Estimation Of Reduced Pressure Build-Up Due To Brine Seepage Using A Convolution Technique
I. Aavatsmark and S.E. Gasda
Event name: ECMOR XVI - 16th European Conference on the Mathematics of Oil Recovery
Session: Discretization I
Publication date: 03 September 2018
Info: Extended abstract, PDF ( 3.17Mb )
Price: € 20
CO2 sequestration involves injection of large volumes of CO2 into a storage reservoir bounded above and below by low-permeability seals. Although capillary effects are usually sufficient to prevent CO2 leakage through the caprock, the native brine will slowly migrate through the over- and underlying units when subject to overpressure in the reservoir. At large scales and over long time periods, diffuse fluid migration may have an important impact on large-scale pressure development. Typically, simulation studies of CO2 injection often omit the possibility of brine migration through the top and bottom boundaries of the reservoir. One reason is that vertical fluid flow requires additional resolution outside of the storage reservoir that poses a large computational burden. Therefore, analytical methods are an attractive approach for capturing diffuse brine leakage. In low permeability layers, the flow is predominantly vertical, and the local system can be reduced to a 1D (vertical) equation. This can be solved on a semi-infinite (vertical) domain for a thick seal (> 10-20 m), with the reservoir overpressure applied as a boundary condition. Because the boundary condition is not constant in time, the resulting solution is a convolution integral that must be computed at regular time intervals. In this paper, we couple the analytical solution for vertical brine leakage with a vertical equilibrium simulator (VESA). The VESA simulator is a reduced-dimension (2D) numerical model for two-phase flow in gravity-segregated systems, which is an appropriate assumption for large-scale CO2 storage. Coupling analytical solution for brine migration into a numerical simulator gives greater flexibility modeling injection into heterogeneous reservoirs. We find that the additional computational burden of the convolution integral is minimal compared to solving the full 3D system at the correct resolution. A solution for pressure development in the adjacent strata is also obtained analytically, leading to a fully 3D representation of pressure in the system. The coupled code is benchmarked with a fully analytical solution, and then applied to large-scale CO2 injection into the Utsira formation. We study the impact of diffuse brine leakage on development of large-scale overpressure in the storage reservoir for scenarios of high-volume CO2 injection.