Parametric, self-segmenting mesh generation tool for the steady-state thermal estimation of induction machines

The purpose of this work is to create an automated, parametric, self-segmenting mesh generation tool to be used within a finite difference (FD) model for the estimation of temperature gradients in induction machines (IM). Rectangular, prismatic, and curved features, typically incorporated in conventional rotor bar geometries, are difficult to mesh in polar coordinates in the R-θ plane. New segmentation algorithms are required to address the modeling of these complex geometries in polar space. The segmentation study shows an appropriate radial step size of 0.5% of the air gap radius for the nodal mesh elements. This step size balances area accuracy to within 1% whilst maintaining computational speed. A 2D half slot comparison between the FD and FEA temperature distribution results in a maximum temperature difference and maximum percentage error of 1.4°C and 0.60% respectively.