Solute transport in porous media studied by lattice Boltzmann simulations at pore scale and x-ray tomography experiments.

With the aid of nondestructive microfocus x-ray computed tomography (CT), we performed three-dimensional (3D) tracer dispersion experiments on randomly unconsolidated packed beds. Plumes of nonreactive sodium iodide solution were point injected into a sodium chloride solvent as a tracer for the evaluation of the dispersion process. The asymptotic dispersion coefficient was obtainable within the experimental scale and was summarized over Péclet numbers from 11.7 to ∼860. Then, the lattice Boltzmann method and moment propagation method were used to elucidate the mechanisms embedded in the dispersion phenomenon. The methods were rigorously verified against the classical Taylor dispersion problem and extended to simulate fluid flow and tracer dispersion in high-resolution 3D digital porous structures from CT. The method of moments, Lagrangian velocity correction function, and dilution index were thoroughly analyzed to evaluate the dispersion behaviors. Numerical simulations revealed ballistic and superdiffusive regimes at the transient times, whereas asymptotic dispersion behaviors appear at longer characteristic times. Besides, the observed transient times unanimously persist over convective length scales of around 12 particles transversely and 16 particles longitudinally. The estimated dispersion coefficients from simulation are in consistence with the experimental result. Furthermore, the simulation also enabled the identification of regimes, including diffusive, power law, and mechanical dispersion. Thus, the proposed experimental and computational schemes are of practical means to study dispersion behaviors by direct pore scale imaging and modeling.

Micro-tomographic analyses of specific interfacial area inside unconsolidated porous media with differing particle characteristics from microscopic to macroscopic scale.

HYPOTHESIS: In capillary trapping in an unconsolidated porous medium, the interphase area is influenced by the distribution of the trapped phase clusters. These attributes, in turn, are affected by particle characteristics, indicating that the interphase area is affected by the particle characteristics. EXPERIMENTS: A micro-tomography technique was used to observe capillary trapping of a nitrogen-water system. The effect of particle characteristics on micro- to macroscopic properties was measured, respectively: pore size distribution (PSD) and porosity, bubble size distribution (BSD) and saturation, and bubble surface area and specific interfacial area. Capillary trapping experiments were carried out for media consisting of different particle shapes, range of sizes, and degrees of uniformity. FINDINGS: Particle characteristics govern not only the PSD but also the pore shape. Then, the PSD and pore shape govern the BSD and bubble surface area, thus, affecting the specific interfacial area. The specific interfacial area of highly angular particles differs from that of highly spherical particles and natural sands. On the basis of these findings, a statistical model of PSD and a bubble morphology model are developed. Comparisons of specific interfacial area with uniform bubble distribution and thermodynamic filling assumption models show that those assumptions predict interfacial area less accurately.

Behavior of CO2/water flow in porous media for CO2 geological storage.

A clear understanding of two-phase fluid flow properties in porous media is of importance to CO2 geological storage. The study visually measured the immiscible and miscible displacement of water by CO2 using MRI (magnetic resonance imaging), and investigated the factor influencing the displacement process in porous media which were filled with quartz glass beads. For immiscible displacement at slow flow rates, the MR signal intensity of images increased because of CO2 dissolution; before the dissolution phenomenon became inconspicuous at flow rate of 0.8mLmin-1. For miscible displacement, the MR signal intensity decreased gradually independent of flow rates, because supercritical CO2 and water became miscible in the beginning of CO2 injection. CO2 channeling or fingering phenomena were more obviously observed with lower permeable porous media. Capillary force decreases with increasing particle size, which would increase permeability and allow CO2 and water to invade into small pore spaces more easily. The study also showed CO2 flow patterns were dominated by dimensionless capillary number, changing from capillary finger to stable flow. The relative permeability curve was calculated using Brooks-Corey model, while the results showed the relative permeability of CO2 slightly decreases with the increase of capillary number.

Velocity measurement of clay intrusion through a sudden contraction step using a tagging pulse sequence.

Magnetic resonance imaging (MRI) with a spatial tagging sequence was used to measure the velocity distribution of clay that was forced past a sudden contraction. A spatial tagging sequence provided magnetic resonance images of clay that allowed measurement of the velocity distribution in the clay, which can provide profound insights on the deformation process of clay during the intrusion process. The experiments were conducted using a specially-designed vessel that could operate at up to 30 MPa. The vessel offers a rectangle test section with a sudden contraction step that had a ratio of contraction of 2:1. The vessel was installed into a commercial magnetic resonance imaging equipment and then the fluid motion of clay flowing into the narrow contracted channel was quantitatively investigated to examine behaviors of flowing clay as non-Newtonian fluid. MRI results are compared with those obtained by computational fluid dynamics (CFD) calculation. Velocity distributions obtained from each tag displacement did not well agree with those predicted by CFD results near the contraction step where the fluid accelerated rapidly. However, a post-processing on calculation results, in which virtual tag displacement is calculated, gave better agreement with experiment and enabled us to compare MRI results with CFD results.