MCSA is primarily collected using special current probes
which allow the data collectors to take current inputs. This current is then
converted from analog to digital form, filtered and transformed to FFT spectra
of amplitude versus frequency.
Most of the recent condition monitoring (CM) research has been focused
on utilizing electrical monitoring of the motors with emphasis on examining the
stator current of the motor 1. Even though thermal and vibration monitoring
have been utilized for decades, large amount of research has been recently
directed toward using MCSA to detect both of electrical and mechanical faults
It is commonly difficult
or even impossible to interpret the information contained in a raw signal to
meaningful information by just looking at it. Hence, all of the
presently available techniques require the user to have some degree of
expertise in order to distinguish a normal operating condition from a potential
failure mode. The raw signals obtained from an instrument measuring a physical process always contain
noise that hampers the useful information contained. This is because
the monitored spectral components can result from a number of sources, including
those related to normal operating conditions due to both the design and
construction of the motor and the variation in the driven load. 8– 11
illustrate few condition monitoring methods that have been proposed for
different types of rotating machine faults detection and localization.
This paper focuses on the MCSA that utilizes results of spectral
analysis of the stator current of an induction motor to spot an incipient or
existing fault of the motor, as the amount of information brought by the use of
advanced signal processing techniques, such as high-resolution spectral
analysis, is lower than that attainable from the use of classical spectral
analysis property 1.
The reminder of
the paper is organized as follows: Section 2 looks into MCSA. In Section 3, the
data acquisition system is described. The
rotor fault is summarized in section 4. The system responses and PSD diagrams
are deliberated in sections 5 and 6, respectively. The conclusion is drawn in
2. motor current signature analysis
Signature Analysis is a powerful monitoring tool for electrical machine and
motor-driven equipment that provides a nonintrusive mean for detecting the
presence of mechanical and electrical abnormalities in the motor and the driven
equipment, including distorted conditions in the process downstream of the
motor-driven equipment 12.
MCSA is an electric machine monitoring
technology developed in 1985 by Oak Ridge National Laboratory (ORNL) 13, 14
as a means for determining the effects of aging and service wear systems, and
it is applicable to a broad range of machinery. It has been used as a condition
monitoring method in electrical machines to detect and diagnose motor bearing
wear, eccentricity problems, stator faults, as well as rotor broken bars and
rings assessments for inaccessible motors during plant operation 12-15.
Variations in the
motor current though are very small compared to the average current drawn by
the motor, can be extracted reliably and non-intrusively at a remote location
from the equipment then processed for studying the machine condition. Another
advantage of motor current signatures is that it is obtainable in time and over
time to alarm any early signs of deterioration of the motor condition. In
conclusion, MCSA has been in the CM arena
for so long, giving enough proof of effectiveness at industrial environments
16. One of the original concepts behind the development of MCSA
was to eliminate the loss of instrumentation in the dangerous areas within
power plants 17.
3. DATA ACQUISITION SYSTEM
components used to create the Data Acquisition (DAQ) set-up is composed of few
sensors, National Instruments DAQ card, and a computer.
experimental work, all the collected data was acquired using the National
Instruments PXI cards. The PXI 6251 is a multifunction data acquisition card;
it accepts 68 pin cables with up to 24 digital inputs and 16 analogue inputs
and is housed within the PXI 1031.
During the data
acquiring process the sampling frequency was 20 KHz. 100,000 sample s were
recorded for each run, which means for such setting the required time to
collect this length of data is 5 Sec. All of these signals are available for
visual inspection of the operator on the PC screen. Fig. 1 shows the schematic
diagram of the current and voltage acquisition system.
Fig. 1 The system setup
4. Broken Rotor Failure
Broken Rotor Bars (BRB) faults can be a serious problem,
although they do not initially cause an induction motor to fail. Therefore,
there can be serious secondary effects. The fault mechanism may result in
broken parts of the rotor hitting the end windings or stator core in high
voltage motors at high speed. This can cause serious mechanical damage to the
insulation and a consequential winding failure may follow, resulting in a
costly repair and lost production. Hence the detection of incipient rotor defects is considered as an
essential part of motors condition monitoring.
The experimental scene, the partial bar fault was created by
drilling one of the bars by making a hole with gradually increased diameters of
2.5, 3.5, and 4.5 mm. Breaking one bar is realized through making a hole of 6 mm diameter and
with a depth that allows the full height of the bar is taken. Taking the full
bar height insures that the rotor bar is partially or completely broken. The last stage of this test was breaking two
adjacent bars of the rotor; by breaking each bar through a hole with 6 mm
diameter. Fig. 2 shows the rotor with one BRB.
Fig. 2 Rotor with 1 BRB fault
are broken or even cracked bars, the rotor’s impedance exhibits an unbalance.
The immediate consequence of such an unbalance is the existence of inverse
sequence currents. Hence the air gap flux density is distorted, and the
currents generate an inverse magnetic field (IMF) that turns counter-motor
Plots of Fig. 3 show the current response of the healthy and
faulty motor. The stator current is effected by the severity of the rotor fault
while the same load was applied as in Fig 3-A.
The plots, of Fig 3-B, show the direct and drastically
effect of loading on the line current of the motor with different BRB
severities. The more load applied to the motor, the more current drawn.
Fig. 3 Motor Current responses
Some signal distortion due to the fault is noticed; hence
the proper way to examine the effect of the fault upon the current and other
signals, in general, is to go through signal processing methods.
pSD FAULT Features
amplitude of IMF depends on the severity of fault and the value of the current
in the rotor bars. The IMF originated in the rotor produces harmonic currents
of frequency (1-2S)fS in the stator windings.
current harmonics interact with the main magnetic field and set a torque over
the rotor, which oscillates with a frequency of 2SfS 18.
The amplitude of this oscillation is a function of the motor’s load inertia. As
a reaction of such speed perturbation, new currents arise in the stator at a frequency
of (1+2S)fS. The new current component at frequency (1+2S)fS
is overlaid on the original, and then modifies its amplitude 16, 20 &21.
methods have exhibited good results for detection of broken rotor bar faults by
investigating the sideband components around the supply current fundamental
(line) frequency 18, 20 and 22.
sidebands in the stator current spectrum are located in the neighbourhood of frequencies
given by (1)-(3) 16 & 20.
fSB = (1± 2KS)fs
K = 1, 2, 3…,
fSB: Side bands broken rotor bar
fS: Electrical supply frequency (50
S: per unit slip speed.
NS: motor synchronous
NR: rotor mechanical
P: number of the stator poles.
Table 1 demonstrates the locations of the fault
features. The tested motor is a 4 pole, and is run at speed of 1442 rpm on
average, with 49.6 Nm load and synchronous
speed of 1500 rpm. The data of the tested motor is shown in the
appendix. For K=7, the faults harmonics are located at 46 Hz and 54 Hz.
In Fig. 4- C, the fault harmonics appear at 53.5 and
46.5 Hz, which are the first entry of Table 1.
Table 1 Locations of frequency side-band fSB
When the Pole Pass Sideband Frequencies (fSB1 and fSB2)
of Fig. 4 are compared to the values in Table 2 17, the condition of the
rotor bars can be easily determined.
Fig 4, it is apparent that the amplitude corresponding to both side frequencies
is in a direct proportionality with the severity of the fault.
Table 2 Rotor Bar Failure Levels
54 – 60
High resistant connection or Cracked Bars
Broken Rotor Bars Will Show in Vibration
Multiple Cracked/Broken Bars, Poss Slip, Ring
Severe Rotor Faults
comparison, three plots were put on the same graph. Fig.5 shows the difference
in amplitude between the healthy and faulty cases of the motor when same load
current plot response to the fault is noticeable. While the blue-coloured plot
characterizes the healthy case, those in green and red plots represent the PSD
of the rotor with one and two BRBs, respectively. The plots agree with the
guidelines laid in Table 2, accurately.
negative dB is about 70 for healthy rotor (>60 dB), this value decreases to
40 and 35 dB for one and 2 BRBs, respectively. The last figures alarm the
presence of broken rotor bars.
Fig. 4 PSD plots for healthy and faulty rotors
The primary failure of IM is the rotor which would overload
and melt when limit switches failed. It was discovered that the rotor bar
failure signature was unique enough that not only could the signature be
quickly identified, but that condition values could be applied easily 17.
Fig. 5 Current PSD for healthy and BRBs
rotor bars fault is presented and the side bands frequencies occurring in the
line current spectra of IM due to broken bars are described.
results showed the efficiency of the MCSA method to distinguish between the
different statuses of the rotor. MCSA can detect abnormal rotor operating
conditions in IMs in an efficient, reliable and cost effective manner.
IMs operated mostly at their rated load torque, in cases of insufficient motor
load, it is advised to repeat the measurements with higher loads when a
determinate decision could not be reached about a fault.
It was discovered that the rotor bar failure signature was
unique enough that not only could the signature be quickly identified, but that
condition values could be applied easily.