Top of Form

Chemical

Chemical

Food & Beverage

Food & Beverage

Hydrocarbons & Energy

Hydrocarbons & Energy

Life Sciences

Life Sciences

LNG

LNG

Power

Power

Pulp & Paper

Pulp & Paper

Refining

Refining

Water & Wastewater

Water & Wastewater

DoctorKnow® Application Paper

Title: Evaluation of the Field Application of Motor Current Analysis

Source/ S.W.Bowers, K.R.Piety, R.W.Piety, Richard J. Colsher
Author:
Product: General
Technolog Motor, Vibration
y:
Classificat Not Classified
ion:

EVALUATION OF THE FIELD APPLICATION

OF MOTOR CURRENT ANALYSIS
S.W. Bowers, K.R. Piety, and R.W. Piety Richard J. Colsher - Program
Manager
Computational Systems, Incorporated Electric Power Research Institute

INTRODUCTION

Electric motors experience a wide range of mechanical problems common to

most machinery, such as, imbalance, misalignment, bearing failures, and
resonance. However, electric motors also experience their own peculiar set
of problems which are a result of electromagnetically generated fields in the
stator and rotor. Electric Power Research Institute (EPRI) Report EL-2678
Volume 1 prepared by General Electric Company summarized data compiled
from a survey which included approximately 4800 motors of over 100 hp. Of
1227 reported failures 47 per cent were either stator or rotor related.
Therefore, it is hardly surprising recent years have seen a proliferation of
discussion regarding on-line testing of electric motors for some of these
phenomenon. While vibration measurements have historically been the
mainstay of most on-line condition monitoring programs, new techniques,
such as those involving spectral analysis of the electric line current powering
the motor, are commanding significant interest.

EPRI Project 1864-4 commissioned Computational Systems, Inc., (CSI) to

take a broad look at the results of electric motor diagnostics (electric current
signature analysis for broken rotor bars, in particular) when incorporated into
on-going predictive maintenance programs at several fossil-fired and nuclear
power generation plants. This paper briefly looks at several conclusions and
observations resulting from this work. The measurement and diagnostic
techniques used are not discussed in detail here due to limitations of time
and space. All techniques are well documented in literature and are
discussed in the final report to be issued by EPRI. All results are from non-
synchronous squirrel cage induction motors.

SUMMARY OF CONCLUSIONS AND OBSERVATIONS

Through the cooperation of a number of utilities over 10,000 electric current

and vibration spectra were acquired from 352 motors.
Results of Electric Current Analysis for Broken Rotor Bars. Of 352
motors tested, 65 were never run at sufficient sustained load for the data to
be considered reliable for rotor bar analysis. Of the remaining 287 motors,
85.7 per cent of the motors were estimated to have less than one broken
rotor bar, 11.5% with 1 to 3 broken bars, and 2.8% were calculated to have
greater than three broken rotor bars. The trend of the data proved to be very
repeatable if the load was maintained above 50 to 60 per cent of full load.
The motors with sufficient data for detailed trend plots showed no significant
degradation over two years even when 4-5 broken rotor bars were present.
However, these units were seldom started or shut down and typically ran at
moderate loads. Motors subjected to maximum loads and harsh operating
conditions would not be expected to be as stable.

Review of Speed Identification Techniques for Use in Rotor Bar

Analysis. Use of a 0-80 Hz, high resolution electric current signature has
been proven to be a reliable source of information for rotor bar fault detection
when used properly (see Appendix A). However, the technique relies on
accurately identifying the sideband of line frequency occurring at the line
frequency (Lf) minus the number of poles (NP) times slip frequency (SF).
This can require very accurate information regarding the rotational speed of
the motor to prevent selection of the wrong peak as the sideband frequency.
The vibration signature, flux spectrum, 0-80 Hz electric current signature,
demodulated signature, and MCSA signature were evaluated for use in
providing an accurate motor speed for rotor bar fault calculations. The ideal
method would allow the speed determination to be made from data collected
at the same location as the electric current signature, be suitable for use in
manual or automated programs, and be applicable to all motors. No single
method was found which met all these requirements. However, most motors
could have an accurate speed determined from one, or more, of the listed
methods.

Does Electric Current Analysis for Broken Rotor Bars Always Work?
There have been a few instances encountered when the reliability of using
the dB difference between the Lf amplitude and the Lf-(NPxSF) sideband
amplitude to predict broken rotor bars has been questioned. On each
occasion, there has been sufficient data missing to prevent a clear indictment
against the procedure's reliability. For example: one instance was
encountered in this study when an inspection reported the majority of the
rotor bars in the motor to be broken or cracked. The vibration and electric
current data would not have supported the prediction of such extreme
electrical degradation. Unfortunately, the most recent electric current data
was collected 6 months prior to repairs. Two possible explanations are:

1 ) The motor underwent severe degradation during the 6 month interval.

2) The electric current technique requires a significant torsional variation to

occur as the result of the torque differences produced when broken bars
pass through the magnetic fields of the motor as opposed to good bars. If
these regions are not distinctly differentiated because there are many areas
with faulty bars distributed around the rotor, then this might cause the
torsional variation to behave differently then expected. Also, high numbers of
broken bars may not have the same additive effect as the first several
display.

It has not been possible to substantiate any of these theories. Without

complete and accurate documentation it will be difficult to evaluate any such
occurrence. The authors continue to solicit any case histories where this
technique appears to have failed to detect the presence of broken bars.

Vibration Versus Electric Current Analysis. Many analysts believe the

presence of NPxSF sidebands around the harmonics of motor speed in the
vibration signature are an indication of rotor bar problems. This was not
supported by the data in this study (see Appendix B). These sidebands did
not prove to be direct indicators for the presence of, or the severity of, broken
rotor bars. Neither the presence, amplitude, nor number of sidebands
correlated with the rotor bar analysis from the electric current data (including
some confirmed faults). While broken rotor bars may contribute to the
existence of these sidebands there seem to be other factors affecting their
presence in the vibration signature.

Correlation of Lf Harmonics to Electric Current Analysis Results. Is

there information in the harmonic distortion of the 0-800 Hz line current
signature which can be correlated to rotor bar faults? A population
distribution analysis of the ratio of the amplitude of line frequency harmonics
to line frequency was performed using data from 190 motors. The amplitude
of the Lf harmonics was not found to relate to the rotor bar condition.
However, similar units tended to have similar values. Therefore, the pattern
of harmonic distortion of line frequency in the electric current signature would
appear to be more directly related to the design/application of a motor than to
the condition.

Correlation of Energy Between the LF Harmonics to the Electric Current

Analysis. EPRI-sponsored research performed by General Electric has
shown certain sidebands of slip frequency present around the higher
harmonics of line frequency to be sensitive to rotor bar faults. The level of
"noise" in the frequency regions between the Lf harmonics in the 0-800 Hz
electric current signature was noted to decrease significantly when some of
the rotor bar faults in the study were repaired (see Appendix C).

An analysis was conducted to see if a correlation could be made between

the energy non-synchronous to Lf, and its harmonics, with rotor bar
condition. The results were mixed. Although some motors seemed to show a
distinct relationship between the prediction of broken rotor bars and the
"noise", many did not.

Population Survey of Vibration Amplitudes at Lf and 2xLf. The vibration

signatures from motors occasionally show some amount of vibration at Lf (60
Hz) and 2xLf (120 Hz). The 2xLf peaks, in particular, can be an indicator of
an electromagnetic fault. In an effort to quantify what amplitudes are
"normal", data was reviewed from the horizontal, vertical, and axial sensor
orientations of approximately 150 motors. At 60 Hz (1xLF), 95% of the motor
population had vibration amplitudes equal to, or less than, 0.035 ips
(horizontal), 0.026 ips (vertical), and 0.036 ips (axial). At 120 Hz (2xLF), 95%
of the motors had vibration amplitudes equal to, or less than, 0.084 ips
(horizontal), 0.070 ips (vertical), and 0.134 ips (axial) included 95% of the
motors.

Population Survey of Vibration Amplitudes for a High Frequency Band

Encompassing the Rotor Bar Pass and Stator Slot Pass Frequencies.
Another commonly occurring set of frequencies which are of interest to the
analyst, especially for identifying rotor gap eccentricity, occur at the number
of rotor bars (slots) or stator slots times motor rotational frequency, along
with associated sidebands of 2xLF (120 Hz) and motor rotational frequency.
What vibration amplitudes are generally seen as a result of these
frequencies? A general scan was made for the amplitude distribution of
energy above 800 Hz in extended range vibration spectra for 316 motors.
These spectra had a maximum frequency of approximately slot pass plus
400 Hz. Amplitudes were calculated as both acceleration and velocity. In this
group, 95% of the motors had amplitudes equal to, or less than, 0.764 g's
(rms.), or 0.040 ips (peak) for this frequency band.

Correlation of High Frequency Vibration to Comparable Electric Current

Vibration Frequencies. The same frequency band was used to compare the
vibration and electric current spectra for correlation. No consistent correlation
was found to exist when the respective amplitudes were compared. Also, no
correlation was found when the amplitude of the frequency band from 800 Hz
to the maximum frequency in the vibration signature was compared to the
ratio of the Lf amplitude to the amplitude of the band from 800 Hz to the
maximum frequency in the electric current signature. In several cases, where
the vibration amplitudes varied significantly, the electric current signature
stayed relatively constant for the same motor.

Would this same observation apply when a known eccentricity was present?
Since no documented examples of eccentricity existed in the study database,
a variable static eccentricity was introduced into a 1.5 horsepower motor.
Lab tests showed a definite change in the vibration amplitudes observed at
the rotor slot pass frequency when the clearance between the rotor and
stator was varied. However, no corresponding change was detected in the
electric current signature.

A substantial volume of data from motors with eccentricity is a definite need

for truly evaluating the effects of this type fault and substantiating these
results. It was rather surprising that working with this number of utilities for
two years did not produce one documented case of eccentricity, much less
cases with the full range of data desired for comparison and correlation. In
light of these results, the most logical course for further evaluation of
eccentricity faults is large-scale lab simulations.

GENERAL RECOMMENDATIONS FOR MONITORING ELECTRIC

MOTORS

As a result of the experience gained with the data seen in this project,
feedback from utilities, and the authors' prior experience with vibration-based
predictive maintenance programs, the following general recommendations
for the routine monitoring of electric motors have been formulated:

I. A minimum of 7 vibration measurement points should be established for

the two bearings of the typical electric motor. Data should be acquired
monthly. Data from the first 5 of these points will be used for standard
vibration analysis and trending, for items such as imbalance, misalignment,
looseness, and bearing faults. Motors having sleeve bearings should be
monitored from 0-15 orders of rotational frequency. Motors with rolling
element bearings should be monitored from 0 to at least 6 times the ball pass
frequency of the inner race (BPFI), or approximately 40-50 times rotational
frequency. A resolution of 1-2 Hz per line of resolution is usually more than
sufficient for this data. These 5 points are:

A. MOH - motor outboard horizontal

B. MOV - motor outboard vertical
C. MOA (or MIA) - motor outboard (or inboard) axial
D. MIH - motor inboard horizontal
E. MIV - motor inboard vertical

Data from two additional points will be used to monitor changes occurring at
the higher frequencies, especially in the areas about rotor and stator slot
pass. These spectra should acquire data out to 2 times rotor, or stator, slot
pass (use the greater of the two) plus 400 Hz. The next section describes
taking data at these two points at a much higher resolution for baselines and
analysis than is needed monthly. For the monthly data, use 400-800 lines of
resolution in order to reduce data storage requirements. These two points
are:

F. EOH - extended range, motor outboard horizontal

G. EIH - extended range, motor inboard horizontal

II. Five measurement points should be established for special analytical

purposes. These points will have data acquired once (preferably soon after
commissioning) to use as a baseline reference, and yearly thereafter, except
when needed to investigate a problem. The points beginning with "H" are
high resolution points used to separate electrical frequencies from other
nearby frequencies. These should span 0-200 Hz with a resolution of at least
0.125 Hz per line of resolution (1600 lines). These three points will be useful
when vibration related to an electrical phenomenon is suspected, such as
running off magnetic center, looseness or flexibility in the stator, and
eccentricity.

A. HOH - high resolution, motor outboard horizontal

B. HIH - high resolution, motor inboard vertical
C. HOA (or HIA) - high resolution, motor outboard (or inboard) axial

Two more measurement points beginning with "E" should be set up to

acquire data to a maximum frequency of 2 times rotor, or stator, slot pass
frequency (use the greater of the two) plus 400 Hz. If the number of
rotor/stator slots is unknown, then the user should collect data out to a
maximum frequency of 5000 Hz. This area is of interest for various stator,
rotor, and eccentricity problems. Use 3200 lines of resolution for baselines
and analysis. For routine data, such as a monthly route, use 400-800 lines
for these two points. The high frequency range for some motors may require
stud mounting of the accelerometer for best results. These points are:

D. EOH - extended range, motor outboard horizontal

E. EIH - extended range, motor inboard horizontal

III. Electric current signatures should be acquired as a baseline and at a

minimum of yearly thereafter (quarterly or semi-annually preferred) for all
motors. Motors suspected of having rotor bar problems or operated under
rigorous conditions (frequent starts/stops, starting under load, shock loading,
running near or above full load, etc.) should have this data acquired quarterly
or even monthly, depending on circumstances. All three phases should be
checked for uniform voltages and loading. One phase should have an
electric current signature acquired which spans 0-80 Hz with at least 1600
lines of resolution. This signature will be used to look for broken rotor bars
and shorting ring problems. The other two phases can be monitored with as
low as 100 lines of resolution to ascertain the balance between phases. The
motors should always be above 50 per cent load (70-100 per cent preferred)
and have been run sufficiently long to reach full operating temperatures.
These points might be labeled as follows:

A. RB1 - phase 1 current signature (Hi-resolution 1600-3200 lines)

B. RB2 - phase 2 current signature (Low resolution 100 lines)
C. RB3 - phase 3 current signature (Low resolution 100 lines)

IV. Temperature - the inlet air, outlet air, and skin temperatures should be
checked routinely to watch for overheating. This is especially important in
very dirty environments where the motors may be subject to clogging with
contaminants. Elevated temperatures shorten insulation life and thereby
motor life.

Trending is an important tool for evaluating the seriousness of electric motor

faults. Many faults might be tolerated as long as the condition is not
degrading. As seen from the data, motors operated under moderate
conditions do not necessarily degrade even when several broken rotor bars
are present. The same can be true for other faults such as eccentricity.
However, keep in mind that the risks must be weighed carefully. If a broken
rotor bar moves out of the rotor and contacts the stator, the cost of repairs
can be significantly increased; this could happen instantaneously with no
time for trending to forewarn the analyst.

The analysis of the electric current signature of the motor has proven to be
very reliable for quantifying broken rotor bars when used properly. Many
other techniques have been shown to be sensitive to rotor bar faults, but
have not been shown to be as reliable or to provide an estimate numbers of
broken bars. During this study, no other motor faults were determined from
an examination of electric current signatures. Unfortunately, the number of
faults reported in the study was small. Nonetheless, our examination of
various line current spectra out to 800 Hz and 5000 Hz did not reveal any
useful findings for fault detection on the set of motors monitored.

APPENDIX A - ROTOR BAR ANALYSIS

Electric current analysis (sometimes referred to as motor current analysis)

applies various techniques to measure variations in the flux of the electric
motor. One of the most commonly applied techniques creates a frequency
signature of the electric line current. Data is collected from the windings of a
motor by installing a current transformer (CT) around one or more of the
conductors which provide power to the motor. The CT may be either
permanently installed or of the clamp-on design. A variation of this technique
uses a flux coil placed in close proximity to the motor frame. The CT has an
advantage in that no transducer is required at the motor so all readings can
be done from the switchgear room, allowing multiple motors to be checked
from one location (2). The use of the CT is also better documented and the
analysis procedure has been more refined than when using the flux coil.

Output from either device is input to a data acquisition instrument capable of

producing a time waveform and frequency spectrum. This instrument should
have a dynamic range in excess of 60 dB and be capable of high resolution
spectral plots of 3200 lines or greater. Sometimes the output may be further
conditioned using special filters or demodulation to reveal very low amplitude
or low frequency effects which might otherwise be masked by the dominant
line frequency peak at 60 Hz (LF).
A broken or cracked rotor bar creates heating and thermal expansion as
cracks heat due to increased current resistance and as adjacent bars now
carry additional current to compensate for the broken bar. These stresses,
which are especially aggravated by frequent starts, can cause adjacent bars
to break, thermal bowing of the rotor, or loosening of the broken bar. If the
broken bar works out of the rotor, then it could rub the stator causing
secondary damage and possibly resulting in catastrophic failure.

Torsional asymmetries created by the uneven distribution of current, and by

modulations of magnetic flux, torque, and rotational frequency, affect the
pattern and amplitudes of the line frequency harmonics and sidebands
present in the vibration and current signatures. High resistance joints and
cracked shorting rings (end rings) can have a similar, though usually much
less pronounced, effect. Since the vibration level of any of these effects can
be quite low, the indicators can often be obscured in the vibration spectrum
even when readily discernible in the electric current signature. The
sidebands of interest for diagnosing faulty rotor bars from an electric current
signature occur at a frequency offset from line frequency equal to the number
of poles times the slip frequency (NPxSF).

A commonly used indicator of the presence and severity of rotor bar faults is
the dB amplitude difference between the amplitude of the NPxSF side band
to the left of the line frequency peak and the amplitude of line frequency. This
sideband results from a prominent torsional asymmetry being produced. This
paper shall follow the convention used by General Electric Company and
EPRI in their article "Non-invasive Detection of Broken Rotor Bars in
Operating Induction Motors", and label this sideband asLSB1. Therefore
where:

Fr = motor rotational frequency

LSB1 = lower slip sideband of line frequency
Lf = fine frequency
NP = number of poles
SF = slip frequency

the following formulae may be used:

LSB1 = Lf - (NP * SF)

SF = (2 * Lf/NP) - F

Figure 1 shows data from a four pole motor with a rotational frequency of
29.74 Hz and a line frequency of 59.98 Hz. Using the given formulae we
obtain:

SF = (2 * 59.98Hz/4) - 29.74 = .25 Hz

LSB1 = 59.98 Hz - (4 * .25 Hz) = 58.98 Hz

Note the LSB1 sideband one hertz below line frequency at 59 Hz.
Fault Criteria and Constraints

A general rule of thumb for motors at operating temperature running over 50

percent load is that if the electric current signature has a delta difference
between the amplitude of the LSB1 peak and the amplitude of the line
frequency peak of greater than 50 to 55 dB the motor is good. The less the
dB difference (remember dB difference -- ratio) between the LSB1 and Lf
amplitudes, the greater the number of broken rotor bars predicted.
CAUTION! This relationship changes with load and may become
undependable at low loads. The closer to full load the better; don't rely on
data under 50 percent load.

Thompson and Rankin (8) formulate an equation based on the work of

Hargis and Gaydon (9) for deriving a conservative estimate of the number of
broken rotor bars. This formula often underestimates the actual number of
broken bars. Changing some of their symbols to correspond to those in this
report gives the following formula:

where:
n = estimated number of broken rotor bars
dB = dB amplitude difference between Line Frequency and LSB 1
NP = number of stator poles
R = number of rotor bars or slots

Using the motor current data in Figure 1 as an example, results in the

following values:
Lf amplitude = 47.3 dB
LSB1 amplitude = 14. I dB
AlP = 4 stator poles
R = 58 rotor bars

Use these values and the equation just given for estimating the number of
broken rotor bars to arrive at the following prediction:

As was stated this formula is conservative; this motor had 4 broken rotor
bars rather than the estimated 2.5 bars.

Effects of Load

The degree to which the unit is loaded affects the temperature of the motor
which in turn affects thermal expansion of cracks and breaks. Reduced load
may not produce enough heat to open cracked bars.

Load also affects the speed of the motor. As the motor becomes more
heavily loaded, the rotational frequency slows and the slip frequency
increases. The greater the slip, the greater is the frequency separation you
will observe between LSB1 and line frequency. The lighter the load, the
larger the ratio between line frequency amplitude and that of the LSB1
sideband; especially, as load moves below 50 percent of full load. From 50 to
100 percent load this effect is less significant. Figure 2 shows the effect as
the load moves from 25 to 115 percent of full load amps. Notice the
sidebands in relation to line frequency.
Figure 3 is from the same motor as Figure 1, except the load has been
reduced to 40 percent. In Figure 3 the line frequency amplitude is 42 dB and
the LSB1 amplitude is 4.5 dB. This gives a delta dB of 37.5 dB at 40 percent
load as opposed to 33.2 dB at 100 percent load. Using this new value results
in the following:

The data obtained when the motor was at 100 % load resulted in a prediction
of 2.5 broken rotor bars. The data at 40% resulted in a prediction of 1.5
broken rotor bars. More dramatic differences can occur. Always acquire data
as near full load as practical for the most reliable results.
Rotor Bar Analysis Results

A primary purpose of this study was to examine the use of current signatures
for rotor bar analysis as part of existing periodic predictive maintenance
surveys. Data was received from 9 utilities on a total of 371 motors over a
period of approximately 2 years. Some motors have 1 set of data, others
have multiple data acquisitions over the period. Table 1 summarizes the
results. Due to the quantity of data being reviewed, a commercial rotor bar
analysis program, MotorCheck®, which attempts to compensate for load
variation, was used to analyze the data. Figure 5 compares the manual
calculation using the given formula with the automated calculations across a
wide load range for a motor with 4 broken rotor bars. Data from motors with
less than 50% load were considered unreliable and therefore classified as
invalid in the table. There were also a few motors on crushers, pulverizers,
etc., where the load was so inconsistent that the data was judged invalid.
The net result was valid electric current data on 287 motors.
 Table 1 - ROTOR BAR ANALYSIS RESULTS

PRIORITIZATION

If the motors with invalid data are excluded, then 85.7% of the motors
indicated less than one broken rotor bar. At this level of fault prediction, the
motor is normally still considered to be in acceptable condition. The 11.5% of
the units which were estimated to have 1 to 3 broken bars may be
considered marginal and require increased monitoring of the electric current
signature for warning of future degradation. There may be a broken rotor bar
but other asymmetries such as high resistance joints or casting voids can
contribute to the prediction. Those units (2.8%) which were calculated to
have greater than three broken rotor bars were considered to have at least
one but probably multiple broken bars; these units should be monitored
closely and inspected/repaired at the next outage. If used in a application
requiring frequent stops/starts, especially if starting under load, then
continued use of these motors was not recommended.

CONSISTENCY OF READINGS AND TREND

Figures 5 through 8 show the trend of the number of broken rotor bars
estimated for four motors for a period of approximately two years. These
motors were chosen as examples due to the differences between the
calculated severity. In general, the data from all motors proved to be very
repeatable if the load was maintained above 50 to 60 per cent of full load.

Figures 5 and 6 show data from two forced draft (FD) fans which the electric
current analysis predicted to have approximately 5 broken rotor bars. While
FD Fan 4A shows a very slight increase in the number of broken bars
predicted, FD Fan 3B shows little or no change. At the end of the study FD
Fan 3B was inspected and found to have 4 broken rotor bars. These motors
are typically run at 65 to 70 per cent of full load and are seldom shutdown or
started.

Data was not acquired which could be used to trend a similar situation in any
motor subjected to shock loading, starting under load, or with frequent starts.
The motors we could trend showed very consistent data with no significant
degradation. Motors subjected to more severe service conditions would be
expected to have a much greater likelihood of rapid degradation and/or
failure.
APPENDIX B - VIBRATION AND ROTOR BAR ANALYSIS

Vibration analysis continues to be one of the most versatile and informative

tools available for on-line condition monitoring and problem analysis. Some
electromagnetically related faults, such as eccentricity, many stator faults,
and rotors off magnetic center, are easier to analyze and monitor using
vibration spectra than with an electric current signature given the present
state of the technology. Furthermore, vibration analysis is often required to
identify the root cause of the problem which can be due to structural or
mechanical sources such as bent rotors, thermal bowing, misalignment,
structural distortion, and resonance.

Rotor bar faults are a clear exception to the statement just presented, and
are difficult, if not impossible, to confirm or quantify using vibration alone. In
addition, rotor bar analysis using the electric current signature, is virtually
unique in that the extent of the fault can be mathematically estimated with
some accuracy.

Sidebands around Rotational Frequency Spaced at the Number of Poles

times Slip Frequency

When sidebands of the NPxSF are visible about the rotational frequency and
its harmonics in the vibration spectrum, is this an indication of a rotor bar
problem? The standard vibration signatures which are acquired often lack
sufficient resolution to see these sidebands. To evaluate the Occurrence of
this characteristic, high resolution (200 Hz at 3200 lines) vibration data was
acquired from 30 motors at a single Site.

Thirteen of the thirty motors had very clear sidebands about 1xRPM spaced
at the NPxSF. Most had the same sidebands about the harmonics of RPM as
well. Of the 13 motors with sidebands, 4 motors had an estimated number of
broken rotor bars of 1 or more. The most severe having an estimated 5.8
broken bars. The 2xLf (120 Hz) and slot pass frequencies in the vibration
signature of these motors were all below 0.02 inches per second peak
velocity and would not seem to indicate the presence of some other electrical
fault, such as an eccentric air gap.
Of the 17 motors which did not have clear sidebands in the vibration spectra,
all but one had readily discernible sidebands of NPxSF about line frequency
in the current signature. Four of the 17 were estimated to have between 1
and 2 broken rotor bars.

All motors (4 units) which had an amplitude difference of less than 40 dB

between line frequency and the NPxSF lower sideband, in the electric
current signature, did have sidebands in the vibration spectrum.

Figures 9, 10, and 11 show vibration data in linear and log scales, and
electric current data from a motor with 0.4 estimated broken rotor bars.
Figures 12, 13, and 14 show the comparable data from a motor with 5.8
estimated broken bars. The motor with less than one estimated broken rotor
bar has NPxSF sidebands of the highest amplitude and greatest number.
Use of these two factors from the vibration signature as the sole criteria
would erroneously indicate the first motor was the most severely degraded.
Neither condition was confirmed by inspection and/or rebuild.

These findings do not support the use of the NPxSF sidebands in the
vibration signature as direct indicators for the presence of, or the severity of,
broken rotor bars. Neither the amplitude nor the number of sidebands
appears to correlate with the electric current data. Very low levels of 120 Hz
and slot pass frequency would not indicate eccentricity as a potential cause.
While broken rotor bars may contribute to the existence of these sidebands
there seem to be other unidentified factors affecting this pattern which
complicate the interpretation. Nonetheless, it is the authors' opinion, that until
additional evidence is presented to substantiate these findings, the presence
of prominent NPxSF sidebands in the vibration signature should still prompt
the acquisition of an electric current signature for rotor bar analysis.
APPENDIX C - Comparison of the 0-800 HZ Electric Current Signature
Before and After the Repair of Four Broken Rotor Bars.
REFERENCES:

1. Improved Motors for Utility Applications Volume 1. EPRI Report EL-2678,

(General Electric Company, Schenectady, New York) October, 1982.
2. A.R. Crawford and S. Crawford, The Simplified Handbook of Vibration
Analysis, Volume I!, CSI, Knoxville, TN, as yet unpublished.
3. James E. Berry, Detection of Multiple Cracked Rotor Bars on Induction
Motors Using Both Vibration and Motor Current Analysis, Technical
Associates of Charlotte, Inc., 1992.
4. Thompson, W.T. et al, "An On-Line, Computer-Based Current Monitoring
System For Rotor Fault Diagnosis in 3-Phase Induction Motors",
Turbomachinery International, November/December, 1987, pp. 17-24.
5. Glenn H. Bate, "Vibration Diagnostics for Industrial Electric Motors Drives",
Bruel & Kjoer Application Note.
6. "Combining Motor Current and Spectral Analysis on a PC-Based System",
Computational Systems Inc. Application Paper, 1991.
7. Kliman, G.B. et al, "Non-invasive Detection of Broken Rotor Bars in
Operating Induction Motors", IEEE-PES, Transactions on Energy
Conversion. Vol. 3, No. 4, Dec., 1988.
8. W.T. Thomson and D. Rankin, "Case Histories of On-Line Rotor Cage
Fault Diagnosis", Conference Condition Monitoring, 1987, Department
Mechanical Engineering, University College Swansea, Pinerich Press,
Mumbles, Swansea, UK.
9. Hargis, C. et al, "The Detection of Rotor Defects in Induction Motors",
Proceedings IEEE Conference on Electrical Machines - Design and
Application, 1982, No. 213, pp. 216-220.
10. MotorCheck® Training Seminar, Computational Systems, Inc., Knoxville,
TN, 1991.
11. Nailen, R.L., Motors, "Power Plant Electrical Reference Series", Volume
6, Electric Power Research Institute, Palo Alto, CA, 1987.
 12. Campbell, W.R., "Alternating Current Electric Motor Problems", Arabian
American Oil Company, Dhahran, Saudi Arabia.
13. Sears, F.W. et al, College Physics, 4th Edition, Addison-Wesley
Publishing Company, 1977.
14. Current Measurements Catalog, Edition CT92-A, AEMC Instruments,
Boston, MA.
15. Elkasabgy, N.M., A.R. Eastman, and G.E. Dawson, "The Detection of
Broken Bars in the Cage Rotor of an Induction Machine", Dept. of Electrical
Engineering, Queen's University, Ontario, Canada, IEEE Paper No,
88CH2565 0/88/0000-0181, 1988.
16. Brancato, E.L., "Estimation of Lifetime Expectancies of Motors", IEEE
Electrical Insulation Magazine, Vol.8, No.3, May/June, 1992.
17. Cameron, J.R. et al, "Vibration and Current Monitoring for Detection of
Air Gap Eccentricity in Large Induction Motors", Proceedings IEEE, Vol. 133,
May, 1986.
18. Thompson, W.T. et al, "Failure Identification of Off-Shore Induction Motor
Systems Using On-Line Condition Monitoring", Proceedings of 4th National
Reliability Conference, Birmingham, UK, 1983.
19. R.C. Kryter and H.D. Haynes, "Condition Monitoring of Machinery Using
Motor Current Signature Analysis", Sound and Vibration, September 1989,
pp. 14-21.
20. Haynes, H. D., "Application of Signature Analysis for Determining the
Operational Readiness of Motor Operated Valves Under Blowdown Test
Conditions", Proceedings of the 16th Water Reactor Safety Information
Meeting at the National Bureau of Standards, October 24-27, 1988.
21. Smith, S. F. et al, "Machine Monitoring via Motor-Current Demodulation
Techniques", Proceedings of the 44th Meeting of the Mechanical Failures
Prevention Group, April 3-5, 1990.

All contents copyright © 1998 - 2006, Computational Systems, Inc.

All Rights Reserved.

Send comments to: Last Updated 05/08/03 © Emerson, 1996-2005

[email protected] Legal and Privacy Statements