Two-stage propagation analysis of safety risks in complex underground engineering: An integrated modeling framework

The complex nonlinear mapping relationships among safety risks in underground engineering projects provide various potential pathways for risk propagation. A lack of understanding of the propagation behavior of risks can lead to improper risk prevention and control, which may trigger chain reactions of safety accidents. To address this issue, this study quantified and visualized the interactions among safety risks in complex underground engineering by coupling association rule mining and complex networks. It also examined the propagation process of risk factors under different intervention measures, considering managers' preferences (intervention intensity and timing), using an improved epidemiological model. The results show that for the same risk, the risk-control effectiveness of different stakeholders as controlled factors varies significantly. Government and regulatory departments, as well as project owners, should play a leading role in safety risk management, whereas contractors' unsafe behaviors, material and equipment factors, and environmental factors, when used as intervention factors, show poorer risk control effectiveness. Managers' preferences affect the peak and convergence times of risk propagation, and a more conservative approach is more favorable for risk control. This study recommends that managers increase the frequency of detecting abnormal conditions to promptly identify latent risks and implement early intervention measures to prevent their spread. This study constructed a dynamic risk propagation analysis framework for complex underground engineering projects. It contributes to the understanding of the formation mechanisms of underground engineering safety risks and provides decision support to implement precise risk interventions, thereby promoting the achievement of engineering safety goals.