To characterize the influence of reservoir conditions upon multiphase flow, we calculated fluid displacements (drainage processes) in 3D pore spaces of Berea sandstone using two-phase lattice Boltzmann (LB) simulations. The results of simulations under various conditions were used to classify the resulting two-phase flow behavior into three typical fluid displacement patterns on the diagram of capillary number (Ca) and viscosity ratio of the two fluids (M). In addition, the saturation of the nonwetting phase was calculated and mapped on the Ca-M diagram. We then characterized dynamic pore-filling events (i.e., Haines jumps) from the pressure variation of the nonwetting phase, and linked this behavior to the occurrence of capillary fingering. The results revealed the onset of capillary fingering in 3D natural rock at a higher Ca than in 2D homogeneous granular models, with the crossover region between typical displacement patterns broader than in the homogeneous granular model. Furthermore, saturation of the nonwetting phase mapped on the Ca-M diagram significantly depends on the rock models. These important differences between two-phase flow in 3D natural rock and in 2D homogeneous models could be due to the heterogeneity of pore geometry in the natural rock and differences in pore connectivity. By quantifying two-phase fluid behavior in the target reservoir rock under various conditions (e.g., saturation mapping on the Ca-M diagram), our approach could provide useful information for investigating suitable reservoir conditions for geo-fluid management (e.g., high CO 2 saturation in CO 2 storage). 2013). No-slip boundary conditions were imposed at all solid nodes via a halfway

No abstract

Given the world's growing demand for energy, a combination of geological CO2 sequestration and enhanced oil recovery (EOR) technologies is currently regarded as a promising solution, as it would provide a means of reducing carbon emissions into the atmosphere while also leading to the economic benefit of simultaneously recovering oil. The optimization of injection strategies to maximize CO2 storage and increase the oil recovery factors requires complicated pore‐scale flow information within a reservoir system consisting of coexisting oil, water, and CO2 phases. In this study, an immiscible three‐phase lattice‐Boltzmann (LB) model was developed to investigate the complicated flow state with interaction between water, oil, and CO2 systems in porous media. The two main mechanisms of oil remobilization, namely, double‐drainage and film flow, can be captured by our model. The estimation of three‐phase relative permeability is proposed using the digital rock physics (DRP) simulations. The results indicate that the relative permeability of CO2 as calculated using our steady state method is not sensitive to the initial oil fraction if the oil distribution is originally uniform. Baker's (1988) empirical model was tested and found to be able to provide a good prediction of the three‐phase relative permeability data. Our numerical method provides a new tool for accurately predicting three‐phase relative permeability data directly based on micro‐CT rock images.

The CO_{2} behavior within the reservoirs of carbon capture and storage projects is usually predicted from large-scale simulations of the reservoir. A key parameter in reservoir simulation is relative permeability. However, mineral precipitation alters the pore structure over time, and leads correspondingly to permeability changing with time. In this study, we numerically investigate the influence of carbonate precipitation on relative permeability during CO_{2} storage. The pore spaces in rock samples were extracted by high-resolution microcomputed tomography (CT) scanned images. The fluid velocity field within the three-dimensional pore spaces was calculated by the lattice Boltzmann method, while reactive transport with calcite deposition was modeled by an advection-reaction formulation solved by the finite volume method. To increase the computational efficiency and reduce the processing time, we adopted a graphics processing unit parallel computing technique. The relative permeability of the sample rock was then calculated by a highly optimized two-phase lattice Boltzmann model. We also proposed two pore clogging models. In the first model, the clogging processes are modeled by transforming fluid nodes to solid nodes based on their precipitated mass level. In the second model, the porosity is artificially reduced by adjusting the gray scale threshold of the CT images. The developed method accurately simulates the mineralization process observed in laboratory experiment. Precipitation-induced evolution of pore structure significantly influenced the absolute permeability. The relative permeability, however, was much more influenced by pore reduction in the nonwetting phase than in the wetting phase. The output of the structural changes in pore geometry by this model could be input to CO_{2} reservoir simulators to investigate the outcome of sequestered CO_{2}.