SPE Members
Abstract This paper describes a technique to model the behavior of three phase low temperature CO2 systems using two phase PVT algorithms, and its application to two field CO2 projects.
An approximate tuning procedure has been developed for using a two phase equation of state to characterize the displacement behavior of CO2 in the three phase domain. This characterization technique results in good prediction of laboratory slim tube displacement tests and the minimum miscibility pressure (MMP) for both pure and impure CO2 streams. Laboratory corefloods and field scale displacements can also be modeled to a practical level of accuracy. With this pseudo two-phase representation, complex three-phase flash calculations and the need for four-phase relative permeabilities are avoided.
This approach was used in forecasting the North Ward Estes CO2 project. Volume of mixing effects characteristic of low temperature CO2 systems are modeled in a vertical fine resolution geostatistical cross section with a compositional simulator. Reasonable agreement with observed field performance is obtained. An immiscible CO2 huff-n-puff test at the McElroy field was also modeled, using a 2-D radial compositional model. The cross section was derived from geostatistics. The simulation also provided a good approximation of field observation
Introduction In low temperature CO2 reservoir fluid systems, complex phase behavior are observed. A small three-phase region is usually found in first contact PVT experiments, where two liquids and a gas (L-L-V) coexist. A solid phase is often observed as a precipitate at high molar concentrations of CO2. Multiple phase generation during CO2 flooding has also been studied by Shelton and Yarborough, Henry and Metcalfe. The L-L-V phase behavior makes the prediction of displacement efficiency difficult because of the complex phase equilibrium and the uncertainty in treating the relative permeability characteristic of the various phases.
For several Permian Basin reservoir fluid and CO2 systems (e.g., Chevron's North Ward Estes and McElroy oils), the three phase region occurs over pressures of importance in CO2 project operations. Exact compositional modeling of these multiple phases for CO2 displacement would require extensive computer time and laboratory data. The need for multiple-phase (involving the second liquid phase or a solid phase) calculations in reservoir simulation has not yet been established in the industry. The effect of a three phase region on the simulation of CO2 displacement efficiency has been studied by Nghiem and Li, and Fanchi. Use of two-phase flash calculations in the three-phase region did not seem to change recovery predictions significantly. However, the conclusions are based on simplified assumptions of relative permeability curves.
At displacement pressures close to the pressure range where the three phase region exists, unusual behavior in slim tube results have been observed by Creek and Sheffield. The displacement can be multi-contact miscible, yet shows non-piston like behavior, as shown in Figure 1 for slim tube displacement of a North Ward Estes oil by pure CO2. Because of this phenomenon, slim tube ultimate recovery can be underpredicted up to 20% OOIP with a two-phase compositional simulator, using a typical tuning procedure for oil (see Procedure A below). A difference of 5% OOIP oil recovery is large enough to lead to wrong conclusions about the miscibility of a CO2 displacement process. Negahban and Kremesec studied the tuning of equation-of-state parameters to match the MMP and three phase regions.
P. 289^