Simulation of Geological Carbon Dioxide Storage

Abstract

We modifed a streamlined-based simulator based on the work of Batycky et al. (1997) [7] to solve CO2 transport in aquifers and oil reservoirs. We then use this to propose design strategy for CO2 injection to maximise storage in aquifers and to maximise both CO2 storage and enhanced oil recovery (EOR) in oil reservoirs. We first extended Batycky et al. (1997) [7]'s streamline simulator from two phases (aqueous phase and hydrocarbon phase) and two components (water and oil) to a three- phase (aqueous phase, hydrocarbon phase and solid phase) and four-component (water, oil, CO2 and salt) simulator specialized for CO2 injection. We solved CO2 transport equations in the hydrocarbon and aqueous phases along streamlines and in the direction of gravity. To capture the physics of CO2 transport, in the hydrocarbon phase, we used the Todd-Longsta® (1972) [112] model to represent sub-grid-block viscous fingering. We implemented a thermodynamic model of mutual dissolution between CO2 and water and resulting salt precipitation [104; 105]. The resultant changes in porosity and permeability due to chemical reaction and salt precipitation were also considered. We accounted for two cycles of relative permeability hysteresis (primary and secondary drainage and imbibition) by applying two di®erent trapping models: Land (1968) [69] and Spiteri et al.(2005) [103]. Therefore, relative permeability changes and variations in the trapped non-wetting phase saturations due to hysteresis can be updated on a block-by-block basis. We then used this streamline-based simulator to design CO2 storage in aquifers. We propose a carbon storage strategy where CO2 and brine are injected into an aquifer together followed by brine injection alone. This renders 80-95% of the CO2 immobile in pore-scale (10s ¹m) droplets within the porous rock; over thousands to billions of

years the CO2 may dissolve or precipitate as carbonate, but it will not migrate upwards and so is e®ectively sequestered. The CO2 is trapped during the decades-long lifetime of the injection phase, reducing the need for extensive monitoring for centuries. The method does not rely on an impermeable cap rock to contain the CO2; this is only a secondary containment for the small amount of remaining mobile gas. Furthermore, the favorable mobility ratio between injected and displaced fluids leads to a more uniform sweep of the aquifer leading to a higher storage e±ciency than injecting CO2 alone. This design was demonstrated through one-dimensional simulations that were verified through comparison with analytical solutions. We then performed simulations of CO2 storage in a North Sea aquifer. We design injection to give optimal storage e±ciency and to minimise the amount of water injected; for the case we study, injecting CO2 with a fractional flow between 85 and 100% followed by a short period of chase brine injection to give the best performance. Sensitivity studies were conducted for different rock wettabilities and comparison with the Land trapping model. We found that the effectiveness of our proposed strategy is very sensitive to the estimated residual CO2-phase trapping. We then extended our study of the design of CO2 storage in aquifers to oilfields. We again constructed analytical solutions to the transport equations accounting for relative permeability hysteresis. We used this to design an injection strategy where CO2 and brine are injected simultaneously followed by chase brine injection. We studied field- scale oil production and CO2 storage for di®erent CO2 volumetric fractional flowrates. While injecting at the optimum WAG ratio gives the fastest oil recovery, this allows CO2 to channel through the reservoir, leading to rapid CO2 breakthrough and extensive recycling of the gas. We propose to inject more water than optimum. This causes the CO2 to remain in the reservoir, increases the field life and leads to improved storage of CO2 as a trapped phase. Again, a short period of chase brine injection at the end of the process traps most of the remaining CO2. Finally, we investigated the e®ect of salt (halite) precipitation during dry, supercritical CO2 injection using our modifed streamline-based simulator. In this study, pseudo one- dimensional and two-dimensional homogeneous and heterogeneous systems were used to

study the sensitivity of di®erent parameters, which include relative permeability, grid size and brine salinity to salt precipitation. In our three-dimensional model, based on a geological model of a CO2 injection site, we constructed a near wellbore fine grid model with almost 1.5 million grid cells. Simulations were conducted successfully, and we found that salt precipitation can be a very important e®ect to consider when dry CO2 is injected into a high salinity reservoir. In this reservoir, after only 2 years of CO2 injection, about 20% of permeability of the reservoir was reduced, which will seriously reduce the injectivity of the injector and fluid flow within the reservoir.