New upscaling technique for compositional reservoir simulations of miscible gas injection

The use of compositional reservoir simulation is necessary to better represent the physical phenomena associated with enhanced oil recovery methods, particularly with miscible or near-miscible gas injection processes. However, upscaling of the geological fine model is required because computational costs can make it unpractical to perform compositional simulations in fine-scale models. Therefore, upscaling of fine-scale geological models is important and the use of coarse-scale models in reservoir simulation is necessary to reduce computational time. However, it can degenerate results due to loss of resolution of the small-scale phenomena, averaging of sub-grid heterogeneity and numerical dispersion, especially in oil fields where miscible gas is injected, continuously or alternatively with water (WAG). In black oil simulations, errors of the upscaling process are usually mitigated with the use of pseudo relative permeability curves when two-phase immiscible displacement is considered. In compositional reservoir simulation of miscible gas injection, the interaction between thermodynamic phase behavior and the sub-grid heterogeneities significantly increase the challenge of the upscaling procedures. This work presents a new technique for a robust upscaling of compositional displacements allowing a better representation of the small-scale results when miscible gas injection is considered, as in cases in which the produced gas is partially or totally reinjected in the reservoir. It can be applied in any compositional simulator with no need to adapt any transport coefficient or to use dual porosity dual permeability models, thus providing a flexible and computationally efficient approach. The proposed technique is based on employing an alternative fluid model with a pseudo minimum miscibility pressure (PMMP) above experimental values, in order to ensure an immiscible displacement. This solution does not violate the phase behavior and ensures formation of two hydrocarbon phases in the reservoir. Therefore, the gas relative permeability plays a role on the simulation results and can be used for an improved fit of the fine model production curves. The technique is applied in two steps, first an alternative fluid model is generated and then pseudo relative permeability curves are used to better fit the production curves. In this study, problems that occur when upscaling is applied in fields with miscible gas injection were investigated. We show that the proposed technique can significantly improve the coarse-scale results - production curves, sweep efficiency and gas saturation map profile - in different scenarios of heterogeneity. Three case studies were evaluated and compared with the standard static upscaling processes. Additionally, we show improvements in the production curve results and physical representation of gas injection behavior.