Numerical analysis of the influence of miscibility in density-driven porous media flows

Purpose Density-driven convection in partially miscible systems plays a critical role in applications such as CO2 sequestration; however, the coupled interaction between buoyancy-driven instability and interfacial dynamics under varying miscibility remains understood. In particular, the conditions governing the transition between plume-like convection and droplet-forming Plateau–Rayleigh instability (PRI) have not been systematically quantified. The purpose of this study is to investigate how miscibility (cs), Rayleigh number (Ra) and viscosity contrast (R) together control the onset, morphology and mass transfer characteristics of density-driven convection. Design/methodology/approach The authors perform non-numerical simulations using the Darcy–Cahn–Hilliard model over a wide range of Ra and R, systematically analysing the onset time of convection, droplet formation dynamics and solute flux evolution. Findings The results demonstrate that immiscible systems with high interfacial tension undergo PRI-induced droplet breakup, whereas partially miscible systems form jet-like dissolution streams. Increasing Ra intensifies PRI and enhances droplet formation, while simultaneously delaying the onset of density-driven convection. For favourable viscosity contrast (R ≤ 0), smaller and more numerous droplets are generated for increasing cs, whereas R 0 suppresses interfacial breakup and promotes plume-dominated convection. The solute flux attains a quasi-steady value, i.e. F ∼ 0.01, that is largely independent of cs and Ra, but strongly dependent on R. Originality/value These findings quantitatively establish how buoyancy, miscibility and viscosity contrast govern the transition between interfacial and density-driven instabilities. This study provides a unified framework for predicting drop formation and mass transfer in partially miscible porous media systems, with implications for subsurface CO2 storage and related multiphase transport processes.