Summary
We investigate oil recovery from multicontact miscible (MCM) gas injection into homogeneous and crossbedded porous media, using a combination of well-characterized laboratory experiments and detailed compositional flow simulation. All simulator input data, including most EOS parameters, were determined experimentally or from the literature produced fluids in all experiments were found not to be in compositional equilibrium. This was not predicted by the simulator, giving a poor match between experimental and simulated oil recoveries. The match was significantly improved for the cross-bedded displacements by using alpha factors derived from the MCM displacements in the homogeneous pack.
Introduction
The recovery of oil by miscible gas injection has been a subject of interest and research in petroleum engineering for more than 40 years (Stalkup 1983). In a first-contact, miscible (FCM) displacement, the gas and oil mix instantly in all proportions. No capillary forces exist, so, in principle, residual oil saturation is zero, and 100% oil recovery should be achieved. In practice, many phenomena conspire to limit the efficiency of the miscible flooding process, including viscous fingering, gravity override, and permeability heterogeneity. Moreover, it is often not economical, and sometimes not technically feasible, to inject a gas that is first-contact miscible with the oil. Instead, the injected gas is designed to develop miscibility with the oil by mass transfer during the displacement. This is a so-called MCM gas injection. If the bulk of the mass transfer is from the gas to the oil, then the displacement is termed a condensing drive. If most of the mass transfer is from the oil to the gas, then it is termed a vaporizing drive. In most cases, however, because of the multicomponent nature of oil and gas, the mass transfer is actually a mixture of both these cases, and the displacement is termed a condensing-vaporizing drive.
Small-scale heterogeneities can have a significant impact on recovery efficiency (Jones et al. 1995; Jones et al. 1994; Kjonsvik et al. 1994), yet they cannot be modeled explicitly in field-scale simulations. Some of the most common small-scale heterogeneities found in sandstone reservoirs are laminations. However, because laminations have a small size and are generally at an angle to the principal flow direction, their influence onfluid flow is one of the most difficult features to predict numerically.
There is a significant amount of literature describing systematic investigations of first-contact miscible and immiscible displacement processes in laminated sandstones (Huang et al. 1995, 1996; Ringrose et al. 1993; Kortekaas 1985; Honarpour et al. 1994; Hartkamp-Bakker 1991, 1993; McDougall and Sorbie 1993; Marcelle-DeSilva and Dawe 2003; Borresen and Graue 1996; Roti and Dawe 1993; Dawe et al. 1992; Caruana and Dawe 1996; Caruana 1997). Both experimental and simulation studies show that significant volumes of oil can be trapped by capillary forces during immiscible displacements (Huang et al. 1995, 1996; Ringrose et al. 1993; Kortekaas 1985; Honarpour et al. 1994; Hartkamp-Bakker 1991, 1993; McDougall and Sorbie 1993; Marcelle-DeSilva and Dawe 2003; Borresen and Graue 1996; Roti and Dawe 1993; Dawe et al. 1992; Caruana and Dawe 1996; Caruana 1997). However, the influence of these heterogeneities on MCM displacements, during which capillary forces change from being very significant when gas is first injected to negligible once miscibility has developed, has not yet been investigated. Indeed, the only comparisons of well-characterized MCM displacement experiments and detailed simulations reported in anywhere in the literature are those of Burger and colleagues (Burger and Mohanty 1997; Burger et al. 1996; Burger et al. 1994).
Pore-Scale Modelling of Solvent-Based Recovery Processes
