Thermodynamically Driven Fluid Mixing Across Phases Induced by Viscous Flow Instabilities in Porous Media

Abstract

Fluid mixing and its interplay with flow instabilities and/or channeling through heterogeneous media have been studied in some detail for mostly fully miscible conditions in which a single phase is generally represented by two components, e.g. a solvent and a solute. In such studies, fluid properties are assumed to be either constant or to follow simple mixing rules depending only on concentration. In particular, viscous fingering, a hydrodynamic instability due to the displacement of a more viscous fluid with a less viscous one, has been studied predominantly for immiscible or fully miscible fluids.

However, many problems of interest, such as gas injection in hydrocarbon reservoirs, involve multiple species and fluid properties, even in single-phase, and depend non-linearly on temperature, pressure, and composition through an equation of state (EOS). Moreover, depending on the minimum miscibility pressure, solubility factor, and in-situ conditions, a two-phase region may develop, e.g. in a partially (multi-contact) miscible system. Fingering in this regime and its interplay with mixing have gained less attention in previous studies, and never been compared to those in single-phase.

This work aims at studying mixing of a finite volume of CO2 with multi-component oil in various regimes of miscibility, with fluid properties determined by rigorous EOS-based (Peng-Robinson) phase-stability and phase-split computations. Fickian diffusion, driven by chemical potential gradients, is modeled to capture the diffusive fluxes across sharp phase boundaries e.g. across a finger perimeter in absence of perfect miscibility. Flow and transport are modeled on very fine grids with our in-house, higher-order, finite element reservoir simulator.

In addition to compositional effects across different degrees of miscibility, we investigate the impact of correlated heterogeneities for a wide range of (geostatistically generated) correlation lengths and lognormal permeability distributions.

Our numerical framework is capable of resolving small-scale fingering patterns on one hand, and the potentially dispersed flow caused by heterogeneity. The results provide a broad perspective into mixing mechanisms coupled with complex fingering patterns in porous media while demonstrating critical differences in dynamics of fluid mixing in single- and two-phase flow through both homogenous and heterogeneous media.

AAPG Datapages/Search and Discovery Article #90258 © 2016 AAPG Eastern Section Meeting, Lexington, Kentucky, September 25-27, 2016