Abstract: In order to address the challenge that frontline commanders cannot accurately assess fire risks in oil tank area fire scenarios, this paper proposes a multi-layer network model for fire risk propagation in oil tank areas based on multi-layer network theory. First, by analyzing critical accident scenarios of oil tanks, we constructed single-layer networks of accident propagation for different types of tank, and subsequently developed a multi-layer network model for risk propagation in oil tank area accidents. Subsequently, we quantitatively evaluated the importance of scenario nodes in tank accidents using node index including degree centrality, betweenness centrality, and closeness centrality, thereby measuring risk levels of different tank accident scenarios. Finally, through case analysis of the 2015 Zhangzhou PX project explosion in Fujian Province, we examined the fire development process and identified critical accident scenario nodes. The study demonstrates that, compared to traditional single-layer networks, the multilayer network can more clearly and effectively identify critical scenario nodes with significant influence, thereby providing valuable support for priority disposal in emergency response.

Key words: oil tank fire, risk analysis, emergency rescue, multi-layer network, scenario evolution