Improving the transformer thermal modeling by considering additional thermal points

Abstract

Early detection of internal faults within the transformer reduces the outage and repair costs by planning the repair procedure according to the circumstances and prior to a catastrophic failure. Thermal faults are types of internal faults, which lead to high temperatures in the transformer causing accelerated aging of the insulation or its short-term degradation. Physical-thermal models can predict the temperature distribution within the transformer and detect its thermal faults. The accuracy of these models depends on heat transfer modeling, experimental factors, modeling elements, and estimation of transformer design-dependent parameters. This paper improves a physical-thermal model by considering extra thermal points in it. The model is applied to three distribution transformers with natural oil and air cooling systems. The accuracy of the model is demonstrated, and an algorithm for fault detection is also introduced. It is shown that considering the radiator thermal points in the model, enhances its accuracy in the calculation of top oil and radiator temperatures. Moreover, it turns the model to an evaluation tool for the temperature gradients on the transformer height measured in the thermography.