During the Cretaceous, most of Texas was a shallow carbonate platform that underwent periodic episodes of siliciclastic (shale) deposition (McFarlan and Menes, 1991; Phelps et al., 2015). The platform to the west of the Marathon–Ouachita tectonic front (Pennsylvanian) has been termed the Comanche Platform, whereas the deeper Gulf Basin, which includes the producing region of south Texas, was located to the east of the ancient tectonic front (Goldhammer and Johnson, 2001). Relict Lower Cretaceous reef margins built up during deposition of the Sligo (Barremian) and Stuart City/Edwards formations (Albian) formed the eastern edge of the Upper Cretaceous Texas shelf, and possibly acted as an oceanographic sill during periods of sea-level transgression (Denne et al., 2016; Tinnin and Darmaoen, 2016). Subbasins of the Gulf Basin active during Eagle Ford deposition include the Maverick Basin in the southern part of the producing region and the East Texas and Brazos basins north of the San Marcos Arch (Figure 1). Siliciclastic input during Eagle Ford deposition was primarily derived from erosion of the Ouachita and Sabine uplifts northeast of the San Marcos Arch and a possible southern source south of the Maverick Basin, leaving the producing region between the Maverick Basin and the San Marcos Arch starved of siliciclastic sediments (Denne and Breyer, 2016). As most Eagle Ford rocks studied in outcrop were deposited on the proximal Comanche Platform or in the clay-rich East Texas Basin, they are not exact analogs for the distal rocks of the south Texas producing region (Denne et al., 2016).

The Eagle Ford Group was deposited during a transgressive–regressive cycle that represents one of the highest known episodes of eustatic sea level (Haq et al., 1988). The transgression began during the late Early Cenomanian and reached its climax near the Cenomanian–Turonian boundary, coincident with oceanic anoxic event 2 (OAE2) (Figure 2). During most of this transgression, the bottom waters of south Texas were oxygen poor (anoxic to euxinic), allowing for preservation of the abundant organic matter found in Lower Eagle Ford marls. Oxygen-rich bottom waters, possibly derived from the Arctic Ocean, flooded the Western Interior Seaway and the Texas shelf near the onset of OAE2 during the Late Cenomanian (Eldrett et al., 2014), marking the contact between the Lower and Upper Eagle Ford. Redox conditions during the subsequent highstand and regression of the Turonian never fully returned to the anoxic conditions of the Cenomanian, so in general organic-matter abundances are lower in rocks from the Upper Eagle Ford than in rocks from the Lower Eagle Ford (Denne et al., 2014). This, coupled with a basin-wide reduction of siliciclastic input and an increase in carbonate deposition during the highstand, produced a diagnostic decrease in the gamma-ray log associated with the Upper Eagle Ford (Denne and Breyer, 2016).

The Eagle Ford is dominated by two lithologies: marl (stone) and limestone. Some authors have chosen to use the Dunham (1962) carbonate classification scheme based on depositional textures observed in thin section for the Eagle Ford (e.g., Frébourg et al., 2016). However, the rocks of the south Texas Eagle Ford are mudrocks predominantly composed of calcareous plankton tests (coccoliths and planktic foraminifera) and were not derived from an active carbonate platform or reef. Therefore, they are classified here using a mudrock, not limestone, terminology. Unfortunately, the definition of a marl varies by author; Pettijohn (1949) defined marls as having 25–75% clay, mudstones with greater than 75% clay were classified as calcareous shale, whereas rocks with less than 25% clay were termed argillaceous limestones. Jackson et al. (1980), in the context of Deep Sea Drilling Project material, defined marl(stone)s as containing 30–60% calcite, and oozes/limestones as more than 60% calcite.

Our examination of Eagle Ford thin sections from south Texas cores identified three basic lithotypes within the marl-limestone spectrum: marls, foraminifera-rich marls, and recrystallized limestones. The marls are typically darkest in color and contained few, if any, laminae of winnowed foraminifera. Although not distinguishable in thin section, the mud matrix is thought to be composed of silt-to-clay–size coccolith tests, clay, organic matter, and silica (Driskill et al., 2013). The sand-grain–size planktic foraminifera found within the marls are usually small, biserial forms (Heterohelix) filled with organic matter, or larger, globose forms filled with kaolinite (Figure 3A). The marls are typically composed of 35–60% calcite and contain a range of 3–9% total organic carbon (TOC), with an average of 5% (Figures 4–6). Foraminifera-rich marls are typified by numerous laminae of winnowed globose, planktonic foraminifera, many of which are filled with calcite (Figure 3B). Thicker laminae may also contain abundant inoceramid prisms, phosphate clasts, and fish debris. As these foraminifera-rich laminae are frequently thick enough to be observed in core, the general color of core sections dominated by this lithotype is usually intermediate between the black marls and the white limestones. Foraminifera-rich marls generally contain more calcite (45–80%) and less TOC (1–6%, averaging 3.2%) than marls without winnowed laminae (Figures 4–6). Thin sections labeled as “recrystallized limestone” contain visible evidence of interparticle calcite cement (Figure 3C) and were generally taken from core sections that are much lighter in color than the marls. Although the limestones sometimes contain winnowed foraminifera and inoceramid laminae, they more commonly contain calcified radiolaria and calcispheres. A jump in calcite content is observed from the lower to upper 70% range that represented the onset of diagenesis. Samples below 75% calcite show no visible signs of recrystallization, whereas those above 75% always show evidence of diagenetic calcite. Nearly all of the thin sections with interparticle calcite cement are made up of greater than 75% calcite (Figures 4, 5), suggesting that 75% is an appropriate cutoff for dividing limestones and marls in the Eagle Ford. We therefore define “limestone” in the Eagle Ford as containing greater than 75% calcite and having undergone recrystallization (Figure 4).

As would be expected, the limestones contain less TOC than the marls, ranging from 0.5 to 2.5% with an average of 1.3% (Figure 6).

Previous studies of limestone–marl couplets in organic-rich mudrocks have postulated various mechanisms for the primary factors controlling their deposition. It has been posited by some researchers that limestones and marls were originally deposited as homogeneous mudrocks that were subsequently altered by diagenesis. Mechanisms for secondary limestone formation include dissolution of aragonite followed by precipitation as calcite within specific layers and hardground formation during periods of nondeposition (Loutit et al., 1988; Munnecke and Samtleben, 1996). A primary origin of limestone–marl couplets has been postulated by other researchers, who have invoked variations in siliciclastic input that either diluted or concentrated a constant supply of carbonate material, changes in redox conditions that produced differing levels of organic matter decomposition and diagenesis, dissolution because of changes in CCD levels, and variations in the productivity of carbonate-producing organisms (Prell and Hays, 1976; Berger, 1979).

Examination of cores from the subsurface of south Texas found that most limestones contain high abundances of calcispheres and calcified radiolaria, which are rare to absent in the marls (Denne et al., 2014, 2016). Intact fecal pellets and uncrushed planktic foraminifera tests suggest that calcite precipitation occurred prior to significant compaction of the sediments, whereas crushed foraminifera are common in the marls. The limestones are preferentially bioturbated in comparison to the surrounding laminated marls and generally contain lower abundances of the elemental redox proxies molybdenum and vanadium, indicating that bottom-water oxygen levels were higher during limestone deposition than during deposition of the adjacent marls. Based on these observations, it was hypothesized that the distal Eagle Ford limestones of south Texas formed during periods of high fertility and enhanced water-column mixing. Water-mass mixing increased bottom-water oxygenation, whereas possible upwelling along the relict Lower Cretaceous reef margins provided nutrients to the surface waters, which resulted in blooms of radiolaria and calcisphere-producing dinoflagellates. Oxidation of organic matter produced alkaline conditions near the sediment–water interface that were conducive to calcite precipitation, initially filling globose planktic foraminifera tests, calcifying pyritized radiolaria, and cementing lag deposits, and then progressing to interparticle calcite precipitation (Denne et al., 2016).

The permeability of tight rocks is typically measured on crushed core material (Luffel et al., 1993). This method cannot consider the laminated structure of the rock, which is destroyed in the crushing process. Sinha et al. (2013) have pointed out the limitations of this method, and Passey et al. (2010) documented the wide variation in permeability values reported by different laboratories utilizing this technique. Rosen et al. (2014) described the dual-pump system for measuring permeability on intact core plugs utilizing an injection pump operating at a constant rate with a back pump maintaining a constant pressure. This system can measure permeabilities below 1 nD using low viscosity, low-compressibility supercritical fluids that are miscible with residual core liquids, which mitigates the requirement to extract and resaturate a sample prior to measuring its permeability. With this system, Rosen et al. (2014) measured permeability on 36 core plugs from five Eagle Ford wells. They obtained x-ray diffraction data from plug end trims on 24 plugs from two of the wells (Table 1). Kosanke and Warren (2016) combined this data with SEM information to demonstrate that the limestones, in which the depositional texture, fabric, and structure have been destroyed by calcite replacement and recrystallization during diagenesis as described above, have the highest permeabilities because of their tendency to contain fractures. However, most of the hydrocarbons in the Eagle Ford reside in the marls, not in the limestones (Breyer et al., 2013, 2016; Driskill et al., 2013; Sahoo et al., 2013). It is important to note, however, that although the highest amounts of TOC are found within the marls, which average 4% TOC, stratigraphy also plays a role in TOC distribution, as Lower Eagle Ford marls typically contain higher amounts of TOC than Upper Eagle Ford marls (Figure 5).

Pores in mudrocks occur as intergranular pores between grains, intragranular pores within grains, and as nanosized organic pores within organic matter (Loucks et al., 2009, 2012), with the development of organic pores being related to both thermal maturity and organic matter type (Curtis et al., 2012). Reed and Ruppel (2012) showed that the Eagle Ford exhibits a range of nanometer- and micrometer-scale pores, with organic porosity in the Eagle Ford developed in higher-thermal maturity samples (R_o > 0.8). Kosanke and Warren (2016) conducted a SEM-based study that showed that the porosity in the Eagle Ford, which in their study ranged from 8 to 12%, was largely limited to the marls and that all of the intergranular pores are lined or filled with hydrocarbon (defined in their paper as either bitumen or pyrobitumen based on visual kerogen analysis and the results of solvent extraction). The authors documented pervasive round pores in pyrobitumen in samples from a well in the thermally mature (R_o = 1.45) part of the play, which presumably formed during gas generation (Figure 7). The pores they observed in the organic matter in samples from an area of lower thermal maturity (R_o = 0.62) were relatively small and slit shaped and likely formed during oil generation (Figure 8). The intergranular pores in an outcrop sample from an area with an even lower thermal maturity (estimated R_o = 0.4–0.55) were filled with a viscous hydrocarbon that had apparently migrated through the intergranular pore system (Figure 9). Kosanke and Warren’s (2016) study revealed that marls with higher matrix permeabilities (50–200 nD) had a greater volume of coarser-grained laminations containing transported, usually broken, foraminifera tests and inoceramid fragments but did not contain any fractures. Rosen et al. (2013) showed that thin sections of laminated marls having permeabilities greater than 200 nD typically had epoxy penetration along fractures, and although it is likely that the fractures were the result of coring, they nonetheless illustrate the presence of a connected pore system once fractures form. Diagenetically altered limestones with natural fractures lined with pyrobitumen have permeabilities on the order of 1000 nD or greater. These fractures may have been closed in the subsurface by the pyrobitumen and “reopened” by the fracturing of the pyrobitumen induced by the coring process. It is important to note that the permeability measured in the laboratory is the total permeability of the plug, so that the permeability measured on samples with fractures reflects the permeability of the fracture plus the permeability of the matrix. It is not possible to ascertain the permeability of the limestone matrix separately. However, in one of the plugs (measured permeability = 265 nD), fractures that are partially filled with calcite cement predominate, indicating that even partially filled fractures are permeable, more so than completely healed fractures.

Economic production from reservoirs with nanodarcy permeability can only be obtained by inducing and maintaining fractures with hydraulic stimulation. Natural fractures already present may be open or cemented and reactivated during hydraulic stimulation and contribute to production (e.g., Blanton, 1982; Gale et al., 2009; Lee et al., 2014). This can generate a hybrid fracture system with larger drainage area and fracture surface area to allow for crossflow from matrix to fractures. Rosen et al. (2014) described the Eagle Ford as a dual-porosity system, with the hydrocarbons stored in the pores of the marls feeding a network of progressively larger natural and induced fractures that carry the hydrocarbons to the wellbore (Figure 10). The interbedded marls and limestones form centimeter-scale brittle–ductile (or stiff-compliant) couplets like those described in Paleozoic gas shales by Slatt and Abousleiman (2011). These couplets influence hydraulic fracturing over a range of scales (Slatt and Abousleiman, 2011) and at the smallest scale may increase production by hosting complex near-wellbore fracture systems (Breyer et al., 2015, 2016). Kosanke and Rosen (2013) and Breyer et al. (2013) found that the geomechanical and acoustical properties of the interbedded marls and limestones comprising the Eagle Ford varied with composition and structure. In most cases, the Eagle Ford will be most productive (all other things being equal) when the right mixture of marl and limestone is present. Too much limestone lowers the storage capacity and hydrocarbon volume of the system, whereas too much marl reduces the complexity and drainage of the fracture system. The distribution of the limestones is also important (Breyer et al., 2015, 2016). Even if the percentage of limestone in two sections is equal, hydraulic stimulation will produce a more complex fracture network when the limestone is present as a series of thin interbeds rather than as a single thick limestone (Breyer et al., 2015). For example, Meek et al. (2013) described a 15-m (49-ft)-thick marl section interbedded with thin limestones as “the most productive and most brittle” part of the Eagle Ford. This interbedding of limestone and marl can be measured using limestone frequency—the number of limestone beds per unit thickness (Breyer et al., 2015, 2016).

Breyer et al. (2015) tied an increase in production along strike across several drilling areas to an increase in limestone frequency (Figure 11). In two wells 25 km (16 mi) apart along strike, the percentage of marl and limestone is essentially constant, but because the limestones become thinner to the northeast, the limestone frequency increases by 33%. Breyer et al. (2016) showed variation in production in wells on the same pad completed with the same treatment but landed in zones of differing limestone frequency (Figure 12). Production in these wells increases with limestone frequency. Also, in a multivariate analysis involving scores of engineering and geologic variables and more than 1000 wells, all measures of limestone frequency reduced to a single factor that correlated positively with production (Breyer et al., 2016). The above reservoir geometry leads to the possibility that highly oblique wellbores may prove more productive that strictly horizontal wells in that it can complete in multiple brittle–ductile couplets.

Natural fractures in the subsurface and in outcrop display self-similar development at multiple scales. A useful schematic distribution of such is given in Figure 13A. Some fractures are individual layers contained, whereas others occur throughout larger layering and eventually into tall and wide corridors of fractures, some related to faults or not. These may be related to faults of no noticeable offset that cut the entire section of interest. Figure 13B, C shows details of such distributions in outcrop. In general, a schematic fracture distribution model based on core observation can be merged with mechanical property logs to give a mechanical property explanation for the fracture distribution (Figure 14). In this figure, the rigidity modulus (G) curve represents variation in stiffness and strength for this section and shows that the observed fracture distribution based on core observations compares favorably with mechanical property variation and layering as seen in the log. This is true in the Eagle Ford as well.

We interpreted and quantified natural fractures in 19 cores from the Eagle Ford trend from De Witt and Gonzales Counties in the northeast to Maverick and Dimmit Counties to the southwest. The data cover the Eagle Ford section and adjacent partial Austin and Buda sections. The cores represent a mix of Eagle Ford Consortium cores and individual company cores. Determination of whether observed fractures are naturally occurring, drilling induced, or core handling artifacts was performed with the aide and guidance provided by Nelson (1993), Kulander et al. (1990), and Lorenz and Cooper (2018). Fracture intensity (FI) is measured in the core as the number of interpreted natural fractures per foot along the core. Therefore, they are a P₁₀ measure of FI. P₁₀, as shown by Golder and Associate, is a one-dimensional measure of fracture occurrence along a line, in this case parallel to the core axis. We observed many kinds of natural fractures in the Eagle Ford and adjacent cores. Three important types are shown in Figure 15A–C. These include bed-contained fractures particularly in the limestone layers, shear fractures related to tectonism, particularly evident in core from the Maverick Basin on the western edge of the Eagle Ford trend where Laramide compression occurred, and exposure and karst-related fractures in the uppermost Buda.

In terms of the distribution of natural fracture abundance, rocks respond to deformational stress and strain, like all materials, in a manner related to their inherent mechanical properties. The rocks’ mechanical properties are a function of its petrology or makeup of the rock. In order of importance, the controls of mechanical properties are bulk composition or mineral makeup, porosity, grain size, and fabric (Nelson, 2001). Ash beds do not occur in great enough thickness to play a role in decreasing fracture intensity. As a result, stratigraphic and sedimentology variations manifest themselves in mechanical property differences and subsequent diagenesis; in specific, natural fracturing. In general, rocks that are relatively stronger, stiffer, and potentially more brittle will contain more abundant natural fractures from a deformational event than those that are less so (Nelson, 2001, 2012). This mechanical-fracture distribution is shown in Figure 14. Therefore, by defining the mechanical property distribution in the subsurface, we may predict relative natural fracture abundance. This requires calibration of fracture distribution and mechanical property distribution. Historically, the mechanical properties used to do this were the elastic properties of Young’s modulus (E) and Poisson’s ratio (γ; PR). Rocks displaying relatively high E and low γ are relatively stronger, stiffer, and potentially more brittle and contain more abundant natural fractures from a single deformational event. The converse is also true. We can combine these two moduli into another one called rigidity modulus (G), also known as shear modulus. In general, higher G represents a relatively strong, stiff, and brittle material and usually contains more abundant natural fractures. These moduli can be measured in a static manner in the laboratory with deformational testing rigs or dynamically in the subsurface with appropriate sonic logs. Rigidity (G) can be calculated from E and γ curves using

or from shear sonic log data by

where ρ is rock density and V_s is shear wave velocity.

Plotting E, G, PR, and gamma ray (GR) values from log data for the Austin, Eagle Ford, and Buda section, we see varied stiffness and compositional behavior (Figure 16A, B). This shows that at a large scale the Buda is very stiff and rigid, whereas the Eagle Ford is relatively compliant, and more “shaley,” and the Austin is intermediate in behavior. In Figure 16C, in an expansion of a portion of Figure 16B, the Eagle Ford internal mechanical structure is depicted using the nomenclature of EF100 to EF500 used in this report. It is evident that the Eagle Ford itself is a mixture of mechanical properties and rock types that manifests itself in natural fracture intensity variations and compositional variations that change noticeably throughout the trend.

The various sedimentary rocks we deal with in the petroleum industry display a wide range in G values. Figure 17 shows relative G values for several productive unconventional shales and the carbonates that generally encase them. As seen on this figure, the shales are less stiff or more compliant than the encasing carbonates but also display a large variation among themselves (60% variation). Indeed, unconventional shale reservoirs are mechanically not the same at all. Also evident is the relation to encasing carbonates where the carbonates above and below are stronger and stiffer than the target unit being hydraulically stimulated.

Similar to the internal mechanical variation in the Eagle Ford is a quantitative stratigraphic correlation in G and FI for a similar unconventional reservoir, the Bakken Formation, constrained by 24 cores, Figure 18. Calibration equations such as shown in Figure 18 can be used to predict natural FI in wells with a shear sonic log but no core. In addition, these differences in calibration can be used to explain and predict details of stratigraphic provenance, layering, and susceptibility to chemical and physical diagenesis. Such predictions are facilitated by mapping G and FI along the trend in both the updip (c. 6000, depth) and downdip (c. 12,000, depth) positions.

When examining the compilation of G and GR curves from all the wells studied, three groups of curve types were defined: green, red, and blue. Upon mapping these log groupings, they were found to be geographically segregated along the Eagle Ford trend (Figure 19). Rigidity and FI data from these groups were mapped for all the Eagle Ford as well as for each of the five layers within the Eagle Ford. This figure shows general trends for all the Eagle Ford layers. Rock stiffness, as shown by G, increases from the downdip trend to the updip green trend. Also, within the downdip trend, G increases from the northeast (red trend) to the southwest (blue trend). These map variations are mimicked by FI with higher FI in updip wells (green trend) and an increase from northeast to southwest in the downdip trend (red trend to blue trend). Similar trends are seen for all of the Eagle Ford internal layers E100–EF500 (not shown here).

The Eagle Ford is largely composed of two interbedded rock types: coccolith-rich marls and recrystallized limestones. Most of the hydrocarbons in the Eagle Ford, regardless of thermal maturity, reside in the pore network of the marls as any primary porosity in the limestones has been filled with calcite cement. The mechanical properties of the Eagle Ford are a function of petrology and diagenesis: bulk composition/mineralogy, porosity, grain size, and fabric, with the hydrocarbons that are stored in the marls accessed via fracturing of the recrystallized limestones. The interbedded marls and limestones form centimeter-scale brittle–ductile (or stiff-compliant) couplets that influence hydraulic fracturing over a range of scales, and at the smallest scale may increase production by hosting complex near-wellbore fracture systems. Economic production from the Eagle Ford, which has nanodarcy permeability, can only be obtained by inducing and maintaining fractures with hydraulic stimulation. Natural fractures already present in the limestones may be open or cemented and reactivated during hydraulic stimulation and contribute to production, which can generate a hybrid fracture system with a larger drainage area and fracture surface area to allow for crossflow from the matrix to fractures. The Eagle Ford is therefore a dual-porosity system, with the marls feeding a network of progressively larger natural and induced fractures that carry hydrocarbons to the wellbore. In most cases, the Eagle Ford will be most productive when the “right” mixture of marl and limestone are present. Too much limestone lowers the storage capacity of the system, and too much marl reduces the complexity of the fracture system. The distribution of the limestones is important. Even if the percentage of limestone in two sections is equal, hydraulic stimulation will produce a more complex fracture network when the limestone is present as a series of thin interbeds rather than a single thick limestone. The interbedding of limestone and marl can be measured using limestone frequency—the number of limestone beds per unit thickness. Variation in production is observed in wells on the same pad completed with the same treatment but landed in zones of differing limestone frequency, with production in these wells increasing with limestone frequency. Once a location is selected where the reservoir geometry and layering are considered optimum, there is a possibility that highly oblique wellbores may prove more productive that strictly horizontal wells in that it can complete in multiple brittle–ductile couplets.