Open Collections

Abstract

Voltage-source converter (VSC)-based AC machine drives are the most widely utilized electromechanical systems. From industrial automation to electric transportation and marine electrical systems, AC machine drives are well-established for their high efficiency and superior control. The design and analysis of these drives rely on computer simulations of AC machine drives in electromagnetic transient (EMT) simulation programs. Commonly used detailed switching models (DSMs) are computational bottlenecks due to the need for small step sizes, which slows simulations. Instead of DSMs, average-value models (AVMs) have been developed that neglect switching, allowing larger step sizes and making them much faster. However, converter physics such as power losses, fundamental voltage drop, and bidirectional power flow – all of which the DSM can capture – are not captured by the conventional AVMs. This is because conventional analytical AVMs are derived under the assumption of an ideal power balance and a fixed mode of operation for the AC machine drive (motoring or generating). This thesis proposes new methods of parameterizing experimentally determined losses and fundamental voltage drop, which can be added to existing voltage-source inverter (VSI) AVMs. Further, this thesis develops bidirectional VSC AVMs capable of accommodating bidirectional power flow by first extending an existing parametric average-value modelling approach and then formulating a novel approach that considers bidirectional operation from the start point. Through extensive computer studies and experimental validation, it is demonstrated that the proposed lossy AVM (LAVM) accurately captures steady-state losses and fundamental voltage drop and produces accurate results in time-domain transients. The proposed bidirectional AVM methodology has been extended to the 120-degree VSI BLDC drives and demonstrated to produce excellent results in the time- and frequency-domain for motoring and generating modes.