Speaker
Description
This work, carried out within the GeoSafe consortium, combines laboratory measurements, imaging, and numerical modelling to demonstrate how pore-scale simulations can constrain upscaling parameters - particularly dispersivity - for continuum-scale reactive transport models. The Digital Rock Physics (DRP)-informed workflow, implemented in our open-source code GeoChemFoam (https://github.com/GeoChemFoam), is applicable to any rock with a suitable CT-image, but is illustrated here for clay-rich rocks.
Clay-rich subsurface rocks are prime host rock candidates for the safe isolation of radioactive waste, due to their low permeability, high sorption capacity, and heterogeneous pore structure. Predicting reactive transport in such materials, however, remains challenging since pore-scale heterogeneities and coupled physicochemical processes strongly influence contaminant migration. Accurate large-scale transport prediction requires numerical models that correctly upscale the pore-scale physics into effective medium properties that characterise the system.
By comparing experimental data with model predictions, we illustrate where simplistic 1D models fall short and demonstrate how a DRP-informed workflow enhances our ability to predict what we measure. We begin with advective flow-through experiments in a reaggregated rock sample to measure fluorescein breakthrough, followed by batch sorption experiments to determine the distribution coefficient, $K_D$. One-dimensional PHREEQC reactive transport simulations employing a linear sorption model and the experimentally determined $K_D$ reproduce the overall effluent concentrations reasonably well, yet a persistent mismatch suggests the influence of dispersion - a key parameter in modelling contaminant and radionuclide migration. The challenge here is that dispersivity is a geometry-specific property and generally unknown a priori.
We utilise GeoChemFoam to compute the dispersivity of the sample, by solving a closure problem within a micro-CT image dataset. Incorporating this DRP-informed dispersivity into the PHREEQC model yields excellent agreement with the experimental breakthrough curve. Under these conditions, we show that DRP-informed upscaling enables us to reproduce experimental breakthrough without the need for parameter tuning. We also illustrate how GeoChemFoam can be leveraged to complement the experimental dataset by simulating different conditions such as varying flowrates. In conclusion, DRP effectively links experiments and models, determining upscaling parameters for input to field-scale models relevant to safety assessments.
| Country | United Kingdom |
|---|---|
| Acceptance of the Terms & Conditions | Click here to agree |
