Mechanistic modelling of erosion and desiccation cracking of swelling clays

The understanding of the coupled hydraulic, chemical, and mechanical behaviour (HCM) of swelling clays has been advanced substantially over the last few decades. Less advanced, however, is the understanding, and from there the predictive modelling, of phenomena involving discontinuous processes such as erosion and desiccation cracking. The ability to describe rigorously the mechanisms leading to erosion and desiccation cracking is fundamental for the formulation of physically realistic models, which are required for safety/longevity assessment of clays as sealant or barrier materials. The research presented in this thesis aims to advance the understanding and predictive modelling of coupled hydro-chemo-mechanical processes involved in erosion and desiccation cracking of swelling clays by developing the necessary theoretical descriptions, which link interactions at particle/micro level to macroscopic events, and the corresponding numerical implementations. Rigorous mathematical descriptions of clay erosion and cracking must include both continuous (swelling/deformation) and discontinuous (damage/fracture, particles detachment) processes. The latter involve sequences of discrete events of finite spatial dimension, and therefore cannot be described in principle by the classical approach based on local/differential formulations of processes. A non-local approach based on Peridynamics theory (PD) is selected for the development of coupled HCM models in this work. It has been previously demonstrated that PD formulations are particularly suitable and effective for solving problems with large deformation and discontinuities. This quality of PD is further confirmed in the present work by comparing models' predictions with experimental data at all stages of model development. A PD model which couples water flow and chemical transport in unsaturated clays is formulated first as a foundation for erosion. This is used to develop a new erosion model by coupling free swelling, detachment of clay particles and transport of detached particles by flowing water. The erosion model is used to investigate the effects of solution chemistry and flowrate on the penetration, extruded mass, and particle release rate of compacted bentonite. This model is extended by a new formulation for co-transport of accessory minerals with clay swelling, which is used to investigate the role of accessory minerals in the erosion of compacted bentonite. The erosion model is applied to study the long-term performance of clay buffer under variable hydro-chemical conditions. The agreement with experimental data strongly supports the applicability of the model to account for the hydro-chemical conditions (composition and velocity) of the eroding environment. The model for erosion with co-transport of accessory minerals is shown to reproduce qualitatively experimental data, provide explanations for the effect of accessory minerals on erosion, and deliver quantitative predictions for this effect. A PD model for desiccation cracking is formulated by coupling deformation and water transport. The validations of this model show excellent agreements between numerical and experimental results. Specifically, the desiccation cracking model shows the correlations between the shrinkage of clay, changes in displacement fields and crack growth as experimentally observed. It is shown that the cracking model captures realistically key hydraulic, mechanical and geometry effects on clay desiccation cracking, including crack initiation, propagation, and final crack patterns. The erosion and cracking models are finally integrated into a single computational framework, which is used to investigate the erosion of swelling clays assisted by piping flow in cracking channels. It is shown that the integrated model can explain the self-healing of swelling clay in piping channels and predict the mass loss of swelling clays under piping flow for backfill and plug-in material. The agreements between model predictions and experimental data with different coupled processes and environment conditions suggest that the newly developed non-local theoretical and computational tools for analysis of clay cracking and erosion can be used under variable hydro-chemical conditions. The proposed multi-physics non-local formulations provide a robust framework for incorporating a wide set of additional couplings, such as gas and heat transport, and geochemical reactions.