Risk assessment of oil impregnated pressboards using the statistical distribution of partial discharge before breakdown

Oil-cellulose insulation system is used widely in electric power transformers. Partial discharges (PD) are considered to be important to destroy the oil-impregnated pressboard insulation system. Many types of defects can lead to PDs. At the same time, PDs gradually destroy the pressboard, product conductive channels in it. Until now, it is still very difficult to diagnosis the severity of the defects by PDs. In this paper, high voltage experiments were carried out on oil-impregnated pressboards with ball-plate electrode configuration, to research features of the PDs just before the breakdown. These features were used to predict if the defect is leading to a breakdown immediately. Several levels of high voltages were put on fifty specimens to produce carbonized trees, and let the trees propagate inside the pressboards until the pressboards were broken down. The PD impulses were recorded during the experiments. The statistical distributions of some characteristics of PDs just before the breakdown, such as the apparent discharge in one second, the total apparent discharge, the phase range width and the pulse number per second, were obtained. Through the cumulative probability function or the probability density function of the statistical distributions, the probability of the electric breakdown of pressboards can be calculated, so the breakdown can be predicted.