Modeling and simulation of failure mechanisms in low-voltage electric motors for predictive maintenance applications

Abstract

Through predictive maintenance of low-voltage electric motors, unexpected failures can be avoided. The multi-physics modeling and simulation framework with MATLAB FEA based on 1000 motor dataset was developed in the present study. Operational parameters like load (0.57-15 kW, mean 7.61 kW), voltage (215-246 V, mean 230.49 V), current (1-59.87 A, mean 30.08 A), vibration (-0.10-1.28 mm/s, mean 0.50 mm/s) and temperature (39.9-102.4 °C, mean 69.42 °C). Fault indicators include insulation degradation (0.1% to 99.9%, average 49.62%); bearing wear (0% to 99.9%, average 48.57%); rotor bar defects (0% to 99.6%, average 48.13%); and stator winding faults (0% to 99.9%, average 49.78%). Some results are that the temperature and insulation degradation correlate positively by 0.12, and that vibration and bearing wear correlate positively by 0.07. According to the simulation results, failure probabilities range from 0.48 to 1.00. In addition, 15% of simulations exceeded the early-warning level, requiring immediate motor maintenance.  According to the sensitivity analysis, if each fault mechanism is increased by 10%, the failure probability will increase by 4.5-5.1%. The framework facilitates condition-based maintenance and identifies motors at risk of failure with over 85% accuracy, as well as appropriate thresholds for early action. Results of the evaluation indicated that a quantitative and reliable approach can help improve motor reliability as well as reduce maintenance cost in industry.