Pipelines are generally a safe and reliable medium in transporting hydrocarbons. In the context of oil and gas transportation, the hydrocarbons are transported from the drilling rig (offshore or onshore) to the refining plant and later on to oil depots. As with other structures, pipelines have a tendency to fail. Although the probability is small but failures do occur. The failure of pipelines can be detrimental and have implications on the environment, social and economic sense. Demonstrably good design, appropriate material selection and best practice are key in ensuring the continued safety of pipelines.
Research has been conducted extensively in predicting the failure of pipelines by employing a probabilistic approach. This approach as compared to the deterministic approach allows for uncertainties to be considered. As with many engineering problems, uncertainties are an inherent value and an inseparable part of real life situations. Thus it is imperative to address these uncertainties in a manageable and consistent manner. There are several approaches that have been developed to consider and represent uncertainties in pipeline failures such as fuzzy logic, neural networks, Bayesian Statistics etc. The values from these assessments are relative which might be useful for risk ranking but are not absolute in the sense that it fails in judging whether the risk is acceptable to the local community.