Abstract
Stress corrosion cracking (SCC) is a type of crack that grows in a corrosive environment, which threatens the integrity of pipeline and may lead to major and/or sudden pipeline failures. The complex mechanism of SCC involves interactions of electrolyte chemistry, coating quality, metallurgy, stress, and other pipeline operating conditions. As a result, it is challenging to estimate the SCC failure probability at a certain location of the pipeline. Additionally, the nature of data uncertainty in the pipeline operation made a precise SCC prediction even more difficult. In this study, a Bayesian network model was developed to integrate the theoretical and empirical knowledge regarding SCC prediction. By combining the relevant parameters and varying mechanisms, both high pH SCC and near-neutral pH SCC probabilities and crack growth rate can be predicted. The initial prediction results are validated by comparing with the field SCC data. The Bayesian network SCC model can serve as a reference tool for the pipeline operators to determine the time or location of SCC inspection, repair, or replacement. The probabilistic results make it feasible to run sensitivity analysis, to determine the impact of uncertainty data.
