Geological carbon storage (GCS) involves unstable drainage processes, the formation of patterns in a morphologically unstable interface between two fluids in a porous medium during drainage. The unstable drainage processes affect CO₂ storage efficiency and plume distribution and can be greatly complicated by the mixed‐wet nature of rock surfaces common in hydrocarbon reservoirs where supercritical CO₂ (scCO₂) is used in enhanced oil recovery. We performed scCO₂ injection (brine drainage) experiments at 8.5 MPa and 45°C in heterogeneous micromodels, two mixed‐wet with varying water‐ and intermediate‐wet patches, and one water‐wet. The flow regime changes from capillary fingering through crossover to viscous fingering in the micromodels of the same pore geometry but different wetting surfaces at displacement rates with logCa (capillary number) increasing from −8.1 to −4.4. While the mixed‐wet micromodel with uniformly distributed intermediate‐wet patches yields ~0.15 scCO₂ saturation increase at both capillary fingering and crossover flow regimes (−8.1 ≤ logCa ≤ − 6.1), the one heterogeneous wetting to scCO₂ results in ~0.09 saturation increase only at the crossover flow regime (−7.1 ≤ logCa ≤ − 6.1). The interconnected flow paths in the former are quantified and compared to the channelized scCO₂ flow through intermediate‐wet patches in the latter by topological analysis. At logCa > − 6.1 (near well), the effects of wettability and pore geometry are suppressed by strong viscous force. Both scCO₂ saturation and distribution suggest the importance of wettability on CO₂ storage efficiency and plume shape in reservoirs and capillary leakage through caprock at GCS conditions.

The irregular morphology of single rock fracture significantly influences subsurface fluid flow and gives rise to a complex and unsteady flow state that typically cannot be appropriately described using simple laws. Yet the fluid flow in rough fractures of underground rock is poorly understood. Here we present a numerical method and experimental measurements to probe the effect of fracture roughness on the properties of fluid flow in fractured rock. We develop a series of fracture models with various degrees of roughness characterized by fractal dimensions that are based on the Weierstrass–Mandelbrot fractal function. The Lattice Boltzmann Method (LBM), a discrete numerical algorithm, is employed for characterizing the complex unsteady non-Darcy flow through the single rough fractures and validated by experimental observations under the same conditions. Comparison indicates that the LBM effectively characterizes the unsteady non-Darcy flow in single rough fractures. Our LBM model predicts experimental measurements of unsteady fluid flow through single rough fractures with great satisfactory, but significant deviation is obtained from the conventional cubic law, showing the superiority of LBM models of single rough fractures.