Risk evolution analysis of gas leakage accidents based on complex network

Abstract

Toxic gas leakage accidents are catastrophic and thus the subject of considerable attention among researchers, regulators, and companies. Thus, this study constructed a complex network evolution model of hazardous gas leakage accidents. It did so by analysing accident reports to then construct 10 fault trees to, in turn, construct an accident chain with causal linkages. This model comprised 67 nodes and 94 edges and effectively described the progression of such accidents from an event causality perspective. Microscopic and macroscopic analyses of the model revealed critical risk events in the network. Moreover, the comprehensive clustering coefficient (0.052), average path length (4.595), and network diameter (10) were calculated in MATLAB. These metrics were used to identify the shortest disaster paths associated with various prevention and control strategies. This study conducted deliberate attack experiments on the aforementioned network model. The findings of the study indicated the most hazardous events and pathways that are most likely to trigger accidents. These findings aligned with statistical data on accident causes, thereby providing insights for proactive prevention and effective management of gas leaks.