Characterisation of layer mechanism in immiscible three-phase flow of water-wet and mixed-wet rocks

Abstract

The drainage layer is an oil layer that sits between gas and water in three-phase fluid systems. Flow in the layer is an important recovery mechanism responsible for reducing the residual oil saturation during immiscible gas injection in porous media. In sandpack and rock systems, where visual observations were previously impossible, the occurrence of drainage layer has long been disputed. In rocks where the drainage layer has only been indirectly observed, the drainage layer was inferred to have occurred when the oil relative permeability curve has a quadratic form with respect to oil saturation.

This study utilized a high-resolution X-ray microtomography technique to investigate drainage layers in both water-wet and mixed-wet sandstone rocks, arising from the simultaneous flow of three fluid phases: oil, water, and gas. Our experimental methodology requires no contrast agent to be added to the oil phase; hence, any potential issues posed by the addition of contrast agents (i.e., Iododecane) were avoided. We introduce a new image processing recipe whereby small features such as a thin oil layer can be effectively extracted and segmented from 3D microtomography images.

We performed a series of displacement tests on sandstone rock, whereby water first displaced oil, followed by gas, and then chase water injections. We studied the amount of oil and gas remaining as well as their respective morphologies and fluid configurations at the end of each displacement stage. The pore occupancy, residual saturation, fluid–fluid curvature, interfacial area, Euler characteristic, and specific surface area of the remaining fluid phases were investigated to elucidate the displacement mechanisms, such as double-drainage and flow in layers. Simultaneously, we studied the thickness of layers that formed in the system. We identified two methods to examine the contribution of layer flow from thickness information: the first was from the analysis of specific surface area of fluid in clusters, and the second was from the frequency distribution of thickness histogram. The results show that in a water-wet system, the drainage layer has a major role in reducing the residual oil saturation to a level approaching zero. In contrast, in a mixed-wet system, the drainage layer is discontinuous, attributed to the discontinuity of water layer in the oil-wet pores. Since the drainage layer requires a simultaneous presence of continuous gas and water phase to form, the discontinuity in water phase contributes to the discontinuity of drainage layer. Therefore, a continuous drainage layer in mixed-wet rock is only possible in small pores that exhibit water-wetness.

In oil-wet pores of mixed-wet rocks, the piston-like flow of water during waterflooding is akin to the water flow in an oil-wet rock. The piston-like displacement of water initiates two important occurrences; the water trapping and continuous flow of oil layer, called the wettability layer. The water trapping occurs when there is a substantial increase in the thickness of wettability layer which triggers the snap-off at pore throats. Additionally, the wettability order of oil wet pores (of mixed-wet rocks) is oil-gas-water from most to least wetting thus justifying the presence of water trapping in the pore centers. Meanwhile, the wettability layer is thick and continuous and contributes considerably to reducing the residual oil saturation. The injection of gas does not disrupt the flow of oil via wettability layer; however, the flow of wettability layer is highly sensitive to the level of water saturation in the pores.

Finally, in exploratory work, flow in layers was modeled as an incompressible liquid flow through a pipe with four different cross-sectional shapes (sphere, cuboid, cylinder, and prism) whose dimensions were determined by matching the specific surface area of layers. Using computational fluid dynamics (CFD) to solve Navier-Stokes equations, we predicted the permeability of layers, thus avoiding assumptions from the use of a simplified Kozeny-Carman equation. The result matches well to the hydraulic conductance empirical model calculated from Ransohoff and Radke (1988). The relationship between specific surface area and hydraulic conductance was developed to simplify the requirement for solving the flow on a complex geometry of fluid-connected pathways. The comparison with Corey’s correlation shows that the generated relative permeability for layer flow matches at low residual oil saturation with the relative permeability modelled as a power-law with exponent 2.0. Using the workflow presented here, we can rapidly estimate the relative permeability of three-phase flow in rocks without requiring input from a pressure transducer. The model however requires further validation.