Risk assessment of liquid ammonia tanks based on Bayesian network and Probit model

Abstract

This paper presents a quantitative risk assessment method to predict the potential risks of liquid ammonia tanks. The method combines probabilistic prediction and consequence assessment: the former is predicted by Bayesian networks derived from bow‐tie mapping, in which the conditional probabilities are determined by analytic hierarchy process, the latter is calculated based on numerical simulation of accident consequences and Probit probability model. The application results show that the method can be used for quantitative risk assessment of liquid ammonia storage tank, and the Bayesian network model is efficient and stable, which can realize risk prediction, accident causes analysis and risk mitigation analysis. Human errors during production or maintenance and corrosion are the main causes of liquid ammonia leakage. Toxicity has the greatest effect on the risk of liquid ammonia tanks. Risk mitigation measures can significantly reduce individual risk and societal risk. The obtained results provide meaningful guidance for the prevention and risk mitigation strategies of liquid ammonia leakage accidents, so as to improve the safety level of liquid ammonia during storage.