Tertiary carbon dioxide injectivity is not always greater than brine injectivity during an initial cycle of CO2 injection above its minimum miscibility pressure (MMP), even though CO2 has a much lower viscosity than brine. This paper goes beyond the approximate analytical models used for previous investigations by examining the effects of dispersive mixing, crossflow between layers, viscous instability and phase behavior on injectivity for reservoir-scale displacement conditions in the limit of large permeability correlation length. A more realistic and permeability correlation length. A more realistic and comprehensive analysis of the influence of relative permeability on injectivity that does not rely upon simplifying assumptions regarding mixing and phase behavior is also provided in the form of a systematic sensitivity study. The sensitivity calculations incorporate recently reported high-pressure two- and three-phase flow relative permeabilities measured at elevated temperature with CO2, decane, and brine in a Guelph dolomite core as a base case. This paper compares analytical calculations with numerical results obtained using a finite-difference, compositional, equation-of-state simulator. Compositional simulations that model reservoir-scale physical dispersion and cross flow are numerically difficult to make accurately, so careful attention has been paid to this aspect of the problem. First, a numerical dispersion control scheme based upon a third-order-correct finite-difference method that has been extensively tested and compared with simple analytical solutions is used. Second, the work presented here is validated by mesh refinement studies that evaluate truncation error and demonstrate that the grid used is adequate for this purpose. The most important conclusion is that the mixing phenomena due to dispersion, crossflow and viscous instability that phenomena due to dispersion, crossflow and viscous instability that simple analytical models neglect can significantly influence injectivity. The results indicate that three-phase flow of gas, oil and brine needs to be modeled and that, contrary to the conclusions of other investigators, three-phase flow effects can have important influences on injectivity, even when CO2 is injected above its MMP. In addition, the most sensitive relative permeability parameters for reservoir-scale, tertiary CO2 flooding conditions in parameters for reservoir-scale, tertiary CO2 flooding conditions in the presence of correlated permeability heterogeneity are identified. Sensitivity to the relative permeability parameters can also be substantial, even at low permeability contrast. The results presented in this paper are particularly pertinent to the hybrid-WAG displacement process since they provide information about the first cycle of CO2 injection.
Field and laboratory experience have shown that a variety of injectivity responses are possible when CO2 is introduced into a previously waterflooded reservoir to recover tertiary oil above the previously waterflooded reservoir to recover tertiary oil above the MMP. Although operators of several tertiary field pilots and full-scale projects have reported no significant CO2 injectivity problems or trends of increasing injectivity with continued throughput, low CO2 viscosity is not a sufficient condition for high injectivity relative to terminal waterflood performance. Observations of dramatically reduced or declining injectivity have also been reported to occur during the initial periods of continuous CO2 injection at tertiary pilots in the Crosset, Levelland, Slaughter Estate and Wasson fields. Furthermore, injectivity measurements conducted in laboratory cores are not direct indicators of injectivity at the field scale.