A Comprehensive Accident Model for Natural Gas Processing Stations: Integrating Improved SHIPP Model With Bayesian Networks for Dynamic Risk Prediction

Abstract Natural Gas Processing Stations (NGPSs) are a crucial component in the production and transportation processes of the oil and gas industry, serving as key hubs for gas processing, storage, and distribution. However, there is currently a lack of comprehensive accident models specifically designed for natural gas processing systems, and few studies have systematically analyzed the interdependencies of safety barriers. This paper proposes an improved System Hazard Identification, Prediction, and Prevention (SHIPP) model, which integrates fault tree analysis (FTA), event tree analysis (ETA), and Bayesian networks (BNs) to construct a dynamic and data-driven accident model tailored for NGPSs. Compared to traditional SHIPP and other system-theoretic models such as STAMP and AcciMap, our approach introduces a quantitative Bayesian framework that dynamically updates fault probabilities based on real-time monitoring data, enabling predictive analytics rather than just retrospective diagnosis. The model incorporates general fault trees for five key prevention stages—leakage prevention, diffusion prevention, ignition prevention, escalation prevention, and harm control — capturing the failure modes of diverse safety barriers. These are mapped to BNs to account for uncertainty and conditional dependencies, allowing continuous probability updates through Bayesian inference. A case study of an NGPS fire and explosion accident demonstrates the model’s effectiveness in reconstructing accident pathways and uncovering causal mechanisms. The results show that the proposed model not only enhances accident analysis capabilities but also offers decision support for proactive risk prediction and prevention, contributing to improved safety management at NGPSs.