MCSA is particularly well-suited to motors operating in harsh environments since the current monitoring sensor can be located in the motor controller cabinet, rather than on the motor. MCSA is a form of electrical signature analysis (ESA), a broader field which includes other methods suitable for induction motors such as voltage signature analysis and motor circuit analysis. ESA can be also be applied to generators, transformers and other electrical hardware. MCSA can be used as a stand-alone method, or combined with vibration and thermal monitoring.
A typical MCSA system consists of a current transformer attached to one of the phases of the motor’s supply cable, a signal conditioner and a spectrum analyser. When only one phase is monitored, as is usually sufficient, this is known as single stator current monitoring. MCSA can detect faults in the rotor as well as the stator, by effectively using the stator windings as a transducer to sense the currents induced by the rotor.
The signal conditioner processes the signal in the time domain, and the spectrum analyser then converts this into a frequency domain signal for analysis. The supply current to the motor is ideally a sinusoidal wave and deviations from this perfect signal can be readily detected in a spectral plot. A typical motor signal will include various sideband harmonics caused by electrical and mechanical imperfections and faults. These combine to present what appears to be fairly random interference in a time domain plot. However, when presented in the frequency domain, the frequency and amplitude of each individual disturbance can be clearly seen, giving valuable insights into the condition of the motor.
Motor faults produce disturbances with frequencies typically between 0-5 kHz. The sampling rate should be at least 10 kHz, since the Nyquist theorem shows that accurate signal characterization requires a sampling rate at least two times the highest frequency measured.
The most common faults encountered in motors are bearing faults and stator faults, with rotor faults also significant. MCSA is most effective for detecting faults in the stator or rotor, as well as significant misalignment between them. Although MCSA is able to detect bearing faults, these are generally more easily identified and characterised using vibration analysis. Similarly, vibration analysis is able to detect short circuits in stators, and loose or broken rotor bars, but these faults will be more readily detected by MCSA. For MCSA, the frequency induced by a particular fault condition depends on the operating conditions of the motor, and its synchronous speed, slip frequency and pole-pass frequency.
The fault conditions that MCSA is best at detecting are shorting or opening of the stator windings, broken or loose rotor bars, or significant misalignment that results in an non-uniform air gap between the rotor and the stator, known as air-gap eccentricity.
Although MCSA is widely used to detect broken rotor bars, it can produce nuisance warnings due to the motor operating at different load levels or a lack of baseline data. More recently-created fault detection algorithms seek to avoid these fault alarms and ensure reliable fault detection.
Dr Dubravko Miljković, of the Croatian Electricity Company, agrees that it is better to “apply vibration analysis for mechanical faults like bearing faults and misalignments.” By contrast, MCSA has, “…better sensitivity to electrical faults; also vibration sensors are expensive and there is an additional problem if such sensors have to work [in] harsh environments. Also MCSA is better (cheaper) for remote applications compared to vibration monitoring. The effectiveness of analysis on electrical problems using vibrational analysis would generally be much lower.
“Electrically-related problems, such as shorting/opening in the stator, broken rotor bars, slip frequency and eccentricity need to apply MCSA. By also monitoring mechanical vibration, you are able to identify the mechanically-related problems like unbalance, misalignment, bearing faults, coupling and load mechanical failures, and also to some extent electrically-related problems such as 2LF (line frequency). For mechanical defects, vibration [sensing] is definitely much better. It also depends on number of vibration/accelerometer sensors as well as their positioning in axes.”
He continues: “While MCSA/ESA works better for rotor bar damage in most cases, I have seen cases where it failed but the fault was detected using vibration analysis. The effectiveness of detecting electrical problems using vibrational analysis is generally much lower. I would say that MCSA and vibration analysis are complementary methods that should be used together when resources permit.”
For motors operating in harsh environments, such as immersion pumps operating down well shafts or within nuclear power plants, it may be problematic to obtain reliable condition monitoring data from sensors attached directly to the motor. MCSA has a major advantage in such situations since the current can be monitored by sensors installed inside the motor control cabinet, far from the harsh environment in which the motor is operating. The same cannot be said for vibration monitoring.
Limitations
On limitation of MCSA is that it is generally not able to determine whether a disturbance is caused by the motor or the power supply. Dr Miljković explains: “Conventional MCSA is unable to distinguish between distortions caused by the voltage waveform and faults that are a genuine indication of developing equipment problems. One solution is model-based MCSA known as model-based voltage and current (MBVI). Another approach is SVCM. The AC level of the space vector current modulation (SVCM) has been used to identify all the additional components, their frequencies and their superposition under distorted voltage supply. [Another option is] using total harmonic distortion (THD) with fuzzy logic. THD works because when stator winding faults occur, it generates harmonics in the line current. But harmonics are also created due to distorted supply voltage. So it is necessary to distinguish the distortion in line current due to motors’ internal problems and distortion due to distorted supply voltage. [Making a] distinction between these two phenomena is very challenging to predict the condition of motor. However, voltage and current signals can be used and harmonic distortion can been calculated for both cases. The harmonic distortion has been used as the input to a fuzzy logic classifier, which provides the condition of the motor.”
In conclusion, MCSA is a highly effective way to detect electrical faults in motors, such as shorting or opening of the stator windings, broken or loose rotor bars, or significant misalignment that results in an non-uniform air gap between the rotor and the stator. MCSA may also have some ability to detect mechanical faults such as bearing failures, although vibration analysis is typically more effective for this. However, MCSA may still be used instead of vibration analysis in situations where cost is an issues or motors operating in harsh environments would make the installation of vibration sensors challenging.