Multi-Physics Modeling and Optimization of Advanced Electric Machinery and Magnetic Gear Development

Abstract

Over the last two decades, the subject of multi-physical optimization and co-design of electro-magnetic devices, such as electric machines for traction vehicles, or more exotic devices such as magnetic gears, has gained significant interest. While electric machines are widely used in industry now, researchers and industry alike continue to seek methods of reducing cost and increasing performance, which has necessitated further research in the development of advanced optimization approaches. New advanced optimization algorithms and the study of the multi-physics modeling of electric machines may be adaptable for application to emerging magnetic gear technology to increase the performance of various magnetic gear topologies. Due to the nonlinear relationship between machine design parameters and electric machine performance, electric machine design typically requires implementation of an optimization algorithm. Conventional optimization algorithms often use thousands of design evaluations. While various design evaluation methodologies have been proposed, the most accurate approach to electric machine design typically employs finite element analysis (FEA) methods. Thousands of design evaluations may be feasible when optimizing in a single physics-domain (e.g. electromagnetic domain), however, this approach quickly becomes untenable as additional physics domains, such as the structural domain, or when considering losses at multiple operating points of an electric machine used in a traction vehicle. A 75% faster method of multi-physically modeling drive cycle losses is proposed and used for optimizing an electric machine for traction vehicle propulsion applications, yielding a machine which exhibits lower losses over the entire drive cycle. Magnetic gears, like mechanical gears, convert power between high-speed, low-torque rotation, and low-speed, high-torque rotation. However, the former does so without mechanical meshing of teeth. Rather, magnetic gears leverage the modulated interaction between the flux generated by permanent magnets on the rotors. Consequently, magnetic gears offer the potential to combine the compact size and cost effectiveness of mechanically geared systems with the reliability and quieter operation of larger direct drive machines. Yet, magnetic gears are a relatively immature technology, and significant work remains before enabling researchers to easily employ multi-physics optimization approaches to co-design magnetic gears. There exists little literature on modeling any physics other than the electromagnetic domain relating to magnetic gears. This manuscript provides insight into manufacturability considerations and structural modeling of magnetic gears to support researchers in the future developing magnetic gear technologies. While numerous magnetic gear topologies exist, this work focuses on the development of analysis and design techniques for axial and radial flux coaxial and cycloidal-type magnetic gears and magnetically geared machines. Prototypes of two radial flux magnetic gears, an axial flux magnetic gear, three cycloidal-type radial flux magnetic gears, a novel hybrid radial and axial flux cycloidal-type magnetic gear, and a novel integrated two-stage radial-flux axial-flux magnetic gear with integrated motor prototypes were developed. All of the above were tested to calibrate and validate the analysis tools and investigate the practical considerations associated with the technology. Despite conservative design practices, the largest of these machines achieved a specific torque higher than the industry mechanical cycloidal analog. Various MATLAB-based software tools were developed to support these optimization approaches.