
signature analysis.
In the case of diagnosis of broken rotor bars, without a correct estimation
of the operating slip, the sidebands due to broken rotor bars cannot be
identified. In the past, slip has been provided by measurement of the
rotational speed with a stroboscope or by measurement of the magnetic flux
with a special coil [3]. For the new CSA tool, the slip is directly deducted
from the stator current signature.
Furthermore, as stated earlier, current components due to mechanical
phenomena may appear at frequencies close to those caused by broken rotor
bars. Hence, correct identification of these components and following
elimination of these from the analysis process allows for reliable diagnosis of
broken rotor bars.
For diagnosis of abnormal levels of air gap eccentricity, correct
identification of the current signature pattern defined by equation (4) involves
correct estimation of the operating slip. Furthermore, it requires knowledge
of the number of rotor slots for the individual machines tested. This number
is not readily accessible via the nameplate data but, nonetheless, known by
the OEM and often listed in the machine specific datasheets.
Finally, the data spectrum has to be of a sufficient resolution allowing for
diagnosis of broken rotor bars at very light loads. The resolution of the Figure 2: Picture of the CS Meter
current spectrum for diagnosis of broken rotor bars is 10 mHz and the Prototype.
resolution of the current spectrum for diagnosis of air gap eccentricity is 0.1
Hz.

INSTRUMENT DESCRIPTION
A picture of the instrument is shown in Figure 2.

Figure 1: Screen Dump of a Prototype Version of CSView

Iris Rotating Machine Conference
June 2003, Santa Monica, CA

The instrument is battery operated (supports

approximately 3 hours of operation) and
weighs approximately 2 kg. A plug
allows for connection of an external
power supply for operation or charging
purposes. The instrument has a LED color
display and a key pad providing an
environment for interaction between the
user and CSMeters software for
acquisition and diagnosis of data. A serial
port is located on the right side of the
instrument allowing for communication
between CSMeter and.
The instrument has 2MB of
internal memory allowing for storage of
data for up to 30 motors. Each data
acquisition
has
a
duration
of
approximately 110 seconds. The input for
the motor supply current, provided via a
current probe in series with an internal
signal condition unit, is located on the top
Page 4 of 8

OEP2095

of the instrument (BNC plug). The

internal signal condition unit helps
insure that the instrument uses its full
dynamic range at all times.

Broken Rotor Bar Components Spectrum

0

supply
frequency

-10

The instrument allows for

onsite diagnosis of broken rotor bars
and, later on, diagnostis of air gap
eccentricity. The diagnosis results are
show on the LCD screen. In the
following, data acquired with the new
instrument will be presented.

Frequency Components [dB]

-20
-30
-40
-50
-60

Noise
lower
sideband

-70

upper
sideband

CASE STUDY I: No Broken Rotor

Bars
58.0
58.5
59.0
59.5
60.0
60.5
61.0
61.5
62.0
Frequency [Hz]
Figure 3 shows part of the
frequency resolved current spectrum Figure 3: Current Spectrum obtained on a healthy motor.
for a 575 V, 100 HP, 60 Hz fan motor,
operating in a cement plant. The full
load speed is 1780 rpm yielding a frequency interval of 58.66 Hz to 61.33 Hz for detection of broken rotor bars.
The motor was operating at 71 Amps, corresponding to approximately 80% full load. As can be seen from Figure 3,
the spectrum is completely free of any current components around main supply frequency, f1 , and consequently, the
frequency range in which current components due to broken rotor bars are expected, are empty. The motor thus
shows no signs of broken rotor bars.
CASE STUDY-II: Signs of Rotor Asymmetry
Figure 4 shows part of the frequency resolved current spectrum for yet another 575 V, 100 HP fan motor
operating in a cement plant. The full load speed is 885 rpm yielding a frequency interval of 58 Hz to 62 Hz for
detection of broken rotor bars. The motor was operating at 50 Amps, corresponding to approximately 50% full load.
Based on the load conditions, the instrument predicted current components due to broken rotor bars to be
positioned at 59.0 Hz and 61.0 Hz. A search band is applied around these positions. Figure 4 shows one current
component to be present in each search band. The components are distributed symmetrically around f1 , as expected,
but different magnitudes, 48.7dB and 60.0dB from the main supply frequency. The components are a sign of initial
rotor asymmetry but not yet indicative of an unhealthy motor.
238438

CASE STUDY III: Broken Rotor

Bars
Figure 7 shows part of the
frequency resolved current spectrum
for coal mill rated 440 V, 180 HP
operating in a utility plant. The full
load speed is 885 rpm yielding a
frequency interval of 58 Hz to 62 Hz
for detection of broken rotor bars on a
50Hz motor.

Broken Rotor Bar Components Spectrum

supply
frequency

0
-10

Frequency Components [dB]

-20
-30
-40
-50
-60

Noise
lower
sideband

-70
57.0

57.5

58.0

58.5

59.0

upper
sideband
59.5 60.0 60.5
Frequency [Hz]

61.0

61.5

62.0

62.5

Figure 4: Current Spectrum showing signs of rotor asymmetry.

Iris Rotating Machine Conference
June 2003, Santa Monica, CA

63.0

Based on the supply current,

the instrument predicted sidebands due
to broken rotor bars to be positioned at
49.70 Hz and 50.57 Hz.
These
frequency positions are close to that of
the supply frequency. Figure 7 show
the supply frequency to have a
Page 5 of 8

Motor With Broken Rotor Bars

0

Supply Current [dB]

-10

2sf1 components
due to broken rotor
bars.

-20
-30
-40
-50
-60
-70
-80
47

48

49

50

51

52

53

Frequency [Hz]

Figure 7: Current Spectrum from a Motor Subjected to

Broken Bars

Figure 5: Dismantled cage rotor showing bar to

end region. The faulty region is obvious. The
spectrum obtained on this motor is shown in
Figure 7.

somewhat wide declining current component. This is caused

by the motor being subjected to smaller changes in load, i.e.
smaller changes in supply current, during the data acquisition
process. However, the peak pick algorithms embedded in the instrument was able to detect the declining slopes of
the supply frequency within the applied search bands and thus disregard these slopes from the analysis thereby
correctly identifying the current components due to broken rotor bars.

CASE STUDY IV: The Influence of Gearboxes

Figure 6 shows part of the frequency resolved current spectrum for coal mill motor rated 575 V, 300 HP, 60
Hz operating in a utility plant. The full load speed is 885 rpm yielding a frequency interval of 58 Hz to 62 Hz for
detection of broken rotor bars. The motor is driving a coal mill through a three-stage reduction gearbox. The output
speed of the gearbox at full load conditions is 19.39 rpm and the individual shaft speeds internal to the gearbox are
52.80 rpm are 141.69 rpm at full load conditions.
The fundamental rotational speed frequencies for these shafts at full load conditions are 0.32 Hz, 0.88 Hz
and 2.36 Hz respectively. Since part of the spectrum, which may contain current components due to broken rotor
bars, span from 58Hz to 62 Hz, only the last two reduction stages may give rise to a series of current components in
the part of the current spectrum where
678325
broken rotor bar components may be
Broken Rotor Bar Components Spectrum
present. Specifically, two components
supply
0
frequency
may be found in the upper and lower
search band.
-10

Iris Rotating Machine Conference

June 2003, Santa Monica, CA

-20

Frequency Components [dB]

The motor was operating at

250
Amp
corresponding
to
approximately 77% load. Based on the
gearbox nameplate data, the instrument
was able to correctly identify the
current components caused by the
shafts in the gearbox. The positions of
these components are displayed in
Figure 6. As can be seen in Figure 8,
the two current components in each
search band are indeed caused by the
gearbox.
Specifically, the two
components in each search band are a
4th harmonic from the 3rd reduction

-30
-40
-50
-60

Noise
lower
sideband

-70
57.0

57.5

58.0

58.5

upper
sideband
59.0

59.5 60.0 60.5

Frequency [Hz]

61.0

61.5

62.0

62.5

63.0

Figure 6: Current spectrum acquired on a coal mill motor connected

to a gearbox.
Page 6 of 8

stage and the 2nd

harmonic from the 2nd
reduction stage. Since
the current components
in the search band are
caused by the gearbox
and not by the presence
of broken rotor bars, the
motor was diagnosed as
not being subjected to
broken rotor bars.
This
example
clearly demonstrates that Figure 8: Current spectrum acquired on a coal mill motor connected to a gearbox.
gearbox
components Several mechanical current components are present in the spectrum.
need
be
correctly
identified and omitted
from the analysis. If the influence of gearbox components is not considered when identifying the presence of broken
rotor bar current components in the current spectrum, otherwise healthy machines may be incorrectly diagnosed as
unhealthy. This may lead to significant costs and may thus raise questions regarding the value of any condition
based monitoring program such a system being a part of.

CONCLUSIONS
A new portable instrument, called the CSMeter, has been developed for reliable on-line detection of broken
rotor bars and abnormal levels of air gap eccentricity. The instrument is the first of its kind to be fully portable and
provide on-the-spot diagnostics about the state of the rotor winding and air gap eccentricity problems.
By entering nameplate data, the instrument allows for reliable detection of current components due to
broken rotor bars and abnormal levels of air gap eccentricity. As shown in the paper, reliable detection of broken
rotor bars, can also involve the correct identification of current components due to mechanical components in the
drive system. The instrument takes these factors into account when performing the analysis.
Based on preliminary field testing, it can be concluded that the CSMeter provides reliable diagnosis about
the state of the rotor winding by estimating the number of broken rotor bars and detecting abnormal levels of air gap
eccentricity.
ACKNOWLEDGEMENTS
The authors would like to thank Greg Stone of Iris Power Engineering for initiating the development
project

REFERENCES
[1]
J R Ca meron, W T Thomson and A B Dow: Vibration and Current Monitoring for Detecting Air gap
Eccentricity in Large Induction Motors, Proc IEE Journal, Part B, Vol 133, No 3, May 1986.
[2]

W T Thomson and D Rankin: Case Histories of Rotor Winding Fault Diagnosis in Induction Motors, 2nd
Int Conf Proc on Condition Monitoring, University College Swansea, March 1987.

[3]

G B Kliman and J Stein: Induction Motor Fault Detection Via Passive Current Monitoring, Proc Int Conf
(ICEM90), MIT, Boston, USA, 1990, pp 13-17.

[4]

W T Thomson, D. Rankin and S. J. Chalmers: On-line Current Monitoring and Fault Diagnosis in High
Voltage Induction Motors Case Histories and Cost Savings in Offshore Installations, Offshore Europe 87,
Conf Proc SPE September 1987, Aberdeen, Scotland, SPE 16577/1 SPE 16577/10.

Iris Rotating Machine Conference

June 2003, Santa Monica, CA

Page 7 of 8

[5]

F Filippetti, G Francheschini and C Tassoni: Neural Networks Aided On-line Diagnostics of Induction
Machines, Transactions IEEE-IAS, July/August, Vol 31, No 4, 1995, pp 892-899

[6]

R R Schoen, B K Lin, F G Habetter, H J Shlog and S Farag: An Unsupervised On-line System for Induction
Motor Fault Detection Using Stator Current Monitoring, IEEE-IAS Transactions, November/December, Vol
31, No 6, 1995, pp 1280-1286

[7]

P Tavner and J Penman: Condition Monitoring of Electrical Machines, Research Studies Ltd, John Wiley
& Sons, England, 1987.

[8]

C Hargis, B G Gaydon and K Kamish: The Detection of Rotor Defects in Induction Motors, Proc IEE
EMDA Conf, London, 1982, pp 216-220.

[9]

W T Thomson, D Rankin and D G Dorrell: On-line Current Monitoring to Diagnose Air gap Eccentricity
An Industrial Case History of Large HV, 3-Phase Induction Motors, IEEE Transactions on Energy
Conversion, Dec. 1999.

[10]

J Penman, J G Hadwick and B Barbour: Detection of Faults in Electrical Machines by Examination of the
Axially Directed Fluxes, Proceedings ICEM78, Brussels

[11]

J Penman, H G Sedding, B A Lloyd and W T Fink: Detection and Location of Interturn Short Circuits in the
Stator Windings of Operating Motors, IEEE Transactions on Energy Conversion, Vol 9, No 4, December
1994

[12]

W.T. Thomson and M. Fenger: Current Signature Analysis to Detect Induction Motor Faults, IEEE
Industry Applications Magazine, July/August Issue, 2001.

Iris Rotating Machine Conference

June 2003, Santa Monica, CA

Page 8 of 8