Probabilistic characterization of seepage‐induced tunnel face failure via a two‐phase double‐point material point method

Seepage‐induced tunnel face collapse and water inrush are major hazards when excavating through water‐rich fault fracture zones, where excavation perturbs the groundwater regime and triggers coupled hydro‐mechanical instability. This study presents a two‐phase double‐point material point method (MPM) framework that explicitly resolves the interaction between the fractured‐rock skeleton and pore water under large deformation. The model integrates a cementation‐based constitutive formulation to capture shear weakening and bond degradation in fault gouge, while dynamically updating porosity and permeability during deformation. Validation against a saturated granular column collapse experiment demonstrates the model's capability to reproduce coupled large‐deformation flow behavior. Numerical simulations of tunnel excavation reveal the full evolution of failure from initial deformation to collapse and subsequent water inrush. Parametric analyses identify permeability−porosity thresholds governing transitions between stable conditions, delayed collapse, and coupled water‐rock inrush. A probabilistic exceedance analysis based on 10 000 simulations further quantifies the likelihood of different collapse extents. The proposed framework provides a physics‐informed and transferable tool for risk‐informed tunnel engineering decision‐making in a water‐rich fault fracture zone.