A New Method for Modeling Bypassed Oil Recovery in EOS Compositional Simulation

Abstract

Abstract The conventional reservoir simulation uses the local equilibrium assumption, where fluids are completely mixed and at equilibrium within individual gridblocks. This assumption allows for no bypassing of oil at the sub-grid scale in gas injection simulation. Oil bypassing by gas, however, always occurs due to micro and macroscopic heterogeneities, gravity, and front instability. There are a few methods proposed in the literature that attempt to model the effects of sub-grid heterogeneity in gas injection simulation. The concept of miscible residual oil saturation (i.e., the Sorm method) excludes the immobile oil from flash calculations and explicitly models bypassed oil. However, the Sorm method cannot model bypassed oil recovery because it allows for no mass flux between the bypassed and flowing fractions. The alpha-factor method uses transport coefficients to adjust components’ flux in compositional simulation. It has been used to retain a desired amount of residual oil in history matching for predominantly single-phase flow. However, its applicability to partial miscibility conditions is not fully understood. This is likely because the presence of more phases with partial miscibility yields more non-linearity and severer non-uniqueness in the history matching process. In this research, we develop an efficient method to model bypassed oil recovery in multiphase compositional simulation. Oil bypassing in gas injection simulation is first explained using dual-porosity-flow (DPF) models. We show that DPF can capture the characteristics of the dispersion-capacitance model, where oil held up in the bypassed fraction gradually migrates to the flowing fraction through intra-block mass flux. Fluid flow in the presence of capacitance is characterized using the DPF parameters; the bypassed fraction, throughput ratio (RT), and Peclet number. A new fluid characterization method is then used to efficiently reproduce DPF characteristics using a single-porosity-flow (SPF) model. An EOS fluid model is corrected for capacitance by adding heavier oil components that are characterized depending on RT. A case study demonstrates successful application of the new method to reflect capacitance effects observed in corefloods in field-scale gas injection simulation. Prediction of residual oil distribution within a reservoir depends significantly on how bypassed oil is modeled in the simulation. Unlike the alpha-factor method, our method requires no changes in the governing equations and in relative permeabilities.