Transformer’s Loss of Life Prediction using a Dynamic Thermal Model

Abstract

Predicting the loss of life of an operating power transformer is of great interest in the economic, technical and social fields. The thermal modelling approach is widely used for estimating the loss of life of a transformer due to its ease of integration into a real-time tracking tool and the relatively small amount of input information. This article proposes a dynamic thermal model to predict the loss of life of a power transformer. The proposed model takes into account the variations in the operation of certain parameters of the transformer, such as the load, the thermal capacity, the oil time constant and the ambient temperature. The formulated model calculates the hot spot temperature by the differential equations method, and the aging rate is calculated by the conventional IEEE technique. The model inputs are derived from information acquired by sensors and stored in an Excel database. The model was tested on two transformers, ONAF 400 MVA transformer and OFAF 605 MVA transformer. The results obtained are compared to two other thermal models and measured values. A loss of life prediction accuracy is 1.96 minutes on average for the first transformer and 6.40 minutes on average for the second transformer during a load cycle of 780 minutes and 1200 minutes, respectively. The comparative study between the results obtained and the other thermal models validates the relevance and reliability of the proposed model.