Most of the industries still employ preventive maintenance programs, while few large industries have migrated to predictive maintenance initiatives.
In preventive maintenance programs, also commonly known as Planned Preventive Maintenance (PPM) scheme, the assets are inspected and maintenance activity is implemented periodically, as per a time schedule, based on the guidelines given by the equipment manufacturer or by the internal maintenance engineers. These could include greasing of a bearing every three months or change of bearing every three years etc. These activities are taken up irrespective of whether the bearing requires greasing or whether the bearing is in a very healthy condition or not. Such programs are also common in the automotive segment, where most of the users opt for periodic maintenance programs like changing the gear oil for every 5000 km of driving or overhauling of the engine for every 10000 km of run.
On the other hand, predictive maintenance (PdM) programs monitor the health of the equipment by assessing the state for various parameters to determine whether any maintenance intervention is required or not. Thus, predictive maintenance philosophy requires periodic or continuous monitoring of the health of the asset by using various technologies, which typically include analysis of vibration, ultrasonic, current, or thermal signatures of the assets. While some schemes employ one of these signature analysis methodologies, there are some schemes where multiple signatures are analyzed for better and more accurate prediction of the fault conditions.
In terms of cost, PdM programs require capital investments in terms of instruments and analytic tools, while the preventive maintenance programs are rule-based and do not require upfront investments; hence industries adopt PPM schemes. With the cost of the instruments related to various signature capture and analytic tools becoming more affordable, predictive maintenance is becoming popular.
The motor being the driver in all the electrically operated systems, with the driven parts being pumps, compressors, fans, etc., any strain in the driven parts will get reflected in the driver characteristics, which in turn distorts the current waveforms. Thus monitoring and analyzing the current signature is an easy and effective scheme to identify the nature of faults, and is popularly known as Motor Current Signature Analysis (MCSA).
MCSA is becoming popular due to the fact that it is sufficient to monitor at only one point – power supply input point, compared to vibration schemes where the monitoring may be required at multiple points in the system – driver and driven parts.
Most of the literature in this domain is still theoretical and little information is available in the public domain about the practical schemes and analytical methodologies. Also, comparative assessment of MCSA with the traditional vibration schemes is not available in the reported case studies.
We had designed an experimental set-up where both MCSA and vibration signatures were captured simultaneously and analyzed for specific fault conditions. This has given a deep insight into the relationship between the current and vibration signatures and proves beyond doubt that MCSA is indeed a good analytic tool for predicting most of the common faults.
The experimental setup is as shown in Figure 1. The system consists of a 3-phase 1HP motor with an operating speed of 1500 RPM, fitted with a 20 mm shaft with two bearings. The current sensors, using 10A clamp CTs, were fixed at the power input. Wireless vibration tri-axial sensors were mounted at four points; a) motor non-driving end, b) motor driving end, c) bearing #1 and d) bearing #2. All the data were collected in the cloud and analyzed for waveform and spectral properties.
Fig. 1 Experimental setup
The vibration velocity spectrum in mm/sec at bearing #2 is shown in Fig. 2. It is observed that the spectral components at 1X, 2X, and 3X are low. However, some stray components appear at 6X (specifically on Y-axis), indicating that the experimental setup is not perfect.
Fig. 2 Vibration spectrum – normal state
The current waveform and its demodulated spectral plots indicate low levels of 3rd harmonic, but with a marginally high level of 7th harmonic.
Fig. 3 Current waveform and spectrum for the normal state
The angular misalignment was introduced by adjusting the shims at bearing #2. The vibration signature shows high levels in the Y-axis direction (axial – in the direction of the shaft), as expected, with medium levels in radial horizontal, and vertical directions.
Fig. 4 Angular misalignment – vibration spectrum
The current waveform and its demodulated spectral plots indicate very high levels of 7th harmonics, with additional components appearing at third and other sub-harmonics; clearly indicating the fault frequencies associated with the angular misalignment fault condition. The distortions are clearly visible in the current waveform too.
Fig. 5 Angular misalignment – Current waveform and spectrum
The parallel misalignment was created by adding shims at both the bearing points. The vibration and current spectral plots are shown in Fig.6 and Fig.7 respectively. Horizontal components are dominant in many points, with the axial components being high at 1X. This defect is reflected in the current spectrum as high 3rd, 7th, and 13th harmonics in the current spectrum.
Fig. 6 Parallel misalignment – vibration spectrum
Fig. 7 Parallel misalignment – Current waveform and spectrum