Laboratory Characterization and Modeling to Examine CO2 Storage and Enhanced Oil Recovery in an Unconventional Tight Oil Formation

Available online at www.sciencedirect.com

ScienceDirect Energy Procedia 114 (2017) 5460 – 5478

13th International Conference on Greenhouse Gas Control Technologies, GHGT-13, 14-18 November 2016, Lausanne, Switzerland

Laboratory characterization and modeling to examine CO2 storage and enhanced oil recovery in an unconventional tight oil formation James A. Sorensena*, Bethany A. Kurza, Steven B. Hawthornea, Lu Jina, Steven A. Smitha, Alexander Azenkenga ªEnergy & Environmental Research Center, University of North Dakota,15 North 23rd Street, Stop 9018 Grand Forks, ND 58202-9018, United States

Abstract The Bakken Formation is a major unconventional tight oil play in North America. The potential for the Bakken to serve as a target for CO2 storage and CO2-based enhanced oil recovery was examined using advanced characterization techniques, laboratory experiments, and modeling. Bakken rocks are dominated by nanoscale pore throats, and although porosity is low, much of the nanoscale porosity is connected. Collectively, the characterization, experimental, and modeling results suggest the nanoscale pore throats make important contributions as transport pathways for CO2 storage and oil mobilization in the Bakken. 2017J.A. TheSorensen Authors.Published Published Elsevier ©©2017 byby Elsevier Ltd.Ltd. This is an open access article under the CC BY-NC-ND license Peer-review under responsibility of the organizing committee of GHGT-13. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of GHGT-13. Keywords: geoligic storage; unconventional geolgic storage; enhanced oil recovery; reservoir characterization; laboratory studies

* Corresponding author. Tel.:+1-701-777-5287; fax: +1-701-777-5181. E-mail address: [email protected]

1876-6102 © 2017 J.A. Sorensen Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of GHGT-13. doi:10.1016/j.egypro.2017.03.1690

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

1. Introduction In recent years, the greatest expansion in North American oilfield development has been in unconventional tight oil formations (d0.1 mD permeability), where fluid flow pathways are dominated by natural and artificially induced fractures. The tight oil resources in the United States are massive, with several hundreds of billions of barrels of oil in place in the Bakken petroleum system (hereby referred to simply as “the Bakken”) of the Williston Basin [1]. The Eagle Ford Formation, a tight oil resource in Texas, appears to be of comparable magnitude, and emerging tight oil plays such as the Montney Formation in Alberta and the Niobrara Formation in Wyoming and Colorado further underscore the growing importance of unconventional oil production to North America’s energy portfolio. Given their size and broad geographic distribution, tight oil formations may hold substantial opportunities to simultaneously store CO2 while increasing the recoverable reserves of oil by injecting CO2 for enhanced oil recovery (EOR). A previous study by Sorensen and others [2] applied the U.S. Department of Energy (DOE) methodology [3] for estimating CO2 EOR and storage capacity the Bakken Formation in North Dakota. The results of that exercise suggest that the Bakken holds a potential storage resource of 1.9 to 3.2 billion tons of CO 2, which could yield 4 to 7 billion barrels of incremental oil [2]. However, it is important to note that the DOE methodology was developed with conventional oil reservoirs in mind, and many of the assumptions upon which that methodology is based (especially with respect to fluid phase behavior and sweep/storage efficiency) may not be as directly applicable to tight oil formations because of their unique porosity, permeability, and organic matter characteristics. The widespread exploitation of tight oil resources is also a relatively recent development (within the last 8 to 10 years); thus the current level of knowledge of mechanisms and factors affecting both oil production from and CO 2 storage in these unconventional tight formations is relatively low when compared to knowledge of conventional reservoirs (over 40 years of history). With these issues in mind, since 2014, a research program to evaluate the potential for storage of CO 2 for greenhouse gas emission mitigation and attendant CO2-based EOR in the Bakken has been conducted by the Energy & Environmental Research Center (EERC) at the University of North Dakota. The ultimate goal of the program is to provide stakeholders with new knowledge that can be applied to the design and execution of a pilot injection and production test in a Bakken reservoir. Routine and advanced characterization techniques were applied to samples of the key shale and nonshale lithofacies to develop a detailed understanding of the pore throat networks of the different lithofacies at scales ranging from macro- to microto nanoscale. This paper will also present the results of laboratory-scale experiments to examine the ability of CO2 to permeate plug samples representing several lithofacies of the Bakken, including the organic-rich shales, and mobilize hydrocarbons from those samples. History match modeling of the plug-scale experiments was also conducted to examine the relative effects of different rock characteristics and mechanisms on hydrocarbon mobilization. 2. Background The Bakken, located in the central region of North America, is an unconventional tight oil resource with oil-inplace estimates of 300 billion barrels (Bbbl) to 900 Bbbl of oil [4,5].The Bakken is characterized by several distinctive lithofacies, each with its own unique properties that may (or may not) significantly affect the mobility and ultimate fate of CO2 within the formation. The fine-grained clastics and carbonates of the Middle Bakken Member are representative of a tight, fractured reservoir rock that is capable of transmitting fluids once it has been hydraulically fractured. In North Dakota, the Middle Member typically comprises between three and seven distinctly different lithofacies that range from silty carbonates to calcite/dolomite-cemented siltstones. In most areas of the Bakken, the Middle Member of the formation is bounded above by the Upper Bakken Shale and below by the Lower Bakken Shale. Both shale members are organic-rich, oil-wet shales that are the source rocks for the oil-productive areas of the Bakken. Some of the key challenges associated with characterization of the Bakken include low porosity (typically <10%), low permeability (typically <1 mD), very fine grained minerals (4 to 60 µm) and clay-size particles (<4 µm) that are hard to resolve both chemically and physically, and a high degree of rock heterogeneity. These factors directly influence the capacity of tight oil formations to transport and store CO 2. They also affect the ability of the injected CO2 to mobilize oil from the matrix into the fracture network and, ultimately, increase oil production. Inadequate identification of these features poses serious challenges to the development of effective injection and production strategies for CO2 EOR and storage in tight, fractured reservoirs.

5461

5462

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

The viability of injecting CO2 into the Bakken for simultaneous CO2 storage and EOR has been the focus of previous laboratory and modeling research activities, the results of which suggest that 1) at the core plug scale, CO 2 can permeate organic rich shales and other tight lithofacies and mobilize significant amounts of oil from those rocks and 2) the oil production response of a Bakken reservoir to CO 2 injection may be delayed, but the increase in oil production rates could approach100% [6–9]. However, an examination of publicly available data from pilot-scale field injection tests using CO2 indicate that those tests have not yielded the improvements in oil production predicted by modeling [10]. The disparity between the laboratory and modeling results and the field tests reflects the large degree of uncertainty when it comes to understanding the mechanisms controlling fluid movement and phase behavior in the Bakken. This is due, in part, to significant data gaps in the identification and characterization of micro- and nanoscale fracture networks and porosity and in understanding the factors controlling CO 2 permeation and transport within the formation [8]. A recent modeling-based study revealed that multiphase fluid behavior and flow in fluid-rich shales vary substantially depending on the size of pore throats and that properties such as fluid viscosity and density are much different in nanoscale pores (mode of 2–3 nm) than in macroscale pores (mode of 11 µ m) [11]. Thus to better understand and model fluid permeation and transport within unconventional reservoirs, it is critical to understand the nature and distribution of nano-, micro-, and macroscale pores and fracture networks within the formation. 3. Reservoir properties and rock characterization Oil is produced from the Bakken Formation over a wide area in the Williston Basin in North Dakota and Montana in the United States and Manitoba and Saskatchewan in Canada. The Bakken Formation is made up of a series of complex lithofacies with variable distribution and properties, as shown in Figure 1 [8,12,13]. The formation was deposited in the Devonian to Mississippian Periods and consists of upper and lower black shales and a middle siltstone/sandstone/carbonate member [12,14–16]. 3.1. Reservoir properties The total thickness of the Bakken Formation is less than 160 ft in the Williston Basin at a maximum depth of 12,150 ft. The maximum thickness of the Upper, Middle, and Lower Bakken units is 23, 85, and 50 ft, respectively [17]. The porosity of the Bakken units is relatively constant across the basin, with the Middle Bakken having an average porosity around 5.5% and Upper and Lower Bakken Shales having approximately 3.6%. Compared to porosity, the permeability varies drastically among the Bakken units, from 0.0001 to 0.1 mD for the Upper and Lower Bakken and 0.0001 to 57 mD for the Middle Bakken [6,18,19]. Kurtoglu and others [6] studied the reservoir properties of different oil fields at the Bakken, including Elm Coulee Field in Montana and Murphy Creek, Bailey, and Reunion Bay Fields in North Dakota. They summarized the reservoir properties in these fields as shown in Table 1, which clearly demonstrates the variation of parameters in the formation. 3.2. Rock characterization Dozens of core plugs were collected from four wells that penetrate through all of the Bakken units in North Dakota. Figure 2 shows white light photographs of slabbed core samples of the major lithofacies that occur in the Bakken in those four wells. The major lithofacies in the study area are, from bottom to top, the Lower Bakken Shale (LBS), Middle Bakken Lithofacies 1 (MB-L1), Middle Bakken Burrowed Lithofacies (MB-L2), Middle Bakken Laminated Lithofacies (MB-L3), Middle Bakken Packstone Lithofacies (MB-L4), Middle Bakken Lithofacies 5 (MB-L5), and the Upper Bakken Shale (UBS). Figure 2 illustrates the high degree of heterogeneity that exists within the different Middle Bakken lithofacies with respect to matrix characteristics, particularly with regard to depositional features and mineralogy distribution. Detailed evaluation of rock properties was conducted using photomicrography,

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig. 1. Stratigraphic column and schematic profile of the Bakken (modified from Kuhn and others [12]).

Scanning electron microscopy (SEM), x-ray diffraction (XRD) mineralogical analysis, and x-ray fluorescence (XRF) analysis to determine the rock composition and chemical elements in the Bakken units. Thin-sections samples of the key Bakken lithofacies were analyzed and photographed using a petrographic microscope. Mineralogical assemblages and prevalence were determined and estimated through the use of standard optical techniques. Photomicrographs were produced at 200× magnification with plane-polarized light, as shown in Figure 3. The photomicrographs clearly show the variation of mineralogy and grain size between the Bakken units. Thin-section slides used for optical microscopy analysis were also used for SEM analysis. Backscattered electron (BSE) images were obtained on the samples to characterize textural and structural features of the different minerals found in the samples. X-ray signals obtained from energy-dispersive spectroscopy (EDS) were used to identify the chemical composition of the different mineral grains. Finally, the combination of textural and structural features observed from BSE images with the chemical elemental composition obtained from EDS analysis were used to determine the mineral composition of the sample. Fig. 4 presents an SEM mineral map showing the mineral composition of the Upper Bakken Shale, where we can easily identify the distribution of mineral components in the

5463

5464

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478 Table 1. Reservoir properties of the Bakken (modified from Kuroglu and others [6]. Montana

North Dakota

Properties

Bakken Unit

Elm Coulee Field

Murphy Creek, Bailey, and Reunion Bay Fields

Formation Type (lithology)

Upper Bakken

Organic-rich shales

Organic-rich shales

Middle Bakken

Silty dolostone

Limestone to sandstone

Lower Bakken

Organic-rich siltstone

Organic-rich shales

Upper Bakken

6–10 (1.8–3.1)

12–18 (3.6–5.5)

Middle Bakken

10–40 (3–12)

30–40 (9–12)

Lower Bakken

2–6 (0.6–1.8)

8–10 (2.4–3.1)

Geological Structure

Whole Bakken

Poplar Dome

Nesson and Billing Anticline

Porosity, %

Middle Bakken

5–7

4–6

Permeability, mD

Middle Bakken

0.01–0.1

0.0001–0.001

Pressure Gradient, psi/ft (kPa/m)

Whole Bakken

0.53 (11.99)

0.78 (17.64)

Temperature, °F (°C)

Whole Bakken

240 (115.6)

240 (115.6)

Oil Saturation, %

Middle Bakken

75

50

Oil Gravity, °API

Whole Bakken

42

39–42

Horizontal Well Length, ft (m)

Middle Bakken

5000 (1520)

9000 (2740)

Net Thickness, ft (m)

Fig. 2. Photos of cores from different Bakken units for lithology identification data acquisition.

rock matrix: grains such as quartz, dolomite, feldspar, pyrite, albite, and organics, etc., fill in the clayey matrix. Lower Bakken Shale has a similar mineral composition to the Upper Bakken Shale, while Middle Bakken samples have significantly more dolomite and quartz than clays, which indicates the lithology of the unit is a mix of sandstone and limestone. An example of a combined BSE image and mineral map of a Middle Bakken (MB-L3) sample is provided in Figure 5. XRD mineralogical analysis was conducted to quantify the bulk mineral composition of the samples using the Rietveld refinement method [20,21]. The XRD results, summarized in Figure 6, show that quartz, carbonate minerals (i.e., calcite and dolomite), clays, and alkali-feldspar are the dominant mineral components in the Bakken Formation; however, there are more organic-rich clays than carbonates in the Upper and Lower Bakken Shales, while there is very little organic matter in the Middle Bakken.

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig. 3. Photomicrographs for selected Bakken samples.

Fig. 4. SEM mineral map of the Upper Bakken Shale.

5465

5466

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig. 5. SEM mineral map of Middle Bakken – laminated lithofacies (MB-L3). Black represents porosity.

Fig. 6. XRD analysis for the selected Bakken samples.

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Mercury capillary entry pressure testing was also done on samples of each of the major lithofacies to determine the pore throat size distribution. Figure 7 shows pore throat size histograms for a Lower Bakken Shale sample and two Middle Bakken samples. The pore throat size distributions are typical of what were observed in other Bakken samples and show that the matrix of the shales and Middle Bakken lithofacies are dominated by nanoscale pore throats. Previous work by Sorensen and others [8] showed that macro- and microscale fractures provide a majority of the naturally occurring fluid flow pathways in the most oil-productive zones of the Middle Bakken. The multiple-scale levels of porosity and permeability within the various lithofacies, combined with the effects of scale on fluid phase behavior in tight formations [11], serve to complicate the ability to predict CO2 permeation and oil mobilization rates within unconventional tight oil formations such as the Bakken.

Fig. 7. Pore throat size distribution based on mercury capillary entry pressure testing.

4. Advanced core characterization Knowledge of the bulk porosity, permeability, and mineralogy of the various Bakken lithofacies derived from conventional analytical methods provides the context to evaluate macro- to microscale formation attributes such as depositional environment. However, the dominance of low-end micro- to nanoscale pore throat sizes suggests that detailed knowledge of nanoscale pore throat networks is necessary to accurately predict fluid phase behavior. That knowledge, in turn, is needed to determine the mechanisms controlling CO 2 permeation and storage in the Bakken as well as attendant hydrocarbon mobilization that can lead to EOR. To improve upon the shortcomings of conventional analytical techniques to identify critical features of tight rocks[22,23], a combination of advanced imaging and microscopy techniques, including whole-core and micro-x-ray computerized tomography (CT and micro-CT) imaging, field emission (FE)–SEM, and focused ion beam (FIB)–SEM, were used to characterize samples in terms of several parameters, including naturally occurring fracture apertures, intensity, and orientation; pore throat mineralogy and connectivity; rock matrix characteristics; mineralogy; and organic content. Whole-core CT and micro-CT scanning, FE–SEM, and FIB–SEM were conducted on core samples from three wells representing the major Bakken lithofacies types. Fracture networks were first identified at the macroscale through visual core descriptions, as described above, and whole-core CT scanning. CT imaging provides a noninvasive way of generating detailed information on the occurrence of fractures, bedding planes, fossils, and bioturbation in core from shales and tight formations [23–25]. CT imaging using x-rays produced at different energy levels allows for continuous whole-core scans that can be calibrated to produce images of bulk density and photoelectric factor distribution, which can be used to interpret porosity, organic content, and mineralogy. Figure 8 shows an example of how whole-core CT scanning data can be processed to provide unique insight regarding the three-dimensional distribution of features that may affect fluid flow. In this case, Figure 8 shows the distribution of fossil worm burrows and brachiopods within a section of the Middle Bakken Burrowed Lithofacies (MB-L2).

5467

5468

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig. 8. CT scan of a 4-inch. diameter core of Middle Bakken Burrowed Lithofacies (MB-L2). The CT data have been processed to highlight the three-dimensional distribution of burrows and a brachiopod fossil that are within the core sample.

The visual descriptions and CT scanning results were used to select locations for further analysis. One-inch-diameter plugs were collected from those locations within the whole core and then evaluated using microCT scanning, optical microscopy of thin sections, SEM, and SEM–EDS imaging to better characterize macro- and, possibly, microscale features. The micro-CT scanning process was followed by FE–SEM analysis of ion-milled samples to determine porosity and organic matter volume fraction for multiple samples. Finally, FIB–SEM imaging techniques were used on portions of selected 1-inch plugs to characterize areas of interest identified in the initial FE– SEM results. The goal of the FE–SEM and FIB–SEM work was to evaluate connective fractures down to the smallest apertures that present technology can determine. FE–SEM is capable of 1,000,000× magnification with a spatial resolution of 1.2 nm with proper conditions and sample preparation [26]. This analytical technique was used for characterization of nano- and microscale features, such as determining fine-grained mineral (i.e., clay) occurrence and grain geometries, nanoscale pore visualization, micropore and pore throat mineralogy, and nano- and microfracture imaging and analysis (aperture, intensity, orientation). FIB–SEM is a technique that couples FE–SEM with FIB into a single instrument that can be used to mill away very thin layers of the sample surface, leaving a fresh, highly polished surface of the sample that can be imaged and analyzed. The images are then stacked to reconstruct a 3-D image of the sample area of interest for enhanced understanding of the properties of the tight rock sample such as fracture networks, porosity and pore-size distribution, connected versus isolated porosity, and distribution of organics and mineral phases. By using very high resolution imaging techniques that were available with these advanced methods, detailed knowledge of the ultrafine fractures and pore networks was determined.

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

5. Advanced characterization of organic-rich Bakken Shales The shale members of the Bakken Formation are known to be organic-rich, serving as the source rock for hydrocarbons in the Bakken petroleum system. The organic-rich nature and extremely low porosity of the shales were confirmed by the FE–SEM analysis of samples of both Upper and Lower Bakken Shales. An example of the data and image generated by the initial FE–SEM analysis on a shale sample, in this case a Lower Bakken Shale, is shown in Figure 9. The FE–SEM analysis for a sample of Lower Bakken Shale shows organic matter content ranging, in volume percent, between ca. 17% to ca. 25%, while porosity is <1%. It is important to note that these estimates suffer from gray-scale segmentation difficulties stemming from a continuum of gray scale between pores, organics, and minerals within the rock matrix. Thus these values should be considered to be semiquantitative. Figure 10 shows two FIB– SEM images of the same Lower Bakken Shale sample: one that uses gray scale to illustrate the distribution of organics, minerals, and porosity and another that uses colors to illustrate the distribution of connected and unconnected porosity as well as organics. Figure 10 not only shows that the Upper Bakken Shale is dominated by organics, as expected, but also appears to have more connected nanoscale porosity than unconnected. These trends were observed in all of the Upper and Lower Bakken Shale samples that were characterized using this method. This is reflected not only by the presence of nanoscale fracture networks that occur within kerogen following thermal maturation (as described by Loucks and others [27], but also by recent work [28] that demonstrates the nanoscale network of “pore” spaces that also occur within the kerogens and could serve as a pathway for fluid transport. This observation suggests that this nanoscale pore network may be the means by which CO 2 can permeate the Bakken shale and mobilize hydrocarbons, as observed by Hawthorne and others [29]. 6. Advanced characterization of Middle Bakken nonshale lithofacies The nonshale (clastic and carbonate) lithofacies of the Middle Bakken are also considered to be unconventionally tight and collectively represent the thickest unit of the formation. The total organic content (TOC) of the Middle Bakken tends to be very low (typically <1%), but oil saturations >50% are not uncommon and can be as high as 70% in the most productive areas. In addition to their relatively higher porosity and permeability (as compared to the shales), they tend to be geomechanically more brittle and are therefore more suitable targets for the application of hydraulic fracturing stimulation techniques which are necessary to economically produce the oil that is otherwise locked in the tight formation. Because of these attributes, the nonshale lithofacies of the Middle Bakken serve as the primary horizontal drilling target and productive reservoir rocks of the Bakken petroleum system. The lack of organic matter and relatively higher porosity of the various Middle Bakken lithofacies, as compared to the shales, were confirmed by the FE–SEM analyses. An example of the data and images generated by the initial FE– SEM analysis of a Middle Bakken sample, in this case a Laminated Lithofacies (MB-L3), is shown in Figure 11. The FE–SEM analysis for this sample of MB-L3 shows organic matter content no greater than 0.2%, with porosity ranging between ca. 1% to ca. 9%. Figure 12 shows two FIB–SEM images of the same MB-L3 sample: one that uses gray scale to illustrate the distribution of organics, minerals, and porosity and another that uses colors to illustrate the distribution of connected (blue) and unconnected (red) porosity as well as organics (green). This sample was selected for FIB–SEM analysis because examination of the FE–SEM images suggested that there was a clay-filled microfracture in the sample and it was thought that FIB-SEM analysis would shed insight into whether the porosity that could be observed between the clay particles within that microfracture was connected or unconnected. Figure 12 shows that the nanoscale porosity observed in the clay-filled microfracture is more connected (blue) than unconnected (red). This, in turn, suggests that the microfracture may serve as a fluid flow pathway for injected CO 2 and subsequently mobilized hydrocarbons. These trends were observed in many of the Middle Bakken samples that were characterized using this method, especially those in the laminated (MB-L3) and packstone (MB-L4) lithofacies. These observations indicate that the microfractures and the nanoscale pore network within them may be a substantial portion of the means by which CO2 can permeate the tight Middle Bakken lithofacies and mobilize hydrocarbons, as observed by Hawthorne and others [29].

5469

5470

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig. 9. Example of the data generated by FE–SEM analysis of a Lower Bakken Shale sample (Phi = porosity; OM = organic matter; PAOM = porosity of organic matter; HD = high-density material; ATR = apparent transformation ratio).

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig. 10. Example of FIB–SEM analysis of a Lower Bakken Shale sample. In the gray-scale image on the left, lighter areas represent mineral grains, darker areas represent organic matter (interpreted to be kerogen and bitumen), and black represents porosity. In the image on the right, green represents organic matter, red represents unconnected porosity, and blue represents connected porosity. The sample shown in these images was taken from within the area identified as 3D AOI 2D19 in the FE–SEM image in Figure 9.

7. CO2 permeation and oil extraction experiments using bakken rocks Experiments to evaluate CO2 permeation into and oil extraction from the Upper and Lower Bakken Shales and key Middle Bakken lithofacies were conducted on samples from several North Dakota wells, including samples from the same cores that were subjected to the advanced characterization program described above. The goal of that work was to expand on the efforts presented in Hawthorne and others [29], which looked at the effects of CO2 on hydrocarbon mobilization in Bakken shales and undifferentiated Middle Bakken samples. The efforts presented here made a point to develop permeation and extraction data for specific major lithofacies types within the Middle Bakken. Another hypothesis driving the work was that the advanced characterization data may help explain the permeation and extraction data generated by these experiments. Core plug samples provided by the North Dakota Geological Survey from a well in North Dakota (designated Well 24123) with Upper, Middle, and Lower Bakken units were submitted for routine core analysis, including porosity, permeability, and oil saturation measurements. The average reservoir properties for each unit can be found in Table 2. The parameters showed that the well penetrated through a representative Bakken reservoir in North Dakota based on the reservoir properties reported in Table 1. The rock properties of the Bakken in this well were also very similar to the properties of the rocks taken from the wells that provided samples for the advanced characterization, and therefore the insight gained from those activities can be directly applied to the Well 24123 rocks. Six plugs representing the Lower Bakken Shale and the Middle Bakken lithofacies MB-L1, MB-L2, and MB-L3 were selected to investigate the ability of CO2 to permeate these tight rocks and subsequently mobilize hydrocarbons.

5471

5472

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig. 11. Example of the data generated by FE–SEM analysis of a Middle Bakken Laminated Lithofacies (MB-L3) sample (Phi = porosity; OM = organic matter; PAOM = porosity of organic matter; HD = high-density material; ATR = apparent transformation ratio).

Unlike flow in conventional reservoirs, oil and gas transport through the tight Bakken matrix is likely via diffusion, which means it requires much more time to observe the oil recovery response than in traditional core-flooding experiments [7,29,30]. Therefore, small sample dimensions were used in order to observe the extraction response in a reasonable time. To mimic the oil recovery process in a real reservoir, the experimental pressure and temperature were set to reservoir conditions. Each core sample (1.1-cm diameter and ca. 4-cm long) was put inside the extraction vessel (1.5-cm diameter and 5.7-cm long), which was placed into an ISCO Model SFX-210 supercritical extractor. The extractor was thermostatically controlled at 230°F (110°C) during the experimental process. The pressure

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig.12. Images generated by FIB–SEM analysis of the smallest-scale Middle Bakken Laminated Lithofacies (MB-L3) sample from the figure shown above. The left image is an FE–SEM image of the MB-L3 sample that was subjected to ion milling and FIB–SEM analyses. The middle image is gray-scale image from the FIB–SEM image, with lighter areas representing grains and the black features representing porosity. The right image is the FIB–SEM data processed to show the three-dimensional distribution of organic matter (green), unconnected porosity (red), and connected porosity (blue). Table 2: Reservoir properties of Bakken units from Well 24123

Bakken Unit

Number of Plugs Analyzed

ɸavg, %

Kavg, mD

So_avg, %

UB

4

1.4

0.00075

62.2

MB-L5

3

4.4

0.031

54.3

MB-L4

15

4.4

0.0081

61.0

MB-L3

9

5.0

0.1035

62.0

MB-L2

15

5.3

0.0295

48.3

MB-L1

3

5.4

0.05

60.4

LB

3

3.8

0.00525

52.3

Note: ɸavg: average porosity; Kavg: average permeability; So_avg: average oil saturation.

throughout the entire system was maintained at 5000 psi by an ISCO Model 260D syringe pump operated in the constant pressure mode. At these temperature and pressure conditions, the CO2 is in a supercritical state. The supercritical CO2 was injected into the extraction vessel to fill the space around the core-and let CO2 penetrate into the core. (It should be noted that the CO2 was not forced through the core as in a conventional core flooding experiment. Rather, the core was essentially exposed to a “bath” of the supercritical CO2.) The recovered hydrocarbons were collected and measured at certain intervals. The detailed experimental procedure for CO 2 extraction was described by Jin and others [19]. Hawthorne and others [7] and Jin and others [19] showed that the extraction of oil from tight oil formation rocks cannot occur unless CO2 first permeates the rock sample. Therefore, the experiments use hydrocarbon recovery over time as a proxy for estimating the ability of CO2 to permeate the various Bakken lithofacies. Figure 13 shows the 24hour oil extraction performance of CO2 for the various Bakken samples. The results clearly demonstrate the ability of CO2 to permeate all of the samples and subsequently mobilize oil. In the Middle Bakken samples, more than 90% of the oil was recovered from the plugs in 24 hours, while over 70% of the hydrocarbons were removed from the Lower Bakken Shale samples. For all of the samples, most of the hydrocarbon mobilization occurred within the first 8 hours of the experiment, with between 85% and 95% removed from the Middle Bakken samples and between 50% and 60% being removed from the shales in that initial time period.

5473

5474

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Fig.13. 24-hour oil recovery using CO2 at 344 bar (110°C) for the Middle Bakken cores from Well 24123.

8. Plug-scale modeling of CO2 permeation and oil extraction in Bakken Shales Jin and others [9] used the same types of experimental data described above and statistical analysis based on a linear regression of those data to investigate the significance of likely parameters that could control the permeation of CO2 and mobility of oil within Bakken rocks. Specific parameters that were evaluated with respect to their sensitivity to oil recovery included porosity, permeability, TOC, pore throat radius, and water saturation. The results of that work showed that the two most important variables correlating with oil recovery in the Middle Bakken are pore throat radius and water saturation, while porosity had a minimum effect [9]. In the shale samples, TOC and pore throat radius were observed to have the greatest effect because kerogen is oil-wet and oil has a strong affinity for kerogen, which makes it difficult for CO2 to displace the hydrocarbon molecules [9]. At the same time, CO2 is known to readily dissolve into oil and have an affinity for absorbing into kerogen, suggesting that high TOC may translate into more material into which CO2 can readily be stored. While these phenomena may make EOR from the shales challenging, it suggests that the Bakken shales may have an affinity for absorbing CO 2, thereby enhancing their storage capacity for CO 2. Furthermore, the connected pore spaces observed in the FIB–SEM results described above indicate that pathways for CO2 permeation, although small in scale, do exist. The knowledge gained from the exercise described in Jin and others [9], combined with the observation reported in Hawthorne and others [7] that diffusion is a primary mechanism for the permeation of CO2 in Bakken rocks were the basis for modeling efforts to simulate CO 2 permeation and oil mobilization in plug-scale Bakken shales from Well 24123. Parameters that were considered in the modeling were bottomhole pressure (BHP), porosity (phi), permeability (K), irreducible water saturation (Swi), diffusion coefficient (Co_Diff), and maximum adsorption (Ads_max), as shown in Table 3. Figure 14 shows the simulation run that most closely matched the experimental results for the Lower Bakken Shale samples shown in Figure 13, and Table 3 presents the values for each of the variable control parameters that were used in the model for that simulation run.

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Table 3: Values for each of the variable control parameters that were used in the model for the simulation run shown in Figure 14.

Parameter

LB

Unit

BHP

4995 (34.44)

Psi (MPa)

ɸ

0.0611

fraction

K

0.0113

mD

Swi

0.11

fraction

Co_Diff

1.67E-5

cm2/s

Ads_max

0.1624 (0.358)

gmole/lb (gmole/kg)

Fig. 14. Results of simulation of oil recovery from CO2 permeation in a Lower Bakken Shale plug.

When comparing the results of the simulation to the actual CO2 permeation and oil extraction experimental data, it appears that although the general shapes of the oil recovery curves are similar, the modeling appears to underpredict both the rate of oil extraction and total amount of oil recovered from the shale plug in 24 hours. One explanation for this discrepancy may be that the model is not set up to account for the connected nature of the porosity that was observed in the FIB–SEM analyses of the shales. While the porosity of those shales is low, what little porosity there is appears to be fairly well connected, and those pathways, however small, may account for the higher CO2 permeation and oil mobility that is observed in the experiments.

5475

5476

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

9. Conclusions The obvious primary challenge of using the Bakken, or any tight oil formation, as a target for large-scale storage of CO2 and EOR is the characteristic low porosity and low permeability of the formation. Furthermore, the presence of complex, heterogeneous lithologies (including organic-rich, oil-saturated shales) complicates the ability to understand and predict the effectiveness of various mechanisms (e.g., diffusion, sorption, dissolution) that will be acting on CO2 mobility and storage. In an attempt to experimentally quantify the ability of CO2 to permeate tight Bakken rocks and mobilize oil from them, a set of laboratory experiments were conducted on small core plugs. While similar experiments on Bakken rocks had been conducted and presented by Hawthorne and others [7], it was thought that the generation of more CO 2 permeation and oil extraction data from samples obtained from other wells was necessary to confirm those earlier studies. Also, the new experiments were designed to specifically generate permeation and extraction data on the key lithofacies that were the subject of advanced characterization. The results of the CO2 permeation and oil extraction experimental tests clearly demonstrate, at the core plug scale, the ability of CO 2 to permeate both organic-rich shales and tight nonshale rocks and subsequently mobilize oil from those rocks. In fact, most of the hydrocarbon mobilization occurred within the first 8 hours of the experiment, with between 85% and 95% of the oil being removed from the Middle Bakken samples and between 50% and 60% being removed from the shales in that initial time period. The characterization efforts confirm that micro- to nanoscale pore throat sizes dominate the fluid flow pathways within both the Bakken shales and the nonshale lithofacies. This underscores the notion that detailed knowledge of nanoscale pore throat networks is necessary to accurately predict fluid phase behavior. That knowledge, in turn, is needed to determine the mechanisms controlling CO 2 permeation and storage in the Bakken, as well as attendant hydrocarbon mobilization that can lead to EOR. With respect to CO2, micro- to nanoscale fracture networks will be the primary means of its movement throughout the unstimulated areas of the reservoir, and the characteristics of those naturally occurring small-scale fracture systems will control the contact time that CO2 has with the oil in the reservoir. By using advanced characterization techniques, including very high resolution images generated by the FE–SEM and FIB–SEM studies, detailed knowledge of the microscale fractures and nanoscale pore networks was obtained. The advanced characterization efforts showed that the Bakken shales are dominated by organics, as expected, but also appear to have more connected nanoscale porosity than unconnected. The presence of nanoscale fracture networks that occur within kerogen following thermal maturation (as described by Loucks and others [27] and Bousige and others [28]) were confirmed, providing support to the concept that a nanoscale pore throat network occurs within the kerogens that could serve as a pathway for fluid transport within organic-rich shales. These nanoscale pore throat networks may be the means by which CO2 can permeate and mobilize hydrocarbons in the Bakken shale, as was observed by the CO2 permeation and oil extraction experiments. With respect to the nonshale rocks of the Middle Bakken, previous work suggests that much of the permeability within unstimulated Middle Bakken lithofacies (i.e., the matrix) is associated with microfractures [8]. However, conventional SEM images of Middle Bakken samples show that microfractures are often filled or partially filled with clays. FE–SEM analysis of a clay-filled microfracture showed the existence of pore spaces within the clays. The use of FIB–SEM analysis on that clay-filled microfracture showed that the nanoscale porosity observed in the clay filling is actually highly connected. This, in turn, suggests that despite the presence of clay, the microfracture may indeed serve as a fluid flow pathway for injected CO2 and subsequently mobilized hydrocarbons. These trends were observed in many of the Middle Bakken samples that were characterized using this method, especially those in the laminated (MB-L3) and packstone (MB-L4) lithofacies. These observations provide compelling evidence that the microfractures and the nanoscale pore network within them may be a substantial portion of the means by which CO 2 can permeate the tight Middle Bakken lithofacies and mobilize hydrocarbons, as observed by the CO 2 permeation and oil extraction experiments. The data generated by the activities presented above yield an improved understanding of the nature and distribution of nano-, micro- and macroscale pores and fracture networks. Results provide previously unavailable insight on nanoscale pore throat mineralogy and connectivity, rock matrix characteristics, mineralogy, and organic content. These efforts suggest molecular diffusion, TOC, and pore throat size perhaps exert more influence on CO 2 permeation and storage in tight oil formations than in conventional oil reservoirs. These findings may be used to support EOR scheme design and refine estimation methods for CO2 storage potential in tight oil formations.

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

Acknowledgements The material presented in this paper is based upon work supported by the U.S. Department of Energy National Energy Technology Laboratory under Award No. DE-FC26-08NT43291. Additional financial support for the project was provided by the North Dakota Industrial Commission through the North Dakota Oil and Gas Research Program, the Lignite Energy Council, Continental Resources, Marathon Oil Company, Hess, and ExxonMobil/XTO Energy. The authors want to acknowledge and thank Ingrain Digital Rock Lab for the advanced characterization images and its support of these efforts. The authors also thank Marathon Oil Company and the North Dakota Geological Survey for providing invaluable access to core samples, and Computer Modelling Group, Ltd., Kinder-Morgan, Schlumberger, and Baker Hughes for their support of the project.

Disclaimer This paper was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government, nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

References [1] Energy Information Administration. Technically recoverable shale oil and shale gas resources—an assessment of 137 shale formations in 41 countries outside the United States. Washington, D.C.: U.S. Department of Energy; 2013. www.eia.gov/analysis/studies/worldshalegas/pdf/ fullreport.pdf (accessed 2013). [2] Sorensen JA, Braunberger JR, Liu G, Smith SA, Klenner RCL, Steadman EN, Harju JA. CO 2 storage and utilization in tight hydrocarbonbearing formations—a case study of the Bakken Formation in the Williston Basin. Energy Procedia 2014; 63:7852–7860. [3] U.S. Department of Energy. Carbon sequestration atlas of the United States and Canada. U.S. Department of Energy Office of Fossil Energy; March 2007. www.precaution.org/lib/carbon_sequestration_atlas.070601.pdf (accessed September 9, 2014). [4] LeFever J, Helms L. Bakken Formation reserve estimates. Survey white paper, Bismarck, North Dakota: North Dakota Geological Survey; 2008 March, 6 p. [5] Continental Resources, Inc., 2012, Bakken and Three Forks: www.contres.com/operations/bakkenand-three-forks (accessed May 30, 2013). [6] Kurtoglu B, Sorensen J, Braunberger J, Smith S, Kazemi H. Geologic characterization of a Bakken reservoir for potential CO 2 EOR. Unconventional Resources Technology Conference, Denver, Colorado, August 12–14, 2013. URTeC 1619698. [7] Hawthorne SB, Gorecki CD, Sorensen JA, Steadman EN, Harju JA, Melzer S. Hydrocarbon mobilization mechanisms from Upper, Middle, and Lower Bakken reservoir rocks exposed to CO2. Paper presented at the SPE Unconventional Resources Conference – Canada, Society of Petroleum Engineers; 2013. SPE 167200-MS. [8] Sorensen JA, Braunberger JR, Liu G, Smith SA, Hawthorne SA, Steadman EN, Harju JA. Characterization and evaluation of the Bakken petroleum system for CO2 enhanced oil recovery. Paper presented at the SPE–AAPG–SEG Unconventional Resources Technology Conference, San Antonio, Texas, 20–22 July 2015. URTeC 2169871. [9] Jin L, Sorensen JA, Hawthorne SB, Smith SA, Bosshart NW, Burton-Kelly ME, Miller DJ, Grabanski CB, Harju JA. Improving oil transportability using CO2 in the Bakken system—a laboratorial investigation. Proceedings of the SPE International Conference & Exhibition on Formation Damage Control, SPE-178948, Lafayette, Louisiana, February 24–26, 2016. [10] Sorensen JA, Hamling JA. Historical Bakken test data provide critical insights on EOR in tight oil plays. The American Oil & Gas Reporter 2016, 59(2):55‒61. [11] Alharthy NS, Nguyen TN, Teklul TW, Kazemil H, Graves RM. Multiphase compositional modeling in small-scale pores of unconventional shale reservoirs. SPE Annual Technical Conference and Exhibition, New Orleans, Louisiana, September 30 – October 2, 2013. SPE 166306. [12] Kuhn PP, di Primio R, Hill R, Lawrence JR, Horsfield B. Three-dimensional modeling study of the low-permeability petroleum system of the Bakken Formation. AAPG Bulletin 2012, 96(10). [13] Nordeng SH. The Bakken petroleum system—an example of a continuous petroleum accumulation. North Dakota Department of Mineral Resources Newsletter 2009, 36(1):21–24.

5477

5478

James A. Sorensen et al. / Energy Procedia 114 (2017) 5460 – 5478

[14] Hayes MD, Holland Jr. FD. Conodonts of Bakken Formation (Devonian and Mississippian), Williston Basin, North Dakota. AAPG Bulletin 1983, 67(8). [15] LeFever JA. Horizontal drilling in the Williston Basin, United States and Canada. Rocky Mountain Association of Geologists, Denver, Colorado; 1992. [16] Fox JN, Martiniuk CD. Reservoir characteristics and petroleum potential of the Bakken Formation, southwest Manitoba. Journal of Canadian Petroleum Technologies 1994, 33(8). [17] Yan C, Luo G, Ehlig-Economides CA. Systematic study of Bakken well performance over three well-completion-design eras: Journal of Canadian Petroleum Technology 2014, 54:2. [18] Burrus J, Osadetz K, Wolf S, Doligez B, Visser K, Dearborn D. A two-dimensional regional basin model of Williston Basin hydrocarbon systems. AAPG Bulletin 1996, 80(2). [19] Jin L, Hawthorne SB, Sorensen JA, Kurz B, Pekot L, Smith SA, Bosshart NW, Azenkeng A, Gorecki CD, Harju JA. A systematic investigation of gas-based improved recovery technologies for the Bakken tight oil formation. Proceedings of the Unconventional Resources Technology Conference, San Antonio, Texas, August 1-3, 2016. URTeC-2433692. [20] Bish DL, Howard SA. Quantitative phase analysis using the Rietveld method. Journal of Applied Crystallography 1988, 21(2). [21] Mittemeijer EJ, Welzel U. Modern diffraction methods. Wiley Online Library, January 2013. [22] Josh M, Esteban L, Delle Piane C, Sarout J, Dewhurst DN, Clennell MB. Laboratory characterisation of shale properties. Journal of Petroleum Science and Engineering 2012, 88–89:107–124. [23] Erdman N, Drenzek N. Integrated preparation and imaging techniques or the microstructural and geochemical characterization of shale by scanning electron microscopy, in Camp W, Diaz E, Wawak B, eds. Electron microscopy of shale hydrocarbon reservoirs. AAPG Memoir 2013, 102:7–14. [24] Walls J, Armbruster M. Shale reservoir evaluation improved by dual energy x-ray CT imaging. Journal of Petroleum Technology 2012, November. [25] Wargo EA, Kotaka,T, Tabuchi Y, Kumbur EC. Comparison of focused ion beam versus nanoscale x-ray computed tomography for resolving 3-D microstructures of porous fuel cell materials. Journal of Power Sources 2013, 241:608–618. [26] JEOL-USA, Inc. JSM-7800f extreme-resolution analytical field emission SEM: www.jeolusa.com/PRODUCTS/ElectronOptics/ ScanningElectronMicroscopesSEM/FESEM/JSM7800F/tabid/869/Default.aspx#227380-specifications (accessed June 28, 2013). [27] Loucks RG, Reed RM, Ruppel SC, Jarvie DM. Morphology, genesis, and distribution of nanometer-scale pores in siliceous mudstones of the Mississippian Barnett Shale. Journal of Sedimentary Research 2009, 79:848–861. [28] Bousige C, Ghimbeu CM, Vix-Guterl C, Pomerantz AE, Suleimenova A, Vaughan G, Garbarino G, Feygenson M, Wildgruber C, Ulm FJ, Pellenq RJ-M, Coasne B. Realistic molecular model of kerogen’s nanostructure. Nature Materials 2016, 15:576–582. [29] Tovar FD, Eide O, Graue A, Schechter DS. Experimental investigation of enhanced recovery in unconventional liquid reservoirs using CO2—a look ahead to the future of unconventional EOR. Proceedings of the SPE Unconventional Resources Conference, The Woodlands, Texas, April 1–3, 2014. SPE 169022. [30] Wan T, Yang Y, Sheng JJ. Experimental and numerical study of the EOR potential in liquid-rich shales by cyclic gas injection. Journal of Unconventional Oil and Gas Resources 2015, 12.