A multiagent-based modeling approach for emergency evacuation plan optimization during toxic gas releases within chemical plants

It is essential to minimize the gas exposure risk during evacuation in the event of toxic gas releases. This study uses a multiagent-based model to simulate the evacuation processes of a chemical plant and surrounding residential areas. An evaluation and optimization method for evacuation plans was proposed. A risk assessment model for ambient evacuation was established, and the evacuation plan optimization process was proposed. Using a chemical plant in Tianjin, China, as an example, the evacuation scenario of workers and surrounding residents after liquid ammonia release was simulated, and the effectiveness of the optimization strategies for the original evacuation plan is verified. The results reveal that using optimization strategies dramatically reduces the gas exposure risk during evacuation. The gas exposure risk during evacuation of some evacuation sub-areas is reduced by optimizing the evacuation route. Meanwhile, issuing the evacuation order sooner and optimizing the warning diffusion model exerts an excellent risk reduction effect on areas with low ammonia concentrations. Evacuation optimization strategies suitable for different evacuation sub-areas are not the same. They need to be determined according to the real situations of the evacuation sub-areas. The results can help facilitate decision-making for the formulation and optimization of emergency plans for toxic gas releases.