Modeling and Detection of Stator Winding Insulation Degradation in VSI-fed Electric Machines Under High dv/dt

Nowadays, in many advanced applications, for example electric vehicles and aircrafts, electric motors are being fed by voltage source inverters (VSIs). These inverters generally utilize fast-switching power electronics devices to provide the modulated pulse voltages for motor operation. Motor’s winding insulation undergoes additional electrical stress due to high voltage slew rate (dv/dt) resulted from the use of VSI, which may cause accelerated insulation degradation. To investigate and quantify this electrical stress, various modeling techniques have been proposed over the years with a goal to optimize the motor stator design for safer and more reliable operation. In this thesis, various winding insulation modeling techniques are investigated, in which a simple case study along with comparative analysis is performed to analyze the electrical stress for different dv/dt levels. Based on this investigation, finite element (FE) based approach is selected for detailed stator winding insulation modeling, parameter determination and analysis. Degradation of winding insulation is inevitable due to various internal and external factors. Therefore, monitoring of the insulation condition becomes necessary to ensure a safe and reliable operation of the VSI-fed electric machines. Detecting degradation in insulation in an early stage can prevent major failure in the machine and the drive system. To this end, an online insulation monitoring technique is required. Based on the existing literature, the high frequency transient current could be a good indicator for online insulation monitoring. In this thesis, how winding insulation degradation affects the machine transient current is investigated. To do so, complex random wound stator winding is modeled at first using the FE based approach, in which detailed arrangement of each wire inside the stator slot and different insulation layers are considered. Thereafter, insulation degradations of various types and severity are emulated within the FE based model, including the turn to ground (TG) degradations and turn to turn (TT) degradations. With the emulated winding insulation degradations, the effectiveness of the detection techniques is investigated. In this work, a new insulation condition detection technique based on current response and wavelet packet decomposition (WPD) is proposed, which has the capability to determine the overall state of health (SOH) of insulation and classify types of insulation degradation. Therefore, the stator winding insulation is modeled and the SOH indicator for insulation degradation detection is also developed through this thesis research