A Comparative Study on Heterogeneity of Clay Rocks Using Pore‐Scale Diffusion Simulations and Experiments

Abstract

Clay rocks such as the Opalinus Clay (OPA) in Switzerland are considered as potential host rocks for deep geological disposal of nuclear wastes. Radionuclide migration in clay rocks is dominated by molecular diffusion highly dependent on the pore network geometries. In the sandy facies of OPA (SF-OPA), this pore network is critically modified due to compositional variability and owing to diagenetic reaction products, for example, carbonate and sulfide minerals (Philipp et al., 2017). Such spatial variability is responsible for heterogeneous diffusion patterns (Kulenkampff et al., 2015). At the continuum scale, the representative elementary volume (REV) is an important parameter for understanding and modeling the reactive transport (RT) of contaminants and radionuclides. An accurate estimation of the REV for diffusion is decisive for a meaningful continuum-scale simulation of radionuclide migration in heterogeneous clay rocks like SF-OPA, thus playing an important role in the safety analysis of deep geological repositories for nuclear waste.In continuum mechanics for a heterogeneous material, the REV is the smallest volume over which a measurement can be made to produce a representative value of the macroscopic property in three-dimensional space (3D) (Bear Abstract Accurate modeling and simulation of radionuclide migration in clay rocks such as the Opalinus Clay (OPA) play a key role in the safety assessment of deep geological repositories for nuclear wastes. At the continuum scale, the representative elementary volume (REV) is a fundamental constraint to quantify the effective diffusivity, which is a key parameter in reactive transport (RT) models. Therefore, an accurate estimation of the REV is essential for a meaningful continuum-scale RT simulation in heterogeneous clay rocks. This study presents a comprehensive analysis of the heterogeneities of porosity and effective diffusivity in clay rocks by using the classical sampling theory and pore-scale simulations. First, in this study, the two-dimensional representative elementary area is correlated with the REV for porosity via a characteristic length. Next, it is shown that the REV for diffusivity is larger than the REV for porosity. Moreover, these two REVs can be correlated using Archie's law. In such a way, the REV for diffusivity can be determined by the developed correlations through analyzing two-dimensional microstructures, thus significantly reducing the computational cost. Finally, the applicability of our approach for clay rocks is validated by experimental data on the diffusion of tritiated water in the heterogeneous sandy facies of OPA. From both the experimental data and the modeling prediction, the REV for diffusivity in the sandy facies of OPA is in the order of cubic centimeters. This study provides critical insights into the diffusion in heterogeneous clay rocks toward an enhanced predictability of radionuclide migration.Plain Language Summary Contaminant migration in clay rocks is dominated by molecular diffusion due to their low permeability....

Pore‐Scale Modeling of Water and Ion Diffusion in Partially Saturated Clays

“…This study extends our previous numerical framework for saturated porous media (Wu et al., 2020; Yang & Wang, 2019; Yuan et al., 2022) to partially saturated conditions by adding (a) the equilibrium water/vapor distribution in pore geometries; (b) a stable method to capture the water transport through the water/vapor interface; (c) the electrokinetic boundary condition of liquid/gas interfaces; and (d) the local solute mobility relating to the distance away from the solid/liquid interface. The commonly used numerical method for pore‐scale simulations is the Lattice Boltzmann Method (LBM) benefitting from its efficient parallelization and flexible handling of various liquid‐solid boundary conditions.…”

Elucidating the role of water films on solute diffusion in unsaturated porous media by improved pore‐scale modeling

“…S1) and its pore size distribution. Since the maximum soil particle size is around 400 µm, this cubic microstructure can satisfy the representative elementary volume requirement based on our previous studies (Yang & Wang, 2018;Yuan et al, 2022). A 100 × 100 × 100 uniform numerical mesh with a lattice size of 20 µm was adopted for the LBM.…”