ИСТИНА

Utilization of CO2 in saline aquifers is widely acknowledged as a viable technique to mitigate the effect of climate change by reducing the greenhouse gas emissions into the atmosphere. The utilization can be implemented by a few wells used to inject CO2 into a confined aquifer overlain by a caprock. The injection leads to the formation of CO2 plume mostly caused by the buoyancy-driven gas flow. The less denser gas, as compared to brine, rises to the upper parts of the aquifer and spreads below impermeable caprock. The modeling of the gas transport in the plume with account for all trapping mechanisms is very important for evaluating the consequences and potential risks of the utilization. A potential risk of the injection is the leakage through old, abandoned wells or faults near the injection site. If the injection occurs into a large regional sloping aquifer without a structural trap, then gas can reach much more distant areas from the injection site than in the case of injection into an anticline reservoir. In a negative scenario, the old wells and faults can serve as potential paths for gas to the Earth surface. The leakage can result in groundwater pollution or, in the case of the leakage to the atmosphere, makes the utilization worthless. To assess the noted risk of leakage we investigate supercritical CO2 injection into a sloping saline aquifer and propose a simple relationship to estimate the maximum gas migration distance in the updip direction. This might be the most hazardous direction to a leakage site. The proposed estimate is derived from the system of governing equations for immiscible flow of gas and formation brine. By writing the equations in non-dimensional form, we guess the scaling law for the migration distance at a late stage of CO2 injection. Then, we verify the scaling law by means of 3-D reservoir simulations of miscible CO2 injection with account for the residual and solubility trapping. We derive an estimate that relates the maximum migration distance with the dip angle, the porosity, the anisotropic permeability, and the end-points of saturation functions. We show that the estimate is rather accurate for different reservoir temperatures and brine salinity and in the case of a flue gas injection. We show that the proposed scaling is useful for a quick assessment of the risk of CO2 reaching a potential leakage site in a large regional aquifer. It can also be applied to estimate the propagation of the uncertainties of reservoir parameters to the uncertainty of the migration distance. We acknowledge the financial support of the Russian Science Foundation (grant no. 19-71-10051). We are also grateful to the reservoir engineers of the Gazpromneft Science and Technology Centre for fruitful discussions. Some simulation work of this study was partly funded by the Gazpromneft STC.