Time-Lapse Diffusion Logs to Discriminate Reservoir Oil from Miscible Gas

Abstract

Abstract A carbonate light oil reservoir surrounded by salt is without an active water or gas cap. The reservoir energy is limited. Without miscible gas flooding the initial production will drop quickly and the ultimate recovery is expected to be 10%. With miscible gas flooding the recovery is hoped to be in excess of 30%. Oil desaturation measurements in observation wells near the injectors may give an early proof of this hypothesis, providing information on the miscible gas front and the efficiency of the sweep. However, the contrast between the reservoir oil and rich injected gas is small for monitoring with conventional cased-hole nuclear measurements. Cased-hole resistivity will not help, because the water saturation is low and not expected to change. An effort was made to qualify another measurement that could distinguish the two fluids to supplement the conventional logs: the diffusion of polarized hydrogen. The hydrogen associated with the longer chain molecules in the oil diffuses slower than the lighter hydrogen in the injected gas. Measuring the diffusion distribution of a hydrocarbon composition is possible with nuclear magnetic resonance (NMR). A diffusion model was created using the liquid and gas compositions that were predicted using a PVT model as the proportion of injected gas to native oil was increased. This diffusion distribution indicated the liquid and gas phases could be distinguished and were therefore separately measurable for all the proportions modeled. This was verified with an NMR laboratory measurement of bulk fluids at various oil and gas proportions at reservoir conditions. The carbonate formation in this reservoir is characterized by heterogeneity; permeability variations of two orders of magnitude, and wide pore-size distribution. The hydrogen diffusion that is intended to discriminate the two fluids is partially suppressed in the formation pore volume due to restricted diffusion. The second phase of laboratory experiments consisted of quantifying the effect of the 90% dolomite 10% anhydrite rock on the diffusion of gas and liquid when constrained by the formation pore structure. The gas phase diffusion was reduced by a factor of four, and the liquid phase diffusion was reduced by 10%. Even with these reductions there was only little overlap of the diffusion distributions between the live reservoir fluid and injected sour gas that saturated the core plugs of porosity 21, 15, and 7. Laboratory tests also validated the NMR-transparency of special casing at reservoir conditions.