ECS-9806023 McInery The objective of this research project is to develop a method for early detection of insulation breakdown in low voltage, PWM controlled, ac induction motors. The method will be empirically based, but the data acquired in the experimental phase of the project will be well suited to validation of model based / predictive methods that may be developed in the future. Insulation breakdown and internal connection failures cause roughly 37% of all motor failures. Insulation breakdown is known to begin in a localized area of weakness. It is posited that the local, small scale changes associated with the initial stages of insulation breakdown will be reflected in a motor's high frequency response. This hypothesis forms the basis of the proposed research project. In standard duty, fixed speed motors, there is nothing to excite the response of the motor at high frequencies. And, whereas it is difficult to generate high frequency sinusoidal voltages of sufficient power to excite most industrial motors, PWM drives by their very nature generate significant energy out to frequencies well beyond 40 kHz. In this project, high bandwidth, high dynamic range measurements of the voltages and currents into and out of a 50 HP PWM drive will be acquired over a wide range of operating conditions. The motor speed (RPM) and output torque will also be recorded. Localized insulation weakness will be induced in various regions of the motor's stator windings using spot heaters. Local motor temperatures will be recorded using RTDS. All full scale experimental tests will be performed in the motor test facility at Oak Ridge National Laboratories. The acquired data will analyzed to determine what physical quantity (e.g., current, power, or impedance) and what feature extraction technique (e.g., spectral, wavelet, or higher order spectral analysis) are optimal for the early detection insulation breakdown in low voltage ac induction motors. Once a method of detecting insulation breakdown is defined, the required sensors and algorithms can be incorporated into adjustable speed drives by their manufacturers. Ideally, no sensors beyond the control sensors currently used in these drives would be required. A technique for detecting insulation breakdown in its early stages and monitoring its progression in low voltage motors would fill a significant gap in existing predictive maintenance programs. The maturity and success of electronic drives suggests that their design principles and the techniques developed in this project could be applied to the development of portable instrumentation for insulation health monitoring in fixed speed motors (in industry, manufacturing, on hermetic chiller motors, electric vehicle motors,...). A POWRE grant will provide the essential equipment, supplies and time to help keep the PI focused in her new research ventures.
