Pipeline risk assessment using artificial intelligence: A case from the colombian oil network

Currently, in order to make decisions regarding the safety of pipelines, the risk values and risk targets are becoming relevant points for discussion. However, the challenge is the reliability of the models employed to get the risk data. Such models usually involve a large number of variables and deal with high amounts of uncertainty. Therefore, there is a strong need for a powerful tool to cope with that uncertainty, and one of the best tools dealing with uncertainty is the implementation of artificial intelligence methods using fuzzy logic. Hence, this study aims to present an artificial intelligence inference system that minimizes the uncertainty of traditional approaches of risk assessment in pipelines. Also, in order to show the applicability of the model developed, this study presents a case from the Colombian oil transportation network. Besides that, this study presents an uncertainty analysis for the risk values, comparing the results of the inference system with traditional approach. The results show that the inference system performs better since the magnitude of the average error and its standard deviation are less than the traditional approach.