CO2 Flooding Strategy in a Communicating Layered Reservoir

Abstract

Abstract A computer simulation study of the effects of crossflow between sublayers of a single major zone of a San Andres dolomite reservoir during injection of a CO2 slug followed by a waterflood shows that the crossflow causes a substantial reduction in oil recovery. This loss can be reduced by use of alternate injection of small increments of CO2 and water during the CO2 slug injection. Introduction The subject of this investigation is the Southeast Seminole San Andres Unit (SESSAU), which is operated by Belco Petroleum Corp. Properties of this reservoir are listed in Table 1. The Southeast Seminole field was discovered in Feb. 1964. Primary recovery through June 1979 was about 10.8% OOIP. It is estimated that ultimate primary recovery would be about 15%. However, a unit was formed in Feb. 1978 and waterflooding was begun in May 1979. Waterflood response has now occurred in most of the producing wells. CO2 flooding could be initiated soon or at a much later date after most of the secondary oil recovery has occurred. The purpose of this study was to determine the approximate range of enhanced oil recovery that might be expected from CO2 flooding, based on a five-spot symmetry element. It was recognized that this information could not be extrapolated directly to estimate such recovery for the entire unit, but it would serve as a basis for determining whether to pursue the possibility. In addition, guidance could be obtained as to the preferred strategy of operation in CO2 flooding based on reservoir and oil characteristics and the extent of waterflooding at the start of CO2 flooding. In this feasibility study, the following factors were investigated: 1. extent of previous oil recovery at the start of CO2 flooding, 2. amount of residual oil saturation left by CO2 flooding, 3. extent of viscous fingering of CO2 through the mobile oil phase, 4. extent of availability of waterflood-trapped oil for reconnection by CO2, 5. permeability reduction for the nonaqueous phase resulting from an asphaltic precipitate reported to occur in CO2 flooding, 6. CO2 slug size (always followed by waterflooding to an economic limit), 7. continuous injection of the CO2 slug vs. alternating injection of small increments of water and CO2 (WAG or WACO2), 8. permeability and porosity variation by layers, and 9. degree of crossflow between layers. The reservoir simulation program used in this study was developed by Todd, Dietrich and Chase Inc. to simulate either compositional mass-transfer processes during a displacement or miscible viscous fingering during a displacement, but not both at the same time. It is a finite-difference simulator, and thus uses a reservoir model in which the composition is considered uniform within each of the discrete grid blocks that comprise the reservoir pore space. Phase equilibrium can be calculated for the composition in each grid block. However, if viscous fingering occurs, then either the fineness of the grid must be smaller than the size of the fingers or a method such as the mixing-parameter method must be used to simulate in a coarser grid the effect of viscous fingering on the mobility of the fingering fluid and of the fluid through which the fingers are penetrating. If the compositional transition zone in which phase separations and mass transfer occur between phases in CO2 flooding is to be represented, the grid system must be even finer than that required to delineate viscous fingers, since this transition zone lies along the side surfaces and at the tips of the viscous fingers. The thickness of the transition zone is about 4 to 8 ft (1.22 to 2.44 m), and 50 grid blocks are required to represent phase equilibrium changes accurately through this zone.