Thermal Distribution Characteristics Related to the Transformer Fine Structure Predicted by CFD-Based Spline Interpolation

Abstract

Insulation overheating is a major cause of transformer failures, which makes it important to quickly and accurately predict the thermal distribution characteristics related to the transformer’s fine structure. In this work, a hybrid model of computational fluid dynamics (CFD)-based spline interpolation (CFD-SI model) is proposed for the real-time prediction of transformer thermal characteristics. Taking a 750-kV oil-immersed power transformer as an example, a CFD model is established from a complete conjugate heat transfer model and validated simultaneously by 24-h experiments and the empirical formulation. A group of 30 CFD simulations is employed to establish a hybrid model. Validation results of eight randomly selected cases demonstrate that this hybrid model can accurately describe heat flow distributions and predict hot-spot locations with a computational speed of 5570 times faster than the CFD model. Furthermore, a series of 42 predictions of hot spot temperatures under various scenarios are performed using this hybrid model to characterize insulation aging due to overheating, which enabled continuous management over its lifetime.