Consequence assessment of gas pipeline failure caused by external pitting corrosion using an integrated Bayesian belief network and GIS model: Application with Alberta pipeline

Corrosion is one of the main reasons for pipeline failure in the oil and gas industry. Because a pipeline failure can result in serious personal injury, monetary loss, and environmental damage, pipeline operators need to make timely, and cost-effective decisions to prevent accidents in high consequence areas. The current study proposed integrating GIS and Bayesian belief network to assess the consequence of transmission pipeline failure on the society (casualty) and environment. To calculate the casualty of the pipe segment, the model incorporates information such as pipe characteristics, failure mode, and population density. An event tree is used to represent all potential outcomes of a gas release based on the two most important variables that have a significant impact on accident evolution: the amount of time between a gas leak and a potential ignition, and the possibility of an explosion due to confinement from the environment. Finally the societal and environmental consequence are estimated based on empirical equations, and subjective judgement, respectively. The spatial GIS capabilities combined with the Bayesian network's reasoning power creates a powerful tool for estimating the severity of pipe failure in a given area based on the information currently available.