Interpretation of a CO2 WAG Injectivity Test in the San Andres Formation Using a Compositional Simulator

Abstract

ABSTRACT It has been well documented that tertiary carbon dioxide WAG cycle injectivity during displacements above the minimum pressure for dynamic miscibility cannot be reliably predicted on the basis of viscosity ratios and waterflood injectivity performance alone. Approximate analytical models based upon gravity-free displacement with nondispersive plug flow of constant composition slugs in noncommunicating radial layers have been used to interpret injectivity during reservoir tests as a function of relative permeability, flow geometry, effective wellbore radius, and layering. We examine compositional simulation as an alternative for incorporating the additional effects of phase behavior, dispersive mixing, gravity, capillary forces, viscous instability, crossflow, and more realistic representations of reservoir heterogeneity in injectivity calculations for field test interpretation. We show how simple modifications can be made to an existing finite-difference equation-of-state simulator to allow for detailed modeling with these mechanisms of reservoir-scale cross-sections that incorporate geostatistical representations of reservoir heterogeneity and radial flow near injection and production wells. The compositional model is applied to simulate and interpret an injectivity test conducted in the Mabee Field in the San Andres Formation, Martin County, Texas. Injectivity during the field test was significantly greater than initial waterflood injectivity during all WAG cycles, increased during each CO2 cycle, and was greater than would be anticipated from other field tests in the San Andres formation- or from available laboratory data. Compositional simulation of the field test indicates that the reservoir response during the first cycles of CO2 and brine injection, including the initial increasing CO2 injectivity trend and higher brine injectivity after the first CO2 cycle, are consistent with measured relative permeability, phase behavior, and reservoir characterization data, but that the observed injectivity increase during the second cycle of CO2 injection cannot be attributed to either the near-wellbore condition of the reservoir at the start of the test- or to the presence of thief zones that are statistically consistent with measured core data and well logs. Injectivity is a key variable in determining the economic incentive associated with a proposed CO2 project, and an important implication of this study is the validation of compositional simulation as a means for interpreting field tests and developing improved predictions of reservoir injectivity performance. Another important conclusion is that geostatistical techniques can be used successfully to characterize high heterogeneity in carbonate reservoirs for injectivity calculations.