Probabilistic failure assessment of oil and gas gathering pipelines using machine learning approach

Oil and gas gathering pipelines leak can cause severe environmental pollution and even lead to fire or explosion accident at gathering stations. Implementing an effective risk assessment is crucial for mitigating the risk associated with leak incidents. However, traditional methods have the limitations in dealing with the a large volume of real time data. In industry 4.0 environment, the digitalization of process operations is becoming an inevitable trend, necessitating the use of big data and artificial intelligence technologies. This paper presents a data-driven approach for failure risk assessment of oil and gas gathering pipeline using machine learning approach. The contributing factors of pipeline leak is identified, and ESD is used to describe the escalation of pipeline leak. A BN model is developed to present pipeline leak accident scenarios, which is mapped into a LightGBM model to build a data-driven model for failure risk assessment of oil and gas gathering pipelines. A case study is performed to illustrate the methodology. It is observed that this study can be helpful for supporting the digitalization of risk management of oil and gas gathering pipelines.