Reservoir Simulation of CO2 Storage in Deep Saline Aquifers

Abstract

Abstract We present the results of compositional reservoir simulation of a prototypical CO2 sequestration project in a deep saline aquifer. The objective was to better understand and quantify estimates of the most important CO2 storage mechanisms under realistic physical conditions. Simulations of a few decades of CO2 injection followed by 103 to 105 years of natural gradient flow were done. The impact of several parameters was studied, including average permeability, the ratio of vertical to horizontal permeability, residual gas saturation, salinity, temperature, aquifer dip angle, permeability heterogeneity and mineralization. The storage of CO2 in residual gas emerges as a potentially very significant issue meriting further study. Under some circumstances this form of immobile storage can be larger than storage in brine and minerals. Introduction Geological Storage Geological sequestration of CO2 is one of the few ways to remove combustion emissions in sufficient volumes1 to mitigate the greenhouse effect. Several groups have reported aquifer-scale simulations of the storage process, usually in order to estimate the volume that can be stored1–14. Most schemes that have been put forward depend on storing CO2 in the supercritical state. In these schemes, buoyancy forces will drive the injected CO2 upward in the aquifer until a geological seal is reached. The permanence of this type of sequestration depends entirely on the integrity of the seal over very long periods of time. Assuring such integrity in advance is very difficult. Our study focuses on three modes of CO2 sequestration that avoid this concern:pore-level trapping of the CO2-rich gas phase within the geologic formation;dissolution into brine in the aquifer; andprecipitation of dissolved CO2 as a mineral, e.g. calcite. All three modes are familiar, though to date the little attention has been paid to the first. Each of these modes is permanent for the time frame of interest in CO2 sequestration. The key issues then become 1) how to maximize these three highly desirable forms of sequestration so that very large volumes of CO2 can be permanently stored in aquifers, without the need for ensuring long-term seal integrity and 2) how long it takes for the injected CO2 to migrate into these modes of storage. The principal petrophysical parameters influencing storage as an immobile gas phase (in this paper, we use the term "gas" as shorthand for "supercritical fluid") are relative permeability, including hysteresis, and the residual saturation of a nonwetting phase. Both depend on the rock making up the aquifer and thus can vary with location. The phase behavior of the CO2/brine mixture controls storage in solution, and this depends upon brine salinity, temperature and pressure. The principal geochemical driver accompanying storage is the acidification of the brine resulting from dissociation of dissolved CO2. Low pH brine10 in turn induces several reactions with minerals in the formation. An obvious example is the dissolution of carbonate cements. Other reactions are analogous to weathering, in which the acid extracts cations from aluminosilicates (feldspars, clays, etc.). The released cations may form relatively insoluble carbonate precipitates such as siderite. The competition between these reactions will determine the potential for additional storage by mineralization. The time scales for these processes vary widely. Once CO2 injection ends, the fluid displacement leading to residual saturations depends on absolute and relative permeabilities, hysteresis, buoyancy forces, the dip of the aquifer, the natural background flow gradient, and the magnitude of the residual saturation. Dissolution of CO2 into brine is rapid, but the overall rate of mass transfer depends on contact between the phases. This is a complicated function of time, especially after injection stops, controlled by the same parameters as the post-injection fluid displacement. Geochemical reactions (mineral dissolution and precipitation) are typically slow1,10, though under some conditions the rate may be comparable to other mass transport processes4,14.