Study On The Modelling And Dynamics Of Safety Risk During Tunnel Construction

Tunnel construction is a high-risk project because of the limited a priori knowledge of geotechnical uncertainties, the potential for causing damages to adjacent structures, facilities and pipelines, and the complex organizational management processes. To prevent the tunnelling-induced major accidents from occurring, it is necessary to implement a reasonable safety risk analysis during tunnelling. Tunnel construction process is a complex socio-techincal system. However, most current risk analysis models lack a systematic view and the very nature of tunnel construction is not completely considered, resulting in limited applications of those models. Based on the system theory and control theory, a safety risk analysis model was established to identify the accident causality underlying the tunnelling-induced major accidents and provide supports for tunnel construction safety management.First, the current trends, stimuli and challenges of establishing a systemic safety risk analysis model were analyzed. The safety problems during tunnelling were identified concerning the features of tunnel construction. Then, the goals, assumptions and boundaries of the model were determined. Based on the modeling assumptions, the tunnelling system was divided into three sub-systems, i.e., technical, environmental and organizational sub-system, which represented the tunnel construction process, the tunnelling-induced damages to the environment and the organizational management process respectively. The modelling aims and challenges of each sub-system were clarified, and the basic safety control structure of tunnel construction was proposed as a representation of the interrelationship among the sub-systems.Second, based on the goals, assumptions and boundaries, a feedforward control model for shield tunnelling was established; a quantitative risk assessment method for tunnelling-induced damages to the environment was proposed; and an analysis for organizational and managerial factors that can influence safety risk was conducted. For the technical sub-system, the complex nonlinear relationships between the geometrical, geological, shield operation parameters and the ground surface settlement during the tunnel construction process were extracted using smooth relevance vector machine with adaptive Gaussian kernel function and particle swarm optimization. The shield operation parameters were then optimized based on the established relationships to form a feedforward control method. For the environmental sub-system, a probabilistic risk assessment framework was proposed to describe the risk propagation. By combining relevance vector classifier and Bayesian networks under the proposed framework, the accident scenarios, consequences and probabilities of tunnelling-induced damages to structures, pipelines and ground surface were identified and assessed using both historical data and expert judgments, which determined the risk level of each tunnel section. For the organizational sub-system, an organizational framework was proposed to describe the relationships between the organizationl factors and the system safety performance. The definitions, dimensions and inter-relationships of safety culture, safety management system and safety performance were reviewed using system dynamics, which gave insights into how the organizational learning, decision-making and management were taking place.Finally, by integrating the three sub-systems, a Dynamic Risk Assessment For Tunnelling system (DRAFTs) was proposed to analyze the risk dynamics and accident causality during tunnelling. The simulation results showed the cost, schedule and safety performance variation during tunnel construction and revealed that accidents during tunnel construction were caused by the interaction between organizational factors and technical factor, which gave foresight to prevention of major accidents during tunnelling. However, due to the complexity of tunnel construction, the proposed model needs to be improved in the future.