Comparison of Feasibility of a Standard Reliability-Based Approach and a Bayesian Network Approach for Integrity Management of a Northern Canadian Liquids Pipeline
Abstract Integrity management of a northern Canadian liquid pipeline required a unique approach due to the presence of methanol-induced, short, axially-oriented, internal stress corrosion cracking (iSCC) adjacent to girth welds. Previous studies focused on investigation of the susceptibility to iSCC, crack detection performance of inline inspection tools, and leak rates due to iSCC. The current feasibility study aimed to identify an approach to estimate the probability of leak due to iSCC that uses information from previous investigations and handles the data gaps. Two probabilistic approaches — a standard reliability-based approach and a Bayesian Network approach — were compared based on a number of criteria related to effective use of existing data, flexibility of the framework, and ability to iteratively improve the results. The results of the comparison indicate that a hybrid of the two approaches is an ideal way to develop a comprehensive integrity management tool.