Assessment of Three-Phase Relative Permeability Models Using Laboratory Hysteresis Data

Abstract Gas injection techniques, such as WAG, frequently require consideration of co-existing oil, gas and water phases and the impact of saturation cycles as water and gas slugs move through the reservoir. These processes may be assessed using numerical simulations. This paper presents the analysis of a detailed laboratory study, designed to provide data for verifying hysteresis models for such simulations. A number of studies have reported evidence for hysteresis in gas relative permeabilities in WAG flooding, leading to lower gas mobilities than predicted by conventional two-phase models. A reduction in gas mobility tends to improve gas sweep and incremental recovery for WAG based IOR schemes. Three-phase hysteresis models have recently been developed to include these hysteresis effects. The models include trapping of gas and reduction of water relative permeability in the presence of trapped gas. In these models, saturation changes can be irreversible, and relative permeability may decrease with each change in direction in saturation. A carefully planned laboratory study investigated secondary and tertiary immiscible WAG floods in both water-wet and intermediate-wet Berea cores, giving four separate sets of experimental data (a total of over 30 individual floods). For each flood, in-situ saturation profiles, mass balance and pressure drop data were measured. The in-situ saturation data ensures that laboratory artefacts (such as capillary end effects) do not influence conclusions. Analysis of the experimental data shows that hysteresis models should include the following features: –Irreversibility of hysteresis cycles;–Potential for the reduction in the residual oil saturation with trapping of gas by water;–Reduction in both water and gas permeability, with potential for the fractional flow to vary with trapped gas saturation;–Variation in Land trapping factor between hysteresis cycles. This study confirms the need for three-phase hysteresis models. Although published models may include some of the observed hysteresis effects, no model includes them all.