In-line inspection of underground pipelines for corrosion damage using “smart” pigs is now quite common. With the advent of high-resolution pigs that can identifi large numbers of potential anomalies, more sophisticated methodologies are required for interpreting the results of an in-line inspection. Of particular interest is the probability that the depth of corrosion in a particular location exceeds a critical depth defined by the local pipe characteristics and maximum operating pressure. In this paper, a Bayesian statistical methodology for determining the probability that corrosion exceeds critical magnitude is presented. The estimated probabilities (from the posterior distribution) are based on an assumed pit depth distribution (the prior distribution), the pig call data produced by the in-line inspection (the data), and the detection and depth accuracy performance characteristics of the pig utilized (the data model). The resulting exceedance probabilities can be used with or without corrosion consequences to make inspection/maintenance policy decisions.
Keywords: pipeline corrosion, in-line inspection, pig performance, Bayesian statistics, risk assessment