Qualitative and quantitative experimental study of convective mixing process during storage of CO2 in homogeneous saline aquifers

The storage of CO2 in deep formations like saline aquifers is being actively considered in order to support the reduction of greenhouse gas emissions. It has been observed that dissolution of CO2 into brine causes an increased density in the mixture and if the corresponding Rayleigh number of the porous medium is high enough to initiate convection flows, then density-driven natural convection occurs and the rate of dissolution increases. As such, an increased contribution by the dissolution mechanism for trapping of CO2 decreases the risk of leakage. However, despite the recognized importance of convective dissolution in geological CO2 storage, there is insufficient experimental data available for studying the accelerated mass transfer rate of CO2 into saline aquifers. In this paper, we focus on performing a series of novel experiments about the density-driven natural convection mechanism in a precise experimental set-up with homogeneous Hele-Shaw cell geometries and by using CO2 and water. Our new approach and procedure for performing the experiments give us an opportunity to have both qualitative (images and video) and quantitative (amount of dissolved CO2 into water) data at the same time. This study examines onset time for convection, critical wavelength of convection fingers and dissolution flux of CO2 into water as objective parameters. The growth and progress of convection fingers after onset time for convection and the effect of model properties on the behavior of the convective mixing process are presented and discussed. Moreover, there are some speeded-up videos from the experiments that are suitable for improving public awareness of the problem facing society. Based on the performed experiments, the calculated dimensionless onset times for convection are 194.90 and 227.80 from quantitative and qualitative measurements respectively and the calculated critical wavelength of convection fingers is 2π/0.0524. Additionally, the dissolution flux of CO2 into water after onset time for convection is related linearly to Δρ·g·cos(θ)·k·C0/μ by a prefactor of 0.021. These results are comparable with the available theoretical and numerical works in the literature. Comparison of the results of the experimental models and their equivalent numerical models shows that while the numerical models are good tools for prediction of growth mechanism, shapes and positions of convection fingers, their effectiveness in the prediction of the onset times for convection and the dissolution fluxes is uncertain.