Multiphysics Analysis to Effectively Evaluate Thermal Performance of Liquid-Cooled Electric Machines

The prediction of the maximum operating temperature in electric machines is very important to ensure that the machine can produce the required power safely. Accurate thermal modeling is required to predict the heat transfer coefficient (HTC) of walls between coolant and heat sources and estimate the temperature in the machine. Analytical calculation of HTC is difficult for sophisticated geometric bodies since the dimensionless correlations are only available for simple geometries. To reduce the effort required to develop a thermal model of the cooling system of an electric machine, a two-way hybrid multiphysics approach using finite-element analysis (FEA) and computational fluid dynamics (CFD) to determine the HTC and evaluate the thermal performance of a liquid-cooled electric machine is presented in this article. In this study, a 60-kW switched reluctance machine (SRM) is used as an example model to evaluate its thermal performance. In the hybrid multiphysics approach, using the HTC estimated by CFD and heat generation in the machine as the inputs, the FEA is used to determine the maximum steady-state temperature in the machine. An analytical approach is also implemented to determine the HTC of the example SRM to correlate with the HTC obtained using a hybrid approach. The analytical HTC is used in FEA to obtain the temperature distribution in the machine. The temperature obtained from hybrid and analytical approaches is compared. The SRM considered in this study is built and tested for different operating points. The machine is tested for a long time to record the steady-state temperature at different operating points. Results obtained from the hybrid approach are validated with the experimental temperature data.

Automated summary

This article introduces a two-way hybrid multiphysics approach using finite-element analysis and computational fluid dynamics to predict the maximum steady-state temperature of electric machines. The method can accurately estimate the thermal performance of the machine and validate the results with experimental data.