Reliability model selection and validation using Weibull probability plot—A case study

• A guideline for reliability analysis is developed based on the Weibull probability plot.
• The application of the guideline is shown by a case study of the power transformer.
• We found two-fold Weibull mixture distribution is the best model for power transformer.

In planning and operation of an electric power transmission system, accurate estimation of the reliability of its different components is a fundamental and challenging concern. Two-parameter Weibull probability distribution has been used frequently for this purpose. However, this model cannot capture all behaviours of the many different components in a transmission system. Over the years several alternative models have been proposed that are either derived from, or in some way related to, the two-parameter Weibull distribution. However, these alternatives have received little attention in reliability analysis of power transmission systems. It seems the lack of a simple and applicable guideline for model selection and validation is one of the main impediments. Many of these derived distributions (Weibull distribution family) have characteristic shape in the Weibull probability plot (WPP). Hence, WPP can be used as a tool for model selection and validation through the Weibull distribution family. This paper provides a WPP-based guideline and flowchart for selection of the most appropriate Weibull method. The application of the guideline is demonstrated by a case study using historical data of power transformer.