Probabilistic Tunnel Collapse Mechanism Based on Smoothed Particle Hydrodynamics and Random Field Theory

In this work, we simulate tunnel collapse in random heterogeneous soils using the smoothed particle hydrodynamics (SPH) method, and probabilistically quantify the volume of failed mass, the amount of ground subsidence, and the runout distance of the collapsing mass within the tunnel cavity, which are all parameters pertaining to tunnel construction safety and risk assessment. We propose a framework that integrates random field theory (RFT) with SPH, and after validation of our SPH implementation with a physical experiment of tunnel collapse, we next conduct parametric analyses to determine the influence of spatial correlation and cross-correlation parameters between two material parameters of the soil, namely the friction angle and the cohesion. Results show that longer correlation lengths tend to moderate collapse severity, while negative correlations between cohesion and friction angle increase the extent of failure. Nonstationary fields further stabilize collapse responses due to the increase in soil strength with depth.