Streamline Simulation of Four-Phase WAG Processes

Abstract

Abstract Miscible or near-miscible water-alternating-gas (WAG) flooding is an attaractive process for medium viscosity (30–300 cp) oils if solvents are available. Four fluid phases can develop in these processes. Very few reservoir simulators and none of the published streamline simulators can handle four fluid phases. In this work, we have developed a four-phase streamline module that works with an existing finite-difference simulator to study gas or WAG injections in medium viscosity reservoirs. 2-D and 3-D simulations of WAG injections in quarter five-spot models have been demonstrated using this simulator. WAG injection simulation of the reservoir oil indicates that four phases exist near the gas-oil displacement front. The second liquid phase (or the third hydrocarbon phase) is present in many grid blocks under the conditions studied. WAG injection improves the sweep over the single slug solvent injection. For the cases studied, the sweep efficiency increases with the WAG ratio, but this result cannot be generalized. Reservoir hetrogeneity also affects the sweep. Gravity override is observed in WAG simulations with the vertical (X-Z) crossection. Oil recovery is low in the lower part of the reservoir. Horizontal production wells affect the fluid flow and thus the sweep of the reservoirs. Introduction There are many moderately viscous (30–300 cp) oil reservoirs where the viscosity is not large enough to warrant thermal recovery techniques. Miscible or near-miscible water-alternating-gas (WAG) flooding can be considered if the waterflood recovery is small and solvents are available. West Sak / Shraeder Bluff is such a resource in the North Slope of Alaska; mixtures of natural gas liquid (NGL), lean gas (LG) and CO2 are available and have been suggested[1–2] for injection. Laboratory studies conducted by Khataniar et al 3 show that mixtures of 85% CO2–15% NGL or 60% LG-40% NGL develop miscibility at the reservoir pressure and temperature. Most viscous oil reservoirs are also at a low temperature. When temperature < 120 ºF and pressure < 2000 psi, oil and gas mixtures can form three hydrocarbon phases (oil, gas and a second liquid phase, L2). This behavior has been shown for several West Texas oils and CO[2.4–5] Mohanty et al [6–7] have conducted slimtube/micromodel studies with the West Sak oil and hydrocarbon solvents. They showed that three hydrocarbon phases coexisted under reservoir conditions (totaling four phases if water is included) and condensation of solvent components into the oil and the resulting viscosity reduction improved oil recovery. Mechanistic modeling of WAG processes in these reservoirs involves modeling of four fluid phases, which most commercial reservoir simulators are not designed to handle. There are two issues related to four fluid phases: computation of phase equilibrium of three hydrocarbon phases and modeling of relative permeability of four phases. Many phase behavior packages have the capability of computing phase equilibria between three hydrocarbon phases. Such computations are, however, much more computationally expensive than two-phase equilibria calculations.[8] Measurement of four phase relative permeability is difficult if not impossible; little experimental data is available on four-phase flow. Four-phase relative permeability models have been hypothesized in the past. Guler et al. [9] have proposed a four-phase relative permeability model based on the Baker model[10] for three-phase flow for water-wet media. Li et al.[11] have proposed a four-phase relative permeability model for mixed-wet reservoirs. UTCOMP[12] is one of the finite-difference-based simulators which can handle three hydrocarbon phases and water. Guler et al[9] and Li et al.[11] have conducted compositional simulations of gas and WAG injection in typical patterns of visous oil reservoirs involving four phases. These simulations are computationally very expensive primarily due to three-phase flash calculations, leading to the use of small number of grid blocks and high numerical dispersion. Thus, there is a need for more efficient computational techniques.