Mass transfer during enhanced hydrocarbon recovery by gas injection processes

In order to estimate the potential incremental hydrocarbon recovery by gas injection, compositional reservoir simulators are commonly used in the industry. Successful design and implementation of gas injection processes (mainly CO2) rely in part on the accuracy by which the available simulation tools can represent the physics that govern the displacement behavior in a reservoir.;In the first part of this thesis, we investigate the accuracy of some physical models that are frequently used to describe and interpret dispersive mixing and mass transfer in compositional reservoir simulation. The calculations from compositional simulation are compared with the experimental observations.;A quaternary analog fluid system (alcohol-water-hydrocarbon) that mimics the phase behavior of CO2-hydrocarbon mixtures at high pressure and temperature has been designed in our research group. A porous medium was designed using Teflon materials to ensure that the analog oil acts as the wetting phase, and the properties of the porous medium were characterized in terms of porosity and permeability. Relative permeability and interfacial tension measurements were also performed to delineate interactions between the fluid system and the porous medium. Displacement experiments at First-contact miscible (FCM), near-miscible and multicontact miscible (MCM) conditions were consequently performed (Rastegar, 2010).;The effluent concentrations from two-component FCM displacement experiments exhibit a tailing behavior that is attributed to imperfect sweep of the porous medium: A feature that is not captured by normal dispersion models. To represent this behavior in displacement calculations, we use dual-porosity (DP) models including mass transfer between flowing and stagnant porosities. The 4-component two-phase displacement experiments (near-miscible and MCM) are consequently simulated using the DP models constructed based on observations from FCM displacements.;We demonstrate that the accuracy of our displacement calculations relative to the experimental observations is sensitive to the selected models for dispersive mixing, mass transfer between flowing and stagnant porosities, and IFT scaling of relative permeability functions. We also demonstrate that numerical calculations substantially agree with the experimental observations for some physical models with limited need for model parameter adjustment.;The second part of this thesis is devoted to diffusion and matrix-fracture interactions during gas injection in fractured reservoirs. Molecular diffusion can play a significant role in oil recovery during gas injection in fractured reservoirs, especially when matrix permeability is low and fracture intensity is high. Diffusion of gas components from a fracture into the matrix extracts oil components from matrix and delays, to some extent, the gas breakthrough. This in turn increases both sweep and displacement efficiencies.;In current simulation models, molecular diffusion is commonly modeled using a classical Fick's law approach with constant diffusion coefficients. In the classical Fick's law approach, the dragging effects (off-diagonal diffusion coefficients) are neglected. In addition, the gas-oil diffusion at the fracture-matrix interface is normally modeled by assuming an average composition at the interface which does not have a sound physical basis.;In this work, we present a dual-porosity model in which the generalized Fick's law is used for molecular diffusion to account for the dragging effects; and gas-oil diffusion at the fracture-matrix interface is modeled based on film theory in which the gas in fracture and oil in the matrix are assumed to be at equilibrium. A novel matrix-fracture transfer function is introduced for gas-oil diffusion based on film theory. Diffusion coefficients are calculated using the Maxwell-Stefan model and are pressure, temperature and composition dependent. A time-dependent transfer function is used for matrix-fracture exchange in which the shape factor is adjusted using a boost factor to differentiate between the transfer rate at early and late times.;Field-scale examples are used to show that our approach, which is based on a sophisticated physical model for gas-oil diffusion (film theory), gives significantly different results from the conventional approach. It is also demonstrated that the dragging effects (off-diagonal diffusion coefficients) and time-dependency of matrix-fracture transfer function can moderately impact the oil recovery during gas injection in fractured reservoirs. We also show that miscibility is not developed in the matrix blocks even at pressures above minimum miscibility pressure (MMP) when molecular diffusion is the main recovery mechanism during gas injection in fractured reservoirs.;In recent years, coalbed methane has become an important source of energy in the United States. Since primary production techniques typically recover less than half of the methane in a coalbed, enhanced coalbed methane (ECBM) recovery processes are needed in which CO2 and/or N2 are injected into the coalbed to recover more CH4.;One of the main mechanisms that govern the dynamics of ECBM recovery processes is the sorption of gases onto the coal surfaces. Despite the well-documented complexity of multicomponent sorption phenomena, adsorption and desorption of CH4/CO2/N2 mixtures in the porous coal is commonly modeled with the extended Langmuir model. The extended Langmuir model has been proven unable to accurately describe the multicomponent sorption that is central to ECBM recovery processes and, therefore, more sophisticated sorption models are needed.;In the third part of this thesis we apply potential theory to describe the multicomponent sorption of relevance to ECBM processes. In this approach for modeling multicomponent sorption, each component is assumed to be affected by a characteristic potential field emitted by the coal surface. We discuss the implementation of potential theory with emphasis on the simulation of ECBM processes where computational efficiency and accuracy must be balanced. The model must be solved by an iterative scheme, and is hence more computationally expensive than the extended Langmuir approach.;The results and analysis presented in this work demonstrate that the application of potential theory of sorption to modeling of ECBM recovery processes can improve the accuracy of calculations. However, the additional complexity of the model and the associated increase in the computational efforts may not balance the gain in accuracy sufficiently to warrant application in general purpose reservoir simulation.