2024

Singh, Shruti; Scherman, Eero; Petrov, Ilya; Sergeant, Peter; Pyrhönen, Juha

A polynomial chaos based surrogate modeling method to quantify the uncertainties and perform the global sensitivity analysis for a thermal model of the electrical machine is presented. The method is demonstrated using a 27kW 100 krpm axially laminated synchronous reluctance machine (ALAsynRM) as a study case. Exact thermal parameters and thermal conditions in any electrical machine are somewhat unpredictable. Sensitivity analysis provides a platform to determine the machine's thermal working condition. However, this analysis can be time consuming and requires a lot of simulations. A probability density function is utilized to show the impact of changing thermal coefficients and input and output cooling fluid temperatures on the machine output temperature. This probability density function can be obtained even with limited number of parameters. Most of the machines, especially those having complicated structures require three dimensional thermal analysis to determine the operating temperature of different components. Therefore, a 3D model of a drum wound liquid cooled stator and the axially laminated air cooled rotor is utilized as a case study. It is shown that before the experimental analysis if the thermal coefficients and fluid temperatures are not available, a probability density function can be utilised to determine the range with less number of simulations.