Reservoir Modeling of Restricted Platform Carbonates: Geologic/Geostatistical Characterization of Interwell-Scale Reservoir Heterogeneity, Dune Field, Crane County, Texas

Petroleum reservoirs typically yield only a fraction of the oil initially in place because geologic heterogeneity causes incomplete drainage of oil. Accurately predicting oil recovery requires realistic estimates of interwell porosity and permeability patterns that are discontinuous in both the horizontal and the vertical direction. Results of this study, which show the effects of heterogeneity on fluid flow and oil recovery efficiency through several simulation experiments, are applicable to many other carbonate and noncarbonate fields worldwide. The experiments were conducted using cross sections of a lower portion of the Grayburg reservoir in Section 15 of the Dune field. A technique combining petrophysics and geophysical borehole logs was used to estimate permeabilities at 1-ft intervals in wells spaced approximately 930 ft apart (20-acre spacing). Detailed interwell permeability patterns were estimated by both deterministic and stochastic geologic interpretation. Deterministic interpretation involved correlating permeabilities between wells and, owing to lack of a better geological model, assuming gradational changes in permeability where lateral discontinuities occurred. Stochastic geologic interpretations generated with a geostatistical method known as conditional simulation accounted for interwell uncertainty and provided the more complex, realistic permeability distributions that are needed to predict performance of infill wells.Simulation of waterflooding in the deterministic permeability distributions illustrated key mechanisms causing bypass of mobile oil. High-permeability strata desaturated rapidly and became thief zones, or conduits for rapid cycling of water from injector to producer, resulting in poor sweep efficiency. Despite horizontal stratification of high- and low-permeability facies, vertical flow in the simulations was significant, sometimes making isolation of thief zones by selective perforation difficult. Simulation of waterflooding in stochastically generated permeability distributions indicated that no more than 40 to 55 percent of the mobile original oil in place (MOOIP) could be produced with the existing well spacing (10 acres). Adding infill wells to achieve an average well spacing of 2.5 acres generally increased the recovery efficiency of MOOIP by at least 27 to 32 percentage points. The infill wells tended to cause high water/oil ratios where the well spacing was less than 2.5 acres or where the high-permeability facies were laterally extensive relative to well spacing. In general, the stochastic permeability distributions giving the most realistic water/oil ratios (WOR) in the simulator before infill also yielded low to moderate WOR's after infill.