Cohesive particle model using discrete element method on the Yade platform

Abstract

The thesis describes implementation of particle-based model of concrete using the Discrete Element Method (DEM) using the Yade platform. DEM discretizes given domain using packing of (spherical) particles of which motion is governed via local contact laws and Newton's equations. Continuity is modeled via pre-established cohesive contacts between particles, while discontinuous behavior arises naturally from their failure. Concrete is modeled as a homogeneous material, where particles are purely discretization units (not representing of granula, mortar or porosity); the local contact law features damage, plasticity and viscosity and calibration procedures are described in detail.This model was implemented on the Yade platform, substantially enhanced in the course of our work and described for the first time in all its aspects here. As platform for explicit dynamic simulations, in particular the DEM, it is designed as highly modular toolkit of reusable algorithms. The computational part is written in c++ for efficiency and supports shared-memory parallel computation. Python, popular scripting language, is used for rapid and concise simulation setup, control and post-processing; Python also provides full access to most internal data. Good practices (documentation in particular) leading to sustainable development are encouraged by the framework.

The thesis describes implementation of particle-based model of concrete using the Discrete Element Method (DEM) using the Yade platform. DEM discretizes given domain using packing of (spherical) particles of which motion is governed via local contact laws and Newton's equations. Continuity is modeled via pre-established cohesive contacts between particles, while discontinuous behavior arises naturally from their failure. Concrete is modeled as a homogeneous material, where particles are purely discretization units (not representing of granula, mortar or porosity); the local contact law features damage, plasticity and viscosity and calibration procedures are described in detail.This model was implemented on the Yade platform, substantially enhanced in the course of our work and described for the first time in all its aspects here. As platform for explicit dynamic simulations, in particular the DEM, it is designed as highly modular toolkit of reusable algorithms. The computational part is written in c++ for efficiency and supports shared-memory parallel computation. Python, popular scripting language, is used for rapid and concise simulation setup, control and post-processing; Python also provides full access to most internal data. Good practices (documentation in particular) leading to sustainable development are encouraged by the framework.