Lifetime prediction and estimation of power transformer

This project presents a method to estimate and predict failure probability related to aging transformer units in power system. Statistically, the sample mean or the average age method is acceptable if it is used in a case where there is a big population. This method obviously is not suitable for power system components with very few samples of end-of-life failures. The essential weakness of the sample mean is that it only uses information of died components. This research proposed an approach to estimate and predict the lifetime of a power transformer by using NGC data. The data with both died and survive transformers will make contribution on estimating the mean life of power transformer. Two methods that used are normal and Weibull distributions. Although the two methods have different estimation approaches and solution techniques, they are related to each other and use the same format of raw data. From this research, the mean life and standard deviation for normal and Weibull distribution estimation should be quite close and also to the shape of the both distribution. Thus, statistical reliability analysis can provide predictions, such percentage of transformer that will fail at a particular time of before a particular age and how many transformers will fail in the next future year by using failure rate model. From that prediction, the forecasted capital expenditure ( the cost of replacement and consequential failure cost) also can be specified. Thus, it will avoid asset harvesting and the possibility of having unforeseen costs