Thermal characterization of an interior permanent magnet electric motor

An engineering approach for thermal characterization of an interior permanent magnet electric motor (EM) with a specific aim to identify the interdependency between its mean coil temperature and the maximum component temperatures at the conditions with constant and variable loads/speeds is proposed. For specifying the convective thermal boundary condition between stator and rotor that spins at variable speeds/directions, an experimental study aimed at devising the empirical heat transfer correlation for the Taylor-Couette airflow with varying rotating speed/direction is performed. It is firstly revealed that the responsive air-gap heat convection to angular acceleration follows the characteristic of a single-degree-of-freedom (SDOF) dynamic system. Along with the heat transfer correlation for the unsteady Taylor-Couette airflow, the nonhomogeneous thermal conductivity model that considers the orthogonality effect of the bent coiled winding at its two axial ends is integrated with an iteration scheme for calculating the air bulk temperature in an entrapped air chamber. The favorable experimental validation with the maximum discrepancies between predicted and measured component temperatures less than 7 % in steady and unsteady running conditions affirms the accuracy of the proposed simulation model. At a rotating speed of 4800 rev/min and power output of 1688 W, the maximum-to-mean temperature differences of stator winding, magnetic stripe, and steel sheet of stator reach respectively 6.75 °C, 3.21 °C, and 6.94 °C. Particularly, all the maximum component temperatures are well correlative with the mean temperature of the stator winding at the various steady/unsteady operating conditions. This finding extends the application of lumped thermal network analysis that predicts the averaged components’ temperatures; and enables the on-line monitoring for the maximum component temperatures using a mean stator-winding temperature as the controlling factor.