An investigation of trapping of non-wetting phase during imbibition using pore-network modeling

Abstract

Flow of two phases in porous media plays an important role in for example carbon storage, oil recovery, industrial processes and groundwater contamination problems. The creation of an immobile region of one phase which is completely surrounded by the other phase is referred to as trapping. Pore-network modeling is a useful tool for analyzing two-phase flow and trapping because it is cheap compared to experiments and solving the Navier-Stokes equation on a pore scale. A post-processor which analyzes the results from an existing pore-network model was developed. The pore-network model uses a dynamic two-pressure approach to simulate main imbibition in a regular lattice network of dimensions 10x10x30. The post-processor CSDA (Connectivity and Spatial Distributions Algorithm) locates three phases. These are the wetting phase, connected non-wetting phase and disconnected regions of non-wetting phase within the wetting phase. After that, CSDA calculates properties such as volumes, pressure and shape of disconnected regions and the (spatial distribution of) saturation and pressure of the three phases. These statistics are used to assess the influence of the flow rate, the non-wetting phase viscosity, the variance in pore size and the size of pore throat radii on the dynamics of the trapping of non-wetting phase during imbibition. Trapping can occur due to the snap-off of a single pore, or due to bypassing of the non-wetting fluid by the wetting fluid. If the second mechanism is important, the disconnected regions tend to extend over more pore bodies than if the first mechanism is important. In contrast to what was expected from literature, lower flow rates led to lower trapping. This is attributed to the wetting front becoming sharper, and the wetting films becoming smaller. Increased non-wetting phase viscosity led to lower flow rates. At a similar flow rate, increased non-wetting phase viscosity led to a higher number of disconnected regions and a higher residual saturation. Also, trapping tended to occur in pore bodies with bigger radii. Decreasing the variance in pore body radius led to a lower pore volume. This led to a higher wetting front velocity. The variance in volume of disconnected regions also decreased. This means that the volume of disconnected regions depends on the pore body size. Increasing the pore throat radius led to a higher wetting front velocity since the conductivity of the pore throats went up. This led to a more fingered front. Consequently, trapping due to bypassing was important and the disconnected regions extended over more pore bodies. On the other hand, the increased pore throat radius led to a lower probability of snap-off. This decreased the total number of disconnected regions. So trapping due to snap-off of a single pore was less important and trapping due to bypassing was more important for this scenario.