No abstract

No abstract

Summary In this work, the Das and Butler flow model was extended to account for the effect of differential pressure across a laterally and vertically spaced horizontal injection- and production-well pair for solvent vapour extraction (SVX) processes. The extended model provides a more-rigorous dispersion-coefficient correlation using interstitial pore velocity and mean particle size. This was used to history match oil-production data from 3D-scaled physical model experiments and to determine the effective dispersion coefficient of the solvent in heavy oil. The SVX experiments were performed with two model sizes and three different permeabilities spanning than two orders of magnitude. Laterally spaced horizontal wells were used to inject an 86:14 mol% mixture of butane and methane, respectively, as a dense vapour. The average dispersion coefficient of solvent (butane) was found to decrease with injected cumulative solvent and increase with reduction in permeability, and both of these relationships could be approximated with an inverse power function.

In miscible displacements encountered in enhanced oil recovery processes, the unfavorable viscosity contrast between injected solvent and oil usually leads to viscous fingering (VF), a hydrodynamic instability which may result in a lower sweep efficiency and oil recovery. This phenomenon can be observed in a wide range of flows in subsurface porous media. This study examined a simple cyclic time‐dependent displacement rate and its effects on the onset and longer development of VF. It is found that such varying displacement rate can either stabilize or destabilize VF, depending on the cycle period, amplitude, and displacement scenarios. The most important mechanism is that such time‐dependent rate can effectively change the competition between convection (destabilizing effect) and dispersion (stabilizing effect). This is different from the widely used constant injection rate where the flow instability is actually determined by the Peclet number and mobility contrast for a given scenario. This study therefore provided a new aspect to control VF, either enhance or reduce, with low additional costs. It is therefore both scientifically and practically important for a wide range of flows in subsurface porous media. © 2017 American Institute of Chemical Engineers AIChE J, 65: 360–371, 2019

Mass transfer plays an important role in influencing the efficiency of miscible displacements in solvent-based processes in enhanced oil recovery. The mass transfer rate because of the pure molecular diffusion is very slow. However, this process can be greatly enhanced by the appearance of frontal instabilities called viscous fingering mechanisms, which are beneficial for improving the mixing and mass transfer between the injected solvent and oil. Instead of a piston-like displacement, the interface between solvent and oil is very convoluted with intricate finger-like patterns of the less viscous solvent intruding into the highly viscous oil. This intrusion significantly increases the surface area of contact of the two fluids and leads to more efficient mass transfer and mixing. Experimental measurements on the diffusion coefficients of two miscible fluids indicate that, instead of a constant diffusion coefficient (CDC), a concentration-dependent diffusion coefficient (CDDC) is more realistic. In the present study, a CDDC relation in which the diffusion coefficient is exponentially proportional to concentration is adopted. Its effect on the development of frontal instabilities is examined through highly accurate nonlinear numerical simulations. The differences between the CDDC case and the widely assumed CDC case are discussed. Furthermore, the enhancement of frontal instabilities on mass transfer when the CDDC is considered is investigated at various mobility ratios and Peclet numbers. The special characteristics for the CDDC case indicate its important role in miscible displacements. Eventually, the relation of breakthrough time to parameters is correlated to accurately predict the breakthrough time in any CDDC scenario.