Fuzzy Bayesian Network–Based Multidimensional Risk Assessment for Leakage of Blended Hydrogen Natural Gas Pipelines

Abstract

Pipeline transportation is an efficient way to transport hydrogen on a large scale. However, many uncertainties lead to the failure of blended hydrogen natural gas (BHNG) pipelines, involving multiple consequence-based losses. The failure factors of BHNG pipelines have hybrid epistemic and aleatory uncertainties. The present study proposes a multidimensional integration approach to predict the failure risk of BHNG pipelines using fuzzy Bayesian networks (BN). The use of BN constructs topological structures to capture uncertain relationships among critical information. An expert-inspired method synthesizing fuzzy set theory can determine relative probabilities in the framework of cognitive uncertainties, and the analytic hierarchy process is introduced to capture the conditional probabilities. Multiple consequence dimensions were considered to incorporate safety barriers into the consequence risk analysis. The results demonstrate the objectiveness and effectiveness of multidimensional risk evaluation in addressing various risks. Failure risk can be mitigated and prevented by considering multidimensional impact assessment, which minimizes losses of leakage accidents.