Effects of Filtering and Current-angle Adjustment on the Multi-Material Topology Optimization of a 3-phase Stator

This paper investigates the multi-material topology optimization of the 3-phase stator of an electrical machine. The density-based optimization framework uses a gradient descent on the physical properties of the candidate materials (magnetic polarization and current density). The results indicate that a smoothing procedure is necessary to avoid non-manufacturable designs and promotes a broader exploration of solutions. They also highlight a current angle shifting due to the presence of permanent magnets from the beginning of the optimization. The current angle is therefore added to the optimization variables to address this issue. An efficient optimization procedure that couples both techniques is proposed to improve the performance of the algorithm. The following hybrid method returns symmetrical results that perform better than the asymmetrical structures obtained with the other techniques applied independently, indicating that the asymmetry represents a local optimum.