Pore-scale network modeling has become an effective method for accurate prediction and upscaling of macroscopic properties, such as permeability. Networks are either mapped directly from real media or stochastic methods are used that simulate their heterogeneous pore structure. Flow is then modeled by enforcing conservation of mass in each pore and approximations to the momentum equations are solved in the connecting throats. In many cases network modeling compares favorably to experimental measurements of permeability. However, computational and imaging restrictions generally limit the network size to the order of 1 mm3 (few thousand pores). For extremely heterogeneous media these models are not large enough in capturing the petrophysical properties of the entire heterogeneous media and inaccurate results can be obtained when upscaling to the continuum scale. Moreover, the boundary conditions imposed are artificial; a pressure gradient is imposed in one dimension so the influence of flow behavior in the surrounding media is not included. In this work we upscale permeability in large, heterogeneous media using physically-representative pore-scale network models (domain ~106 pores). High-performance computing is used to obtain accurate results in these models, but a more efficient, novel domain decomposition method is introduced for upscaling the permeability of pore-scale models. The medium is decomposed into hundreds of smaller networks (sub-domains) and then coupled with the surrounding models to determine accurate boundary conditions. Finite element mortars are used as a mathematical tool to ensure interfacial pressures and fluxes are matched at the interfaces of the networks boundaries. The results compare favorably to the more computationally intensive (and impractical) approach of upscaling the media as a single model. Moreover, the results are much more accurate than traditional hierarchal upscaling methods. This upscaling technique has important implications for using pore-scale models directly in reservoir simulators in a multiscale setting. The upscaling techniques introduced here on single phase flow can also be easily extended to other flow phenomena, such as multiphase and non-Newtonian behavior. / text

A numerical reservoir model was created to optimize CO₂ Huff-and-Puff operations in the Bakken Shale. Huff-and-Puff is an enhanced oil recovery treatment in which a well alternates between injection, soaking, and production. Injecting CO₂ into the formation and allowing it to “soak” re-pressurizes the reservoir and improves oil mobility, boosting production from the well. A compositional reservoir simulator was used to study the various design components of the Huff-and-Puff process in order to identify the parameters with the largest impact on recovery and understand the reservoir’s response to cyclical CO₂ injection. It was found that starting Huff-and-Puff too early in the life of the well diminishes its effectiveness, and that shorter soaking periods are preferable over longer waiting times. Huff-and-Puff works best in reservoirs with highly-conductive natural fracture networks, which allow CO₂ to migrate deep into the formation and mix with the reservoir fluids. The discretization of the computational domain has a large impact on the simulation results, with coarser gridding corresponding to larger projected recoveries. Doubling the number of hydraulic fractures per stage results in considerably greater CO₂ injection requirements without proportionally larger incremental recovery factors. Incremental recovery from CO₂ Huff-and-Puff appears to be insufficient to make the process commercially feasible under current economic conditions. However, re-injecting mixtures of CO₂ and produced hydrocarbon gases was proven to be technically and economically viable, which could significantly improve profit margins of Huff-and-Puff operations. A substantial portion of this project involved studying alternative numerical methods for modeling hydraulically-fractured reservoir models. A domain decomposition technique known as mortar coupling was used to model the reservoir system as two individually-solved subdomains: fracture and matrix. A mortar-based numerical reservoir simulator was developed and its results compared to a tradition full-domain finite difference model for the Cinco-Ley et al. (1978) finite-conductivity vertical fracture problem. Despite some numerical issues, mortar coupling closely matched Cinco-Ley et al.'s (1978) solution and has potential applications in complex problems where decoupling the fracture-matrix system might be advantageous. / text

Many engineering and scientific applications of flow in porous media are characterized by transport phenomena at multiple spatial scales, including pollutant transport, groundwater remediation, and acid injection to enhance well production. Carbon sequestration in particular is a multiscale problem, because the trapping and leakage mechanisms of CO2 in the subsurface occur from the sub-pore level to the basin scale. Quantitative and predictive pore-scale modeling has long shown to be a valuable tool for studying fluid-rock interactions in porous media. However, due to the size limitation of the pore-scale models (10-4-10-2m), it is impossible to model an entire reservoir at the pore scale. A straightforward multiscale approach would be to upscale macroscopic parameters (e.g. permeability) directly from pore-scale models and then input them into a continuum-scale simulator. However, it has been found that the large-scale models do not predict in many cases. One possible reason for the inaccuracies is oversimplified boundary conditions used in this direct upscaling approach. The hypothesis of this work is that pore-level flow and upscaled macroscopic parameters depends on surrounding flow behavior manifested in the form of boundary conditions. The detailed heterogeneity captured by the pore-scale models may be partially lost if oversimplified boundary conditions are employed in a direct upscaling approach. As a result, extracted macroscopic properties may be inaccurate. Coupling the model to surrounding media (using finite element mortars to ensure continuity between subdomains) would result in more realistic boundary conditions, and can thus improve the accuracy of the upscaled parameters. To test the hypothesis, mortar coupling is employed to couple pore-scale models and also couple pore-scale models to continuum models. Flow field derived from mortar coupling and direct upscaling are compared, preferably against a true solution if one exists. It is found in this dissertation that pore-scale flow and upscaled parameters can be significantly affected by the surrounding media. Therefore, using arbitrary boundary conditions such as constant pressure and no-flow boundaries may yield misleading results. Mortar coupling captures the detailed variation on the interface and imposes realistic boundary conditions, thus estimating more accurate upscaled values and flow fields. An advanced upscaling tool, a Super Permeability Tensor (SPT) is developed that contains pore-scale heterogeneity in greater detail than a conventional permeability tensor. Furthermore, a multiscale simulator is developed taking advantage of mortar coupling to substitute continuum grids directly with pore-scale models where needed. The findings from this dissertation can significantly benefit the understanding of fluid flow in porous media, and, in particular, CO2 storage in geological formations which requires accurate modeling across multiple scales. The fine-scale models are sensitive to the boundary conditions, and the large scale modeling of CO2 transport is sensitive to the CO2 behavior affected by the pore-scale heterogeneity. Using direct upscaling might cause significant errors in both the fine-scale and the large-scale model. The multiscale simulator developed in this dissertation could integrate modeling of CO2 physics at all relevant scales, which span the sub-pore or pore level to the basin scale, into one single simulator with effective and accurate communication between the scales. The multiscale simulator provides realistic boundary conditions for the fine scales, accurate upscaled information to continuum-scale, and allows for the distribution of computational power where needed, thus maintaining high accuracy with relatively low computational cost. / text

Flow and transport phenomena in the subsurface often span a wide range of length (nanometers to kilometers) and time (nanoseconds to years) scales, and frequently arise in applications of CO₂ sequestration, pollutant transport, and near-well acid stimulation. Reliable field-scale predictions depend on our predictive capacity at each individual scale as well as our ability to accurately propagate information across scales. Pore-scale modeling (coupled with experiments) has assumed an important role in improving our fundamental understanding at the small scale, and is frequently used to inform/guide modeling efforts at larger scales. Among the various methods, there often exists a trade-off between computational efficiency/simplicity and accuracy. While high-resolution methods are very accurate, they are computationally limited to relatively small domains. Since macroscopic properties of a porous medium are statistically representative only when sample sizes are sufficiently large, simple and efficient pore-scale methods are more attractive. In this work, two Eulerian pore-network models for simulating single-phase flow and solute transport are developed. The models focus on capturing two key pore-level mechanisms: a) partial mixing within pores (large void volumes), and b) shear dispersion within throats (narrow constrictions connecting the pores), which are shown to have a substantial impact on transverse and longitudinal dispersion coefficients at the macro scale. The models are verified with high-resolution pore-scale methods and validated against micromodel experiments as well as experimental data from the literature. Studies regarding the significance of different pore-level mixing assumptions (perfect mixing vs. partial mixing) in disordered media, as well as the predictive capacity of network modeling as a whole for ordered media are conducted. A mortar domain decomposition framework is additionally developed, under which efficient and accurate simulations on even larger and highly heterogeneous pore-scale domains are feasible. The mortar methods are verified and parallel scalability is demonstrated. It is shown that they can be used as “hybrid” methods for coupling localized pore-scale inclusions to a surrounding continuum (when insufficient scale separation exists). The framework further permits multi-model simulations within the same computational domain. An application of the methods studying “emergent” behavior during calcite precipitation in the context of geologic CO₂ sequestration is provided. / text