An improved POD-Galerkin method for rapid prediction of three-dimensional temperature field for an IGBT module

Xiang You Feng, Fan Bai, Hao Ding, Wen Quan Tao*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


With the continuous improvement of power density and integration level of power devices, the temperature prediction of power devices becomes increasingly crucial for reliable operation and thermal management. Existing thermal analysis methods cannot meet the needs of fast and accurate calculation of detailed temperature fields. A novel three-dimensional fast prediction model of full temperature field for the Insulated Gate Bipolar Transistor (IGBT) module is established based on the improved POD-Galerkin method. The reliability and performance of the proposed model are assessed via the comparison with the finite volume method (FVM) on four types of problems, including two steady-state problems with the first and third boundary conditions, respectively, and two unsteady-state problems with a fixed and variable power loss of chips, respectively. The perfect agreement between the results of the two methods confirms the accuracy of the proposed model, and the comparison of the calculation time shows its ability to greatly reduce the computation effort. The proposed model can obtain the detailed information on the whole temperature field of the IGBT module accurately in a much shorter time, which is significant for temperature and thermal stress monitoring and the development of its digital twinning technology.

Original languageEnglish
Article number107241
JournalInternational Communications in Heat and Mass Transfer
Publication statusPublished - Mar 2024
Externally publishedYes


  • Fast temperature prediction
  • Heat conduction
  • Insulated Gate Bipolar Transistor (IGBT)
  • POD-Galerkin method


Dive into the research topics of 'An improved POD-Galerkin method for rapid prediction of three-dimensional temperature field for an IGBT module'. Together they form a unique fingerprint.

Cite this