Pseudo-Spring smoothed particle hydrodynamics (SPH) based computational model for slope failure

Abstract

Large deformation and strain localization are the common physical processes that a soil slope may encounter when it becomes unstable and fails. Numerical modelling of such phenomenon is generally difficult through mesh-based methods. While the Smoothed Particle Hydrodynamics (SPH), a particle-based method, has emerged as a potential alternative in modelling failure, it still suffers some computational pitfalls mostly ascribed to the use of material independent kernel function. Moreover, in the standard implementation of SPH, the support of the kernel function may significantly affect the computation resulting in unphysical prediction. In this study, an improved SPH based computational framework has been developed for studying stability and failure of soil slopes. Herein, unlike the existing practice, the kernel function is continuously modified and so is the particle interaction depending on the deformation and failure state of the material. The varying particle interaction has been achieved via a pseudo-spring analogy. The soil has been modelled as an elastic-plastic material with Drucker–Prager plasticity and associative flow rule. The factor of safety of the slope, determined by the algorithm has been found to remain unaffected even with different choices of the smoothing length unlike the standard SPH implementation.