Scholarlink

<p>The complex interaction between seepage and mechanical fields in saturated soils plays a crucial role in ground stability, structural safety, groundwater management, and environmental protection. The inherent hydraulic anisotropy commonly found in natural strata further complicates the spatial distribution of the seepage field. To investigate the seepage field around tunnels in anisotropic, water-saturated strata, and its influence on the mechanical behavior of the soil, a hydro-mechanical one-way coupled approach (i.e., drained analysis) was established, accounting for both soil self-weight and tunnel permeability. A hybrid physics-informed neural networks (PINNs) with hard constraints–finite element method (FEM) framework was then developed. In this approach, the steady-state seepage field (i.e., steady-state pore pressure exceeding the hydrostatic, SSPPEH) surrounding the tunnel was first obtained using the PINNs, and the resulting SSPPEH distribution was subsequently interpolated at finite element mesh nodes and incorporated into the equilibrium equations to compute the mechanical response via the FEM. Results are presented for varying permeability and elasticity degrees of anisotropy, which testify that the PINN-derived solutions can effectively serve as a bridge for addressing new classes of problems involving different fields, thereby providing a practical computational framework.</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

PINN–FEM-based approach to hydro-mechanical analysis of tunnels in anisotropic strata

  • 期刊论文

  • 发表日期：

    2026年6月16日

查看全文
  • Senlin Xie,
  • Changqing Xia,
  • Hefu Pu,
  • Xiangsheng Chen,
  • Anfeng Hu,
  • Stefano Mariani

摘要

The complex interaction between seepage and mechanical fields in saturated soils plays a crucial role in ground stability, structural safety, groundwater management, and environmental protection. The inherent hydraulic anisotropy commonly found in natural strata further complicates the spatial distribution of the seepage field. To investigate the seepage field around tunnels in anisotropic, water-saturated strata, and its influence on the mechanical behavior of the soil, a hydro-mechanical one-way coupled approach (i.e., drained analysis) was established, accounting for both soil self-weight and tunnel permeability. A hybrid physics-informed neural networks (PINNs) with hard constraints–finite element method (FEM) framework was then developed. In this approach, the steady-state seepage field (i.e., steady-state pore pressure exceeding the hydrostatic, SSPPEH) surrounding the tunnel was first obtained using the PINNs, and the resulting SSPPEH distribution was subsequently interpolated at finite element mesh nodes and incorporated into the equilibrium equations to compute the mechanical response via the FEM. Results are presented for varying permeability and elasticity degrees of anisotropy, which testify that the PINN-derived solutions can effectively serve as a bridge for addressing new classes of problems involving different fields, thereby providing a practical computational framework.