A Preliminary Exploration of the Coupling Algorithm of the POD-Galerkin Method and FVM in the Temperature Simulation for an IGBT Module

The proper orthogonal decomposition-Galerkin method (PGM) is an effective reduced-order method, which can predict in-situ physical fields rapidly and accurately based on a series of ex-situ results. However, the PGM suffers from difficulties in the application to the complex flow with openings and turbulence. This study preliminarily explores the coupling algorithm of the PGM and the finite volume method (FVM) to avoid these difficulties. A temperature simulation for an insulated gate bipolar transistor (IGBT) module is selected as an example for exhibition. The PGM is adopted to calculate the temperature field of the chip and substrate while the FVM is adopted to calculate the temperature fields of the external heat transfer. Moreover, the PGM and FVM performed in different computational domains are coupled at the interface. Two coupling algorithms are analyzed: (1) D-D (Dirichlet-Dirichlet): the temperatures at the interface are selected as the transmission information. (2) D-N (Dirichlet-Neumann): the temperature and the heat flux at the interface are selected as the transmission information. Results suggest that both the above coupling methods can obtain a continuous temperature field within the maximum deviation of 0.0045 degrees C compared with the results calculated by the FVM on the entire computational domain. The D-N coupling algorithm with additional source term method achieves the best convergence.