How does three fluid phases move inside a porous rock? What is the fate of oil and gas ganglia in three-phase fluid displacements, in terms of trapping, recovery and topological structures? This project has addressed these research questions by developing, validating, and applying a pore-scale simulator directly on imaged porous rock structures for computing three-phase capillary pressure curves with trapping and hysteresis behavior for relevant wetting states. These properties are required to assess three-phase methods like water-alternate-gas (WAG) injection for improved oil recovery in the field, as they will affect the ability of the fluids to flow through the reservoir rock (i.e., relative permeability) at large scale.
We have developed a new method for describing capillary-controlled motion of three fluid phases in porous rock with uniform or mixed wettability. The method preserves volumes of isolated fluid ganglia while computing their energy and pressure states during displacement. We have combined the volume preservation approach with equations of state to account for compressible gas. We include adaptive mesh refinement and parallelism by coupling our codes to an existing software framework (SAMRAI) that includes these functionalities. This permits accurate three-phase simulations on large rock samples, using the national high-performance computing e-infrastructure. Our methods can simulate fluid invasion controlled by either pressure, or saturation (i.e., low flow rate), and it can also track exactly three-phase saturation paths.
Oregon State University has performed successful three-phase flow experiments of WAG cycles with oil present on both water-wet and mixed-wet bead packs at Advanced Photon Source (Argonne IL) for experimental validation of our model. The fluid distributions are imaged during the flow experiments with synchrotron-based micro-CT scanning. Simulations that track the experimental saturation path exactly, and that include a spatially and temporarily varying contact angle to describe interface relaxation toward equilibrium, show excellent accordance with the experimental results. This is the first study ever that has performed a one-to-one comparison of high-resolution 3-D pore-scale fluid distributions from experiment and simulation of three-phase flow in porous media.
Simulations on large rock samples show that saturation-controlled drainage gives rise to cooperative behavior (spontaneous fluid redistribution) and abrupt pressure drops during pore filling. Drainage capillary pressure curves are lower when saturation, rather than pressure, controls displacement. These effects are less significant in imbibition when uniform swelling of the wetting phase occurs ahead of the imbibing front, consistent with experimental observations. These results suggest that the choice of displacement protocol is important when measuring macroscopic flow parameters in the lab. In three-phase displacements, fluid ganglia give rise to dynamic behavior due to fluctuations of the disconnected phase pressure. When gas displaces oil through wide and narrow pore channels, oil displaces water from narrower pore regions. This leads to fluctuating gas/oil capillary pressure and increasing oil/water capillary pressure. Gas/water capillary pressure is lower in the presence of mobile oil.
Simulations of WAG cycles after imbibition show that double displacements (gas displaces oil that displaces water) can recover oil during gas invasion, but it can also push oil into narrower pore spaces accompanied by oil ganglia fragmentation. A large fraction of these oil ganglia contacts both gas and water with lower gas/oil capillary pressure and higher oil/water capillary pressure than the continuous phases. In the following water invasion, water displaces oil ganglia that displaces gas (double displacement), while water snaps off gas/oil interfaces and traps some oil ganglia in water, in similar configurations as before gas invasion. After the WAG cycle, most ganglia trap in contact with the other two fluids, often in contiguous configurations formed by several gas and oil ganglia, consistent with pore-scale experiments.
With respect to relationships between initial and residual saturations, we obtain the following conclusions, all of which are consistent with core-scale measurements in literature: Residual oil saturation after three-phase displacement is smaller than after a two-phase displacement; residual oil saturation decreases and residual gas saturation increases with increased initial gas saturation; the total trapped saturation from three-phase displacement is higher than from two-phase displacements; and residual oil saturation decreases with increasing residual gas saturation. Thus, we conclude that our model can be used to develop reliable trapping relationships for hysteresis models in reservoir simulation tools.
Core-scale laboratory measurements of three-phase capillary pressure curves with trapping and hysteresis behaviour are challenging and extremely time-consuming, with typical time frames of a year. This project has developed and validated a pore-scale model that can generate such data within a few days on sufficiently large rock samples when it is used in a high-performance compute environment. Thus, the model represents a cost-cutting and extremely time-saving technology for the industry. For example, the use of the developed simulator as part of digital rock physics analyses in an initial planning of a WAG project will reduce the number of required three-phase measurements and speed up the work flow leading towards field implementation. We anticipate that pore-scale simulations with our model can also be used to develop more practical flow parameter correlations that can be implemented in reservoir simulators to improve EOR predictions in the field.
Three-phase capillary pressure and relative permeability curves that account for hysteresis and phase trapping are required to assess three-phase enhanced oil recovery (EOR) methods, such as depressurization and water-alternate-gas (WAG) injections, in mi xed-wet reservoirs. These properties are usually included in reservoir simulators in terms of correlations that are based on two-phase data. Measuring three-phase capillary pressure is time-consuming and technically challenging, few three-phase hysteresis data exist, and core-scale water floods show large variation of trapped oil and gas. In this project we will develop improved techniques for three-phase pore-scale modelling based on the variational level set method, and simulate three-phase capillary-co ntrolled displacement at mixed-wet conditions directly on segmented 3D rock images. Experimental three-phase distributions in 3D porous media will be generated via synchrotron-based X-ray microtomography and compared quantitatively against pore-scale simu lations. The validated model will then be utilised directly on sandstone and carbonate rocks to gain increased insight into capillary pressure and relative permeability curves, hysteresis and trapping behaviour (including the structure and amount of resid ual oil) in three-phase EOR processes. In particular, we will investigate how three-phase pore-scale mechanisms, such as oil layer existence and multiple displacement events, affect these properties in complex 3D pore geometries. Water chemistry effects w ill be included to investigate the impact of interfacial tension and wettability change during WAG cycles. Simulation results obtained with the novel pore-scale model can be used subsequently to develop practical and reliable three-phase capillary pressur e and relative permeability correlations, including hysteresis loop logic and trapping models, that can be implemented in reservoir simulators to improve the predictions of three-phase EOR processes in the field.