Speaker
Description
Mass transport in compacted bentonite buffers dominated by sodium montmorillonite (Na-MMT) for deep geological repositories is diffusion-controlled and governed by the hydration state and microstructure of Na–MMT. While experiments constrain bulk behavior, they do not resolve the pore- and platelet-scale mechanisms linking interparticle interactions to transport and mechanical response.
We develop a physics-informed, energy-based coarse-grained (CG) model in which platelet center and edge interactions are described by Morse potentials augmented with a Gaussian Process Regression (GPR) correction trained on atomistic potentials of mean force. Implemented as a single tabulated potential, the model captures hydration-induced oscillations and reproduces interlayer energy minima across geometries and layer-charge variants, while remaining transferable beyond the training set.
Using this model, we simulate Na–MMT assemblies over dry densities of 0.8–1.3 g/cm-3 for both monodisperse and experimentally derived polydisperse platelet systems. We quantify pore structure, tracer-accessible porosity, tortuosity, effective diffusion, and quasi-static elastic properties. The model captures (i) the transition from three- to one-water interlayers with increasing density, (ii) the loss of non-interlayer porosity, (iii) diffusion trends consistent with compacted Na-bentonite experiments, and (iv) the corresponding evolution of stiffness.
These results demonstrate that a single, transferable energy-based CG potential can jointly predict transport and mechanical behavior while explicitly resolving microstructural variability in compacted Na–MMT systems.
| Country | Canada |
|---|---|
| Acceptance of the Terms & Conditions | Click here to agree |
