Ten to possibly thousand fold ratios of horizontal permeability to vertical permeability are required in two "clean" sand reservoirs to match actual water-free product ion of oil and. gas of wells having bottom water. Layer models with average permeabilities of cores at similar stratigraphic position in the reservoir rather than values selected from the permeability frequency distribution and with highly restricted cross flow yield the best match between actual and calculated dilution of wet gas by injected gas in two cycling projects. These same reservoir modeling concepts yield reasonable agreement with actual performance of the Seeligson enriched gas drive test without assumption of serious viscous fingering with gravity segregation.
The effects of heterogeneity off reservoir rocks on fluid displacement efficiency remain as a very important unsolved problem in this period of rapidly growing application of fluid injection for increased recovery of oil. Hypotheses of reservoirs ranging from non-connecting layers characterized by permeability frequency distribution, irrespective of spatial origin of cores, to reservoirs so heterogeneous they are nearly homogeneous have been offered as bases for evaluation of actual oil recovery projects. Since we are forever barred from knowing the actual variation arid continuity of rock properties, it is imperative that we test our simplified models against performance of actual reservoirs to select those assumptions most appropriate as guides to judgment. This paper presents two types of field observations having a bearing on these models:
Determination that 10 to possibly1000 fold ratio of horizontal permeability to vertical permeability is required to match water coning performance of a number of wells in "clean" sand reservoirs in contrast with very low ratios usually observed in small core plugs. Possibly cross flow between layers in laboratory fluid flow models or calculated mathematically for multilayer systems with equal horizontal and vertical permeabilities may not be representative of actual reservoirs in which gross anisotropy appears to be much greater than that of small cores.
Agreement between actual dilution of gas streams by injected gas in two cycling projects and that calculated with assumption of parallel flow in separate layers is reasonable if permeabilities assigned to individual layers are average permeabilities of cores from different wells at the same relative vertical position in the reservoir section rather than values from the permeability frequency distribution.
Most performance features of the current Seeligson enriched gas drive project are adequately matched by this parallel layer type analysis.
