Convective mixing is a process that greatly enhances the transport of mass between two miscible fluids and plays an important role in many engineered and natural processes. In the context of subsurface CO2 storage, density-driven convection of CO2-rich brines promotes the dissolution of the CO2-rich supercritical gas phase, thereby contributing to the storage capacity of the host reservoir. The process is triggered by the occurrence of a gravitational instability induced by the dissolution process, which further develops into perpendicular elongated finger-like patterns that sink in deeper parts of the reservoir. This thesis presents modelling and experimental approaches to study these convective structures and to evaluate their impact on the CO2 storage process.
By presenting a systematic quantitative assessment of the mixing process, this work fills a knowledge gap in previous work on the subject. To this end, a two-dimensional CFD model is used to investigate the density-driven convection over a range of Rayleigh numbers that is relevant to CO2 storage applications. The macroscopic evolution of the convective process is quantified by employing fundamental measures of mixing that use the local spatial structure of the solute concentration field. These are implemented to obtain scaling relationships that may be used for the appraisal of potential CO2 sequestration sites.
This thesis also addresses the limitations of previous experimental work that has largely focused on two-dimensional domains. An approach is presented that applies X-ray Computed Tomography to generate three-dimensional imagery of the mixing process in both unconsolidated and consolidated porous media. The analysis of this rich dataset reveals that the individual plume structures are strongly affected by the characteristic microscopic morphology of the porous space, with an impact observed on the macroscopic spreading process. By providing direct observations of the mixing process in natural porous media, these results represent a useful benchmark for future numerical and experimental studies.