Estimation on Degradation Rate of Insulating Paper in Power Transformers Using Historical Load and Thermal Data

Abstract

Life expectancy of power transformers is limited by the integrity of insulating cellulosic paper wound around transformer winding. The insulating paper is subjected to thermal, electrical and mechanical stresses. There are three main factors causing degradation in the paper, i.e. moisture, oxygen and temperature. In this work, the degradation rate due to temperature is only discussed. That means that transformer life could be shortened or extended by adjusting loads which results in higher or lower temperature conditions. The temperature, namely the hot-spot temperature is typically used to estimate the relative ageing rate and loss of insulation life. Being able to estimate the hot-spot temperature accurately given load and weather data helps operators to plan about transformer replacement properly. This could also reduce costs of early replacement due to a conservative estimation. In this work, thermal models have been developed using machine learning techniques to learn thermal behavior from historical data. The data are comprised of the hot-spot temperature estimated by winding temperature indicator (WTI), load profiles, ambient temperature, rainfall, wind speed and direction and solar radiation. The proposed method has been validated using data for five 400kV/275kV, four 400kV/132kV and two 275kV/66kV autotransformers and has been shown to work effectively.