Contents lists available at ScienceDirect

Journal of Petroleum Science and Engineering

journal homepage: www.elsevier.com/locate/petrol

Characterization of a reservoir ooid shoal complex and Artificial NeuralNetworks application in lithofacies prediction: Mississippian St. Louisformation, Lakin fields, western Kansas

Keithan G. Martin, Matthew W. Totten, Abdelmoneam Raef⁎

Department of Geology, Kansas State University, 108 Thompson Hall, Manhattan, KS 66506, USA

A R T I C L E I N F O

Keywords:Artificial Neural Network (ANN)Oolitic ShoalWestern KansasSt Louis limestoneLakin fieldHugoton embaymentLithofaciesMeramecian

A B S T R A C T

Residing in the Hugoton embayment of western Kansas, the Lakin field has produced over 1.8 million bbl of oilfrom reservoir quality zones within the St. Louis Limestone (Mississippian, Meramecian) oolitic deposits. Thisstudy focuses on improving the understanding of the orientation, geometry, and spatial distribution of ooidshoal complexes in Kearny County, Kansas. To this end, the interpretation of petrophysical log data integratedwith a facies analysis conducted on 400 ft (122 m) of core from three separate intervals provided the basis, inthis study, for understanding the vertical stacking patterns and how facies transitions relate to reservoir qualityzones within the ooid shoal complexes. A supervised Artificial Neural Network (ANN) has been trained andvalidated, based on input-layer of geophysical well-logs and an output-layer of core-lithofacies, for lithofaciesprediction where core samples were not available. The porous ooid grainstone is the principal reservoir facies,with an average log-derived porosity measurement of eighteen percent. Ooid shoal complexes are present in St.Louis Zone B and C, with thicknesses reaching up to 10 ft (3 m) and 24 ft (7 m) respectively. These complexesare NW-SE trending and laterally extend up to 16 mi (26 km), with economically viable patches covering up to8 mile (13 km), and record roughly 2 mile (3 km) in width. The apex of the shoal contains the highest porosityvalues, with significant porosity reduction towards the shoal margins due to the increased cementation. Thisstudy provides information on the characteristics of oolitic deposits for the purpose of understanding controlson reservoir heterogeneity to aid in finding additional hydrocarbon reserves in the St. Louis Limestone inwestern Kansas.

1. Introduction

Mississippian-aged rocks in western Kansas represent two distinctevents; 1) the flooding of a craton by epicontinental shallow seas anddeposition of Kinderhookian-Meramecian aged skeletal and ooliticlimestone, and 2) a relative sea level fall during the LateMississippian with associated deposition of interfingered marine andcontinental sediments (Handford, 1988). During the EarlyMississippian, epicontinental seas covered the entirety of Kansas, butthe transition into the Late Mississippian records that epicontinentalseas inundated only the western and southern portions of the state.During this period, an amalgamation of factors such as globalpaleoclimate, sea level change, and geochemical factors, along withoptimal regional paleogeographic and tectonic conditions, resulted inextensive deposition of marine oolitic limestones across the continentof North America (Qi, et al., 2007; Wilson, 1975; Wilkinson et al.,1985; Handford, 1988; Ettensohn, 1993; Keith and Zuppann, 1993).

These oolitic limestones make excellent reservoir facies, resulting innumerous producing fields across the Mid-Continent region.

The study area within Kearny County is in the northwestern portionof the Hugoton embayment, which is a low relief extension of theAnadarko Basin (Fig. 1). A structurally high, subsurface uplift north-east of the Hugoton embayment is referred to as the Central Kansasuplift. This uplift separates the Hugoton embayment from two othersignificant basins in Central Kansas; the Salina and Sedgwick basins(Goebel, 1968). The total thickness of the Mississippian Limestoneranges between 300 and 500 ft, deepening towards the Anadarko Basinin southwestern Kansas (Goebel, 1968). In western Kansas, the St.Louis Limestone reaches a thickness of 200 ft. (60 km), and istraditionally divided into four informal zones (St. Louis A-D). Onlyzones B and C will be referred to in this study, as these are the onlyoolite-bearing zones (Fig. 2).

Within the widespread St. Louis Limestone oolitic deposits, reser-voir quality zones are of high interest to oil and gas companies, which

http://dx.doi.org/10.1016/j.petrol.2016.11.028Received 24 May 2016; Received in revised form 24 October 2016; Accepted 18 November 2016

⁎ Corresponding author.E-mail address: abraef@ksu.edu (A. Raef).

Journal of Petroleum Science and Engineering 150 (2017) 1–12

Available online 24 November 20160920-4105/ © 2016 Elsevier B.V. All rights reserved.

MARK

has resulted in numerous studies over hydrocarbon fields in westernKansas. Residing in Kearny County, Kansas, the Lakin field is produ-cing from reservoir quality zones within the St. Louis Limestone ooliticdeposits. Previous to this study, an in-depth study on the internalframework of these producing zones in Kearny County has never beenconducted. The purpose of this study is to create an understanding ofthe orientation, geometry, distribution, and continuity of depositionalfacies as pertaining to the reservoir and non-reservoir quality zoneswithin the St. Louis oolitic deposits. The objectives of this studyinclude; (1) investigate what the environmental setting was duringdeposition of Ste. Genevieve/St. Louis (Upper Meramecian) sediments,(2) determine the depositional facies within the parameters of asequence stratigraphic framework, (3) establish an understanding ofthe distribution and continuity of the reservoir quality lithofacies, and(4) attempt to train and apply an Artificial Neural Network (ANN)using an input layer of geophysical well-logs and respective core-lithofacies to utilize wire-line logs to predict differentiate lithofacies,specifically reservoir-quality flow units, and to assess whether signifi-cant sequence stratigraphic surfaces can be correlated throughout thestudy area using wire-line log curves.

2. Methods

Three wells in the Kearny County study area were chosen based onsatisfying the following conditions: i) availability of extensive coreintervals available ( > 100 ft.), ii) the available core samples repre-sented the Upper Meramecian intervals, and iii) the core intervals werelocated within a proximal distance relative to each other to limitvariability in the subsurface (Table 1; Fig. 3). Following the guidelinesof the AAPG Sample Examination Manual by Swanson (1981), a totalof 401 feet of core was described following Dunham (Scholle et al.,1983) to define the lithofacies and their respective depositionalenvironment. In order to augment the initial hand-sample scaleevaluation, thirty-six thin sections were prepared for detailed faciesanalysis. The thin sections were impregnated with blue epoxy, whichprovided an effective way to identify porosity of each facies. To helpdifferentiate carbonate mineralogy, one half of each thin section wassaturated and subsequently stained by alizarin red S and potassiumferricyanid. Generally, in the presence of dolomite or calcite, this

Fig. 1. Mississippian paleogeography of western Kansas (Lane and De Keyser, 1980). The starred square represents Kearny County. Kearny County, enlarged on the right, resides in theHugoton and Panoma Gas Area which contains numerous oil fields, many of which are oolitic reservoirs. Well symbols same as Fig. 3. Latitude and longitude (NAD27) coordinates ofstudy area; −101.21 to −101.3 and 37.79 to 37.88, respectively.

Fig. 2. Upper Paleozoic stratigraphic column, created from study area, illustrating thecarbonate dominated Ste. Genevieve/St. Louis Limestone. Quartz-rich and oolitic zonesare widespread in Kearny County, Kansas. (modified from Qi. and Carr, 2003).

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

2

mixture will stain the minerals various shades of red based onmagnesium and/or iron content. Quartz grains remain unstained,and thus this application is useful for quick mineral identification inSte. Genevieve/St. Louis Limestone thin sections.

A petrophysical analysis was conducted using wire-line logs.Gamma ray, sonic, neutron porosity, and density porosity were thechief wire-line logs used for this study due to their availability in mostwells. Upon defining the chief lithofacies through core analysis,lithofacies data were uploaded into IHS Petra® (geologic software) toground-truth wire-line log data with core interpretations. Once cali-brated to core, wire-line logs were used to make regional correlations oflithofacies and significant sequence stratigraphic surfaces in 137 wellsthroughout the study area. Structural, stratigraphic, and isopach mapswere subsequently created following the wire-line log-core calibration.

3. Lithofacies and depositional environments

Core and thin section photographs of lithofacies 1–3 and 4–6 canbe seen in Figs. 4 and 5 respectively. Information on lithofacies

occurrence for each of the three evaluated core intervals is representedin Fig. 6.

3.1. Lithofacies 1: porous ooid grainstone

The porous ooid grainstone is the main reservoir facies within theSt. Louis Limestone. It is characterized as a very light gray, very wellsorted, well rounded, medium (upper) grain size facies. The porousooid grainstone is poorly cemented by equant calcite cement withprimary-depositional interparticle porosity, and less commonly, oo-moldic and fracture porosity. Heavy hydrocarbon staining was found innumerous core samples within the facies intervals. Ooid concentrationsvaried throughout the interval of the facies, with areas of higherconcentrations containing heavier hydrocarbon staining. The indivi-dual ooids are concentrically laminated with their radial internalstructure remaining, recording an average size from 0.3 mm to0.5 mm. Articulated and disarticulated echinoderms, brachiopods,bryozoans, as well as high concentrations of peloids and low amounts( < 2% whole rock volume) of detrital quartz grains combine to make upthe remaining constituents of the facies. Of the chief fossils identified,echinoderms are the most prevalent. The echinoderm fragments arerelatively large, ranging from 2.0 mm to 3.0 mm, and numerousechinoderm fragments act as nuclei for the ooids. Varying degrees ofmicritization were found during analysis, with many grains havingthick micritic rims or envelopes. Sedimentary structures within theporous ooid grainstone are rare, which could be due to excellent sortingor bioturbation (Handford, 1988). Lithofacies 1 and 2 are both ooidgrainstones, although, cementation is variable. As concluded in Cluff(1984), cementation has a relationship with the degree of ooliticcoating, where higher degrees of coating results in great porosity.This relationship was also found in this study, and can be viewed bycomparing thin section images from Lithofacies 1 and 2 (Fig. 4) Of thethree cored intervals evaluated, the thickness of Lithofacies 1 reachesup to roughly 24 ft (7 m), occupying up to 24% of the St. Louis coreintervals. Lithofacies 1 was deposited as a high-energy ooid shoal.

3.2. Lithofacies 2: highly cemented ooid grainstone

The highly cemented ooid grainstone has similar characteristicswith the porous ooid grainstone (Lithofacies 1), although, the primaryporosity in Lithofacies 1 is replaced with equant calcite cement inLithofacies 2, significantly reducing effective porosity. Majority of ooidshave internal structure, although the internal structure of many ooidshas been abolished through micritization and they are classified aspeloids. The fossil concentrations are low, with the most prevalentbeing gastropods, foraminfera, and crinoids. The highly cemented ooidgrainstones contain 1–5% of detrital quartz grains (whole-rock vo-lume). This facies is defined as a non-reservoir flow unit. Of the threecored intervals evaluated, the thickness of Lithofacies 2 ranges between9 and 21 ft (3–6 m), occupying up to 22% of the St. Louis coreintervals.

Lithofacies 2 is the product of a high-energy environment ofdeposition and is interpreted as residing along the margins of theshoal, where porosity is reduced due to cementation.

Table 1Evaluated core interval details. Study no. correlates with labeled wells in Fig. 3.

Study no. API no. County Field name KB (ft) Cored interval in miss (ft) Thin sections

Top Bottom Total

1 15–093-20147 Kearny Lakin south 3039 4830 5009 179 112 15–093-00317 Kearny Hugoton gas area 3036 4858 4976 118 113 15–093-20011 Kearny Hugoton gas area 3163 4963 5067 104 14

Cumulative total 401 36

Fig. 3. Map of well locations in Lakin NW, Lakin (Mid), and Lakin S fields, T24-26S andR36-37W. Well status of all wells used in the study are displayed. Three St. LouisLimestone core intervals are displayed and labeled as 1,2, and 3. These numbered labelscorrelate with Fig. 4, which highlights details of the core intervals. Coordinate system isUSA PLLS (Township and Range). All of the wells are vertical, perforated through casing,and completed in a similar fashion.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

3

3.3. Lithofacies 3: quartz-carbonate grainstone

The quartz-carbonate grainstone facies is characterized as a verylight gray, medium-to-very fined grained, subrounded, well sortedinterval with individual laminations being very well sorted. Very fewfossil fragments are present, detrital quartz grain abundances arebetween 20–36% of the whole rock volume, with an average grain sizeof 0.227 mm. Along with detrital quartz grains, the quartz-carbonategrainstone contains high concentrations of peloids. Micrite and blockycalcite cement make up the matrix, with portions of the facies having astrong peloidal fabric. Very fine ( < 1 mm) dark laminations cyclicallyrepeat throughout the intervals, with larger, 3 mm dark laminations

occurring as well. The dark laminations are made up of very fine quartzgrains, with the grains ranging between 0.06 and 0.1 mm in size. Smallscale (15 mm) fining upward graded bedding cyclically repeats throughlarge portions of the interval. Lithofacies 3 is declared as a non-reservoir facies, due to compaction and pore filling cementation.Within the three cored intervals evaluated, the thickness ofLithofacies 3 ranges between 10 and 34 ft (3–10 m), occupying up to30% of the St. Louis core intervals.

The quartz-carbonate grainstone is interpreted as a carbonateeolianite dune, which is defined as any rock formed by the lithificationof sediment deposited by eolian processes and contains > 50% CaCO3

(Fairbridge and Johnson, 1978). The following acted as the key

Fig. 4. Thin section and core photographs of lithofacies 1–3. (LF1) The core slab has been stained extensively by liquid hydrocarbons. The thin section has been impregnated with blueepoxy to illustrate porosity. Ooids dominate the thin section, with many echinoderm fragments making up the ooid nucleus. Following ooids, echinoderm fragments and peloids makethe remaining constituents. As shown in the thin section, equant calcite poorly cements the intraparticle matrix. API: 15-093-20147, 4986 ft (LF2) Similarly to LF1, ooids and peloidsdominate the thin section. The ooids have a lesser degree of coating relative to LF1, and calcite cement dominates intraparticle space. API: 15-093-20011, 4973 ft (LF3) The core sampleillustrates numerous laminations, which are made up very fine grained detrital quartz. Thin section shows detrital quartz grain dominance and peloidal fabric. API: 15–093-20147,4922 ft.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

4

Fig. 5. Thin section and core photographs of lithofacies 4–6. (LF4) Thin laminations made up of very fine to fine detrital quartz grains is illustrated in the core slab. The thin section isdominated by peloids, and contains 8% detrital quartz grains. API: 15–093-20147, 5007 ft (LF5) Lithofacies 5 core intervals are often disarticulated. Absence of detrital quartz grains,peloids, ooids, and skeletal fragments with an exclusively micritic matrix characterizes this facies. API: 15–093-20011, 4966–4967 ft (LF6) Large articulated and disarticulated skeletalfragments heavily present. Detrital quartz grains are absent and the matrix is exclusively micritic. API: 15-093-00317 ft.

Fig. 6. Bar graph illustrating the lithofacies abundance in each of the three evaluated core intervals. Lithofacies 3 and 4, peloid-rich grainstone and quartz-carbonate grainstonerespectively, dominate the core intervals, and are characterized by very gradational lithofacies contacts. Well no. 1 is the only core interval that contains Lithofacies 1, porous ooidgrainstone, the principle reservoir flow unit.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

5

diagnostic features used to identify this facies: 1) intervals thatcontained over 20% (total rock volume) detrital quartz, 2) having aclimbing translatent stratification and indistinctly stratified sedimen-tary structures, 3) well sorted; medium to fine grained, and crossstratified, and 4) lacking larger marine shells.

3.4. Lithofacies 4: peloidal packstone

The peloidal packstone is characterized as a very light gray, mediumgrained, rounded, well sorted interval. Peloids make up the over-whelming majority of the facies and intraparticle void are filled withsparry calcite cement. Very few ooids have visible internal structure.Many of the peloid are interpreted as past ooids that have beenmicritized, losing all internal structure. Dark, quartz-filled laminationsappear throughout most intervals as well as styolites. Detrital quartzgrains make up around 2–8% of the rock volume, with an average grainsize of 0.09 mm. Fossil concentrations are very low, with foraminiferaand gastropods being the most prevalent. Lithofacies 4 is the mostvertically extensive lithofacies, with thicknesses ranges between 40 and60 ft (12–18 m), occupying up to 55% of the St. Louis core intervals.

The peloidal packstone facies is interpreted as being deposited in alagoonal/tidal shelf environment, landward of the ooid shoal. Lagoonsare often shallow bodies of water in a completely enclosed, restrictedenvironment with no connection to the sea but can infrequently reachup to several tens of meters deep. The key diagnostic criterion forrestricted shelves or lagoons is vertical sequences and lateral transi-tions, which normally are represented by a low-energy shoaling upwardcycle (Enos, 1983). Lagoon-type environments form from the presenceof a physical barrier, thus a lateral change into a barrier-type faciesassists in the determination of a restricted, lagoon environment and inconjunction with a barrier-type facies, in the opposing direction, atransition into low-energy shoreline facies is characteristic (Enos,1983). The laterally equivalent lithofacies to the Lithofacies 4 (lagoonalfacies) within the study area is Lithofacies 1 (ooid shoal facies). Theooid shoal body acted as a “barrier facies”, or immediate cause ofrestriction, and contributed heavily to the formation of the lagoon.

3.5. Lithofacies 5: clay-rich mudstone

The clay-rich mudstone is characterized as a grayish black, veryfined grained, micritic interval. Detrital quartz grains make up between0 and 3% of the rock volume. Very little fossil content was identifiedand moderate quantities of highly micritized peloids were present.Some clay-rich mudstone intervals have a strong green tint, which isinterpreted as represents glauconite. Using a handheld X-ray fluores-cence spectrometer, the green tinted areas were scanned intensely.Relative to other X-rayed intervals, very high iron (Fe) values, between10–100 ppm, were recorded from the green tinted areas. The mineralglauconite is an iron potassium phyllosilicate and is a part of the micagroup. The high iron content accompanied by a strong green tintreinforced the presences of glauconite. Of the three cored intervalsevaluated, the thickness of Lithofacies 5 ranges between 3 and 9 ft (1–3 m), occupying up to 7% of the St. Louis core intervals. Lithofacies 5 isa non-reservoir flow unit. Many of the clay-rich mudstone coreintervals were disarticulated, as seen in the core image below.

Lithofacies 5 signifies the deepest environment found in the St.Louis limestone, determined to be deposited in a low energy, open-marine shelf. Key criteria for this interpretation includes; 1) little-to-nodetrital quartz grains, 2) little-to-no skeletal fragments, 3) lackingevidence of subaerial exposure, 4) the matrix is exclusively made up ofmicrite, and 5) it is a clay-rich characteristic.

3.6. Lithofacies 6: skeletal wackestone/packstone

The skeletal wackestone/packstone is characterized as a mediumlight gray, fine grained, poorly sorted, fossil-rich interval. Fossils

identified include echinoderms, crinoids, bryozoans, foraminifera,and tabulate corals, with many being fully articulated. Detrital quartzgrains are in very low abundance ( < 2%) of the rock volume.Lithofacies 6 is declared a non-reservoir flow unit. Of the three coredintervals evaluated, the thickness of Lithofacies 6 ranges between 4 and5 ft (1–1.5 m), occupying up to 4% of the St. Louis core intervals.

Lithofacies 6 was interpreted to be deposited in a low energy, open-marine shelf environment. Key criteria for this interpretation include;1) little-to-no detrital quartz grains, 2) abundant, large, disarticulatedskeletal fragments, 3) lacking evidence of subaerial exposure, and 4)the matrix is exclusively made up of micrite. Lithofacies of this natureare thought to be in waters that are between 10 and 30 m (30–100 ft)deep, located seaward from grainstone shoals, such as the Lithofacies 1and 2 (Cluff, 1984).

4. Geophysical well-logs and core-lithofacies for ANNtraining and application

Many authors, (e.g., Helle, 2001; Trappe and Hellmich, 2000;Oldenziel et al.,2000; and Ohl and Raef, 2014), have reported success-ful application of Artificial Neural Networks, ANN, to various types ofparameter estimation problems on the basis of geophysical data. Infact, the application of ANN to a classification of geophysical well-logsfacies, as in this study, falls in the area of statistical pattern recognition.In this application every new input into a trained ANN is assigned toone of discrete classes or categories. The various classes (six core-lithofacies as described above) and the respective well-log responseshave been used in training an ANN; the trained ANN has subsequentlybeen used to classify the well-logs responses into their respective core-lithofacies, when core samples are not available. Advantages of usingANN lie in that they neither require an explicit mathematical model norassume linearity relationship between input and output, (Helle et al.,2001)). In addition to that, ANN can perform well in situations ofhaving inaccurate/noisy data with no assumption of normal datadistribution, and can be trained on example data, so that they can begeneralized/applied to an entire data set for classification or parameterestimation.

The ANN is “trained”, where predictor variables (geophysical logvalues in this study; Gamma Ray, Sonic Interval Transit Time, DensityPorosity, and Neutron Porosity) and a categorical response variable(core lithofacies) are defined (Doveton et al., 2004). During the trainingprocess, the hidden layer nodes and dampening parameters valueswere repetitively adjusted. A higher number of hidden layer nodes(larger network), a lower dampening parameter, and a larger numberof iterations all lead to a more complex network. Inversely, a lowernumber of hidden layer nodes (smaller network) and a higherdampening parameter, and a smaller number of iterations all lead toa smoother, less complex network. The input values of these threevariables were repetitively adjusted until the predicted facies matchedthe core facies as closely possible, with a final value of 50 hidden layernodes, a 0.05 dampening parameter, and 100 iterations. The result ofthis neural network model, for one of the three cored study wells withGR, Neutron Porosity, Density-Porosity, and Sonic well-logs, are shownin Fig. 7. This figure illustrates the agreement of facies between thepredicted and core facies. Facies Probability, the plot to the farthestleft, is equivalent to the “maximum” probability within the neuralnetwork. Within the model, the value at a specific point is checked foragreement with each facies, giving a probability or likelihood for eachfacies. The facies with the highest probability or “maximum” prob-ability will then be the facies that is predicted, which is shown in thesecond plot in Fig. 7.

Adding geologic significance to the probabilities can create a lesserratic model that closely replicates the Core Facies. Transitionsbetween physically adjacent depositional environments occur morefrequently opposed to environments which are significantly distantfrom each other (Bohling and Dubois, 2003). To apply this concept to

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

6

the data, a TPM (Transition Probability Matrix) was computed andapplied. The TPM is derived from the stacking pattern of the corefacies, which calculates the probability of one facies underlying oroverlying one another. After computing and applying the TPM mod-ification, the very thin ( < 1 ft), erratic predicted facies were eliminated,which ultimately enhanced the agreement between the FaciesPrediction and Core Facies. The result of applying the TPM can beseen in Fig. 7, by comparing the “Facies Prediction”. This process wasrepeated for all three of the main study wells, which created a total ofthree neural network models.

In order to quantitatively analyze the accuracy of the neuralnetwork models, a facies prediction “scorecard” was created usingExcel's PivotTable, with Core Facies as the row variable, predictedfacies as the column variable, and count of predicted facies in the tableitself. The Core Facies refers to the “actual” facies that was evaluatedfrom three separate core intervals (Fig. 8). During the creation of thisscorecard, values were taken from all three of the main study wells. TheMaximum Probability was calculated by summing the maximumprobability per facies for the three main study wells combined, andaveraging those values per each facies. It is important to note thatpredicted facies are derived from the maximum facies probability, withthe difference being that maximum facies probability is the highestdecimal value between zero and one relative to the remaining faciesand the predicted facies is a value of either zero or one, with the valueone indicating the presence of a facies and zero indicating it's absence(predicted facies is the equivalent of taking the decimal value of themaximum facies probability to the nearest whole number, one).

Combining all of the Maximum Probabilities per each facies andaveraging the values gives the average percent of confidence in specificfacies. The Maximum Probability Modified was calculated just as theMaximum Probability, with the only difference being that the values

taken had the Transitional Probability Matrix applied to them.After the training process was completed for each of the three main

study wells, the three neural network models were then applied to atotal of 30 wells that contained the same input variables (log curves) asthe model. For wells that contained all input variables needed (the fourmain geophysical logs; GR, Neutron, Density, and Sonic) one of thethree neural network models was applied. Determining which of thethree models to apply to a specific well was controlled by the distancefrom the well to the three neural network models. The model with theshortest distance to the well was the chosen model to be applied.

5. Depositional system evaluation

A schematic interpretation of the environment during the deposi-tion of the St. Louis and Ste. Genevieve limestone is illustrated inFig. 9. This depositional environment reconstruction was created byHandford (1988), and was proven to be analogous within the deposi-tional history of another St. Louis limestone study of western Kansas byQi and Carr (2003). The reconstruction of the depositional history,gathered from the evaluation of three cored intervals and petrophysicalwell logs, concludes that this schematic interpretation is also analogouswithin the Kearny County study area.

The landward-most depositional environment is the coastal eolia-nite, which is represented by lithofacies 3 (quartz-carbonate grain-stone). Moving seaward, the lagoon, represented by lithofacies 4(peloidal grainstone), is restricted by the ooid shoal, represented bylithofacies 1 and 2 (porous and highly-cemented ooid grainstone).Further seaward from the ooid shoal, a transition into an open shelfenvironment, represent by lithofacies 6 (skeletal wacke/packstone) anda final transition into a relatively deep marine environment, repre-sented lithofacies 5 (clay-rich mudstone), concludes the relative

Fig. 7. Four plots of the facies probability, facies prediction, facies prediction modified, and core facies (actual facies) that illustrate the neural networks models facies agreement.

Fig. 8. Predicted facies scorecard for 3 key trained wells with core containing GR, Neutron, Density, and Sonic curves.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

7

position and the depositional profile of each lithofacies during the timeof deposition in the Upper Mississippian.

Sea level rise and fall throughout the Early Devonian-LateMississippian is referred to as the Kaskaskia sequence, which ischaracterized as a second-order cycle, and thus is made up ofnumerous, third order-cycles (Handford, 1988). The St. Louis- Ste.Genevieve sequence represents a third-order cycle, with duration ofless than 5 m.y., and is a result of sea level rise and fall (Handford,1988). As interpreted from data gathered from the study area, thedepositional history and sequence stratigraphic interpretation can bepresented in seven stages (Fig. 10).

The initial ooid shoal, within the St. Louis C zone, formed at theintersection of the wave base and the very low angle sea floor. A slight

relative sea level increase caused thickening of the ooid shoal (up to24 ft (7 m)) until the rising sea level outpaced carbonate productionand drowned the shoal. A relative sea level fall initiated the prograda-tion of the lagoonal facies and, with continual sea level fall, furtherprogradation of the carbonate eolianite facies. At the end of thelowstand, relative sea level rise flooded the carbonate platform, causinga landward shift of the successive parasequence, onlapping the openmarine facies onto the prograded carbonate eolianite facies. The St.Louis B zone ooid shoal was deposited at the intersection of the wavebase, overlying the open marine facies. Relative sea level continued torise, thickening the St. Louis B zone shoal (up to 14 ft (4 m)), but wasultimately outpaced and drowned, and the deep marine facies onlappedthe ooid shoal facies. The deep marine facies deposit is overlain by the

Fig. 9. Schematic illustration of the depositional environment of the St. Louis Limestone. Depositional facies interpreted from core and thin section analysis and their correspondinglithofacies are specified in this graphic; 1 – porous ooid grainstone, 2 – highly-cemented ooid grainstone, 3 – quartz-carbonate grainstone, 4 – peloidal grainstone, 5 – clay-richmudstone, 6 – skeletal wackestone/packstone. (modified from Handford, 1988).

Fig. 10. Seven stages of deposition as result of relative sea level rise and fall. 1) formation of an ooid shoal, 2) relative sea level rise, thickening and subsequent drowning of the ooidshoal, 3) relative sea level fall, landward facies prograde seaward, 4) relative sea level rise, 5) seaward facies retrograde landward, 5) continual sea level rise, deep marine facies blanketsthe carbonate platform, 6) relative sea level fall, erosional surface due to subaerial exposure, and 7) the carbonate platform is flooded due to a relative sea level rise and supratidal andintertidal facies are deposited.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

8

restricted lagoon facies. This is a significant stratigraphic relationship,as it shows a deep marine facies overlain by an intertidal-to-supratidalfacies. This depositional facies relationship is indicative of a sequenceboundary, where the transitional depositional facies between the deepmarine facies and intertidal-to-supratidal facies, which originally over-laid the deep marine micritic mudstone, were subaerially exposed dueto relative sea level fall and subsequently were eroded away. As stated,the sequence boundary is capped by a restricted lagoon, which isoverlain by the coastal eolianite, which concludes the UpperMississippian's St. Genevieve/St. Louis limestone sedimentary pack-age.

6. Ooid shoal complex

Lithofacies 1 (porous ooid grainstone) was deposited as an ooidshoal on a shallow-marine, open shelf. An ooid shoal is defined as alinear landform completely within or extending into a body of waterwith high concentrations of ooids that may accumulate in carbonateplatforms on ramps, shelves, and rimmed shelf edges (Ahr, 1973: Read,1981). As stated by Handford (1998), ooid shoals largely form wherethe sea floor crosses the fair-weather wave base generating a high-energy zone that can be found in the distant offshore or at theshoreline.

Orientation and geometry are key diagnostic features when identi-fying ooid shoal deposits. Within the St. Louis C zone, Lithofacies 1forms an outstretched, northwest-southeast oriented, lenticular bodyroughly 16 miles (6 km) long and 2 miles (1 km) wide. Handford(1998) determined the Great Bahama Banks marine sand shoals areanalogous with a Mississippian ooid shoal in Damme Field, FinneyCounty, Kansas. Similar to the Damme field's shoal geometry, the ooidshoal within St. Louis C zone is oriented with the depositional strike, orperpendicular to the paleoslope (northwest-southeast trending).Fig. 11, an isopach of Lithofacies 1, illustrates the geometry of the St.Louis ooid shoal within the study area (7.3 mile (11.7 km)), with the

similar, elongate geometry of the Great Bahama Banks shoal (5 mile(8 km)). These similarities reinforce the analogy between the ancientMississippian shoal of the study area in Kearny County and Cat Cay'smodern shoal.

The significance of the oolitic deposits within the St. LouisLimestone in western Kansas is due to their reservoir quality porosity,which reaches up to 16% (Fig. 12). Lithofacies 1 is the principlereservoir flow unit in the Upper Mississippian (Meramecian) within thestudy area. Within the St. Louis Limestone intervals, the highly-porousooid shoal deposits reside in St. Louis B and C zone (Fig. 13). The St.Louis B zone is relatively thin ( < 10 ft) throughout majority of thestudy area and records lower porosity values, while St. Louis C zonereaches up to 28 feet in the southern portions of the study area and isdefined as the principle St. Louis Limestone pay zone in the study area.

7. Geologic controls on production

All recorded hydrocarbon production from the MississippianLimestone within the Lakin area is recovered from Lithofacies 1 in

Fig. 11. Located along the west margin of the Great Bahama Banks, an aerial view of the Cat Cay platform margin sand-shoal complex, illustrating the lenticular geometry of modernsand shoals with the gray contours (Handford, 1998). Geometry of the ooid shoal within the St. Louis C zone (pay zone), strongly resembling the geometry of the Cat Cay sand-shoalcomplex.

Fig. 12. Cross plot of neutron porosity (%) and density porosity (%). Neutron anddensity log values were taken from the three cored study wells. This cross plot illustratesthat Facies 1 (Porous Ooid Grainstone) (circles on graph) exhibits the highest porosityvalues and confirms that it is the chief reservoir facies.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

9

the St. Louis C (pay) zone. Reports on production from this pay zone inthe Lakin field totals over 1.83 million barrels. Production in barrels isposted over each St. Louis C zone producing wells in Fig. 14 (isopachmap of the porous ooid grainstone from the St. Louis C zone). Asdisplayed in Fig. 14, the thickest portion of the isopach map is directlyrelated to higher production values, and as the pay zone thins,cumulative production values diminish. Within the thickest portionsof the isopach map, the highest cumulative production values arerecorded (412,267 and 345,589 bbl), although significantly smallervalues (94,892 and 174,690 bbl) can be seen in the thickest area of thepay zone as well. Production variability of this nature is a commonoccurrence that numerous factors can attribute to, including which wellwas completed first. The wells with the highest cumulative productionwere typically drilled first, and the remaining wells were drilled afterproven production as offset wells. If the wells with higher productionwere drilled at a later date, it could be due to less than optimal wellplacement of the original well. In this scenario, the first well drilledcould have missed the structural high, hindering the ability to drain anyoil that is further up dip. The second well drilled (with higherproduction values) may have been placed further up dip at the apexof the structural high, which would explain higher production values.In this specific situation, the date of the well completion is determinedto be responsible for production variation. The reservoir had beendepleted, or “played out”. This can be expressed by correlatingproduction with the completion year in the Table 2.

Directly to the west of Laken NW field, eight dry holes and two gas

wells were drilled. These ten wells are located on the west margin of theooid shoal (Fig. 14), where the pay zone thins out to less than six feet.Kearny County's ooid shoal is naturally a lenticular, elongate body, withthe thickest portions being towards its apex. Poor placement of eightwells along the western flank of the ooid shoal body resulted in little-to-no hydrocarbon production from the respective wells.

As previously illustrated, the St. Louis C zone has been the targetedpay zone within the study area. The Lakin field has been a prolific oilproducing field, although toward the north/northwest, dry holesbecome more frequent (Fig. 15). South of Lakin NW field, excludingwells without production data, 76.7% of wells drills produced economicquantities of hydrocarbons, while north of Lakin NW field success rateaverages only 21.7%. In the center of Lakin field, the St. Louis C zonereaches a thickness of 24 ft. North of the Lakin field, gradual thinningbegins, resulting in a thickness of 6 ft or less in the northernmostportion of the study area. The thinning of the pay zone contributedheavily to the decrease in drilling frequency and wells that were drilledare recorded as dry holes. Along with a thinning pay zone, low recoverycould be a result of diagenetic alterations, as porosity is significantlyreduced through infilling of pores with calcite cement. The southern-most and south central core intervals (Study no. 1 and 2 wells) hadexcellent porosity within the St. Louis C zone. Within the northernmostcore interval, porosity was significantly reduced by the infilling ofcalcite cement. Two chief factors that contributed significantly to lowersuccess of producing hydrocarbons north of Lakin NW field include: (1)significant thinning of the St. Louis C zone (pay zone), and (2) increase

Fig. 13. The core interval from Study No. 1 well (15-093-20147) was the only interval that was cored through the Ste. Genevieve Limestone, St. Louis Limestone, and St. Louis B and Czone. Lithofacies, lithology, GR, Neutron/Den log curve responses, and depositional environment and sequence stratigraphic interpretation of the Ste. Genevieve/St. Louis core intervalfrom API: 15–093-20147.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

10

in diagenetic alteration (the infilling of pores with calcite cement).

8. Reservoir considerations and conclusions

Future exploration for ooid shoal reservoirs within Kearny Countyand other western Kansas counties may prove to be difficult, as manyknown hydrocarbon fields in this region have been drilled extensively.The limited areal extent and outstretched, lenticular shape of the ooidshoal complexes within St. Louis B and C zone places difficulty on

predicting new areas of exploration. With ooid shoal reservoir thick-ness only 24 feet at its maximum, the ooid shoals are too thin and lacksufficient velocity contrast for direct seismic mapping (Parham andSutterlin, 1993). While there are difficulties when exploring thesecomplexes, positive findings for exploration have been gained throughthis study, which include the following; 1) all known hydrocarbon-bearing shoals in Kearny County are NW-SE trending, 2) their verticalextent covers up to 16 miles, with economically viable patches coveringup to 8 miles in length, with roughly 2 miles of width, 3) the thickestand best reservoir quality is along the apex of the shoal, with reservoirquality diminishing along the margins, and 4) the shoal complexesform along the depositional strike of the low angle ramp.

This study increases the understanding of the internal complexity ofthe St. Louis limestone in Kearny County, Kansas, through detailedanalysis of three core intervals to determine facies successions, lateraldistributions, and continuity. Investigating reservoir quality and po-tential per lithofacies and analyzing their predictability was concludedby integrating depositional facies and sequence stratigraphy to char-acterize reservoir potential within the study area. Collectively, thismultifaceted study concludes the following;

1. Six lithofacies were identified in the three St. Louis cored inverals: 1)Lithofacies 1- Porous Ooid Grainstone, 2) Lithofacies 2 – Highly-Cemented Ooid Grainstone, 3) Lithofacies 3 – Quartz-CarbonateGrainstone, 4) Lithofacies 4 – Peloidal Grainstone, 5) Lithofacices 5-Micritic Mudstone, and 6) Lithofacies 6 – Skeletal Wackestone/Packstone.

2. The following depositional environments were interpreted from coreevaluation and thin section analysis: 1) Lithfacies 1 and 2 – highenergy ooid shoal, 2) Lithofacies 3 – Eolanite, 3) Lithofacies 4 –

Back barrier restricted lagoon, 4) Lithofacies 5 – Deep marine, and5) Lithofacies 6 – Open marine.

3. Lithofacies 1 is the only reservoir quality facies, with porosity valuesup to 16%. The porous ooid grainstone occurs in St. Louis B (up to10 ft) and C zone (up to 24 ft). The ooid shoal deposits are verydifficult to predict, and porosity is often reduced through diagenesis,where calcite cement infills the pore space. The apex of the shoalcontains the highest porosity values, while the back shoal (landward)porosity is of good quality as well. Porosity quality diminished on theforeshoal (seaward) and collectively porosity diminished the furtheraway it is from the apex of the shoal. (a) Within the St. Louis C zone,the porous ooid grainstone (ooid shoal) is the pay zone within thestudy area. St. Louis B zone is the chief producer in other counties ofwestern Kansas, but was found to be very thin < 10 feet, absent, orhighly cemented in this locality.

4. The southeastern side of the study area has accumulated the highesthydrocarbon production, producing chiefly out of the porous ooidgrainstone in the St. Louis C zone. The following contribute to thesuccess of hydrocarbon production on the southeastern side of thestudy: 1) the ooid shoal (pay zone) is at its thickest, up to 28 feet, 2)porosity values are high, between 12 and 18%, and 3) local structuralhighs allows for a combination trap, containing both structure andstratigraphic traps.

5. The use of wire-line log signatures within the St. Louis limestone: (a)Lithofacies log signatures were difficult to differentiate, as different

Fig. 14. Isopach of the Lithofacies 1 (porous ooid grainstone) within the St. Louis C zonewith production data (cumulative oil production in bbls) from this zone posted.Production is from each lease, which might contain multiple wells.

Table 2Table correlating production and completion year of St. Louis Limestone wells.

API no. S-T-R Field Oil production(bbls)

Completion year

15–093-20147

4-26S-36W

LakinSouth

412,267 1975

15–093-20150

34-25S-36W

Lakin(Mid)

350,399 1975

15–093-20464

9-26S-36W

LakinSouth

345,589 1978

15–093-20466

4-26S-36W

LakinSouth

174,690 1978

15–093-20461

9-26S-36W

LakinSouth

94,892 1978

Fig. 15. Bar graph which illustrates the drilling results south of Lakin NW field (27.1%success rate) and north of Lake NW field (76.7% success rate) showing the number of dryholes, oil wells, gas wells, and unknown per geographic location.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

11

lithofacies gave very similar wire-line log readings, especially in thecase of Lithofacies 3 and 4, which exhibits a gradational variation.Favorably, the porous ooid grainstone or the “pay zone” gives adistinctive signature, as the neutron/density values increase sig-nificantly in its presence. (b) The St. Louis sequence boundary canbe tracked throughout the study area via log curves, although, othersignificant sequence stratigraphy surfaces are difficult to correlate.As Carr and Lundgren (1994) found, gamma ray spectral logs can beused to recognize exposure surfaces at reservoir scale within the St.Louis limestone of western Kansas. Gamma ray spectral logs werenot available, complicating the correlation of significant surfaces inwells that did not have core available for evaluation.

6. The prediction of the productive facies is greatly aided by use of anArtificial Neural Network (ANN), calibrated to core description. TheANN is sensitive to petrophysical log variations in uncored wells,and accurate in predicting facies using these data.

References

Ahr, W.M., 1973. The Carbonate Ramp: an Alternative to the Shelf Model: Gulf CoastAssociation Geological Societies Transactions, vol. 23, pp. 221–225.

Bohling, G., Dubois, M.K., 2003. An Integrated Application of Neural Network andMarkov Chain Techniques to Prediction of Lithofacies from Well Logs, KansasGeological Survey, Open-file Report, no. 2003-50.

Carr, T.R., Lundgren, C., 1994. Use of gamma ray spectral log to recognize exposuresurfaces at the reservoir scale; Big Bow field (St. Louis Limestone), Kansas. In:Dolson, J.C., Hendricks, M.L., Wescott, W.A. (Eds.), Unconformity-RelatedHydrocarbons in Sedimentary Sequences; A Guidebook for Petroleum Explorationand Exploitation in Clastic and Carbonate Sediments. Rocky Mountain Associationof Geologists, Denver, CO, 79–88, 298 p.

Cluff, R.M., 1984. Carbonate Sand Shoals in the Middle Mississippian (Valmeyeran)Salem-St. Louis-Ste. Genevieve Limestones, Illinois basin; in, Carbonate Sands – ACore Workshop, P.M. Harris, ed.: Soc. of Economic Paleontologists andMineralogists, Core Workshop no. 5, pp. 94–135.

Doveton, J.H., Watney, W.L., Bohling, G., 2004. Theory and Practice of A Web-basedIntelligent Agent in the Location of Pay Zones on Digital Well Log Files, KansasGeological Survey, Open-file Report, no. 2004-28.

Enos, P., 1983. Shelf. In: Carbonate depositional environments. Tulsa, Okla., U.S.A.:American Association of Petroleum Geologists, pp. 267–296.

Ettensohn, F.R., 1993. Possible flexural controls on the origins of extensive, ooid-rich,carbonate environments in the Mississippian of the United States. In: Zuppann,C.W., Keith, B. (Eds.), Mississippian Oolites and Modern Analogs 35. AAPG Studiesin Geology, 13–30.

Fairbridge, R.W., Johnson, D.L., 1978. Eolianite. In: Fairbridge, R.W., Bourgeois, J.(Eds.), The Encyclopedia of Sedimentology, Stroudsburg Pennsylvania, Dowden.Hutchinson & Ross, 279–282.

Goebel, E.D., 1968. Mississippian system. In: Zeller, D.E. (Ed.), The StratigraphicSuccession in Kansas. Kansas Geological Survey, Bulletin, 17–21, no. 189.

Handford, C.R., 1988. Review of carbonate sand-belt deposition of ooid grainstones andapplication to Mississippian reservoir. Damme Field Southwest. Kans.: Am. Assoc.Pet. Geol. Bull. 72 (10), 1184–1199.

Helle, H.B., Bhatt, A., Ursin, B., 2001. Porosity and permeability prediction from wirelinelogs using artificial neural networks: a North Sea case study. Geophys. Prospect. 49,431–444.

Keith, B.D., Zuppann, C.W., 1993. Mississippian oolites and petroleum reservoirs in theUnited States — an overview. In: Zuppann, C.W., Keith, B. (Eds.), MississippianOolites and Modern Analogs 35. AAPG Studies in Geology, 1–12.

Lane, H.R., De Keyser, T.L., 1980. Paleogeography of the late Early Mississippian(Tournaisian3) in the central and south-western United States. In: Fouch, T.D.,Magathan, E.R. (Eds.), Paleozoic Paleogeography of West-central United States.Rocky Mountain Section SEPM, 149–159.

Ohl, D., Raef, A., 2014. Rock formation characterization for carbon dioxidegeosequestration: 3D seismic amplitude and coherency anomalies, and seismicpetrophysical facies classification, Wellington and Anson-Bates Fields, Kansas, USA.J. Appl. Geophys. 103, 221–231.

Oldenziel, T., DeGroot, P., Kvamme, L.B., 2000. Statfjord study demonstrates use ofneural networks to predict porosity and water saturation from time-lapse seismic.First Break 18 (2), 65–69.

Parham, K.D., Sutterlin, P.G., 1993. Oolite shoals of the Mississippian St. LouisFormation, Gray County, Kansas; a guide for oil and gas exploration. In: Keith, B.D.,Zuppann, C.W. (Eds.), Mississippian Oolites and Modern analogs: AmericanAssociation of Petroleum Geologists, Studies in Geology 35, 185–197.

Qi., Lianshuang, Carr, T.R., 2003. Reservoir Characterization of Mississippian St. LouisCarbonate Reservoir Systems in Kansas, Stratigraphic and facies ArchitectureModeling, Kansas Geological Survey Open File report, #2003-27.

Qi, Lianshuang, Carr, T.R., Goldstein, R.H., 2007. Geostatistical three-dimensionalmodeling of oolite shoals, St. Louis limestone, southwestern Kansas. Am. Assoc. Pet.Geol. Bull. 91 (1), 69–96.

Read, J.F., 1981. Carbonate platforms of passive (extensional) continental margins:types, characteristics and evolution. Tectonophysics 81, 195–212.

Scholle, P.A., Bebout, D.G., Moore, C. H., (eds.), 1983. Carbonate DepositionalEnvironments. AAPG Memoir. vol. 33, Tulsa, Okla., U.S.A

Swanson, R.G., 1981. Sample Examination Manual. AAPG Methods in Exploration,Tulsa, Okla., U.S.A.

Trappe, H., Hellmich, C., 2000. Using neural networks to predict porosity thickness from3D seismic. First Break 18 (9), 377–384.

Wilkinson, B.H., Owen, R.M., Carroll, A.R., 1985. Submarine hydrothermal weathering,global eustacy, and carbonate polymorphism in Phanerozoic marine oolites. J.Sediment. Petrol. 55, 171–183.

Wilson, J.L., 1975. Carbonate Facies in Geologic History: New York. Springer-Verlag,471.

K.G. Martin et al. Journal of Petroleum Science and Engineering 150 (2017) 1–12

12