The overall objective of this project is to gain fundamental understanding of the hydrocarbon fluid phase behavior and flow properties in nanoporous shale formations. Hydrocarbon fluids (as well as water) exhibit significantly different behavior in unconventional reservoirs due to nanopore confinement effects and complex, heterogeneous fluid-rock interactions. The traditional, continuum models used for describing fluid behavior and for developing production strategies for conventional reservoirs, are inadequate for shale and other tight formations. Thus, acquiring new knowledge of fluid flow properties and phase behavior in nanoporous media is essential for maximizing the ultimate oil/gas recovery from unconventional reservoirs.
The researchers will employ an integrated experimental and numerical approach to investigate fluid behavior and fluid-solid interactions in nanoporous media. Experimentally, they will characterize the compositions and microstructures of shale core samples (obtained from Chevron — an industrial partner) using a suite of analytical techniques. Small-angle neutron scattering (SANS), combined with custom-designed high-pressure cells, is used to examine in situ the fluid behavior and related changes in nanopores at reservoir pressure (P) – temperature (T) conditions. The Lattice Boltzmann method (LBM) is utilized to develop physics-based relationships among permeability, nanopore structure, and flow conditions informed by experiments.
Los Alamos National Laboratory (LANL), Los Alamos, NM 87545
Although shale oil/gas production in the U.S. has increased exponentially, its low energy recovery (less than 10 percent for oil and 30 percent for gas) is a daunting problem that needs to be solved for continued growth. The fundamental mechanisms underlying the low oil/gas recovery rates lie in gaining a better understanding of the hydrocarbon fluid behavior in shale nanopores. This information is critical for estimating effective permeability of tight rocks, assessing potential for multi-phase flow, and optimizing operational parameters such as well spacing, bottom hole pressures, and pumping rate, to maximize hydrocarbon recovery. Predictive models based on the new knowledge of nanopore hydrocarbon behavior are required to optimize production from unconventional reservoirs.
LANL will integrate experimental characterization and observation with pore-scale modeling in this research. Integration between laboratory measurements and numerical calculations will facilitate the understanding of nanoscale fluid behavior and development of effective predictive relationships that can be ultimately used to design better production approaches. Previous integrated experimental/modeling studies have been scarce and have focused on description of nanopore topologies and connectivities at ambient conditions. This projects integration of SANS measurements with LBM modeling will be first-of-its-kind in examining nanoscale hydrocarbon fluid behavior at relevant reservoir P-T conditions.
This project consists of two phases, each lasting nine months. In Phase 1, LANL characterized the compositions and microstructures of field shale core samples using various analytical techniques, measured shale nanopores and their interactions with fluids (water and hydrocarbon) as a function of P-T using SANS, and simulated single-phase hydrocarbon transport behavior with LBM. In Phase 2, LANL will complete the core sample characterization and hydrostatic pressure SANS data analyses, and characterize the fluid flow behavior in nanoporous shales under in-situ loading conditions with SANS coupled with an oedometer cell designed in Phase 1. LBM modeling will be extended to simulate hydrocarbon phase change and multiphase flow, and integrated with the experiments.
This project will lead to a fundamental understanding of the hydrocarbon fluid phase behavior and flow properties in nanoporous shale formations. In particular, successful completion of both the experimental and modeling tasks on fluid flow studies will provide important insights into the fundamental mechanisms underlying shale matrix diffusion and will ultimately help develop effective strategies for long-term unconventional oil/gas production.
LANL acquired a Wolfcamp core sample (from a depth of 10,500 feet, corresponding to a pressure of 5,200 psi and a temperature of 70 °C) from Chevron and characterized the sample using a variety of analytical methods including quantitative X-ray diffraction, X-ray fluorescence, differential scanning calorimetry/thermogravimetry, and scanning electron microscopy coupled with the focused ion beam technique. The results reveal that the core is made of organic-matter (OM)-rich and OM-lean layers that exhibit different chemical and mineral compositions, and microstructural characteristics.
Using the hydrostatic pressure system and gas-mixing setup, researchers conducted several sets of in-situ high-pressure SANS experiments at pressures up to 20 kpsi using water and methane as the pressure media. Initial data processing has been completed, and further analyses are ongoing.
Researchers performed the first numerical study to calculate the correction factor (ratio of apparent permeability to intrinsic permeability) for complex kerogen nanoporous structures using the lattice Boltzman method (LBM). The results show that the correction factor is always greater than one, indicating that the non-Darcy effects play an important role in the gas flow in kerogen nanopores. In addition, the correction factor increases with decreasing pore size, intrinsic permeability, and pressure, which is in good agreement with the nanopore Knudsen correction.
The researchers conducted SANS and ultrasmall-angle neutron scattering (USANS) measurements of a Wolfcamp shale sample (provided by Chevron) and developed/tested an oedometer system for later high-pressure SANS/USANS experiments.
The researchers completed characterization of a Marcellus shale sample using focused ion beam scanning electron microscopy and conducted high-pressure small-angle neutron scattering measurements using our custom-made oedometer system.
The researchers also completed LBM simulations on the effects of fracture density on effective permeability in a shale matrix-fracture system.
The researchers have measured the nanopore size distributions of two samples of Marcellus shale provided by Noble Energy; one has a high total organic carbon (TOC) and the other has a low TOC. The goal of the characterization is to provide a 3D description of the pore structure of these materials that can then be used in a simulation of gas adsorption. The researchers characterized the microstructures and mineralogy of two Marcellus shale samples with different amounts of TOC using focused ion-beam scanning electron microscopy, with an emphasis on characterizing kerogen and inorganic materials, including both pyrite and clays (illite).
Experimentally, the researchers conducted high-pressure SANS experiments on Marcellus shale samples using water as a pressure medium at the National Institute of Standards and Technology Center for Neutron Research (NCNR). These experiments were to address the question of where the water goes in the shale matrix during hydraulic fracturing.
LANL researchers have found that the relative permeabilities of both water and oil phases in fractionally wet porous media (FWPM) exhibit very different characteristics from those in the purely wet porous media. Particularly, the simulations indicate additional flow resistance in FWPM at an intermediate water (oil) saturation. Through detailed analysis, LANL has concluded that this additional flow resistance is mainly caused by the extremely tortuous flow paths.
The researchers have applied a regularized multiple-relaxation-time LBM model to analyze gas flow in a 2-dimensional reconstructed micro-porous medium at the pore scale. The velocity distribution inside the porous structure was analyzed. The effects of the porosity and specific surface area on the rarefied gas flow and apparent permeability were investigated. The simulation results indicate that the gas exhibits different flow behaviors at various pressure conditions and the gas permeability is strongly related to the pressure.
LANL researchers previously (as part of their quantification of the effects of nano/meso-scale processes in the Marcellus shale-matrix pores) developed an LBM model for flow in straight nanochannels based on slip length and effective viscosity, which is applicable for both gas and liquid flow. Based on that model, researchers have further considered surface diffusion by combining the Maxwell-Stefan approach and Langmuir adsorption theory. These simulations indicate that surface diffusion of adsorbed gas can enhance apparent permeability even at high pressure. To enable the model to simulate hydrocarbon flow in nanopores of shale matrix, LANL researchers extended the boundary treatment to arbitrarily complex geometry and considered interaction forces for various hydrocarbon-organics pairs.
LANL researchers found that the enhancement of permeability due to the nanoscale effect in complex nanoporous media is less significant than in long straight nanochannels, and it is more so for liquid flow than for gas flow. Researchers suspected that this was caused by the bending of streamlines resulting from the tortuosity of porous media (end effect). The end effect may lead to additional flow resistance in complex nanopores. Researchers further investigated the different mechanisms contributing to permeability enhancement in nanopores and found that for gasses, the permeability enhancement is roughly equally caused by viscosity decrease near the solid surface and slip at the solid surface, while it is mainly caused by slip for liquids. Because the end effect mainly affects the slip, it counteracts the permeability enhancement more significantly for water flow than for gas flow in complex nanoporous media.
As part of their quantification of gas-water distribution in the Marcellus shale matrix pores, LANL researchers investigated the interactions of methane gas with shale nanopores at high pressures using SANS at NCNR. The sample used was Marcellus shale from the Marcellus Shale Energy and Environment Laboratory (MSEEL) in the form of a wafer cut from the MSEEL core. Results from this investigation suggest that there may exist more hydrocarbon gases than currently estimated without considering the nanopore confinement effect.
In follow-up experiments, LANL determined the nano-porosity variation as a function of pressure using high-pressure SANS with deuterated methane (CD4) as the pressure medium using Marcellus Shale samples obtained from MSEEL. With increasing pressure, more methane fills in nanopores and thus the nano-porosity decreases. However, a large portion of the total porosity (~59 percent) is still inaccessible to methane, which suggests that a significant portion of nanopores are either closed or have narrow throats, where the hosted hydrocarbon may not be accounted for in the current estimate of original hydrocarbon in place using traditional methods. To better characterize the nanoporosity, LANL is currently teaming with Sandia National Laboratory to derive nanopore structures from focused ion beam scanning electron microscopy (FIB-SEM) images of Marcellus shale samples.
Further SANS experiments were conducted with Marcellus shale from the MSEEL using pressure cycling of deuterated methane (CD4) as the pressure medium. The MSEEL sample was pressurized with CD4 in two pressure cycles: In a low-pressure cycle, the pressure was increased to 3,000 psi and then decreased to ambient. In a high-pressure cycle, the pressure was increased to 6,000 psi and then decreased to ambient. In both cases, there is a hysteresis in the changes in scattering intensity, indicating existence of some residual methane gas in the nanopores. In other words, the methane recovery from the shale matrix may be more efficient when drawn down from a higher pressure (6,000 psi) than from a lower pressure (3,000 psi).
LANL has been working with internal and external collaborators to derive nanopore structures from high-resolution images of Marcellus shale samples for LBM simulation of gas flow in real shale samples under different pressure gradients. LANL has initiated the systematic work to predict effective permeability of computationally generated nanopore structures to provide the required data to develop machine learning emulators for fast/real-time prediction of matrix transport properties. LANL has also initiated the work to develop the ML emulators to predict effective diffusivity based on pore structures of two-dimensional porous media. The work has shown that the trained convolutional neural networks (CNN) can predict the effective diffusivity orders of magnitude faster than the physics-based LB simulation with very good accuracy. We are working on extending the machine learning approach to 3D porous structures and other matrix transport properties.
LANL has completed neutron experiments to quantify the fraction of closed pores in a range of shale samples. These include the development of a neutron method (SANS) to probe open/closed nanopores and measured open/closed ratio (C) in MSEEL core. LANL was also able to identify damage in matrix that can occur at high ΔP and determine the minimum pressure that can cause damage.
LANL has also completed a physics-based analysis (based on lattice Boltzmann methods - LBM) to quantify the effects of pore-scale mineralogy and heterogeneity on apparent permeability, thus extending their initial results to a range of shale-matrix properties. Complimenting this, LANL validated the LBM method to quantify gas transport (D) in matrix using real nanostructures; up to 100x ΔP effect.
Lastly, LANL integrated pore-scale results into a revised set of parameters for dfnWorks, which will initially be calibrated to the MSEEL-I samples. Using this information LANL developed an accurate site/basin-specific predictive model for the matrix transport rate (MTR):
MTR = C X D X ΔP
Although this task (Task 5) is complete, LANL continues to work on the integration of large-scale fractures, tributary fractures, and the matrix. LANL has been building graph-based reduced order models for their DFN framework for the shale application. The idea is to build a graph from a DFN and the perform flow and transport simulations on the graph itself.
Budget Period 1 – DOE Contribution: $700,000
Budget Period 2 – DOE Contribution: $700,000
Budget Period 3 – DOE Contribution: $800,000
Budget Period 4 – DOE Contribution: $973,100
Budget period 5 – DOE Contribution: $800,000
Planned Total Funding:
DOE Contribution: $3,973,100
1The original FWP was designated FE-406/408/409-14-FY15 and lasted two budget periods through 3/0216. Subsequent FWPs that extended this research include FE-722-16-FT18 for budget period 3 through 3/2018; FE-954-18-FY18 and FE-954-18-FY18 R1 for budget period 4 through 12/20; and FE-954-20-FY21 for budget period 5 through 9/21.