A New Modeling Approach to Optimize Methane-Propane Injection in a Field After CHOPS

摘要

Although cold heavy oil production with sands (CHOPS) is an economically attractive method, ultimate recovery does not exceed 10%. Cyclic solvent injection (CSI) has been under consideration as a follow-up EOR application in the industry. This method targets extracting large amounts of remaining oil in the matrix by solvent diffusion, taking advantage of its high contact area with wormholes. Methane and propane are two potential solvents to be used in this practice. Methane is preferred due to its availability and stronger foaming characteristics while propane has lower foaming but better mixing capability. A far-reaching-core to field scale-study was conducted in this paper to test out the potential of pure methane and its mixture with propane as prospective CSI solvents. A 1.5 m long and 5 cm diameter sand-pack was initially saturated with brine. Then, live oil (saturated with methane and methane-propane mixture at different ratios) was transferred to drain the brine out. Typical field scale pressure decline rates were applied and pressure was monitored through eight equally spaced transducers placed onto the core holder. The experimental data from core and PVT tests were matched to a core scale simulation model to obtain K-values. These data were carried to a field scale model to analyze the CSI performance for methane and methane-propane mixture. In field scale modeling, 15-well data from CHOPS field from Alberta, Canada were history matched and 6-cycle CSI performances were followed as post-CHOPS with different well patterns (central, peripheral, all-wells). As a result of these experiments, methane showed about 14% oil recovery, but with additional CO2 huff 'n' puff around 15% recovery was added, totaling 29% recovery. Methane-propane mixture resulted in a lower oil recovery, which was about 5%, due to decreased foamy effect. Valid core-scale simulation was completed by tuning K-values and considering non-equilibrium or equilibrium impact depending on solvent type, showing mostly less than 5% error. In field scale modeling, central and peripheral well patterns yielded oil recoveries consistent with the experiments while all-well huff 'n' puff-type pattern showed a slightly higher value.