Hongyang Cheng

Grain Learning: Bayesian Calibration of DEM Models and Validation Against Elastic Wave Propagation

The estimation of micromechanical parameters of discrete element method (DEM) models is a nonlinear history-dependent inverse problem. In order to reproduce the experimental measurements with high accuracy, this work aims to develop a machine learning-based calibration toolbox named “Grain learning”, which can extract grains from X-ray computed tomography (CT) images and perform Bayesian parameter estimation for DEM models of dry granular materials.

A Simple Multiscale Model for Granular Soils with Geosynthetic Inclusion

Due to the heterogeneity and anisotropy caused by geosynthetic inclusions, the design of geosynthetic-reinforced geostructures has been largely empirically-based. The presence of geosynthetic reinforcement complicates the stress history and fabric characteristics of the reinforced soil, posing an obstacle to the development of constitutive models for geosynthetic-reinforced soils. To circumvent this difficulty, we employ a simple multiscale model based on a coupled FEM/DEM approach. The displacement in granular soil domain is solved in the hierarchical multiscale framework, while the geosynthetic inclusion that prescribes the boundary conditions are modeled concurrently by discrete bar elements. The responses of both multiscale domains are communicated and updated in an explicit time integration scheme. The main objective in this work is to extend the existing multiscale framework for modeling soil-geosynthetic interactions. The predicative capacity of this model is examined in two numerical examples, i.e., shape-forming and pull-out tests. The multiscale characters of the response of geosynthetic-reinforced soil facilitate a better understanding of the reinforcement mechanisms.