The onset of oil mobilization during imbibition has been imaged with pore-scale resolution under dynamic flow conditions viscocapillary balance, by using fast synchrotron-based X-ray computed microtomography. Oil mobilization under unsteady-state displacement has been studied for sintered glass, sandstone and carbonate rock, which demonstrate distinctly different behavior with respect to the cluster-size distributions and respective time evolution.
For the sandstone sample, which showed the largest saturation change between the three samples, over 50% of the oil was mobilized during imbibition. The cluster-size distribution computed from the tomography images, but also detailed visualization at the pore scale revealed that during imbibition the largest (initially connected) cluster breaks off into smaller segments. Over time, successively larger segments break off, which increases the frequency of intermediate-size clusters. In most cases, each segment breaks off into even smaller segments, but also in fewer cases clusters merge and increase in size.
These findings support the view that at the onset of oil-mobilization clusters prefer to break off instead of moving as a large cluster, which gives further insight into the ganglion dynamics flow regime. It is also shown that imbibition dynamics is more complex than assumed under common percolation models, where, for instance, disconnected nonwetting-phase clusters become immobile and remain static. The experimental data instead clearly show time evolution for disconnected clusters that can lead to reconnection or further break off.
During the production of an oil reservoir by waterflooding (in a water-wet rock, e.g., sandstone) the nonwetting phase (oil) is displaced by the wetting phase (water). During this imbibition process, more than 50% of the initial oil can be trapped in the pore space, which is referred to as remaining or residual oil. This trapping is mainly attributed to capillary-dominated processes, which until very recently (Georgiadis et al., 2013), have been directly observed only in model systems like 2D micromodels (Lenormand et al., 1983). As a consequence, very little is known about how trapping actually occurs in 3D real rock and what the transient process is from an overall connected oil phase to disconnected and trapped oil clusters. Most of the current macroscopic descriptions are either phenomenological, based on invasion percolation models (Dias and Wilkinson, 1986; Wilkinson and Willemsen, 1983; Wilkinson, 1986) or pore network modeling (Patzek, 2001), which with few exceptions (e.g., Hammond and Unsal, 2000; Niasar et al., 2010), consider only local elementary pore-scale processes (Lenormand et al., 1983) like snap-off (Roof, 1970) but do not honor the complexity of pore-scale displacement physics like ganglion dynamics (Avraam and Payatakes, 1995) and cooperative processes (Armstrong and Berg, 2013) that involve both capillary and viscous forces at the same time (Mohanty et al., 1987).