Paul R. Tortora1, Shelley L. Anna1, Martin Ferer2, Grant Bromhal3, and Duane H. Smith3. (1) Mechanical Engineering, Carnegie Mellon University, 5000 Forbes Ave., Pittsburgh, PA 15213-3890, (2) Physics, West Virginia University, P.O. Box 6315, Morgantown, WV 26506, (3) National Energy Technology Laboratory, 3510 Collins Ferry Rd., Morgantown, WV 26597
Fluid flows in porous media are of long-standing interest in carbon sequestration, oil recovery, and aquifer remediation. When a fluid flows into a porous medium containing a second immiscible fluid, the first (invading) fluid can form finger-like patterns, displacing only a fraction of the second (defending) fluid, thus limiting the efficiency of these processes. In this talk we present observations of fluid displacements in idealized two-dimensional microchannel networks designed to investigate fundamental aspects of these flows. We compare our observations with fluid displacements in real sandstone cores that have been imaged using x-ray computed tomography. Results from a pore-level computational model and analytical models are compared to the experimental results.
Several types of flow patterns can be observed depending on the capillary number, the ratio of fluid viscosities, and the size of the porous network. Common patterns include capillary fingering and viscous fingering (both of which follow fractal models), stable advance of a fluid front, and intermediate behaviors. The ability to identify and characterize each of these flows, including determining the fractal dimension, is important in optimizing flows for specific applications. We demonstrate the ability to characterize flows in each of these regimes, considering the configuration of greatest interest in applications – that in which a non-wetting fluid displaces a wetting fluid. However, to further elucidate the fundamental flow behavior, the opposite case is also investigated, along with other variations in network configuration and properties, including hydrophobic vs. hydrophilic networks and pressure vs. displacement driven flows.