Piecewise and Physics-Enhanced Topology Optimization for the Design of Electric Machines

Topology optimization has been receiving increasing attention in the past decade for the design of electric machines. Instead of relying on machine designers to come up with intuitive topologies, topology optimization evolves the topology by following the gradient descent direction of the objective function step by step. It may be used to optimize torque production by including the average torque in the objective function. This paper is to propound a new approach to formulating the optimization problem, aiming to simplify the convoluted relationship between the objective function and the design variables. Instead of optimizing the whole design domain at the same time, a piecewise strategy resorting to lumped parameters, such as that appearing in equivalent circuit models, is proposed. The physical connections between lumped elements and the design domain are leveraged to enable the methodology proposed herein. In this paper, results using an ideal setup and a synchronous reluctance machine are shown and demonstrated its effectiveness.