Pengfei Liu

Bentonite and polymer‐modified bentonites, as ubiquitous clay minerals in geological formations and widely utilized barrier materials in engineered systems, significantly influence the hydraulic properties of porous media due to their high swelling capacity and ultralow hydraulic conductivity ( k ). Accurate prediction of k is crucial not only for critical natural processes, such as rainfall infiltration, groundwater flow, and solute transport in subsurface aquifers, but also for engineering applications, such as contaminant containment, nuclear waste disposal, and CO ₂ geological storage. However, existing predictive models often rely on empirical assumptions and non‐physical fitting parameters. This study employs large‐scale molecular dynamics simulations integrated with the original Kozeny‐Carman (K‐C) equation to predict k across a wide range of dry densities for both pure montmorillonite (MMT, = 134.6–1759.4 kg/m ³ ) and carboxymethyl cellulose‐modified MMT (CMC‐MMT, = 147.5–1550.7 kg/m ³ ). Our approach fundamentally challenges the conventional understanding that original K‐C equation is unsuitable for clayey soils: molecular‐scale parameterization of free water porosity (), tortuosity (), and water‐accessible surface area ( S ₀ ) yields predictions within a factor of five of experimental data. Critically, k reduction was proven to stem from bound water immobilization, where polymer modification amplifies this effect through three synergistic mechanisms: (a) enhanced water adsorption thickens bound layers (reducing by 18%–76%), (b) pore filling increases by up to 6 times, and (c) elevated S ₀ further restricts flow pathways. This work establishes a physically rigorous framework for hydraulic conductivity prediction in clay‐rich porous media, resolving long‐standing controversies in clay hydraulics and offering insights applicable to both natural and engineered systems.

Understanding nanoparticles, like carbon nanotube (CNT), transport and retention in porous media is critical for environmental risk assessment and remediation. Millifluidic visualization, spectral induced polarization (SIP), and deep neural network (DNN) were integrated to resolve spatio‐temporal CNT deposition dynamics under complex geochemical conditions. Millifluidic visualization revealed concentration‐dependent retention mechanisms: irreversible straining dominated at high CNT concentrations, while site‐blocking enabled delayed redistribution during flushing. SIP detected real‐time retention by chargeability (R² = 0.82–0.96), while a DNN decoded SIP signals to predict spatially resolved CNT deposition (R² = 0.926), outperforming phenomenological Cole‐Cole modeling. Surface coatings (e.g., montmorillonite, MMT) and divalent cations modulate retention by electrostatic repulsion and pore‐throat constriction, with CNT concentration governing transport efficiency. The hierarchy of controlling factors: CNT concentration > surface properties > cation valence > ionic strength (IS), provides a predictive framework for nanoparticle fate in subsurface environments. This integrated system bridges lab‐scale insights to field‐scale monitoring, advancing non‐invasive SIP applications for nanomaterial management.

Gas transport in porous media has been widely investigated in the geo-environmental, geotechnical, and chemical engineering fields. However, the mechanism for the transition of the gas flow pattern under the influence of the gravity field is not clear. This study gives some insights into this issue by using a hypergravity microfluidic experiment and pore-scale simulations based on the phase-field method. As the centrifugal acceleration increases, the gas transport pattern undergoes a gradual transition from a continuous finger flow to a discontinuous bubbly flow. An increase in pore size results in a weakening of the capillary effect, allowing the buoyancy force to gradually become the dominant force. The numerical simulation method reproduces experimental results of hypergravity microfluidics for chips with different pore sizes. Enhanced buoyancy effect leads to a significant reduction in sweep efficiency. Hypergravity field enhances separation of gas and liquid phases, thereby facilitating the generation of discontinuous bubbles. The dividing line between continuous and discontinuous flow is defined based on dimensionless numbers, which are supplemented by the numerical simulation results.

In the petroleum engineering and chemical industries, fluids engaging in displacement often have non-Newtonian properties, even though many former studies assume constant viscosities in the defending fluid. In this study, the computational fluid dynamics approach was performed in a two-dimensional model with uniformly distributed disks. This arrangement helps reveal the phenomenon and mechanics of how non-Newtonian characteristics of defending fluid affect two-phase displacement in porous media. Both global (in the whole medium) and regional (in the pore throat) studies revealed that shear-thinning makes capillary force and the pressure in the invading fluid decisive and leads to a uniform pattern. Meanwhile, the shear-thickening causes fingering due to the pressure drop in the defending fluid that becomes decisive. Cases of increasing injection rates were investigated to verify their ability to improve efficiency. The results verified that increased injection rates are effective in shear-thinning cases but energy-intensive when it comes to costs in shear-thickening cases. Finally, the viscosity ratio and capillary number (M-Ca) diagram were extended by plotting non-Newtonian cases as lines to consider viscosity variation. An estimation method was presented, which calculates the characteristic viscosity and locates non-Newtonian cases on an M-Ca diagram. This work can serve as a reference for enhanced oil recovery method development and microfluidic manipulation.