2-D Steady-State Heat Transfer Prediction in Rotating Electrical Machines Taking into account Materials Anisotropy: Thermal Resistances Network, Exact Analytical and Hybrid Methods

Abstract

This paper presents two-dimensional (2-D) thermal resistances network (TRNM), exact analytical (AM) and hybrid (HM) methods for calculating steady-state temperature and heat flux distribution in rotating electrical machines considering materials anisotropy (i.e., different thermal conductivities in both directions). They are based on the thermal equivalent circuit (TEC), the improved exact subdomain (SD) technique where the solution and thermal conductivities depend on both directions (r, theta) and the coupling between the two methods. TRNM is known as a semi-analytical method that can predict the heat transfer in the machine in less time than finite element method (Fem). The implementation of TRNM by considering the difference between the thermal
conductivities in (r, theta)
using its equivalence with Fem is presented. The SD technique is improved to consider the difference between thermal conductivities in the directions (r, theta). It is known that the SD technique with non-homogeneous boundary conditions (BCs) is very sensitive to the dimensions of SDs where the harmonics number and the accuracy are lower in small subdomains. Hence, the HM from the TRNM and AM is given to answer these inaccuracies especially in electrical machines with a high number of stator slots and rotor poles. The heat sources are volumetric power losses due to hysteresis, eddy-current, Joule losses and windage losses in all the regions of the machine obtained by a simplified method. The studied problem is conductive with conductive interface conditions (ICs) and convective heat transfer between the machine and the external air and at the rotor internal air.
The semi-analytical results are compared between them as well as with those obtained by Fem.