Non-steady-state modelling method for multi-material temperature field in dry-type transformers based on Proper Orthogonal Decomposition

• Published in: Electric power systems research Vol. 246; p. 111671
• Main Authors: Zhang, Zeyi, Lan, Haijuan, Gong, Jiahao, Xu, Xiongwen, Tang, Wenhu
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.09.2025
• Subjects: Multi-material reduced-order model; Proper Orthogonal Decomposition; Temperature fields; Transformer; Temperature fields; Transformer; Proper Orthogonal Decomposition; Multi-material reduced-order model
• ISSN: 0378-7796
• DOI: 10.1016/j.epsr.2025.111671

Power transformers play a crucial role in power systems, and the accurate analysis of transformer temperature fields, particularly hot-spot temperatures, is a key factor in ensuring the stable operation of transformers. A method for rapid prediction of temperature fields in fluid–structure interaction problems based on POD is presented. An SCB11-800/10kV dry-type transformer serves as the research subject, with both a full-order model and a reduced-order model constructed to solve for the temperature field and its unsteady temperature rise process at a given inlet wind speed. The discontinuity in heat flux density at the junctions of different materials is addressed by creating virtual vertices at these points, enabling the handling of heat transfer in multi-material reduced-order models. Steady-state temperature fields and the unsteady temperature rise process of hot-spots are used as indicators to validate the accuracy and efficiency of the proposed reduced-order method by computing the spatio-temporal distribution of transformer temperature fields at different load rates. Results demonstrate that reduced-order computations based on POD not only ensure the computational accuracy but also enhance the computational speed, suggesting a new effective method for the rapid prediction of the temperature field, especially the hot-spot temperature under different loading rates of the transformer. •Virtual vertices solve grad jumps, let ROMs predict thermal fields in multimaterial.•ROM accurately predicts temperature under rated/varied loads, matching FOM’s result.•ROM achieves 23.7 times speedup vs FOM, reaching seconds-level computation.