Convective dissolution in porous media: three-dimensional imaging experiments and numerical simulations

Abstract

Convective dissolution is a phenomenon induced by a buoyant instability between two fluids resulting in characteristic finger-like mixing patterns. One important example is a key trapping mechanism during CO2 sequestration in deep saline aquifers. The Rayleigh number, Ra, which is a measure of convective vigour and the Sherwood number, Sh, which indicates the strength of mass transport are used to parameterise this process. We present a novel methodology to image convective dissolution using X-ray CT in a three-dimensional porous medium formed of glass beads with a model fluid pair MEG/water and BEG/water. 3D reconstructions allow us to visualise the spatial and temporal evolution of the plume from onset to the shutdown of convection while quantifying the macroscopic quantities such as the rate of dissolution and horizontal concentration profiles. We investigate convective dissolution with and without permeability heterogeneity over the Rayleigh range, Ra = 2000 - 5000, in a spherical, cuboidal and cylindrical geometry. Simple heterogeneity patterns such as single inclined layers and a series of discontinuous layers were included. It was concluded that the spatial configuration of the less permeable layers was the most important factor in either enhancing (due to flow focusing) or impeding (due to compartmentalisation of the plume) the rate of dissolution compared to the homogeneous case. However, linear trends in the Sherwood and Rayleigh numbers were consistency reported in both the homogeneous Sh = 0.025Ra and heterogeneous studies Sh = 0.0236Ra. 2D COMSOL numerical simulations were performed to extend the Ra range explored by the experiments and allow an investigation of permeability heterogeneity configurations not possible in the laboratory. The results confirmed the experimental findings with very similar scaling observed Sh = 0.022Ra. The results also shed light on the plume structures responsible for flow focusing. Finally, we present suggestions for future work in CO2-brine-rock systems.