A Multiphysics Simulation Framework for CO2 Huff-n-Puff in Shale Oil Reservoirs: Coupling Stress-Dependent Maxwell–Stefan Diffusion and Nanopore Confinement

Abstract

Summary Diffusion is non-negligible in shale reservoirs due to the ultra-tight matrix, while advection is also strongly stress-sensitive due to well-developed fractures. Yet, most CO2 huff-n-puff (HnP) models still employ Fickian diffusion and static rock properties. This work develops a unified, fully implicit compositional multiphysics framework that couples flow, geomechanics, nanopore confinement, and multicomponent Maxwell–Stefan (M–S) diffusion, and quantifies how their interactions govern CO2 HnP performance and net CO2 storage in shale oil reservoirs. First, a geomechanical model dynamically links effective stress to porosity, permeability, tortuosity, and characteristic pore radius. This radius updates capillary pressure via the Young–Laplace relation, embedded into an equation-of-state (EOS) flash calculation. To overcome severe nonlinearities in confined phase behavior, a robust hybrid Newton–bisection algorithm with modified fugacity evaluations ensures unconditional VLE convergence. Multicomponent M–S equations are reformulated into a chemical-potential-explicit matrix form and discretized using a finite-volume scheme. The global fully implicit system is solved using Newton–Raphson iterations within a multiple-interacting continua (MINC) formulation to resolve transient, diffusion-controlled fracture–matrix mass transfer. Comparative analyses of cyclic injection reveal distinct mechanistic effects: stress sensitivity has a pronounced negative impact, degrading fracture conductivity and lowering production rates relative to the base HnP case. Conversely, multicomponent M–S diffusion plays a highly beneficial role, significantly accelerating fracture–matrix mass transfer and yielding the highest production peaks. In the fully coupled case, diffusion effectively mitigates the detrimental impacts of stress compaction, yielding production rates substantially higher than those of the base and stress-only cases, although still slightly lower than those of the diffusion-only case. Furthermore, nanopore confinement is theoretically integrated to ensure consistent phase behavior and chemical-potential driving forces across these regimes. Ultimately, neglecting the interplay between stress-induced flow degradation and diffusion-driven matrix recharge leads to a significant underestimation of CO2-EOR cyclic performance. To the authors' knowledge, this work is the first to integrate nanopore-consistent multicomponent M–S diffusion with stress-dependent pore geometry in a fully implicit compositional multiphysics and MINC framework for CO2 HnP in shale oil reservoirs. It provides new quantitative insights into how coupled effects of geomechanical, confinement, and diffusion control recovery and net CO2 storage, offering a more reliable basis for designing CCUS-EOR projects in unconventional assets.