Using Discovery Science To Increase Efficiency of ... - ACS Publications

Dec 16, 2015 - *E-mail: [email protected]. Shale gas is an unconventional fossil energy resource that is already having a profound impact on United Sta...
0 downloads 0 Views 2MB Size
Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Chapter 3

Using Discovery Science To Increase Efficiency of Hydraulic Fracturing While Reducing Water Usage H. S. Viswanathan,* J. D. Hyman, S. Karra, J. W. Carey, M. L. Porter, E. Rougier, R. P. Currier, Q. Kang, ́ ez, N. Makedonska, L. Zhou, J. Jimenéź -Martín L. Chen, and R. S. Middleton Los Alamos National Laboratory, P.O. Box 1663, Los Alamos, New Mexico 87545, United States *E-mail: [email protected]

Shale gas is an unconventional fossil energy resource that is already having a profound impact on United States (US) energy independence and is projected to last for at least 100 years. Production of methane and other hydrocarbons from low permeability shale involves hydraulically fracturing rock, establishing fracture connectivity, and multiphase fluid-flow and reaction processes, all of which are poorly understood. The result is highly inefficient extraction that also raises many environmental concerns. A science-based capability is required to quantify the governing mesoscale fluid-solid interactions, including microstructural control of fracture patterns, and the interaction of engineered fluids with hydrocarbon flow that is required for increasing efficiency and decreasing the potential of environmental impacts. These interactions depend on several complex coupled thermo-hydro-mechanical-chemical (THMC) processes over scales ranging from nanometers to kilometers. Determining the key mechanisms in subsurface THMC systems has been impeded due to the lack of sophisticated experimental methods to measure fracture aperture and connectivity, permeability, and chemical exchange capacities at the high temperature, pressure, and stresses present in the subsurface. In

© 2015 American Chemical Society In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

this chapter, we describe innovative experimental techniques and simulation methodologies to address these issues. We use high-pressure microfluidic and triaxial core flood experiments on shale to better constrain fracture-permeability relations and the extraction of hydrocarbon. These data are integrated with simulations including lattice Boltzmann modeling of pore-scale processes, finite-element/discrete element approach for fracture initiation and propagation in the near-well environment, discrete fracture network modeling at the reservoir-scale for modeling transport through large-scale fractures, and system-scale models to assess the economics of alternative fracturing fluids. The ultimate goal is to make the critical measurements needed to develop models that can be used to determine the reservoir operating conditions necessary to gain a degree of control over fracture generation, fluid flow, and interfacial processes over a range of in situ subsurface conditions.

Introduction Shale gas is an unconventional fossil energy resource that is already having a profound impact on the US energy sector, with reserves projected to last for nearly 100 years (1). The increased availability of shale gas (i.e., methane), which produces 50% less carbon dioxide (CO2) than coal when used as a fuel, is primarily responsible for US CO2 emissions dropping in 2011 to their lowest levels in 20 years (2). However, these unconventional resources formations, e.g., tight sandstones; shales; and coal beds, have very low permeability (microdarcy-nanodarcy), and the extraction of methane and other hydrocarbons from these low permeability formations involves the hydraulic fracturing of the shale rock to increase fracture connectivity and liberate the in place methane. The process of hydraulic fracturing is poorly understood in part due to the wide range of length scales involved. The result of this ignorance is inefficient extraction that raises many environmental concerns (3, 4). The sustainability of this unconventional energy source depends on improving our understanding of why and how hydraulic fracturing works. Using this knowledge, new technologies for optimizing and enhancing the gas production while simulations decreasing environmental risks must be developed to reach this goal. At Los Alamos National Laboratory (LANL), we have been developing experiments and models that can characterize many of the coupled phenomena involved in hydraulic fracturing, e.g., fracture generation; multiphase fluid flow; and chemical processes, under in situ conditions at the reservoir, core, pore-structure and individual pore scales (Figure 1). Our goal is to reveal the fundamental dynamics of fracture-fluid interactions and transform hydraulic fracturing from an ad hoc tool to a safe and predictable approach based on solid scientific understanding. Determining the key mechanisms in subsurface thermo-hydro-mechanical-chemical (THMC) systems has been impeded due to lack of sophisticated experiments that make direct 72 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

observations at (in situ) high temperatures (T), pressures (P), and stresses present in the subsurface. To address these challenges, we use unique LANL microfluidic and triaxial core flood experiments integrated with state-of-the-art numerical simulation approaches. The consequence of this work is the development of alternative fracturing fluids and fracturing techniques that enhance production, reduce waste-water, and mitigate environmental impacts (4). In this chapter we provide an overview of a variety of preliminary results based upon these experiments and simulations.

Figure 1. Shale gas processes across multiple scales.

Background Although hydraulic fracturing and horizontal drilling are changing the energy landscape, the extraction process is inefficient and poses serious threats to the environment. For example, production rates at a typical hydraulic fracturing site decline rapidly (50-60%) in the first couple of years (Figure 2), and natural gas recovery rates are also very low – usually 10–15% of the gas in place (3, 5). Furthermore, the hydraulic fracturing process requires millions of gallons of chemically treated water per well, much of which is not recovered during flow back. Whatever waste water is recovered becomes an enormous liability – it must either be re-injected at depth (potentially generating earthquakes) or be subjected to expensive wastewater treatment (6, 7). Additionally, the contaminated water that remains in the subsurface could potentially leak to drinking water aquifers through faults and natural fractures. The lack of control of the hydraulic fracturing processes adds concern of hydrocarbons leaking to the overlying aquifers or even to the surface, another significant enviromental risk. With tens of thousands of wells spread across a dozen states environmental concerns are on the rise. In conjunction, these issues have lead to a negative outlook among the public and 73 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

lawmakers. Development of shale-gas reservoirs has yet to win widespread public support, e.g., a moratorium banning hydraulic fracturing exists in New York State, because of perceived risks associated with current practices (8).

Figure 2. Average production rate for the Barnet shale play plotted as a function of years of production. All curves follow a similar trend where initial high production rates decline rapidly and are followed by lower rates. The slight increase around 6 years might be due to fracturing that occurred again. Industrial scale shale gas production using hydraulic fracturing is less than ten years old and the techniques, based on this short experience, rely significantly on aqueous fracturing fluids composed of gels, surfactants, biocides, scale and corrosion inhibitors, as well as solid particulates (proppants) to keep fractures open and maintain flow (9). Deployment in these years has been rapid and undertaken without basic research concerning effective fracturing techniques and/or determining the properties of an ideal fracturing fluid. This foundational research has been lacking because the high pressure and temperature experimental capabilities required for in situ measurements are not widely available. With regard to ideal effective fracturing techniques, fracture extent and proppant effectiveness cannot be accurately measured in the field because extraction typically occurs at depths of 2000 to 3000 meters. A combination of experience, sporadic seismic surveys, and various simulation techniques are currently used to inform fracturing operations. However, the consistency between these approaches is questionable and the high extent of uncertainty in the measurements further undermines their credibility. Because of this uncertainty, it is not surprising that only one out of three wells is profitable. Industry believes that the current state-of-the-art fracture propagation models are inadequate for highly heterogeneous layered shales and for hydraulic fractures created by fluid pressure (10). Efficacy of proppants to enhance and maintain flow in hydraulically 74 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

generated and re-activated fractures is difficult to determine as well (11). One hypothesis is that proppants settle out and do not reach many of the fractures where gas exists; slurry transport calculations in our study, in fact, support this hypothesis. It is thus possible that these anthropogenic additives, some of which are deemed carcinogenic and could potentially contaminate groundwater, may be unnecessary after all (10). Although there is speculation, the industry has yet to reach a consensus as to what makes a good hydraulic fracturing fluid. Currently, water with additives is the primary hydraulic fracturing fluid, but a lack of water in some regions and a belief that more stringent hydraulic fracturing regulations and a carbon-constrained economy are on the horizon has renewed interest in non-aqueous hydraulic fracturing fluids. A recent MIT report3 states that CO2-based fluids provide an interesting, although as yet unproven, possibility for enhancing gas recovery, reducing the amount of water required while simultaneously sequestering CO2. The use of CO2 as a hydraulic fracturing fluid, although used sporadically in the past, has shown promising results. CO2 has favorable properties over water to enhance hydrocarbon recovery (12). CO2 also exhibits more effectiveness in fracturing rock due to coupled THMC effects (13). Furthermore, CO2 is miscible with hydrocarbon and exchanges with the hydrocarbon adsorbed to shale organics. This amenable property prevents flow blocking, which is a major challenge with water. If CO2 is the base fluid, additives (such as biocides and surfactancts) might be unnecessary, but an increase in viscosity would be required if CO2 must carry proppants (14). In a few U.S. Department of Energy (DOE) sponsored experiments conducted well in the past before the natural gas boom, CO2 showed up to five times more gas production compared to aqueous fluids, required no additional toxic additives, and greatly minimized water usage. However, the results shown were not consistently positive (15–18). Within this basic research, there are highly non-linear coupled processes that play a major role, such as control fracturing and hydrocarbon extraction, and characterization of these processes is important for accurate prediction. For example, it is possible that the compressibility of CO2 will enhance rock fracturing due to a positive feedback created by thermal stresses when CO2 expands into a new fracture volume and cools the crack tip (13). Thus it needs to be determined if cycling fluid pressure can generate non-hydraulic shear fractures that remain open during the production phase when pressures are reduced. However, this requires analysis of T, P, and stress conditions using modified aqueous- and CO2-based working fluids, a task that is beyond what is currently feasible. Nonetheless, the work presented in this chapter could provide the foundational work for these important and relevant scientific questions to be addressed. We detail the various integrated experimental-modeling aspects of our work that aim at making critical measurements for determining reservoir operating conditions required to control fluid flow and fracture generation in the following sections. An overview of the experiment and modeling capabilities at different scales is presented along with a discussion of how these scales are linked. We also provide references to manuscripts written by the authors for a comprehensive discussion. Specifically, we discuss details regarding: 1) reservoir-scale modeling using a high performance computational suite based on 75 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

the discrete fracture network approach for understanding transport mechanisms and the integration of this modeling approach with data from a real shale site; 2) core-scale experiments using a tomographic triaxial system that can obtain measurements of permeability change due to fracture generation under in situ conditions; 3) microfluid experiments at high-pressure conditions for measuring sweep efficiency of water and CO2 in shale; 4) pore-scale numerical experiments using the lattice Boltzmann method to simulate gas flow.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Reservoir Scale In contrast to the ad hoc methods currently in use, we seek a detailed analysis of the fundamental mechanics behind gas transport to better understand the reason for the rapid decline in the production. Gas production analysis is frequently done in the industry, but the methods implemented are either highly empirical or are based on simplified analytical models that depend on gross idealizations of the reservoir (19). In contrast, we simulate gas transport using advanced physics-based computational models that are built on realistic conditions utilizing well-characterized fracture datasets. Our hypothesis is that different physical and chemical mechanisms control production as a function of time (Figure 3) and we can test this hypothesis with reservoir modeling and compare our simulation results to field data.

Figure 3. A typical production from the Haynesville formation. Our hypothesis is that different physical and chemical mechanisms control production as a function of time, the primary mechanism reduces in scale at time progresses. 76 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

We use the new computational workflow dfnWorks that is built on a suite of high-performance computing (HPC) modules and visualization tools developed at the LANL and other DOE laboratories. The general workflow involves generating and meshing discrete fracture networks (DFN) using field data from geological surveys (20), solving for flow using the massively parallel subsurface simulator PFLOTRAN (21), and evaluating gas flow pathways and gas particle travel times using a unique particle tracking method developed specifically for DFN (22). In our reservoir model, a DFN is generated using fracture properties from the Pottsville shale formation in Alabama such as fracture spacing, orientation, and aperture (Figure 4). A horizontal well, colored blue, along with hydraulically generated fractures, shown in brown, is included at the center of the domain to simulate production from the reservoir. Natural fractures in the system are colored according to pressure, with red being high values (21MPa) and blue low (17MPa). Operational pressure, aperture and porosity are used in the simulation. Figure 4a shows the obtained pressure distribution in the domain highlighting drawdown towards the well. Transport pathways are determined by tracking advective nonreactive particles from locations in the DFN to the well.; trajectories of a selected number of particles are shown in Figure 4b. Further details of the computational suite and the simulation can be found in Karra et al. (23)

Figure 4. Reservoir scale calculations of the production curve. a) DFN generated base on site data from the Pottsville formation, where a well is placed at the center of the domain to simulate production. b) Draw down is created to bring methane packets, represented as particles, to the well. c) The simulation production curve (green) matches the site data (blue) for the first year and then underestimates the production rate because small scales, which govern long term production rates, are not included in the simulation. (Adapted from reference (23).) Based on the particle travel times, a production curve is obtained for the simulation (green, Figure 4c) and is compared to site production curve (blue, Figure 4c). The simulation production curve compares well to the field data especially for the first part. The tail of the simulation curve is less than that of field data. This difference in the tail is expected because the small scale transport mechanisms are not considered in our simulation (Figure 3). This comparison suggests that early production is controlled by the large hydraulic and natural fractures represented by the DFN. 77 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Our results are in line with industry observations of an early fast decline in production rates due to the drainage of hydrocarbons from the large fractures (10). Characterization of the production curve tail is critical because wells often produce for decades, and improvement of the tail will significantly impact long term production. However, the tail of the production curve is the integrated result of nonlinear combinations of 1) mass transport from the damage zone between small fractures and the shale matrix, 2) matrix diffusion, 3) desorption and 4) multiphase flow blocking – all of which have to be incorporated in our reservoirscale simulations. The work discussed in the remaining of the chapter, at the coreand pore- scales, address these complex transport mechanisms with the aim that once they are well-characterized, they can be incorporated into the reservoir scale simulator to aid in decision making.

Core Scale The knowledge gap of fundamental fracture network properties in shale is a debilitating limitation in the improvement of efficient and effective fracturing. Furthermore, core-scale (10 µm to 10 cm) fracture network processes are crucial in characterizing gas transport from intact rock matrix to small fractures that connect to the larger natural and hydraulically generated fractures, modeled as a DFN in the reservoir-scale model. To address these uncertainties we use triaxial coreflood instruments to generate and characterize fracture formation and permeability at in situ temperature, pressure and stress conditions, see Carey et al. for details (24). Figure 5 shows the experimental setup for a shear fracture experiment. These in situ tomographic measurements of fracture propagation are integrated with a finite discrete element model (FDEM) (25, 26). This integration gives a basic framework to predict fracture propagation and estimate hydrocarbon extraction under various conditions such as different fluids, rock properties, and injection/pressurization schemes. The FDEM model we have developed also simulates fracture propagation due to fluid pressure. Figure 6 compares a direct shear fracture experiment using the triaxial coreflood experimental setup with the FDEM model using a Utica shale sample provided by Chesapeake Energy. Overall good qualitative agreement is seen. Both approaches show a vertical fracture in the middle dominates the network along with two fracture arcs on each side of the vertical fracture, although, a stronger fracture system is observed in the experiments. An important lesson from this comparison is that if fluid pressure is not used or if the layers of the shale are not considered in our fracture propagation simulations, the model does not show good qualitative agreement to the experiment. This indicates that interface and fluid flow are key processes to consider in predicting fracture extent in shale.

78 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Figure 5. Experimental setup for a shear fracture experiment. (Adapted from reference (24).)

Figure 6. Comparison of a representative triaxial coreflood experiment (a) with the FDEM model (b) for a direct shear fracture experiment that uses a Utica shale sample provided by Chesapeake Energy. The FDEM model used generic rock material properties for Utica shale rather than properties from this specific experimental sample, with the goal being to determine if qualitative agreement between model and experiment was possible without detailed shale characterization. (Adapted from reference (24).)

79 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Because the permeability of shale is vital in evaluating hydrocarbon extraction but is poorly known under in situ conditions, we have developed the triaxial experiments to have the capability to measure the permeability of the shale during fracturing under in situ pressure, temperature and stress conditions. The stress and permeability as a function of strain for a shale sample is shown in Figure 7. There is no measureable permeability until failure and permeabilities up to 30 mD are measured once fractures initiate. Further deformation of the sample results in closure of fractures and healing leading to reduction of permeability.

Figure 7. Stress and permeability as a function of strain of the experiment shown in Figure 6. Initially, there is no measureable permeability. Once fractures begin to form, permeabilities up to 30 mD are measured. As the sample continues to deform under varying hydrostatic conditions the fractures begin to close resulting in fracture healing and a reduction of permeability. The permeability measurements are critical because they characterize fluid flow through shale fractured under in situ conditions. Moving forward, we are beginning to hydraulically fracture shales and are in the process of moving the triaxial experimental apparatus to a tomographic facility. This will allows us to capture in situ fracture formation images. Finally, acoustic and tracer capabilities are being appended to the experimental apparatus with a coupled flow FDEM capability concurrently being developed along with threedimensional fracture simulation capability.

Fracture-Matrix Interaction For effective hydraulic fracturing and hydrocarbon extraction, high density fracture networks at the mesoscale at the order of millimeters must be created. Although larger fractures are the major transport pathways for the hydrocarbons, a 80 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

high density fracture network of small fractures is required to increase surface area with the shale matrix and allow more methane to be liberated (27). Our approach to characterize the migration of hydrocarbon from the matrix into these small fracture networks is based on integrating microfluidic experiments with Lattice Boltzmann Method (LBM) simulations. Several key phenomena that we seek to investigate in microfluidics experiments and LBM simulations include: 1) flow blocking of hydrocarbon by residual water – on the other hand, supercritical CO2 (scCO2) may facilitate hydrocarbon migration due to its miscibility with oil and gas; 2) wettability between shale and working fluid; 3) dead-end pores can trap hydrocarbon in aqueous systems – scCO2 again will dissolve and free the hydrocarbons; and 4) some components of natural gas can condense as a liquid and block flow. At the micro-fracture scale, at the order of millimeters, we are studying two main issues: 1) the difference between glass and shale micro-models to accurately mimic sweep efficiency in shale; and 2) compare the sweep efficiency of scCO2 and water based fracturing fluids. Through these studies mass transfer in the micro-fractures will be evaluated and then upscaled for use in reservoir model simulations. Figure 8a shows the work flow for obtaining fracture geometry from a triaxial experiment, which is then used to etch the fracture pattern into a shale microfluidic wafer. We have also developed the capability to conduct microfluidic experiments under in situ high temperature and high pressure reservoir conditions. This is a critical capability for characterizing sweep efficiency because shale wetting properties are not easily replicated by synthetic materials and fluid properties are greatly influenced by reservoir temperatures and pressures. The fast flow pathway of scCO2 in a fracture that is initially filled with water is shown in Figure 8b (black). Moving on to the pore structure scale (100 nm – 100 µm), surface tension dominates fluid transport dynamics for the hydrocarbon-brine system. In addition, flow blocking due to multiphase flow could prevent effective extraction to the small scale fracture. The LBM approach is ideal to simulate these processes because it can simulate complex flows in complicated intra-pore geometries and resolve relevant physicochemical processes with high computational efficiency. As an example, we compare a microfluidic experiment with a fishbone fracture pattern shown in Figure 9a to a LBM simulation in Figure 9b. The experiment uses a sample of Utica shale. As seen in Figure 9b, the LBM simulation of the experiment is able to capture the fingering as the invading immiscible water displaces hydrocarbon. It also captures the bypassing of the hydrocarbon in dead end fractures that leads to poor sweep (Figure 9b). Although at first glance, this example appears to be a simple one, there are several complex processes and parameters that control the flow. These include: a) flow rate and fluid viscosity ratios control finger width, and b) network geometry that affects the finger width because fluid from the side channel narrows the finger. More details can be found in Middleton et al. (18)

81 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Figure 8. Schematic of our work-flow for extracting fracture geometry from a triaxial experiment and then etching the fracture pattern into a shale microfluidic wafer. (Adapted from reference (18).)

Figure 9. a) A microfluidic experiment in which a simple fishbone fracture pattern has been etched into Utica shale, saturated with oil and water blocking the exit. b) A LBM of the experiment. The simulation captures the fingering as the invading immiscible water displaces hydrocarbon, but bypasses the hydrocarbon in dead end fractures resulting in poor sweep. 82 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Acquiring and imaging shale samples for use in LBM simulations is also a challenge due to the length scales in shale that must be resolved (28). To make matters worse, numerous shale samples are required to obtain statistics with tight confidence intervals. An alternative approach is to stochastically generate samples with desired properties. On the left side of Figure 10 is a binarized image of a shale and the image on the right is generated to have similar statistical properties as the rock sample using the techniques of Hyman and Winter (29). The algorithm to generate these samples is fairly computationally fast and can generate a large number of them quickly. Another important factor that increases the permeability of shale is crack density, which leads to improved gas migration from kerogen to damage zone (27). Characterizing and evaluating the permeability change due to cracks is done by including fractures into synthetic pore structures similar to Figure 10b.

Figure 10. Left) A two-dimensional slice of a binary image of shale where black is higher permeability organic inclusions. Right) A synthetic sample generated to have the similar statistics, namely geometric observables, as the image on the left.

Figure 11a shows a representative example, where we generate a synthetic medium based on an image of shale, and a solved linear diffusion equation for pressure therein. Dark grey represents low permeability in-organic material, while the white region is higher permeability organic material. Pressure contours obtained from the diffusion solver are shown in color. Flow velocity is represented using arrows whose length is proportional to the magnitude of the velocity vector. In Figure 11b we have included four micro-cracks into the same shale structure and determine the difference in effective permeability. Merely including these four cracks that are aligned with the pressure gradient into the system nearly doubles the effective permeability of the sample! We are further investigating this dependence of effective permeability on crack density based on this novel methodology. 83 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Figure 11. Preliminary investigation into the influence of microcracks on permeability. A pressure gradient is imposed to determine flux, and therefore permeability. Contours denote pressure. The underlying geometry of both samples are identical, but the sample in (b) has four high permeability micro-cracks. The sample in (b) has an effective permeability that is nearly double that of the sample in (a).

Pore Scale The physics occurring at the scale of individual pores (10-100 nm) in shales is remarkable because the length scales involved are so small that standard diffusion breaks down. Specially, the mean free path of a gas particle is larger than characteristic pore size of the medium and unrestricted Brownian motion cannot occur. When the Knudsen number (Kn), the ratio of the mean free path of a gas particle over the characteristic pore size of the medium, is relatively large, gas molecules tend to slip on the pore walls and the effective permeability of the medium can be significantly larger than the intrinsic permeability of the medium. This influence is known as the Klinkenberg effect. The effect of this gas slippage on permeability in shales was recently studied by Chen et al. (30), using our LBM simulations. The authors determined that a significant deviation from the intrinsic permeability occurs due to gas slippage. Figure 12 shows how the permeability of a medium is underestimated if the Klinkenberg effect is not included. If it is neglected, then estimations of permeability can differ by up to two orders of magnitude. This influence is more pronounced at low pressures.

84 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

Figure 12. Relationship between global permeability and porosity, plotted at three different pressures and without consideration of the Klinkenberg effect. If the Klinkenberg effect is not included permeability is underestimated, more so at low pressure.

Conclusions In this chapter, we have described several scientific approaches developed to improve our understanding of the complex processes associated with hydraulic fracturing. First, we described a novel reservoir scale discrete fracture network model that forms as a framework to determine the key physical mechanisms governing the initial decline in natural gas production. Next, we described our approach at the core-scale with triaxial experiments to focus on understanding fracture-permeability behavior and compared it to FDEM modeling of fracture propagation. We benchmarked these triaxial measurements with FDEM model and will upscale the results to the near wellbore environment. Using a high-pressure/temperature microfluidics experiments conducted in shale micro-models, our aim was to extract the details of pore-scale multiphase flow processes within fractures. These details lead to evaluating sweep efficiency of the working fluid which will help the community to design more effective working fluids for the recovery of hydrocarbons. Additionally, by simulating 85 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

these results with the LBM approach, we extended the experiments to consider a wider spectrum of conditions and potential working fluid properties. The goal of this integrated experiment-simulation approach is to expose how differing stress conditions and shale properties govern fracture growth, penetration, and hydrocarbon extraction. The results of pore- and core-scale experiments and simulations provide basic characterization of fracture properties that will be used to populate the reservoir-scale DFN model. This large-scale simulation reveals the mechanics of how processes within the matrix, fracture damage zone, fine-scale fracture network eventually drain to the dominant fracture network connected to the horizontal wellbore. By using field data, we will relate these results to actual production curves with the aim to provide critical data on the efficiency of hydraulic fracturing. The integration of multi-scale experimental measurements and computational modeling of coupled THMC systems is critical to understand and eventually control fracture formation due to fluid flow in the hydraulic fracturing process. Such results provide key insights into the physics behind the complex processes that can lead to more efficient hydrocarbon extraction. More efficient wells lead to a smaller environmental footprint, and improved understanding of fracture generation and propagation leads to better control and containment of fractures in the shale layer. Furthermore, these results increases confidence in the protection of groundwater resources close to the shale layer.

References 1.

2.

3.

President Obama State of the Union Address 2012, “We have a supply of natural gas that can last America nearly one hundred years, and my Administration will take every possible action to safely develop this energy. Experts believe this will support more than 600,000 jobs by the end of the decade. And I’m requiring all companies that drill for gas on public lands to disclose the chemicals they use. America will develop this resource without putting the health and safety of our citizens at risk. The development of natural gas will create jobs and power trucks and factories that are cleaner and cheaper, proving that we don’t have to choose between our environment and our economy. And by the way, it was public research dollars, over the course of thirty years, that helped develop the technologies to extract all this natural gas out of shale rock - reminding us that Government support is critical in helping businesses get new energy ideas off the ground.” Begos, K., Associated Press, Aug. 16, 2012. “In a surprising turnaround, the amount of carbon dioxide being released into the atmosphere in the U.S. has fallen dramatically, to its lowest level in 20 years, and government officials say the biggest reason is that cheap and plentiful natural gas has led many power-plant operators to switch from dirtier-burning coal.” Moniz, E. J.; Jacoby, H. D.; Meggs, A. J. The Future of Natural Gas; MIT: Cambridge, MA, 2012.

86 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

4.

5.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

6. 7.

8. 9.

10. 11. 12.

13. 14.

15.

16.

17.

18.

Soeder, D. J.; Kappel, W. M. Water Resources and Natural Gas Production from the Marcellus Shale; Fact Sheet 2009-3032; U.S. Geological Survey, 2009. Hu, Q. In Integrated Experimental and Modeling Approaches to Studying the Fracture-Matrix Interactions in Gas Recovery from Barnett Shale. Proceedings of the RPSEA Unconventional Gas Conference 2012: Geology, the Environment, Hydraulic Fracturing, Canonsburg, PA. April 17-18, 2012. http://www.rpsea.org/files/3013/ (accessed August 18, 2015). Fountain, H. Disposal Well Halted after New Quake in Ohio. New York Times January 1, 2012. Fomel, S. In Multiazimuth Seismic Diffraction Imaging for Fracture Characterization in Low Permeability Gas Formations. Proceedings of the RPSEA Unconventional Gas Conference 2012: Geology, the Environment, Hydraulic Fracturing, Canonsburg, PA. April 17-18, 2012; http://www.rpsea.org/media/files/files/c5fb4aaa/EVNT-PR-2012UR_08122-53_Multiazimuth_Seismic_Diffraction_Imaging_FractureFomel-04-17-12.pdf (accessed August 18, 2015). Kramer, D. Shale-gas Extraction Faces Growing Public and Regulatory Challenges. Phys. Today 2011, 64, 23. Kargbo, D. M.; Wilhelm, R. G.; Campbell, D. J. Natural Gas Plays in the Marcellus Shale: Challenges and Potential Opportunities. Environ. Sci. Technol. 2010, 44, 5679–5684. King, G. Apache Corporation, Houston, TX. Personal communication, 2014. Calfrac Well Services, Hess Corporation, Shell. Personal communication, 2014. Ishida, T.; Aoyagi, K.; Niwa, T.; Chen, Y.; Murata, S.; Chen, Q.; Nakayama, Y. Acoustic Emission Monitoring of Hydraulic Fracturing Laboratory Experiment with Supercritical and Liquid CO2. Geophys. Res. Lett. 2012, 39, L16309. Wang, H.; Li, G.; Shen, Z. A Feasibility Analysis on Shale Gas Exploitation with Supercritical Carbon Dioxide. Energy Sources 2012, 34, 1426–1435. Enick, M. A Literature Review of Attempts To Increase the Viscocity of Dense Carbon Dioxide; Report DE-AP26-97FT25356; United States Department of Energy, 1998. Wright, T. R. Frac Technique Minimizes Formation Damage Dry Frac. World Oil [Online] 1998. http://business.highbeam.com/1886/article-1G120387355/frac-technique-minimizes-formation-damage (accessed July 21, 2015). Mazza, R. L. In Liquid-Free Simulations − CO2\Sand Dry-Frac. Proceedings of the Conference of Emerging Technologies for the Natural Gas Industry, Houston, TX, March 24-27, 1997. www.netl.doe.gov/KMD/cds/Disk28/ NG10-5.pdf (accessed August 18, 2015). Gupta, D. V. S.; Bobier, D. M. In The History and Success of Liquid CO2 and CO2/N2 Fracturing System; SPE Gas Technology Symposium, Calgary, Canada, March 15-18, 1998. Middleton, R. S.; Carey, J. W.; Currier, R. P.; Hyman, J. D.; Kang, Q.; Karra, S.; Jiménez-Martínez, J.; Porter, M. L.; Viswanathan, H. S. Shale 87 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.

19.

20.

Downloaded by CHINESE UNIV OF HONG KONG on December 24, 2015 | http://pubs.acs.org Publication Date (Web): December 15, 2015 | doi: 10.1021/bk-2015-1216.ch003

21.

22.

23.

24. 25.

26.

27. 28.

29.

30.

Gas and Non-Aqueous Fracturing Fluids: Opportunities and Challenges for Supercritical CO2. Appl. Energy 2015, 500–509. Patzek, T. W.; Male, F.; Marder, M. Gas Production in the Barnett Shale Obeys a Simple Scaling Theory. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 19731–19736. Hyman, J. D.; Gable, C. W.; Painter, S. L.; Makedonska, N. Conforming Delaunay Triangulation of Stochastically Generated Three Dimensional Discrete Fracture Networks: A Feature Rejection Algorithm for Meshing Strategy. SIAM J. Sci. Comput. 2014, 36, A1871–A1894. Lichtner, P. C.; Hammond, G. E.; Lu, C.; Karra, S.; Bisht, G.; Andre, B.; Mills, R. T.; Kumar, J. PFLOTRAN User Manual: A Massively Parallel Reactive Flow and Transport Model for Describing Surface and Subsurface Processes; Technical Report No: LA-UR-15-20403; Los Alamos National Laboratory, 2015. Makedonska, N.; Painter, S. L.; Gable, C. W.; Bui, Q. M.; Karra, S. Particle Tracking Transport on Discrete Fracture Networks: Methods and Results. Comput. Geosci. 2015, DOI: 10.1007/s10596-015-9525-4. Karra, S.; Makedonska, N.; Viswanathan, H. S.; Painter, S. L.; Hyman, J. D. Effect of Advective Flow in Fractures and Matrix Diffusion on Natural Gas Production. Water Resour. Res. 2015, DOI: 10.1002/2014WR016829. Carey, J. W.; Zhou, L.; Mori, H.; Viswanathan, H. S. Fracture-Permeability of Shale. J. Uncon. Oil Gas Res. 2015, 11, 27–43. Lei, Z.; Rougier, E.; Knight, E. E.; Munjiza, A. A Framework for Grand Scale Parallelization of the Combined Finite Discrete Element Method in 2D. Comput. Part. Mech. 2014, 1, 307–319. Zubelewicz, A.; Rougier, E.; Ostoja-Starzewski, M.; Knight, E. E.; Bradley, C.; Viswanathan, H. S. A Mechanisms-Based Model for Dynamic Behavior and Fracture of Geomaterials. Int. J Rock Mech. Min. 2014, 72, 277–282. Bažant, Z. P.; Salviato, M.; Chau, V. T.; Viswanathan, H. S; Zubelewicz, A. Why Fracking Works. J. App. Mech. 2014, 81, 101010. Dewers, T. A.; Heath, J.; Ewy, R.; Duranti, L. Three-Dimensional Pore Networks and Transport Properties of a Shale Gas Formation Determined from Focused Ion Beam Serial Imaging. Int. J. Oil Gas Coal Tech. 2012, 5, 229–248. Hyman, J. D.; Winter, C. L. Stochastic Generation of Explicit Pore Structures by Thresholding Gaussian Random Fields. J. Comput. Phys. 2014, 277, 6–31. Chen, L.; Fang, W.; Kang, Q.; Hyman, J. D.; Viswanathan, H. S.; Tao, W. Generalized Lattice Boltzmann Model for Flow through Tight Porous Media with Klinkenberg’s Effect. Phys. Rev. E. 2015, 91, 033004.

88 In Hydraulic Fracturing: Environmental Issues; Drogos, Donna L.; ACS Symposium Series; American Chemical Society: Washington, DC, 2015.