A Multiscale Methodology for Simulating Miscible Gas Injection Projects Applied to the ROD Field in Algeria

摘要

We present a methodology for the simulation of miscible gas injection applied to a water alternate gas (WAG) project being considered for the ROD (Rhourde Oulad Djemaa) field in Algeria. First we run one-dimensional (1D) simulations using an eight-component Peng-Robinson equation-of-state and tune the fluid characterization to match the results of slim tube experiments at different pressures. We use the simulations to determine the minimum miscibility pressure (MMP) by finding the pressure at which a piston-like advance of the injected gas is observed. We find that while acceptable recoveries that match the experiments may be obtained using 100 grid blocks, even when using 10,000 blocks a converged saturation profile is not obtained, especially close to the MMP. We then run 2D compositional simulations above the estimated MMP to model viscous fingering. We compare the results with those of a black-oil model that assumes a first-contact miscible displacement (FCM). The compositional simulation predicts much less fingering than the FCM black-oil model. This is because the resolution of the 2D compositional simulation, even with a very fine grid, is insufficient to resolve the sharp transition in viscosity from injected gas to reservoir oil and thus severely underestimates the apparent viscosity contrast. We recommend using a simpler FCM simulation if the system is above the MMP. The degree of fingering in the FCM model is accurately predicted using the Koval model. At the field scale we use FCM simulation with a Todd & Longstaff mixing parameter omega that accounts for small-scale fingering. For WAG floods the value of omega is adjusted to account for the reduced mobility ratio between injected gas and water and displaced oil. Overall we show how to account systematically for the development of miscibility and sub- grid-block fingering in field-scale simulation of a miscible gas injection project.