Three-Phase Flow in Fractured Porous Media: Investigation of Matrix/Fracture Interactions

You have access to this full article to experience the outstanding content available to SPE members and JPT subscribers.

To ensure continued access to JPT's content, please Sign In, JOIN SPE, or Subscribe to JPT

The authors present the results of a detailed experimental study in which underlying pore-level-displacement physics of two- and three-phase flow in a fractured rock sample is investigated with high-resolution X-ray microtomography techniques. A unique, three-phase coreflooding setup integrated with a microcomputed-tomography (micro-CT) scanner is used to conduct flow experiments on a miniature, partially fractured sandstone sample to shed light on subtle displacement mechanisms governing matrix/fracture interactions in the presence or absence of spreading oil layers.

Experimental Methodology

Rock and Fluid Properties. A water-wet Berea sandstone core plug, 38 mm in diameter and 100 mm in length, was subjected to continuous, nonuniform stress to create a fracture parallel to the stress axis. The artificial fracture was induced such that it ran through only half of the core plug. This configuration was used to study pore-fluid occupancies and displacement physics at different locations of the medium including matrix-only sites, the fracture, and the matrix adjacent to the fracture. After inducing the fracture, a miniature sample, 10 mm in diameter and 40 mm in length, was cut from the core plug.

The coreflooding tests were carried out with two different fluid systems representing the spreading and nonspreading conditions. X-ray dopants were added to liquid phases to establish sufficient contrasts among all phases (i.e., brine, oil, and gas) during the image-processing steps. All fluids were recirculated through the coreflooding setup before starting the experiment, bypassing the core sample at the experimental conditions for a period of time (i.e., 48 hours) sufficient to equilibrate all the phases and minimize mass transfer between the fluids in the core sample during the experiments.

Experimental Procedure. The experimental apparatus included three major modules: (1) a three-phase closed-loop miniature coreflooding setup, (2) a high-resolution micro-CT scanner, and (3) a data-acquisition system. The flow tests were carried out at pore pressure of 3.45 MPa and ambient temperature (20°C). Each fluid was retracted from a specific section of the separator on the basis of the density of the fluid and then injected into the core sample. The authors used the unsteady-state approach, in which only one fluid is injected at a time.

After mounting the core holder inside the scanner chamber, the core was flushed with carbon dioxide gas and vacuumed for 1 day. Subsequently, a dry scan image of the entire length of the core sample was obtained. Then, the core sample was vacuum saturated with brine to establish 100% water saturation. Doped brine was replaced by low-salinity brine to generate an accurate pore-space map. Then, a wet reference scan was obtained at a resolution of 2.5 μm while the flow of brine was kept at a low flow rate. Afterward, the low-salinity brine was replaced again by the doped brine. At this point, the core sample was subjected to an oilflood with an initial flow rate of 0.01 mL/min to establish a water saturation of approximately 31%. During drainage, the pressure drop across the core sample was recorded and water and oil saturations were monitored. This process was continued until the system reached steady state. At this point, the authors had established the desired two-phase condition. Subsequently, gas was injected into the core at various flow rates to reach different oil saturations.

Data Analysis. Before starting the experiments, the dimensions of the fracture and its distance from the inlet of the core were measured to prepare the scan recipes and determine the locations of interest. It was intended to study the pore-scale displacement physics in three different locations of the sample. The authors selected the sites to allow the study of pore-fluid occupancies at topologically different areas of the sample while using different fluid systems.

An image-processing software visualized the reconstructed 3D images obtained from the micro-CT scanner.  The tomograms had to be processed before final visualization. This data-processing procedure produced sharp boundaries of the pore structure and the phase interfaces in each stage of the flow experiments.

Results and Discussion

After establishing 100% water saturation, oil was introduced to the core at an initial flow rate of 0.01 mL/min followed by incremental increases up to 0.45 mL/min. Oil, as the nonwetting phase, first invaded the fracture because it was the path of least resistance. Once the capillary pressure increased to that required to invade matrix pores, oil displaced water from the larger water-filled pore elements in the surrounding matrix. However, water maintained its connectivity by forming thin wetting layers on the roughness of the fracture walls and the crevices of interconnected pore elements of the matrix. In a pore with an angular cross section, water resides in the corners while oil occupies the center. During secondary gas injection, gas rapidly fills the fracture and displaces the oil phase through the center of the fracture. Subsequently, it starts to extract more oil from the pores of the neighboring matrix with a higher capillary pressure. However, displacing water required a higher threshold pressure of the gas phase. The spreading coefficient of the fluid systems used plays an important role in the displacement mechanisms at different oil saturations of the medium.

Spreading Fluid System. During the gas invasion into pores filled with spreading oil, oil left a thin layer sandwiched between brine in the corners and gas in the center of the pore element. These spreading layers were present not only in the pores of the matrix but also in the fracture, between the wetting layers of brine on the walls and the gas in the center of the fracture. The spreading layers have nonnegligible hydraulic conductivity. This means that the oil phase in the matrix had established a hydraulic connectivity through these connected layers with the oil in the fracture. It was observed that these layers were stable during the gas injection and allowed the oil saturation in the matrix to reduce.

Nonspreading Fluid System. During the invasion of gas into pores filled with nonspreading oil, the authors observed thinner oil layers at lower gas-flow rates in some of the pore elements invaded by gas and even in the fracture roughness. However, as the capillary pressure was increased by increasing the gas-flow rate, these thin oil layers collapsed. This implies that the hydraulic conductivity of the oil was not well-maintained, which hinders the oil movement from the neighboring matrix to the fracture, leading to lower oil recovery compared with that of the spreading fluid system. This means that those layers contributed to the flow of oil for only a limited span of capillary pressure. It is worth mentioning that any layer that was smaller than the spatial resolution used could not be observed. However, by comparing the oil-saturation profiles of the spreading and nonspreading systems at different flow rates, it was evident that the thinner layers in the nonspreading system had little or no effect on the medium’s saturation.

The authors compared the pore fluid occupancies and the displacement mechanisms in two randomly selected pore elements during gas injection at different oil saturations: (1) a pore element that is directly connected to the fracture and (2) a pore that is far away from the fracture section.

Higher Oil Saturations. The authors investigated three-phase pore fluid occupancy generated at lower gas-flow rates with the spreading fluid system in a pore element that is directly connected to the fracture. The oil phase is sandwiched between the brine in the corner and the gas in the middle. The oil maintained its connectivity to the fracture through stable and thick spreading oil layers. However, with the nonspreading system, gas invaded the same pore element by pushing the oil to the crevices without forming any oil layers. It was observed that some of the pores had very thin oil layers that became thinner or collapsed as the capillary pressure was increased. Figs. 1a and 1b show a pore that is far away from the fracture and not invaded by gas at lower flow rates under both the spreading and the nonspreading conditions. The only difference observed is a thicker brine layer under spreading conditions in that particular pore.

Fig. 1—2D cross-sectional views of three-phase segmented images showing the fluid-occupancy map obtained during gas injection under different conditions with a resolution of 2.5 μm. Blue, red, yellow, and gray colors represent brine, oil, gas, and solid, respectively. (a) Low-flow-rate gas injection under spreading condition, (b) low-flow-rate gas injection under nonspreading condition, (c) high-flow-rate gas injection under spreading condition, and (d) high-flow-rate gas injection under nonspreading condition.


The main mechanism of fluid displacement at higher oil saturations is the movement of the connected bulk oil in the form of a piston-like displacement. The oil saturations under both spreading and nonspreading conditions are almost the same at low gas-flow rates, but there was a relatively significant change in the oil saturations at higher gas-flow rates. In addition, it was observed that the average gas-saturation profile in the direction perpendicular to the gas-flow direction was not uniform in both spreading and nonspreading cases. The oil in the larger pores surrounding the fracture was extracted earlier than that located far away from the fracture.

Lower Oil Saturations. At higher flow rates of gas, the spreading layers played a substantial role in displacement. The drainage of oil through the spreading layers is the main governing mechanism of flow at lower oil saturations. Fig. 1c and 1d show the gas invasion into the previously oil-filled pores under the spreading and nonspreading conditions. Fig. 1c, which was generated under the spreading condition, shows that the oil spreads and forms a layer around the gas in the center, while Fig. 1d, which was obtained under the nonspreading condition, shows two oil lenses in two different corners of the same pore element. The oil saturation was measured at different gas-flow rates. The measurements indicate a relatively higher oil recovery for the spreading system when compared with that of the nonspreading system.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 181891, “Three-Phase Flow in Fractured Porous Media: Experimental Investigation of Matrix/Fracture Interactions,” by M. Sabti, A.H. Alizadeh, and M. Piri, University of Wyoming, prepared for the 2016 SPE Annual Technical Conference and Exhibition, Dubai, 26–28 September. The paper has not been peer reviewed.

Three-Phase Flow in Fractured Porous Media: Investigation of Matrix/Fracture Interactions

01 September 2017

Volume: 69 | Issue: 9


Don't miss the latest content delivered to your email box weekly. Sign up for the JPT newsletter.