Modeling and simulation of residents’ response in nuclear disaster

It is important to model and predict residents’ behaviors in an emergency in order to establish good evacuation schemes during disasters. This research presents modeling and simulation of residents’ behaviors in a nuclear disaster focusing on residents’ decision-making processes: information acquisition, situation assessment, and selecting actions. We selected qualitative causal relations between residents’ behaviors and the attributes of information, human, and situations from 57 reviews of the past 12 disaster cases. We then constructed a conceptual model of residents’ behaviors in a conventional stimulus–organism–response (S–O–R) model of human information processing. We adopted probabilistic reasoning (Bayesian belief network) to simulate the situation assessment of a resident in a nuclear disaster. We carried out a simulation using the announcement log of the JCO criticality accident and confirmed that the model could simulate the tendencies in residents’ behaviors observed in the actual disaster and can reflect various features of the conceptual model.