Dynamic Evaluation of the Safety Risk During Shield Construction near Existing Tunnels via a Pair[sbnd]Copula Bayesian Network

Abstract

Shield tunnel construction near existing tunnels can lead to the deformation and settlement of adjacent tunnels, posing safety risks. To analyze and evaluate the key factors influencing such safety risks, a risk assessment method based on the pair-copula Bayesian network (PCBN) model is proposed. The PCBN model, which integrates a Bayesian network (BN) with a pair copula (PC), is applied to simulate a complex risk factor system and related risk factors. Using the constructed PCBN model, a case study was conducted focusing on excavation between Wuhan Metro Lines 8 and 3, and it was determined that the deformation and settlement risk during shield tunnel construction falls into the safe risk category (Level 3). The key risk factors that significantly impact deformation and settlement risk were identified through correlation analysis and were ranked by importance as follows: advance rate, total thrust, and grouting pressure. Compared with traditional risk assessment methods and BNs, the PCBN model can flexibly handle complex relationships among risk factors, especially nonlinear and tail dependencies, due to the integration of PCs. This approach not only enhances the accuracy of risk assessment but also improves the interpretability of the model, with transparent internal mechanisms, largely due to the use of the BN. On the basis of these findings, decision-making recommendations are proposed to reduce the risk of shield tunnel construction near existing tunnels. The implementation of these recommendations in the Wuhan Metro project has yielded positive outcomes.