Clay micromechanics: Mapping the future of particle-scale modelling of clay

¹ University of Glasgow, James Watt School of Engineering, United Kingdom, Glasgow
² The University of Sydney, School of Engineering, Sydney, Australia
³ King Fahd University of Petroleum and Minerals, Center for Integrative Petroleum Research, Saudi Arabia
⁴ Laboratoire de Mécanique et Génie Civil, CNRS, Université de Montpellier, Montpellier, France
⁵ Imperial College, Department of Civil and Environmental Engineering, United Kingdom, London

^* Corresponding author: arianna.pagano@glasgow.ac.uk

Abstract

Geotechnical engineers need to predict the macroscopic behaviour of clays in terms of strength, stiffness, and permeability and the variation in these properties during deformation. Engineers also need to predict the influence of environmental variables (temperature, pressure, pore-fluid chemistry) on these engineering properties. Particle-based methods, which explicitly model individual clay platelets and their interactions, can help identify the fundamental mechanisms that govern the engineering behaviour. Virtual samples can be created, and simulations can consider application of mechanical loading or change in the environmental conditions to generate data on particle kinematics and interparticle interactions. Particlebased models can be used to simulate stress and strain paths that are not easy to reproduce in physical experiments. This approach to simulation also enables parametric studies to understand the sensitivity of the overall behaviour to various particle characteristics and the nature of the particle interactions. This contribution provides a review of the state-of-theart of existing particle-based models for clays, namely Discrete Element Method (DEM), Monte Carlo method (MC), and Molecular Dynamics (MD). The technical challenges, advantages, and disadvantages of each method for the simulation of clays are presented and discussed, together with the technical developments we would like to see over the next decade to realize the full potential of these modelling tools.