Advanced Dynamic Model of E-motor for Control Rapid Prototyping

The thesis deals with the improvement of electric motor simulation for control calibration under healthy and faulty conditions, including the effects of PWM voltage and electric motor harmonics fields. The design study of an electric motor (eMotor) starts with the electromagnetic design, using Finite Element Analysis for the evaluation of the motor output figures and parameters. This is the original purpose of the SyR-e (Synchronous Reluctance-evolution) open source design environment. More recently, the SyreDrive add-on was introduced, whose purpose is to generate a Simulink model for control calibration and accurate waveform simulation, starting from the eMotor design results. Therefore, the thesis aims at developing a unified circuital Simulink and PLECS model for eDrive (e-motor + inverter +control) suitable for both Permanent Magnet Synchronous Machine (PMSM) and Synchronous Reluctance (SyR) machines. The starting point of the work is the non-circuital, discrete-time average Simulink model available in SyreDrive that is considered as the benchmark for comparison of the new findings. The benchmark main limitations are of being time averaged, thus neglecting the instantaneous PWM evolution; of being non-circuital, and therefore not compatible with the analysis of fault and uncontrolled scenarios; and of requiring the inversion of the flux map tables (dq currents function of dq flux linkages). The goal of the thesis is to set up a new circuital model of the PMSM and the inverter, valid for instantaneous and time-averaged simulations, and covering faulty operating conditions of the inverter and motor sides. Two motor modeling approaches are considered, the Controlled Current Generators (CCG) model and the Voltage Behind Reactance (VBR) model. The CCG model represents the motor as three controlled current generators. The currents that pilot the generators come from the Inverse Flux Maps of the machine (dq flux linkage to dq currents). The VBR model represents the motor as an RLE load, with coupled inductors and controlled voltage generators as back EMFs. The model need the Inductance Maps (dq current to dq incremental inductances) and the Flux Maps (dq currents to dq flux linkage). Both CCG and VBR models have been comparatively implemented using the Simscape library of Simulink and in PLECS, using a torque control scheme, with evaluation of the respective computational times. The VBR model is the most promising candidate for general use, although it is heavier computational wise. The key downside of the CCG model is the use of the Inverse Flux Maps, which limit the operating domain with respect to the domain of identification of the experimental or FEA evaluated Flux Maps. Experimental validation has been performed using a traction PMSM (rated 70 kW, 130Nm, 6 poles) and it relates to comparing the phase current ripple in steady-state operating conditions and in the simulation of selected fault conditions. In addition, a FEA software is also used for comparison of phase currents in fault condition. At the end of the thesis, one of the two models will be embedded into public repository of SyR-e on GitHub.