We have proposed the "inject low and let rise" strategy of storing CO_2 in deep saline aquifers as a means of minimizing the risk of leakage. The idea is to maximize the amount of CO_2 held as a residual saturation, trapped in pores of the rock by capillary forces. Supercritical CO_2 is less dense than brine, so when CO_2 is placed in the lower part of the aquifer, it will naturally rise, leaving residual saturation behind. The volume of injected CO_2 is chosen so that the rising CO_2 never reaches the top seal. The CO_2 stored in this way will have very small probability of escaping the aquifer on the time scale of interest for climate change. The distance that the CO_2 rises depends on the uniformity of the displacement front and the CO_2 saturation below the front. Here we explore whether the inherent instability of a buoyancy-dominated CO_2/brine displacement front leads to gravity fingers. Such fingers could reduce the volume of rock through which the rising CO_2 passes. This in turn would reduce the overall amount of CO_2 held as a residual phase and could lead to formation of a "bubble" of CO_2 in the top part of the aquifer: a large volume of CO_2 held at saturations larger than residual and thus potentially mobile. Ensuring the long-term integrity of the seal above such a bubble remains a technical and research challenge, so avoiding bubble formation would reduce the risk of eventual escape. To quantify this behavior and evaluate its large-scale implications, we conducted a series of very high resolution simulations. Under conditions that are conducive to countercurrent flow (rising CO_2, sinking brine), the front locally behaves as if it were one-dimensional with only one phase flowing, and its speed and saturation can be estimated by Buckley-Leverett theory. Capillary forces easily disrupt the buoyant plume, and consequently the first order influence on global behavior of the front is the correlation length of the permeability field. The CO_2 rises along channels, not fingers. The channels are a manifestation of spatial heterogeneity in the rock properties (permeability, drainage capillary pressure curve, and anisotropy). Coarse grid simulations adequately capture the behavior as long as the grid resolves correlated features of the domain. The results indicate that exploiting residual trapping and minimizing CO_2 migration will require good understanding of aquifer heterogeneity and its engineering implications.
展开▼