In China, food-freezing plants that use liquid ammonia, which were established in the suburbs in the 1970s, are being surrounded by urban built-up areas as urbanization progresses. These plants lead to extremely serious casualties in the event of a liquid ammonia leakage. The purpose of this thesis was to explore the key factors of personnel protection failure through the scenario evolution analysis of liquid ammonia leakage. The chain of emergencies and their secondary events were used to portray the evolutionary process of a full scenario of casualties caused by liquid ammonia leakage from three dimensions: disaster, disaster-bearing bodies, and emergency management. A Bayesian network model of liquid ammonia leakage casualties based on the scenario chain was constructed, and key nodes in the network were derived by examining the sensitivity of risk factors. Then, this model was applied to a food-freezing plant in Beijing. The results showed that inadequate risk identification capability is a key node in accident prevention; the level of emergency preparedness is closely related to the degree of casualties; the emergency disposal by collaborative onsite and offsite is the key to avoiding mass casualties. A basis for emergency response to the integration of personnel protection is provided.
Keywords: Bayesian network; emergency management; liquid ammonia leaked; personnel protection; scenario construction.