A review on Bayesian modeling approach to quantify failure risk assessment of oil and gas pipelines due to corrosion

To forecast safety and security measures, it is vital to evaluate the integrity of a pipeline used to carry oil and gas that has been subjected to corrosion. Corrosion is unavoidable, yet neglecting it might have serious personal, economic, and environmental repercussions. To predict the unanticipated behavior of corrosion, most of the research relies on probabilistic models (petri net, markov chain, monte carlo simulation, fault tree, and bowtie), even though such models have significant drawbacks, such as spatial state explosion, dependence on unrealistic assumptions, and static nature. For deteriorating oil and gas pipelines, machine learning-based models such as supervised learning models are preferred. Nevertheless, these models are incapable of simulating corrosion parameter uncertainties and the dynamic nature of the process. In this case, Bayesian network approaches proved to be a preferable choice for evaluating the integrity of oil and gas pipeline models that have been corroded. The literature has no compilations of Bayesian modeling approaches for evaluating the integrity of hydrocarbon pipelines subjected to corrosion. Therefore, the objective of this study is to evaluate the current state of the Bayesian network approach, which includes methodology, influential parameters, and datasets for risk analysis, and to provide industry experts and academics with suggestions for future enhancements using content analysis. Although the study focuses on corroded oil and gas pipelines, the acquired knowledge may be applied to several other sectors.