Risk Estimation of Chlorine Leakage from Storage Tanks via Integrated Fault Tree, Bayesian Network, and Dempster-Shafer Theory
DOI:
https://doi.org/10.18502/jhsw.v16i2.22913Keywords:
Chlorine, Dempster–Shafer theory, Uncertainty, Risk assessment, Bayes’ theory, Storage tanks, Bayesian networAbstract
Introduction: The petrochemical industry faces substantial human and economic risks due to high‑temperature, high‑pressure operations and hazardous chemicals, such as chlorine. Given chlorine’s acute toxicity, oxidizing properties, and severe corrosivity, its accidental release poses a risk of widespread and irreversible damage to human health and safety systems. This study proposes a risk framework by integrating BN and FTA with SAFETI 8.4 software and the Dempster–Shafer theory.
Material and Methods: This study was conducted in 2025 at a petrochemical plant. After identifying the hazards associated with chlorine storage tanks, a chlorine gas leakage scenario was selected and analyzed using FTA. The probabilities of basic events were estimated using the Dempster–Shafer theory to address uncertainty. The results were modeled using a Bayesian network in GeNIe 5.0. Risk estimation was performed with SAFETI 8.4 software incorporating meteorological data. Individual and societal risk curves were plotted using probit equations to determine the areas affected by individual risk and estimate the potential fatalities.
Results: Three chlorine gas release scenarios — leaks of 100 mm and 200 mm, and a catastrophic tank rupture — were analyzed using FTA for the top event. The FTA constructed for the top event “chlorine leakage” comprised 18 intermediate events and 50 basic events. Using the Dempster–Shafer theory, the estimated probabilities were Bel=1.7405606×10−6 and Pl=8.294732×10−6. Social risk curve analysis showed that the Pl curve had higher frequencies than the Bel curve, and that Pl exceeded the unacceptable risk level. At the same time, Bel remained mostly within the tolerable level (ALARP).
Conclusion: The results indicated that, under optimistic conditions, the social risk of chlorine leakage remained within ALARP, whereas under pessimistic conditions, it reached the unacceptable level. The integrated BN and D‑S T reduced uncertainty, enabled accurate risk estimation and comprehensive evaluation, supported evidence‑based decisions, and reduced accident probability while enhancing industrial safety and resilience.
