Multilevel Boundary-Coupled Method for the Efficient Thermal Prediction of High-Frequency Transformers (HFTs) With Cylindrical Windings

Efficient and accurate thermal prediction is important for the optimal design of high-frequency transformers (HFTs). The configuration of HFTs is typically nonrotationally symmetrical. A 3-D finite-element model (FEM) is utilized for thermal predictions due to its high accuracy. However, its prediction efficiency is unsatisfactory due to three factors. First, the thin wires of HFT windings require a dense mesh in the 3-D FEM, increasing the computation time for each FEM iteration. Second, solving the 3-D physical field requires more computation time than solving 1-D or 2-D fields for each FEM node. Third, temperature-dependent HFT losses increase FEM iterations. The proposed multilevel boundary-coupled method tailors two strategies to enhance thermal prediction efficiency while maintaining satisfactory thermal prediction accuracy: 1) <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">multilevel boundary-coupled strategy</i>: solving the thermal field for the winding with a 2-D rotationally symmetrical FEM, solving the thermal field for the remaining region of the HFT with a 3-D FEM, and coupling the 3-D FEM and the 2-D FEM with their boundary temperatures and 2) <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">temperature-dependent HFT loss estimation strategy</i>: approximating the temperature-dependent HFT losses with the temperature estimated by a lumped parameter thermal model for the FEMs. The first strategy reduces the number of nodes and the dimension of the thermal field in the winding region of the FEM, thereby saving computations for each FEM iteration. The second strategy minimizes the additional FEM iterations for the calculation of the temperature-dependent HFT losses. Five HFTs with different winding and core configurations are employed for the comparison of the state-of-the-art method and the proposed method. The comparison results indicate that the proposed method improves thermal prediction efficiency while maintaining satisfactory thermal prediction accuracy.