Scholarlink

<p>The frequent occurrence of water–soil gushing poses severe risks to the safety of shield tunnels. Consequently, a detailed fragility analysis of tunnels under gushing hazards is imperative. To this end, a machine learning-based probabilistic fragility assessment framework for shield tunnels under water–soil gushing hazards is proposed. First, a Material Point Method (MPM)-Finite Element Method (FEM) coupling model is established to simulate large soil deformations and tunnel structural responses during gushing incidents. Second, uncertainties in soil resistance and joint stiffness are assessed through Latin Hypercube Sampling (LHS). To enhance computational efficiency, a Multi-Layer Perceptron (MLP) model is developed to predict gushing-induced tunnel responses radial convergence ratio (RCR) and segmental joint deformation (SJD). Subsequently, leveraging the trained MLP model, fragility curves are generated via log-linear regression linking the gushing intensity measure (IM) to the damage measures (DM). Finally, the effects of varying gushing locations and soil gushing mass (SMG) on tunnel fragility are analyzed. Key findings reveal that lower gushing locations significantly amplify collapse risks: a 500 kg/m SGM at gushing location 180° yields 100% collapse probability, whereas the same SGM at gushing location 90° results in only 18.9%. This research bridges geotechnical modeling and probabilistic risk analysis through a detailed understanding of water–soil gushing hazards, offering a tool for urban planners to prioritize mitigation strategies in high-risk tunnel sections.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fragility assessment of shield tunnels under water and soil gushing hazards

  • 期刊论文

  • 发表日期：

    2026年2月12日

查看全文
  • Xiao-Chuang Xie,
  • Dong-Mei Zhang,
  • Zhong-Kai Huang,
  • Zhao-Geng Chen,
  • Xue-Liang Zhang

摘要

The frequent occurrence of water–soil gushing poses severe risks to the safety of shield tunnels. Consequently, a detailed fragility analysis of tunnels under gushing hazards is imperative. To this end, a machine learning-based probabilistic fragility assessment framework for shield tunnels under water–soil gushing hazards is proposed. First, a Material Point Method (MPM)-Finite Element Method (FEM) coupling model is established to simulate large soil deformations and tunnel structural responses during gushing incidents. Second, uncertainties in soil resistance and joint stiffness are assessed through Latin Hypercube Sampling (LHS). To enhance computational efficiency, a Multi-Layer Perceptron (MLP) model is developed to predict gushing-induced tunnel responses radial convergence ratio (RCR) and segmental joint deformation (SJD). Subsequently, leveraging the trained MLP model, fragility curves are generated via log-linear regression linking the gushing intensity measure (IM) to the damage measures (DM). Finally, the effects of varying gushing locations and soil gushing mass (SMG) on tunnel fragility are analyzed. Key findings reveal that lower gushing locations significantly amplify collapse risks: a 500 kg/m SGM at gushing location 180° yields 100% collapse probability, whereas the same SGM at gushing location 90° results in only 18.9%. This research bridges geotechnical modeling and probabilistic risk analysis through a detailed understanding of water–soil gushing hazards, offering a tool for urban planners to prioritize mitigation strategies in high-risk tunnel sections.