9: Outcrop-Based Reservoir Characterization and Modeling of an Upper Mississippian Mixed Carbonate–Siliciclastic Ramp, Northeastern Oklahoma
Published:January 01, 2019
- PDF LinkChapter PDF
Scott Shelley, G. Michael Grammer, Matthew J. Pranter, 2019. "Outcrop-Based Reservoir Characterization and Modeling of an Upper Mississippian Mixed Carbonate–Siliciclastic Ramp, Northeastern Oklahoma", Mississippian Reservoirs of the Midcontinent, G. Michael Grammer, Jay M. Gregg, James Puckette, Priyank Jaiswal, S. J. Mazzullo, Matthew J. Pranter, Robert H. Goldstein
Download citation file:
Production-scale variability in Mississippian reservoirs of the U.S. midcontinent is poorly understood, largely due to distant spacing of subsurface wells and the lack of outcrops that provide a 3-D distribution of potential reservoir facies. This study utilizes exposures of Upper Mississippian (Meramecian to Chesterian in age) lithofacies in a quarry to develop a 1st-order 3-D facies model at a production or enhanced-production scale (<40 ac [<16.2 ha]).
By utilizing photogrammetry to stitch and georeference high-resolution aerial photos, 3-D representations of outcropping walls and pavement were created at a submeter resolution and serve as valuable tools for the visualization of bed and facies relationships in 3-D space. Drone-based aerial and orthogonal photography was used to capture images and create 3-D models of inaccessible outcrop areas. These models were then imported as base surfaces to geostatistical reservoir modeling software (Petrel), in which they were integrated with petrographic and sequence-stratigraphic data to model facies and porosity relationships. Various algorithms and variogram lengths were tested and compared to high-resolution outcrop data to determine the most appropriate workflow for future subsurface modeling. The Petrel-based facies and porosity models illustrate the lateral and vertical variability that exists in outcrop while providing detailed approximations of subsurface reservoir heterogeneity.
The Mississippian limestone play is an unconventional (low permeability) oil and gas reservoir in the North American midcontinent that has historically produced large volumes of oil and gas and has seen an increase in horizontal drilling activity with advances in drilling and completion technology. The combination of shallow reservoir depths (3000–6000 ft [914–1829 m]) and low drilling costs ($3–3.5 million) has made the Mississippian limestone an appealing target for exploration (CoreLab, 2015). However, exploitation of the Mississippian limestone has proved challenging due to inconsistent hydrocarbon production, high water–oil ratios, and short economic lives of wells (CoreLab, 2015). Production inconsistencies are inferentially tied to internal heterogeneity within the Mississippian limestone with respect to facies, flow units, and regional factors like depositional environment, hydrocarbon migration, and structural features.
The commonly encountered meter-scale vertical and lateral heterogeneity within carbonates has provided a challenge for decades to academic and industry geologists who attempt to characterize these complex depositional systems (Goldhammer et al., 1993; Grammer et al., 1996; Kerans and Tinker, 1997; Pranter et al., 2005; Daniels, 2009; Donovan et al., 2016; Okyay et al., 2016). Mississippian reservoirs of the North American midcontinent provide another example of complex facies and flow unit relationships within carbonate strata (several chapters in this volume) while also offering the opportunity to study their association in the context of a mixed carbonate−siliciclastic system. Approaching reservoir characterization from an integrated perspective can help predict these relationships while also identifying controls on larger-scale depositional geometries and environments.
This study applies an outcrop-based approach, integrating stratigraphy, sedimentology, and modeling to evaluate Upper Mississippian reservoir analogs of the North American midcontinent, which exhibit significant heterogeneities at both reservoir and basin scales. This modern workflow is designed to translate data from outcrop to the subsurface via 3-D modeling of facies and porosity, while also incorporating depositional environment and sequence-stratigraphic relationships for correlation and comparison to time-equivalent strata within subsurface reservoirs.
Outcrop-derived models are commonly used as analogs for modeling the distribution of carbonate reservoirs and reservoir properties in the subsurface (Lucia et al., 1992; Kerans et al., 1994; Eisenberg et al., 1994; Grammer et al., 1996; Kerans and Tinker, 1997; Pranter et al., 2005; Janson et al., 2007; Amour et al., 2013; Pope, 2014). Outcrop analogs provide 2-D and occasionally 3-D views of facies and petrophysical relationships and offer researchers the opportunity to collect detailed information on diagenetic features that may be controlling factors for reservoir development. Outcrop analogs may be affected by depositional and diagenetic factors not present in their subsurface equivalents. However, given that subsurface reservoirs are often controlled by meter-scale facies and petrophysical heterogeneity, understanding the lateral and vertical relationships of potential reservoir facies within time-equivalent outcrops provides valuable data toward predicting hydrocarbon recoveries. By studying and modeling this small-scale variability in outcrop, reservoir quality, geometry, and communication of subsurface reservoirs can be more accurately predicted.
DESCRIPTION OF OUTCROP (QUARRY)
This study utilizes three outcrop walls within a quarry near Pryor Creek, Oklahoma (Figure 1). Each wall ranges from 365 to 700 ft (110–225 m) in length and from 65 to 85 ft (20–25 m) in height. The Meramecian- and Chesterian-aged Moorefield Formation makes up the lower portion of the outcrop and is overlain by the Chesterian-aged Hindsville Formation (Huffman, 1958; Figure 2). The east (Wall 3) and west (Wall 2) walls depict dip-oriented depositional profiles, whereas the southern wall (Wall 1) shows a strike-oriented profile. Blasting conducted in the last decade produced vertical outcrop faces characterized by dense fracturing and little to no weathering profile, providing researchers with obstacles to sample collection and bedding delineation. Inability to safely rappel or scale outcrop faces due to fracture-based instability necessitated the use of high-resolution ground-based and aerial photography in tandem with hand-sample collection to trace bedding and facies continuity across each wall. The area between these outcrops covers roughly 25 ac (10.1 ha), providing the opportunity to study facies and petrophysical relationships across three 2-D orthogonal sections at a scale that is between the well spacing of 40 ac (16.2 ha; distance between wells = 1320 ft [402 m]) and the meter-scale heterogeneity that typically characterizes and complicates carbonate reservoirs (Kerans, 1988; Eisenberg et al., 1994; Lucia et al., 2003).
The stratigraphic nomenclature used in this study is derived from Huffman (1958) and Mazzullo et al. (2013; Figure 2). The Moorefield Formation makes up roughly the lower 50 ft (15.2 m) of each wall. The formation is generally characterized as a mixed argillaceous and cherty limestone with intermixed siltstone and shaly beds in Oklahoma and Arkansas (Huffman, 1958). The Moorefield Formation has previously been interpreted as a lowstand wedge deposited on a carbonate ramp (Handford, 1986). At the study area, the Moorefield Formation is made up of burrowed mudstones and bedded cherts overlain by mixed clay-rich siltstones to calcareous siltstones. Conodont biostratigraphy by Godwin (2010, 2018) has shown that the Moorefield may transgress the Meramecian−Chesterian series time boundary, representing conformable deposition within both series.
The Hindsville Formation (Figure 2) makes up the upper 15–30 ft (4.6–9.2 m) of each wall. The Hindsville Formation was first described by Purdue and Miser (1916) as an often fossiliferous limestone interbedded with shale. The Hindsville Formation at the Pryor Quarry is characterized by skeletal wackestones to grainstones interbedded with soft, thinly bedded burrowed mudstones and silty peloidal packstones. Centimeter-scale trough cross-bedding localized within oolitic and skeletal grainstone beds is indicative of deposition under high-energy conditions. The Hindsville Formation is Chesterian in age (Huffman, 1958; Godwin, 2010).
SAMPLE COLLECTION AND CLASSIFICATION
Given the sheer and highly fractured nature of the Pryor Quarry outcrops, hand-sample collection was limited to one measured section on the flank of each wall. The data set gathered from the quarry included 109 hand samples, 95 thin sections, and 40 core plugs that were used to classify depositional facies and to determine porosity, permeability, and pore architecture. The variation in allochems and their relative abundance were visually estimated from thin sections using standard visual estimation charts. Mineralogy was determined using 35 representative XRD samples and confirmed through visual identification from thin sections. Porosity and permeability were measured from core plugs through routine core analysis by Weatherford Labs. In addition, porosity and permeability measurements were acquired at Oklahoma State University. Transects were supplemented with high-resolution bed tracing via Gigapan© photos and unmanned aerial vehicle (UAV) or drone-based photogrammetric images.
Geologists have historically used photography and 3-D visualization to document, image, and map geological features. Recent advances in technology combining photogrammetric techniques with complex computer-based algorithms have provided the opportunity for large-scale data collection, modeling, and processing in relatively short periods of time (Bemis et al., 2014). Using photogrammetry, 3-D topographic information can be extracted from 2-D photographs. Modern photogrammetric modeling software is able to recognize millions of individual surface points across multiple aerial and orthogonal photos to build 3-D surfaces that can be textured and colored based on pixel values from the source photos. The resulting models exhibit high accuracy and precision in terms of both structure (surface dip angles) and image resolution.
One challenge of photogrammetry is the acquisition of high-resolution orthogonal photos of inaccessible outcrops or structures. Sheer or unstable exposures provide difficulties for ground-based photography, whereas photos taken at high angles display significant distortion, which not only skews subject geometries but also inhibits accurate point matching of photos (Fitzgibbon, 2001). Recent innovations in civilian drone technology provide a fast, cost-effective method for acquiring high-resolution orthogonal photos of high-elevation subjects such as sheer outcrop walls. Drones offer the capability of taking both orthogonal and aerial photographs at multiple altitudes and distances. The greatest advantage of drone-based 3-D modeling over similar point cloud-based technology such as LiDAR (light detection and ranging) and TLS (terrestrial laser scanning) is speed. In this study, an area of approximately 25 ac (0.1 km2) was captured at a centimeter-scale resolution in less than 5 hours. More conventional techniques with poorer resolution and coverage may take weeks of data collection and processing to cover a similar study area (Bemis et al., 2014).
This study utilized a DJI Inspire 1 quadcopter drone, which was equipped with a 12 megapixel camera mounted on a stabilizing gimbal. Aerial photos (N = 950) were acquired at 33 ft (10 m), 67 ft (20m), and 100 ft (30m) altitudes to provide high pixel resolution and accurate point recognition and stitching. An additional 850 orthogonal photos were taken of the outcrop walls, resulting in a total of 1800 photos. The drone automatically georeferences every photo using on-board GPS, which significantly reduces processing times. Photos were imported into Agisoft Photoscan©, a photogrammetric modeling software that creates 3-D orthomosaic models. Agisoft generates a point cloud from the photos, which it uses to create a solid 3-D mesh. The mesh is then colored and textured based on the photo EXIF data to produce the final model. The model can be more accurately scaled by inputting ground control points (GCPs) or known distances. The model generated in this study exhibits a 2 in. (5 cm) pixel size (higher resolutions are achievable through greater computing power) and a structural accuracy of ± 3° dip.
The dense point cloud generated by the photogrammetric modeling was exported to Schlumberger Petrel© software as the base surface for geostatistical facies and porosity modeling.
Tracing of bedding geometry and continuity provides a valuable tool for delineating sequence boundaries, identifying depositional environments, and the influence of local or regional structural events. A Gigapan© Epic Pro camera system was used to stitch approximately 3000 individual photos together to create gigapixel panorama images of each wall with centimeter-scale resolution. These high-resolution images provide a means to identify sedimentary structures and bedding boundaries on inaccessible outcrop walls such as those in the Pryor Quarry.
Outcrop measurements allow for the identification of distinct depositional facies. These facies typically have similar geological and petrophysical characteristics, including porosity and permeability. A 1st-order approximation of subsurface fluid flow units can be created by recognizing the predictable patterns and characteristics of facies in outcrop (Kerans et al., 1994).
Facies 1: Calcareous Peloidal Siltstone–Packstone
The calcareous peloidal siltstone−packstone facies is characterized by tan-colored, massive, meter-scale bedding with scattered oxidized pyrite. Coarse silt to very fine sand-size quartz content ranges from 20% to 70% and is typically inversely proportional to peloid content, which can range from 5% to 30%. Facies 1 contains abundant calcite and silica cements (20–40%) as well as common benthic foraminifera tests and other fragments of skeletal material (Figure 3).
Multiple sedimentary features described from the Pryor Creek outcrops are unique to Facies 1. Dense Palaeophycus and Skolithos burrows and centimeter-scale symmetrical ripples present at the top of the Moorefield Formation indicate shallow-marine deposition under high-energy conditions (Dott and Bourgeouis, 1982; Gaillard and Racheboeuf, 2006). Localized millimeter-scale mud layers indicate occasional storm influence. The dip-oriented Wall 3 exhibits horizontal beds that grade laterally into meter-scale imbricated to shingled bed geometries. These shingled beds display a progradational geometry, partially filling a hemispherical break in bedding that may represent a storm channel cut. The cut is further filled by thinly bedded (0.25–2.0 cm thick), horizontal siltstones. Facies 1 is interpreted to have been deposited in the lower shoreface to ramp−crest environment. The meter-scale shingled beds at the top of the Moorefield Formation may have been deposited as a shallow subtidal bar cut by a storm channel (Figure 4). Horizontal beds represent the main body of the bar, whereas shingled beds reflect current-driven progradational filling of the storm channel, similar to geometries observed in modern offshore Texas environments (McCubbin, 1981). More peloidal-rich beds of Facies 1 may represent deposition in a low-energy lagoon created by the presence of the offshore bar. Facies 1 is more extensively cemented with calcite and silica than the other siliciclastic facies, but still exhibits an average porosity of 3% and permeability of 0.2 µD associated with fractures and intergranular pores.
Facies 2: Calcareous Quartz Siltstone
Facies 2 is a calcite- and silica-cemented quartz siltstone (Figure 3). Silt-size, angular quartz composes 30–50% of Facies 2, along with 40–60% calcite and silica cement, 1–5% peloids, <1% authigenic muscovite, and <1% pyrite. Facies 2 is distinguished from Facies 3 by a smaller proportion (<10%) or lack of clay minerals and a corresponding increase in calcite cement.
Significant postdepositional cementation is present in the form of porosity-occluding calcite and silica cement, the latter commonly occurring as quartz overgrowths. Facies 2 lacks recognizable sedimentary structures, skeletal content, and bioturbation, indicating deposition in the mid to outer ramp or lower shoreface to offshore environment in waters that may have experienced periods of increased circulation or fluctuations in detrital input. The relative lack of clays may indicate continued transgression, as increasing water circulation results in higher dispersion of clays (Carson and Arcaro, 1983). Fracture and intergranular pores contribute to an average porosity of 5% and permeability of 2 µD.
Facies 3: Clay-Rich Quartz Siltstone
Facies 3 is a thinly bedded (0.5–2.0 cm) clay- and calcite-cemented quartz siltstone to very fine-grained sandstone (Figure 3). Constituents include 40% very fine sand- to coarse silt-size angular quartz grains, 35% clay cement, 5–10% calcite cement, 5–10% peloids, 5% plagioclase, <1% pyrite, and scattered brachiopod fragments that most often occur in what are interpreted to be transgressive lags.
Facies 3 is interpreted to have been deposited under deep marine conditions, likely during late transgression. Bedding ranges from thin and fissile to massive. Similar to Facies 2, the lack of recognizable cross-bedding, bioturbation, or abundant skeletal content indicates low-energy deposition in waters that were not conducive to normal-marine fauna. Angular, moderately sorted quartz sand and silt reflects initial deposition likely via eolian processes (Dodd et al., 1993), whereas feldspar and clay constituents may have been derived from fluvial or deltaic processes. The presence of brachiopod lag beds (0.33−2.5 ft [0.1−0.75 m] in thickness) and rare carbonate intraclasts may represent incorporation of underlying carbonate-rich sediments during particularly rapid transgression (Handford and Loucks, 1993). Fractures and intergranular pores contribute to an average porosity of 5% and permeability of 3 µD.
Facies 4: Burrowed Mudstone−Wackestone
Facies 4 is a burrowed mudstone to wackestone with interbedded chert-rich layers (Figure 3). Muddy layers contain approximately 85% carbonate mud, 15% quartz silt, and <1% pyrite. Cherty layers range up to 50% microcrystalline quartz, 45–50% carbonate mud, and 1–5% pyrite. Chert content may be related to abundant sponge spicules that are observable in thin section (Facies 4 only) and are commonly partially replaced by pyrite. Blocky calcite, chert, and chalcedony cements occur within fractures, which are common in the chert-rich layers.
Facies 4 was likely deposited in relatively deep water under low-energy conditions based on the high carbonate mud content, extensive burrowing, and spicule-dominant fauna observed in both thin section and outcrop. The presence of sponge spicule-dominated fauna may be an indicator of restricted, deep-water deposition (Goldhammer et al., 1991). Alternating mud- and chert-rich layers suggest spicules were cyclicly deposited due to allocyclic or autocyclic processes, and then postdepositionally altered to chert through spicule or mixing-zone certification (Knauth, 1979; Rogers et al., 1996; Neuweiler et al., 2014). Facies 4 contains no visible pores, with micro- to nanoscale pores likely contributing to its average porosity of 2%. The average permeability value of Facies 4 was below the resolution of the core analysis (<0.1 µD).
Facies 5: Skeletal Wackestone
Facies 5 is a skeletal wackestone composed of 45% carbonate mud, 35% skeletal fragments, 10% blocky calcite cement, and 10% quartz silt. The skeletal content includes a normal-marine faunal assemblage of (in order of abundance): brachiopods, crinoids, bryozoans, echinoderms, and trilobites (Figure 3). Facies 5 typically occurs in decimeter-scale beds without obvious sedimentary structures but is also observed with centimeter-scale burrow fills within fissile siltstone beds underlying grainier carbonate beds, indicating depositional conditions that were conducive to large burrowing organisms.
The normal-marine faunal assemblage indicates open marine conditions in intermediate water depths and energy regimes. Facies 5 was likely deposited in a middle-ramp environment, seaward of the ramp crest. Facies 5 exhibits moldic porosity possibly related to subaerial exposure, with an average porosity of 2%. The average permeability value of Facies 5 was below the resolution of the core analysis (<0.1 µD).
Facies 6: Skeletal Packstone–Grainstone
Facies 6 consists of both skeletal and nonskeletal carbonate grains within calcite cement and carbonate mud matrix. Fabrics vary from a mud-lean packstone to grainstone cemented with blocky calcite. The faunal assemblage within Facies 6 is similar to that of Facies 5, containing bryozoans, crinoids, brachiopods, benthic foraminifera, trilobites, and echinoderms. Facies 6 differs from Facies 5 in that it contains well-preserved gastropods and radial ooids (Figure 3). In outcrop, Facies 6 exhibits centimeter-scale bidirectional cross-bedding.
Low mud content, allochem type, and cross-bedding indicate that Facies 6 was deposited in shallow, high-energy, and likely tidally influenced waters, in a skeletal or oolitic shoal environment that was more proximal relative to Facies 5 deposition. Similar to Facies 5, Facies 6 exhibits minor moldic porosity possibly related to subaerial exposure, with an average porosity of 1%. The average permeability value of Facies 6 was below the resolution of the core analysis (<0.1 µD).
The sequence-stratigraphic architecture represented in this study includes a twofold hierarchy of depositional sequences that represent cyclical changes in relative sea level and corresponding changes in depositional conditions (Figure 5). Biostratigraphic data from Godwin (2010; 2018) were used to constrain the age of these outcrops to a resolution of only 4–6 million years, thus complicating any attempt to temporally resolve higher frequency cyclicity, which occurs on the order of 400 k.y. or less. The possible unconformity between the Hindsville and Moorefield formations may lead to a more complicated stratigraphic architecture than presented here, but was not considered when developing the generalized sequence-stratigraphic framework used as a foundation for reservoir modeling.
The Pryor Creek outcrops contain two 3rd-order sequences and three to four 4th-order sequences. The 3rd-order sequence at the base of the outcrop is incomplete, recording only the regressive phase of an idealized sequence. Sequence-stratigraphic variability between walls is due to differences in erosion at the top of the outcrop exposures. The quarry floor was mined along an unconformity due to specific mineralogical requirements for Oklahoma aggregate production, which provides some confidence that the lowermost and uppermost sequences correlated between outcrop walls are bounded by geologically time-equivalent surfaces.
Sequence boundaries are characterized by either a significant increase in water depth from inner- to mid-ramp facies below the surface to outer-ramp facies above or by geologically significant surfaces such as lag deposits or rip-up surfaces. The Moorefield Formation contains a high percentage of siliciclastic facies relative to the Hindsville Formation.
The transgressive phase of an idealized sequence is characterized by siliciclastic-dominated facies (1–3), deposited as sea-level rises. Increasing water depth and currents rework eolian quartz silt, fine sands, and lowstand clays. The initial transgression is commonly marked by rip-up of underlying carbonates or shell lags, most noticeably where siliciclastic facies overly Facies 4. Facies 4 represents the deepest water deposition. As carbonates begin to dominate the system, syndepositional cementation may have been limited due to water depths and clay content. Facies 4–6 represent a typical regressive carbonate sequence, culminating in the skeletal and oolitic inner-ramp shoals of Facies 6. Repeated transgressions and regressions of sea level create both vertical and lateral facies heterogeneities that compartmentalize reservoirs in carbonate systems (Grammer et al., 1996).
Recent work has suggested that the likely depositional setting of the Mississippian in the midcontinent region is a distally steepened ramp, as evidenced by prograding low-angle clinoforms and debris flow-supported outrunner blocks (Boardman et al., 2010; Childress and Grammer, 2015, 2018). The observations made from the Pryor Quarry are consistent with this general model; however, the introduction of a source for terrestrial eolian silt and cyclic siliciclastic−carbonate depositional sequences within the system add another layer of complexity to the sedimentologic record of the ramp system (Figure 6). Siliciclastic input into the basin likely occurred through multiple sources. The well-sorted, angular quartz silt that is present in various amounts in all facies likely represents eolian processes that were active throughout the interval of deposition recorded at the Pryor Quarry. In addition to eolian quartz silt, fluvial or deltaic derived siliciclastic sediments may have contributed to the feldspar and clay content of Facies 2 and 3 (Handford et al., 2014). Siliciclastic input into the basin likely suppressed carbonate sedimentation, creating muddier water conditions that would not be conducive to carbonate-producing organisms that rely on photosynthesis or filter feeding (Bathurst, 1975).
The rocks in the Pryor Quarry represent deposition ranging from the ramp crest, as indicated by the presence of the interpreted bar and shoal geometries of Facies 1 and 6, respectively, to the outer ramp, evidenced by the increased clay content and lack of normal marine fauna of Facies 2 and 3. The siliciclastic bar of Facies 1 likely formed an antecedent topographical high for Facies 6 carbonate shoal development (Figure 7). This is a significant line of evidence to support the idea that sea level forms the main control on the overall shift from siliciclastic to carbonate sedimentation.
RESERVOIR AND PORE CHARACTERIZATION
The classic carbonate ramp model suggests highest depositional energy located nearest to shoreline, barring the development of a ramp crest (Ahr, 1973; Burchette and Wright, 1992). A distally steepened ramp geometry should follow this same pattern, generally characterized by grainy, initially high-porosity facies in proximal positions and muddier, low-porosity facies in distal positions. As an example, the mixed carbonate−siliciclastic systems of the Paradox Basin (Pennsylvanian) and Leonardian deposits of the Permian Basin exhibit the highest porosity intervals within shallower water carbonate facies, whereas the siliciclastic facies often act as seals (Mazzullo and Reid, 1989; Grammer et al., 1996; Ruppel and Ward, 2013). Patterns of porosity development within the Pryor Quarry strata, however, deviate significantly from this model. Porosity within the shallowest carbonate facies, which likely contained abundant primary porosity (Facies 5 and 6), is occluded by calcite and silica cements. These strata now act as vertical seals, whereas deep-water siliciclastic facies (Facies 2 and 3) represent the highest porosity reservoir facies (Figure 8) within the Pryor Quarry samples.
The dominant pore types within the siliciclastic reservoir facies are intergranular, intragranular-moldic, and fracture pores (Figure 9). All siliciclastic samples with visible porosity show signs of dissolution enhancement, likely as a result of feldspar dissolution (Figure 9C). Despite low permeability values, pores often appear connected by dissolution-enhanced microfractures.
Individual pores, pore networks, and pore fills were evaluated using a scanning electron microscope to better characterize the Facies 3 reservoir potential (Figure 10). Although it is possible that pore geometries were affected by surface diagenesis, the quarry walls have been exposed for less than 10 years, and efforts were made to collect fresh samples 4–6 inches (10–15 cm) into the rock face. Although mesopores (62.5 μm to 4 mm) were observed, the majority of the pores evaluated fall within the micropore (1–62.5 μm) to nanopore (1 nm to 1 μm) range. Individual pores are typically coated or filled with clay material, typically montmorillonite or illite-smectite mixed-layer clay. Pore throats tend to be narrow, often bridged or occluded by clay minerals; however, many pores appear to be fairly well-connected via a network of microfractures, with apertures ranging from 1 to 3 μm. No calcite crystal growth was observed in any sample of Facies 3. Porosity is inversely proportional to the percentage of calcite cement within a sample. Facies 3, the clay-rich quartz siltstone, contains the smallest calcite percentage and highest clay content of any facies, while also exhibiting the highest porosity. The majority of clays within Facies 3 are likely depositional, though some clays may be authigenic, sourced from the dissolution of feldspar grains. Clay coatings are frequently cited as inhibitors of quartz overgrowth cements in deep sandstone reservoirs, preventing the nucleation of cements on grain faces and preserving porosity (Pittman and Lumsden, 1968; McBride, 1985; Dixon et al., 1989). Inhibition of calcite cement by clays has been less commonly documented, but likely acts through a similar process, preventing calcite nucleation on grains (Buxton and Sibley, 1981; Moraes and De Ros, 1990).
Although clay minerals seem to preserve feldspar dissolution-derived porosity, they also act to severely reduce permeability (Moraes and De Ros, 1990). The primary pores are largely kept open, but pore throats and pore-connecting fractures are generally occluded, resulting in the extremely low range of permeability values recorded from the Pryor Quarry samples (0.1–7 µD).
ANALOG SUBSURFACE RESERVOIR–SHAFFER 1-23 CORE, STACK PLAY
The siliciclastic facies within the Pryor Quarry serve as a lithological analog for the STACK (Sooner trend, Anadarko [Basin], Canadian and Kingfisher [counties]) play of Oklahoma, which consists of stacked reservoirs of the Woodford Shale and Mississippian mixed carbonate–siliciclastic sections. The STACK play is among the more active plays in the southern midcontinent with completions targeting the Mississippi limestone, targeting the thick intervals of silt-rich siliciclastics and carbonates.
To compare the Pryor Quarry Facies 3 to typical Mississippian STACK reservoir lithologies, thin section photomicrographs from the Shaffer 1-23 core (Figure 1) were examined and compared to outcrop samples (Figure 11). The Shaffer 1-23 is located in Blaine County, Oklahoma, on the north flank of the STACK Play. Both samples contain similar constituents and matrix, as well as similar porosity types and values. Lithologies similar to Facies 2 and 3 also occur within the Shaffer 1-23, but, similar to what is observed in outcrop, greater calcite cementation leads to lower porosity content. Higher depositional clay content may inhibit calcite cementation in the Facies 3 reservoir type, and preserve void space. Pryor Quarry and Shaffer 1-23 clay minerals seem to have similar morphologies, resulting in the development of micro- and nanopores between pore coating clays. Shaffer pores exhibit pore throats mostly occluded by clays, but visually appear to be characterized by greater overall connectivity than similar pore systems from Pryor Quarry samples. The Shaffer 1-23 lies in a distal position relative to the Pryor Quarry and does not contain clean carbonate facies. The Shaffer 1-23 and Pryor Quarry are not direct stratigraphic equivalents and may demonstrate differences in burial depths and pressures.
OUTCROP FLOW UNIT MODELING
Although there are numerous examples of large-scale outcrop-based modeling studies in carbonates that cover kilometers of exposure (see for example Goldhammer et al., 1993; Kerans et al., 1994; Jennings et al., 2000; Janson et al., 2007), the majority of Mississippian-aged rocks exposed in the midcontinent region consist of small-scale (<1000 ft [<300 m] in length) 2-D road cuts. The outcrop exposures at the Pryor Quarry provide a unique opportunity to study the facies and petrophysical relationships of midcontinent Mississippian carbonates with orthogonal 2-D perspectives (Figure 12). The quarry area provides a basis for modeling rock characteristics over an area similar to that of an average well spacing in a carbonate reservoir (40 ac [16.2 ha]; distance between wells = 1320 ft [402 m]). The 3-D model grid consists of 8.2 million cells. Each cell is 6.6 ft (2 m) in length, 6.6 ft (2 m) in width, and 1.6 ft (0.5 m) in height.
Outcrop measured sections were used as pseudowells for the modeling and include facies and porosity logs. Two additional pseudowells were created through Gigapan-based bed tracing and used as a constraint during facies modeling (Figure 13). By assuming that facies would remain consistent within a continuous bed at the relatively small scale of each outcrop wall, one additional measured section and pseudowell was created on the southern flank of Wall 3 and one between Walls 1 and 2. These additional pseudowells were only used for facies modeling.
The reservoir model utilized the photogrammetric model as the base surface, as well as georeferenced sequence–boundary points collected using a GPS-enabled laser rangefinder (Figure 14). Convergent interpolation was used to interpolate boundary structure between data points. Four zones representing the four 4th-order sequences observed in outcrop were created using proportional layering, which most closely reflects the outcrop geometry. The model is constrained by the sequence-stratigraphic framework so that each sequence is modeled individually. This ensures that the algorithm only takes into consideration strata that were deposited over the same (interpreted) time interval.
Facies Modeling and Spatial Distribution
Multiple iterations of both the facies and porosity models were created using various algorithms, variogram ranges, and nugget values. Each iteration of the models was compared to the facies and porosity values observed in the outcrop walls. By comparing the modeled facies relationships and geometries to those actually present in the outcrop, modeling parameters could be iteratively modified to best fit the outcrop observations. Truncated-Gaussian simulation, sequential-indicator simulation, and indicator-kriging algorithms were used to explore their utility for facies modeling. Indicator kriging most accurately reflected the relationships present in outcrop, producing laterally continuous beds with gradational facies contacts, both laterally and vertically (Figure 15).
Horizontal and vertical variogram ranges varied by sequence to reflect the continuity of facies in outcrop walls. A 1640 ft (500 m) horizontal variogram range and 2 ft (0.6 m) vertical range most accurately reflected the facies data obtained from outcrop. A nugget of 0.0001 was used.
Porosity Modeling and Spatial Distribution
Due to their susceptibility to diagenetic alteration, porosity values within carbonate rocks are variable, often at submeter scales (Moore, 1989; Eisenberg et al., 1994; Kerans et al., 1994; James and Jones, 2015). Calcite cement occludes much of the porosity within the carbonates. Because the siliciclastic facies are less heterogeneous than the carbonates, a smaller nugget (0.01) more accurately models porosity within them (Figure 15). Horizontal variogram ranges of 330 ft (100 m) and a 1 ft (0.3 m) vertical variogram range produced a porosity model with spatial continuity that accurately reflects the geometries observed in outcrop.
Gaussian random function simulation (GRFS) was used to create a 3-D porosity model that was constrained to the sequence-stratigraphic framework, facies model, porosity pseudowells, and general outcrop observations. The highest porosity is present in Facies 2 and 3, and lowest porosity is associated with Facies 1 and 4–6. Facies 1 porosity distribution is variable due to its proximity to large belts of Facies 2 and 3 high-porosity zones. The 3-D porosity model honors the observed heterogeneity associated with zones that exhibit varying degrees of cement, fractures, and dissolution-enhanced pore space.
The porosity model closely matches the observed variability between facies and outcrop walls. The porosity model illustrates the scale of vertical and lateral heterogeneity that is possible in lithologically equivalent subsurface reservoirs. Reservoir compartmentalization within subsurface analogs such as the STACK Play is likely occurring vertically on a meter scale and horizontally at a subreservoir development scale. Petrophysical characteristics vary both between and within facies controlled by both depositional and diagenetic processes.
When extrapolating this model to the subsurface and to a field or basin scale, it is important to take the relationship of porosity and sequence stratigraphy into consideration. Although variability will exist within facies and sequences, the best reservoir quality will most likely be located at the base of sequences, which can be recognized in both core and well logs. The porosity model shows an excellent correlation between transgressive facies and high-porosity zones. By recognizing the relationship of reservoir quality to sequence stratigraphy, the vertical and lateral compartmentalization that exists within strata and between wells can be more accurately predicted, as can volumetric calculations that predict overall recoveries.
This study integrates multiple datasets and techniques to identify likely controls on facies deposition and distribution, depositional geometries, and reservoir development within the Mississippian Moorefield and Hindsville formations in outcrop. Thin sections and core plugs from outcrop were used to identify distinct facies and their petrophysical characteristics. An idealized vertical facies stacking pattern and sedimentologically significant surfaces were incorporated to create a sequence-stratigraphic framework of the three outcropping walls in a quarry. These data were then integrated with gigapixel and drone-based orthogonal photography to develop 3-D geostatistical models of outcrop facies and porosity. The key conclusions of this study are as follows:
Three carbonate and three siliciclastic depositional facies were identified from thin section and hand sample, reflecting changes in dominant sedimentation type.
Siliciclastic sedimentation was likely dominant during lowstands and transgressions of sea level, whereas carbonate sedimentation was likely dominant during highstand. Two partial 3rd-order sequences, constrained by published conodont biostratigraphy, were identified using a vertical facies stacking pattern. Three to four 4th-order high-frequency sequences are superimposed on the 3rd-order sequences.
Sequence stratigraphy can be used as a predictive tool within the study area. The best reservoir quality in the midcontinent Meramecian strata will likely be found at the base of sequences and can be predicted through the use of sequence stratigraphy and the recognition of the Facies 1 log signature in the subsurface (high gamma ray response).
Deposition of the Moorefield and Hindsville formations likely occurred near the ramp crest of a distally steepened ramp. This is inferred from the presence of a siliciclastic bar and carbonate shoal, as well as the close juxtaposition of deep-water and shallow-water facies.
Reservoir quality hinges on the effects of diagenetic calcite cementation, which affects every facies except Facies 3, in which depositional and authigenic clay minerals likely inhibit calcite cements. Facies 3 provides the highest reservoir quality, whereas the highly cemented carbonate facies (4–6) act as the primary seals to vertical fluid flow.
Drone photogrammetry provides a fast, simple method to capture outcrop geometries in three dimensions, then directly import them to modeling software.
An indicator-kriging algorithm using a horizontal variogram range of 500 m (1640 ft), vertical range of 0.6 m (2 ft), and nugget of 0.0001 most accurately recreates the facies relationships present at the Pryor Quarry. Both sequential Gaussian simulation and Gaussian random function simulation create reasonable approximations of porosity relationships. A horizontal variogram range of 100 m (328 ft), vertical range of 0.3 m (1 ft), and nugget of 0.01 most accurately recreate the porosity data from the Pryor Quarry.
The Shaffer 1-23 core provides a subsurface lithological analog that ties the siliciclastic lithologies seen at the Pryor Quarry to those observed in the STACK Play of the Anadarko Basin, indicating that siliciclastic deposition was widespread across Oklahoma at times during the Mississippian.
Exploration efforts targeting Meramecian reservoir intervals of the Oklahoma subsurface are constantly evolving. These efforts can be enhanced through sequence stratigraphically constrained mapping, a better understanding of diagenetic controls on reservoir quality, and greater data control in reservoir modeling. More sophisticated knowledge of the deposition, diagenesis, and variability of subsurface reservoirs will lead to more refined and consistent production models, especially in innovative oil and gas plays where data are often sparse.
This work was completed as a part of Oklahoma State University’s Mississippian Consortium, funded by Devon Energy, Chesapeake, Marathon Oil, Sinopec, Maverick Brothers Energy, SM Energy, Newfield, Redfork Energy, Unit Petroleum, Chaparral Energy, Longfellow Energy, Samson Energy, American Energy Partners, and Trey Industries. Many thanks to the OSU students who helped with field work and data collection, as well as the OSU faculty for their valuable review and comments. We are grateful for the donation of Petrel from Schlumberger. The contents of this chapter were derived primarily from the first author’s master’s thesis.