A dynamic risk-based routing approach for multi-source and multi-sink evacuation problem in chemical industrial parks

Chemical industrial park (CIP) is a safety-critical system composed of various hazardous materials related units. Once a chemical accident occurs, evacuation is a typical strategy employed to mitigate the casualties. Currently, the emergency evacuation plans in CIPs are lack of quantitative basis risk. Meanwhile, most related studies ignored the temporal-spatial evolution of chemical accidents. The present work raised a multi-source and multi sink evacuation model (MMEM), which is driven by the dynamic risk assessment. The human injury model and the evacuation behavior model are adopted to quantify the dynamic evacuation risk. Moreover, the spatial temporal intersection is considered to avoid the path conflicts caused by multiple evacuation crowds, An efficient prior knowledge-based Dijkstra algorithm (PKDA) is proposed for model solving, and the proposed method is demonstrated by a case study. Simulation results show that PKDA is superior to the traditional Dijkstra algorithm (DA) and the Ant Colony algorithm (ACO), which is competitive in both effectiveness and efficiency. The proposed MMEM model is beneficial to the evacuees select a reasonable evacuation path according to the dynamic evolution characteristics of chemical accidents, which effectively improves the efficiency and safety of emergency evacuation.

Automated summary

Chemical industrial parks (CIPs) are critical systems consisting of various units related to hazardous materials. Evacuation is a common strategy to mitigate casualties in chemical accidents, but current emergency evacuation plans lack quantifiable risk assessment and most studies ignore the temporal-spatial evolution of such accidents. This study proposes a multi-source and multi-sink evacuation model (MMEM) driven by dynamic risk assessment, which improves efficiency and safety by considering dynamic evolution characteristics and path conflicts caused by multiple evacuation crowds.