Lumped Parameter Thermal Network Modelling of Power Transformers

Abstract

The trend of increasing power densities of modern day power converters is pushing components like power transformers to their thermal limits, furthering the need for accurate thermal modelling. Numerical methods like Computational Fluid Dynamics (CFD) and Finite Element Analysis (FEA) are commonly used for the thermal characterization of electronic components like power transformers. Although, such numerical methods provide accurate thermal results they possess the drawback of having a large computational head. The present paper aims to address this issue by investigating a low cost Lumped Parameter Thermal Network (LPTN) model that can provide reasonably accurate results with greatly reduced computational effort. Further, the thermal network modelling methodology employed can be easily automated with a simple and intuitive method for thermal resistance calculation. In order to compare the accuracy of the proposed thermal network model to conventional numerical models, a coupled electromagnetic and CFD (multiphysics) analysis is conducted. Finally, the proposed thermal network model and the multiphysics model are experimentally validated on a PQ 40/30 transformer operating in a 3.3 kW Switch Mode Power Supply (SMPS). The proposed thermal network model is able to predict transformer operating temperatures within 10 % of the experimental results, with only a fraction of the computational time of the detailed multiphysics numerical model, providing a means of quick estimation of transformer thermal management requirements in the initial design phase.