PNNL

Abstract

Unstable immiscible fluid displacement in porous media affects geological carbon sequestration, enhanced oil recovery, and groundwater contamination by nonaqueous phase liquids. Characterization of immiscible displacement processes at the pore-scale is important to better understand macroscopic processes at the continuum-scale. A series of displacement experiments was conducted to investigate the impacts of viscous and capillary forces on displacement stability and fluid saturation distributions in a homogeneous water-wet pore network micromodel with precisely-microfabricated pore structures. Displacements were studied using seven wetting-nonwetting fluid pairs with viscosity ratios M (viscosity of the advancing nonwetting fluid divided by the viscosity of the displaced wetting fluid) ranging four orders of magnitude from logM = -1.95 to 1.88. The micromodel was initially saturated with either polyethylene glycol 200 (PEG200) or water as a wetting fluid, which was then displaced by a nonwetting alkane fluid under different flow rates. Capillary numbers (Ca) ranged over four orders of magnitude for the reported experiments, from logCa = -5.88 to -1.02. Fluorescent microscopy was used to visualize displacement and measure nonwetting fluid saturation distributions. These experiments extend the classical work by Lenormand et al. by using water-wet micromodels, high-precision fabrication, and enhanced image analysis of the saturation distributions. In the micromodel experiments initially saturated with PEG200, a viscous wetting fluid, unstable displacement occurred by viscous fingering over the whole range of imposed capillary numbers. For the experiments initially saturated with water, unstable displacement occurred by capillary fingering at low capillary numbers. When the viscous forces were increased by increasing the injection rate, crossover into stable displacement was observed for the fluid pairs with M > 0. For unstable displacement experiments applying the same capillary number for the various fluid pairs, nonwetting fluid saturations were higher when capillary fingering was the dominant fingering process compared to viscous fingering. Our saturation distributions are consistent with other published experimental work and confirm the numerical results obtained by Lenormand et al.