Material point methods (MPM) are widely used to tackle large displacement problems in geotechnical engineering. However, current multiphase MPM formulations are mainly implemented on the basis of explicit time integration schemes and are inefficient for simulating long time processes, such as the consolidation and seepage in saturated and unsaturated soils. To improve that, this study proposes a new strategy for the initialization of MPM simulations, where kernel interpolation is used to map the previously derived field variables (e.g., soil properties, pore pressure, stress and strain) to the material points of the model for the large deformation analysis. These field variables can be the results with a generic distribution obtained from model experiments, in-situ tests, or numerical simulations. Two typical kernel functions, i.e., the cubic spline and Gaussian functions, are used for the mapping procedure. Parametric analysis indicates that the smoothing length of kernel functions highly depends on the distribution of field variables. The proposed strategy is applied to investigate the runout behavior of an instrumented slope failure caused by artificial rainfall. It shows a good agreement between results obtained from the strategy and the experiment.
10th European Conference on Numerical Methods in Geotechnical Engineering (NUMGE2023)
2. Finite element, finite difference, discrete element, material point and other methods