2. Visualization and Numerical Investigation of Natural Convection Flow of CO₂ in Aqueous and Oleic Systems

Optimal storage of carbon dioxide (CO₂) in aquifers requires dissolution in the aqueous phase. Nevertheless, transfer of CO₂ from the gas phase to the aqueous phase would be slow if it were only driven by diffusion. Dissolution of CO₂ in water forms a mixture that is denser than the original water or brine. This causes a local density increase, which induces natural convection currents accelerating the rate of CO₂ dissolution. The same mechanism also applies to carbon dioxide enhanced oil recovery. This study compares numerical models with a set of high pressure visual experiments, based on the Schlieren technique, in which we observe the effect of gravity-induced fingers when sub- and super-critical CO₂ at in situ pressures and temperatures is brought above the liquid, i.e., water, brine or oil. A short but comprehensive description of the Schlieren set-up and the transparent pressure cell is presented. The Schlieren set-up is capable of visualizing instabilities in natural convection flows; a drawback is that it can only be practically applied in bulk flow, i.e., in the absence of a porous medium. All the same many features that occur in a porous medium also occur in bulk, e.g., unstable gravity fingering. The experiments show that natural convection currents are weakest in highly concentrated brine and strongest in oil, due to the higher and lower density contrasts respectively. Therefore, the set-up can screen aqueous salt solutions or oil for the relative importance of natural convection flows. The Schlieren pattern consists of a dark region near the equator and a lighter region below it. The dark region indicates a region where the refractive index increases downward, either due to the presence of a gas liquid interface, or due to the thin diffusion layer, which also appears in numerical simulations. The experiments demonstrate the initiation and development of the gravity induced fingers. The experimental results are compared to numerical results. It is shown that natural convection effects are stronger in cases of high density differences. However, due to numerical limitations, the simulations are characterized by much larger fingers.