Handling of variability in probabilistic and possibilistic failure analysis of corroded pipes

Abstract

The evaluation of the safe maximum working pressure and the probability of failure of corroded pipes are important steps in the development of risk-based strategies for the inspection and maintenance of pipes carrying oil and gas. These parameters are often calculated based on the inspection results and using probabilistic analysis. To account for the variations in the dimensions of the inspected corrosion pits, the probabilistic approach combines a corroded pipe integrity model, like the ASME B31G, with the probability density functions of the input parameters—pipe wall thickness, corrosion pit length, corrosion pit depth and applied pressure—to predict the probability of failure. This paper compares the traditional probabilistic approach with an alternative possibilistic approach. In the possibilistic approach the input variables are regarded as fuzzy variables whose membership functions are developed from the statistical inspection data. The calculations are carried out using fuzzy arithmetic based on extended interval analysis. Instead of the probability of failure, the possibilistic approach gives the possibility and necessity measures of the likelihood of failure. The possibility and necessity measures are, respectively, more and less conservative than the probability of failure.