New insights on the physics of salt precipitation during injection of CO2 into saline aquifers

Abstract

Salt precipitation associated with injection of dry CO2 into saline aquifers has caused reduced injectivity at CO2 storage sites such as Ketzin and Snøhvit. The present-day reservoir scale models of this phenomenon include processes such as evaporation of water into CO2 and capillary backflow of water into the dried zone, and have the salt formation up-scaled to grid-size. However, salt precipitation via these mechanisms has been suggested to only fill a fraction of the pore network and to not significantly affect the permeability. We report lab-on-chip experiments to provide new insights into the dynamics of salt precipitation at the pore-scale and also to find the possible explanations for the large-scale salt precipitation observed in the field. The results of this study introduce two interrelated and so far unreported phenomena –self-enhancing of salt growth and water film salt transport – which together dramatically intensify the precipitation rate and amount of salt precipitated. Our experiments demonstrate that salt crystals, although at different rates, grow in both the liquid and gas phases. Aggregates of micrometer sized crystals in the gas phase create a micro-porous medium with massive capillarity that can strongly imbibe brine over long distances to the evaporation front via capillary connected water films. The imbibed brine in the salt structure is several orders of magnitude thinner than trapped brine in the pores, and due to its high surface area to volume ratio, it rapidly becomes highly super-saturated, leading to a high nucleation rate and the formation of additional salt crystals. The produced salt provides additional surface area for evaporation, and therefore enhances the overall rate of precipitation. This mechanism is active even at extreme CO2 flow rates and therefore salt formation could be more severe than previously concluded, which was based only on core flooding experiments and numerical simulations. The paper raises a serious need for reconsideration of the currently implemented physics in the simulation tools and also for properly designed laboratory experiments, to ensure that the reservoir volume outside the drainage area, which over time contributes to the bulk of precipitated salt, is also included.