Dynamic Bayesian network-based risk assessment for gas treatment station

Risk assessment in gas processing plants plays a critical role in preventing hazardous events that may escalate into catastrophic accidents and severe economic losses. Conventional approaches, such as Event Tree (ET) analysis, have long been employed to identify potential outcomes and design preventive or protective barriers. Despite their usefulness, ET models remain limited by their static structure and their inability to manage uncertainty adequately. To overcome these drawbacks, this study proposes the use of a Dynamic Bayesian Network (DBN) framework, which enables the explicit representation of event dependencies and the continuous updating of probabilities, features not achievable with traditional ET methods. Moreover, the DBN approach can model the temporal progression of sequential events while incorporating safety barriers into a unified analytical framework. This dynamic framework provides greater flexibility and depth in risk analysis. Its applicability is demonstrated through a case study of a gas processing treatment unit, highlighting the model's effectiveness in representing complex accident scenarios and supporting more reliable risk-informed decision-making. The case study shows that implementing this framework holds advantages in managing safety barriers in a unified way, considering the interplay among safety barriers, making continuous risk-treatment adaptations to sustain the safety and security of the gas processing system. This study provides guidance to practitioners for evaluating the performance of safety barriers in gas processing units.