Impact of Reservoir Mixing on Recovery in Enriched-Gas Drives Above the Minimum Miscibility Enrichment

Summary Gas enrichment is an important variable used to optimize oil recovery in enriched-gas drives. For slimtube experiments, oil recoveries do not increase significantly with enrichments greater than the minimum miscibility enrichment (MME). For field projects, however, the optimum enrichment required to maximize recovery on a pattern scale may be different from the MME. The optimum enrichment is likely the result of greater mixing in reservoirs than in slimtubes. In addition, 2D effects, such as channeling, gravity tonguing, and crossflow, can impact the enrichment selected. Numerical simulation is often used to model the effect of physical mixing on oil recovery in miscible gasfloods. Unfortunately, numerical dispersion can cloud the interpretation of the results by artificially increasing the level of mixing in the reservoir. This paper investigates the interplay among various mixing mechanisms, enrichment levels, and numerical dispersion. The mixing mechanisms examined are mechanical dispersion, gravity crossflow, and viscous crossflow. The U. of Texas Compositional Simulator (UTCOMP) is used to evaluate the effect of these mechanisms on recovery for different grid refinements, reservoir heterogeneities, injection boundary conditions, relative permeabilities, and numerical weighting methods, including higher-order methods. The reservoir fluid used for all simulations is a 12-component oil displaced by gases enriched above the MME. The results show that for 1D enriched gasfloods, the recovery difference between displacements above the MME and those at or near the MME increases significantly with dispersion. The trend, however, is not monotonic and shows a maximum at a dispersivity of approximately 4 ft. The trend is independent of relative permeabilities and gas trapping for dispersivities of less than approximately 4 ft. For 2D enriched gasfloods with slug injection, the difference in recovery generally increases as dispersion and crossflow increase. The magnitude of the recovery differences is less than that observed for the 1D displacements. Recovery differences for 2D models are highly dependent on relative permeabilities and gas trapping. For water alternating gas (WAG) injection, the differences in recovery increase slightly as dispersion decreases. That is, the recovery difference is significantly greater with WAG at low levels of dispersion than with slug injection. For the cases examined, the magnitude of recovery difference varies from approximately 1 to 8% of the original oil in place (OOIP). Introduction Gas enrichment is an important optimization variable in enriched-gas drives. Recoveries from slimtube experiments often give a sharp bend at the MME. Above the MME, slimtube recoveries do not increase significantly with enrichment. The optimum enrichment required to maximize recovery on a pattern scale in the field, however, is likely different from the MME. The difference in the optimum enrichment may largely be the result of greater mixing in the reservoir than exists in slimtubes. In addition, enrichment may impact sweep efficiency in 2D displacements. Oil and gas mixing in a reservoir can include mechanisms such as molecular diffusion, mechanical dispersion, gravity crossflow, viscous crossflow, and capillary crossflow. There are several reasons why recovery could increase for enrichments beyond the MME. First, the density and viscosity of the gas will increase with enrichment, which may improve sweep efficiency. Second, mixing can cause an otherwise multicontact miscible (MCM) flood to develop some two-phase flow.1–4 Richer gases mix closer to the critical locus in the two-phase zone, which causes a smaller and slower lean-gas bank. A smaller lean-gas bank could improve sweep efficiency. Last, richer gases, which mix near the critical locus, decrease miscible residual oil by increasing the velocity of the trailing evaporation fronts. Several authors have examined the effect of mixing and enrichment above the MME on oil recovery. Johns et al.5,6 recently considered the effect of dispersion on recovery in 1D displacements. They showed that the "knee" in the recovery curve from slimtube experiments depends on the level of dispersion. For small dispersivities typical of slimtubes, the knee occurs at the MME. For greater levels of mixing, they showed that the knee in recovery could occur at enrichments much greater than the MME. Chang et al.7 matched coreflood displacements with reservoir simulations at different enrichments and showed that recovery increased sharply for enrichments above the MME. Chang concluded that the increased recoveries were caused by higher displacement and sweep efficiencies as the enrichment level increased. The better sweep efficiency was attributed to increased gas density with enrichment. Jerauld8 also observed an increase in recovery above the MME. Giraud et al.9 observed that the highest recovery occurred at pressures above the minimum miscibility pressure (MMP). Stalkup10 showed that significant additional recovery might be obtained by injecting enriched gases above the MME. A significant increase in recovery occurred for longitudinal dispersivities of as low as 0.3 ft when the solvent and water were injected in slugs. He also concluded that mixing solvent and oil by viscous crossflow during WAG might dominate other mixing mechanisms in the reservoir (i.e., dispersion). Thus, he suggested that predictions of oil recovery from coarse gridblocks might be accurate as long as the vertical gridding is sufficient to capture the effect of viscous crossflow during WAG. Stalkup also showed that simultaneous grid refinement in horizontal and vertical directions caused numerical dispersion to overestimate the recovery with gas enrichment beyond the MME.11–13 Other papers have also examined the effect of viscous crossflow, capillary pressure, diffusion, gravity, heterogeneities, and numerical gridding on recovery.14–16 Pande17 obtained contradictory results from Stalkup. For the 2D cases considered, she showed that the optimal enrichment might be below the MME, not above it. Pande, however, did not examine the effect of gridding or gas trapping on the results. She also used a fluid characterization that exhibited a small sensitivity of displacement efficiency to dispersion, making her results sensitive only to sweep efficiency. Nevertheless, her results indicate that the optimal enrichment is highly problem-dependent.