Diffusion-Based Multiphase Multicomponent Modeling of Cyclic Solvent Injection in Ultratight Reservoirs

Abstract The cyclic solvent (gas) injection has been proved as an economical and effective method to enhance oil recovery in ultratight reservoirs such as shales. However, accurate modeling of cyclic solvent injection has been challenging due to the complex nature of fluid transport in these nanoporous media. Most models are developed based on Darcy's and Fick's laws, which do not capture some critical transport phenomena within nanopores at reservoir conditions. Accordingly, we develop a predictive numerical model encapsulating key transport mechanisms for cyclic solvent injection in ultratight reservoirs. The model is developed based on the binary friction concept that incorporates friction between different fluid molecules as well as fluid molecules and pore walls. The Maxwell-Stefan approach is employed to account for the friction among fluid molecules. The friction between molecules and pore walls is incorporated through partial viscosity and Knudsen diffusivity. A general driving force, chemical potential gradient, is considered for the transport of non-ideal fluid mixtures in ultratight reservoirs. The Peng-Robinson equation of state with confinement effect is used for the phase behavior calculations. The total flux consists of multicomponent molecular diffusion flux resulting from the chemical potential gradient and pressure diffusion flux driven by the pressure gradient. The governing equations for composition and pressure are solved implicitly using the finite difference method. After conducting time-step and grid-size sensitivity analysis, the developed model is validated against analytical solutions and experimental data. The primary production and solvent injection process are then simulated for a trinary oil (CH4, C4H10, and C12H26) and two solvent types (CH4 and CO2). The results show that the transport of hydrocarbon components in the vapor phase is faster than in the liquid phase due to the higher component transmissibilities in the vapor phase. Accordingly, light and heavy components are produced at different rates during primary production since the vapor phase mainly consists of lighter components. For the single-cycle solvent injection cases, CO2 and CH4 improve hydrocarbon recovery, with CO2 slightly performing better than CH4. This is attributed to CO2's ability to extract more intermediate and heavy components into the vapor phase as compared with CH4. The recovery factor of heavy components after CO2 injection (6.2%) is higher than that of CH4 injection (5.9%). For multi-cycle solvent injection cases, the incremental hydrocarbon recovery (0.7%) is slightly better for CO2 injection than CH4 injection (0.3%). Furthermore, the results reveal that CO2 cyclic injection results in producing more intermediate and heavy components from the matrix region in the vicinity of the fracture, while CH4 cyclic injection extracts more light components. The bottomhole pressure sensitivity analysis results indicate that the CH4 injection performance is better under single-phase conditions, while CO2 performance is better under two-phase conditions. Finally, the soaking-time sensitivity analysis results show that the solvent recycling rate decreases and the incremental recovery per cycle increases as the soaking time increases.