Life Model for PWM Controlled Induction Motor Insulation using Design of Experiments Method

Abstract

According to numerous industrial surveys, the failure rate of stator insulation is much higher for Pulse Width Modulated (PWM) controlled induction motors than for the motors with sinusoidal voltage supply. Voltage spikes, high slew rate of spikes, and repetition and harmonics of PWM voltages generate higher electrical and thermal stresses in motor insulation that result into premature failure. In this paper, the Design of Experiments (DoE) method has been proposed to estimate the life of PWM controlled induction motor insulation. The insulation stress under PWM voltages is quantified by three factors viz. spike magnitude, spike frequency and voltage harmonics factor. These stress factors are experimentally computed for PWM controlled induction motor at different switching frequencies. Keeping these stress factors as input, a fuzzy logic model for insulation life is subsequently developed. This model is fine tuned to match the results obtained through insulation aging test with PWM voltages. The fuzzy logic data is analyzed by statistical techniques such as, ANOVA and Standardized Pareto Chart. A full factorial Design of Experiments technique is then applied to the fuzzy logic results to get the first order model relating the insulation life of induction motor with the stress factors. This model can be directly used for insulation life estimation of PWM controlled induction motors.