Geochemical evaluation of CO2 injection and containment in a depleted gas field

The short- and long-term geochemical impact of CO2 injection into a depleted gas reservoir (DGR) is investigated using reservoir/geochemical modeling with TOUGH2/TOUGHREACT and 1D kinetic diffusion modeling with PHREEQC (caprock/well-cement). Simulations of CO2 injection into the reservoir predict displacement and buoyancy of post-production CH4, as well as dry-out of the near-well zone. We computed that the areal extent of the CH4/brine dominated zone and the dry-out zone are relatively small compared to the CO2/brine dominated zone after well-closure. For the current DGR model we therefore conclude that it is reasonable to model geochemical reactions in the reservoir without taking into account post-production CH4. Although the CO2 dissolution capacity of the studied DGR is smaller compared to a deep saline aquifer of similar size, the modeling predicts that dissolution and subsequent CO2 mineral trapping proceed faster. Precipitation of dawsonite and magnesite were yet predicted at initial CO2 partial pressure (PCO2) of 9.3 bar, while these minerals were not identified in reservoir samples. This could indicate that their tendency of precipitation is overestimated by the model and hence the database used. This has significant impact on long-term modeled bulk porosity and PCO2. Simulations of CO2 diffusion through the caprock show that mineral reactions significantly retard the total dissolved carbon (TDC) plume. After 10,000 years, 99% of the TDC is present within the first 6.4 m above the reservoir contact. The progression of the TDC plume in the caprock is sensitive to the composition, kinetic rates, and surface area of primary and secondary minerals. Cement alteration modeling shows progressive carbonation of cement phases, resulting in three zones of distinct mineralogy and porosity. The three zones are predominantly characterized by: (i) unaltered cement, (ii) portlandite dissolution, and (iii) calcite precipitation. The simulated thickness of the affected zone is 3.8 cm after 100 years. This distance is sensitive to kinetic rate constants of C-S-H phases, but less sensitive to kinetic rate constant of portlandite. In summary, our applied methodology provides quantitative predictions of the geochemical impact of CO2 on the DGR storage complex. The methodology can be used for screening of potential DGR storage locations and to define criteria for minimal caprock and cement sheet thickness, for assuring short- and long-term integrity of the storage location.