Pipeline Failure Analysis: Bayesian Network Approach from Fault Tree Analysis

Abstract

Pipeline failures can have catastrophic consequences, posing significant environmental and safety risks. In order to address the pipeline failure issues, a diagnostic Bayesian Network (BN) model was applied for the study of the Root Cause Analysis (RCA). The study aimed to identify the factors responsible for pipeline failures and offer a predictive tool for risk assessment. This study used the Bayesian Network (BN) model using existing Fault Tree Analysis (FTA) data from a past case study on pipeline failures. Translating the potential failure modes and the identified causes using Fault Tree Analysis (FTA) creates the Bayesian Network (BN) model. The results and discussions showed the BN model's accuracy and effectiveness in predicting pipeline failures. Sensitivity analysis highlighted critical factors with substantial influence on system reliability. The model's validation against FTA ensured its reliability in representing dependencies and relationships in the system. The sensitivity analysis revealed that the model categorized the pipe defects into commissioning and material defects, the primary causes of a pipeline to be ruptured. By incorporating a wide range of factors and failure mechanisms, the model offers a comprehensive approach to assessing the risk of pipeline failure incidents. This BN model offers a practical tool for conducting Root Cause Analysis and making informed decisions to improve pipeline system safety.