SPE 114925 Integrated Petrophysical Evaluation of Shale Gas Reservoirs D. Jacobi, M. Gladkikh, SPE, B. LeCompte, SPE, G. Hursan, SPE, F. Mendez, J. Longo, S. Ong, SPE, M. Bratovich, SPE, and G. Patton, SPE, Baker Hughes, and P. Shoemaker, Shoemaker Exploration Company
Copyright 2008, Society of Petroleum Engineers This paper was prepared for presentation at the CIPC/SPE Gas Technology Symposium 2008 Joint Conference held in Calgary, Alberta, Canada, 16–19 June 2008. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract Gas shales are economically viable hydrocarbon prospects that have proven to be successful in North America. Unlike conventional hydrocarbon prospects, gas shales serve as the source, seal, and the reservoir rock. Generating commercial production from these unique lithofacies requires stimulation through extensive hydraulic fracturing. The absence of an accurate petrophysical model for these unconventional plays makes the prediction of economic productivity and fracturing success risky. This paper presents an integrated approach to petrophysical evaluation of shale gas reservoirs, specifically, the Barnett Shale from the Fort Worth basin is used as an example. The approach makes use of different formation evaluation data, including density, neutron, acoustic, nuclear magnetic resonance, and geochemical logging data. This combination of logging measurements is used to provide lithology, stratigraphy and mineralogy. It also differentiates source rock intervals, classifies depositional facies by their petrophysical and geomechanical properties, and quantifies total organic carbon. The analysis is also employed to locate optimal completion intervals, zones preferable for horizontal sections, and intervals of possible fracture propagation attenuation. Resistivity image analysis complements the approach with the identification of natural and drilling induced fractures. We compare results from three different wells to show the effectiveness of the method for shale gas characterization. The methodology presented provides a means to understand the geomechanical and petrophysical properties of the Barnett Shale. This knowledge can be used to design a selective completion strategy that has the potential to reduce fracturing expenses and optimize well productivity. Though developed specifically for the Barnett Shale, the underlying ideas are applicable to other thermogenic shale gas plays in North America. Introduction Numerous organic-rich shale sections located in some North American basins have been proven as productive natural gas plays (Jarvie et al., 2007; Martini et al., al., 2003; Pollastro et al., 2003; Pollastro Pollastro et al., 2007; Pollastro, Pollastro, 2007). They extend over large geographical areas and offer sustainable reservoirs with attractive exploration and development costs (Hill and Nelson, 2000). Economic production from these complex, kerogen-rich formations, which typically possess poorly-defined gas-water contacts, natural fractures, and very low matrix permeability, depends heavily on the completion technology implemented for recovery. The primary strategy used for stimulating production is hydraulic fracturing, the scale of which can pose a major cost and challenge to producers (Mayerhofer et al., 2006). The challenge is mainly related to the difficulties involved in monitoring and predicting the propagation of the fracturing process through the strata in order to recover potential recover potential reserves (Le Calvez et al., 2006; Mayerhofer et al., 2006; Moore and Ramakrishnan, 2006). This uncertainty can be traced to the varying geomechanical properties associated with the complex lithofacies inherent in many shale gas sections. For this reason, shale gas lithofacies and their relation to reservoir stratigraphy and productivity has recently become a focus of producers (Bowker, 2007; Hickey and Henk, 2007). As a result, the ability to define and categorize in situ the complex lithofacies associated with shale gas plays according to kerogen content, mineralogy, and geomechanical properties has the potential to aid in in reducing the costs costs involved in hydraulic hydraulic fracturing and at the same same time improve hydrocarbon hydrocarbon recovery. To meet that challenge, a shale gas facies expert system was designed to differentiate between the complex facies of the Barnett Shale located in the Fort Worth basin. By using density, neutron, acoustic, nuclear magnetic resonance, and geochemical logging sonde data, the model first determines the lithology and mineralogy, then establishes the lithofacies of the various intervals, and finally selects those zones considered either favorable or non-favorable for hydraulic fracturing, based on their computed geomechanical properties and kerogen content. Lithofacies considered favorable for hydraulic
2
SPE 114925
fracturing are those that possess both a relatively rich organic content and a low value of unconfined compressive rock strength, which is related to their siliceous composition and mineralogy. When these types of facies are fractured, the fractures have a higher probability of propagating throughout the rock. The expert system also selects zones which could cause problems with hydraulic fracturing, those designated as “fracture energy attenuators”, such as carbonates with matrix anisotropies as well as phosphatic hard grounds. Thus an integrated petrophysical evaluation that both differentiates and selects facies favorable for production can potentially reduce well completion costs and enhance production in shale gas plays. Before we discuss the aspects of the expert system, we present some background to aid in understanding the complexity of shale gas systems, with special emphasis on the Barnett Shale to show the relevance of our proposed model for shale gas characterization.
Lithofacies of the Barnett Shale The Barnett interval is dominated by fine-grained (clay- to silt-size) organic-rich sediment with an unusually high siliceous content that is due in part to the abundant tests of radiolarians, sponge spicules and silt composing the matrix (Loucks and Ruppel, 2007; Papazis, 2005). Though it can appear homogeneous, within the bulk section various lithofacies and depositional and erosional bedforms have been recognized that are proposed as due to possible eustasy cycles (sea level changes) (Singh et al., 2007) and or periodic autocyclic depositional events (debris and mud flows) (Jarvie et al., 2005; Hickey and Henk, 2007; Loucks and Ruppel, 2007; Papazis, 2005; Singh et al., 2007). Many of these lithofacies have been described according to four major rock parameters: mineralogy and organic content, paleobiota, rock fabric (laminated/nonlaminated) and rock texture. However, the number and type of facies reported have varied. Most likely this is related to the inhomogeneity of the Barnett section, as the lithofacies are reported to vary according to their proximity to different sediment source areas throughout the Fort Worth basin (Bowker, 2007; Milliken, 2007; Montgomery et al., 2005). As much as the studies do differ, all of them have recognized the following facies to be present: organic siliceous mudstones, which are reported as targets to exploit for gas production (Bowker, 2007; Jarvie et al., 2007); phosphatic (apatite) mudstones that in some instances are condensed “hard grounds” that may be related to flooding surfaces (Loucks and Ruppel; Singh et al., 2007); pyritic facies, containing both euhedral crystals and framboidal masses of pyrite, a mineral that provides further evidence of the reducing, euxinic conditions that were prevalent during the deposition of the Barnett (Loucks and Rupple, 2007; Papazis, 2005); carbonate intervals, both as concretions, mudstones and other carbonate lithofacies have also been identified, the mineral content dominantly composed of either calcite and or dolomite, and in some cases ferroan carbonate, both siderite and ankerite have been reported reported (Jarvie et al., 2005; Hickey and Henk, 2007); and finally finally shelly lithofacies of in situ fossil and transported shelly debris composed o f pelycypods, cephalopods and brachiopods, are also reported prevalent in thin intervals (Hickey and Henk, 2007; Loucks and Ruppel, 2007; Singh et al., 2007). These varied lithofacies lithofacies also contain abundant microfossils of forams, agglutinated forams and condonts composing the sediment (Hickey and Henk, 2007; Loucks and Ruppel, 2007; Papazis, 2005; Singh et al. 2007). Lithofacies Effects on Empirical Methods Used for Shale Characterization The total organic carbon (TOC), the thickness of the section, and the maturation of the kerogen associated with the strata of shale gas plays are among many important factors used to assess a reservoir’s potential productivity (Johnston, 2004; Jarvie, 2007, Bowker, 2007). To characterize these formation properties, conventional logging applications and petrophysical methods have been used to indentify and quantify the kerogen of shale sections along with their corresponding maturities. For example, gamma-ray, as well as overlay methods using resistivity and acoustic data, has been demonstrated effective for determining the kerogen content in shale source rocks (Passey et al., 1990; Fertl and Rieke, 1980). Bulk density measurements have also been suggested as a method for deriving the TOC in the Barnett in some intervals (Hickey and Henk, 2007). And, in some applications, combinations of resistivity, neutron and density have been found to be useful in assessing the kerogen maturity of the Barnett Shale (Zhao et al., 2007). Anomalous high resistivity readings and gamma rays are signature responses for some organic-rich, black, shales (Olson, 1982). These logging responses as well as other methods using conventional logging measurements may be useful to determine the TOC and the maturation of source rocks composing shale gas strata. However, their effectiveness are subject to the complex mineralogical lithofacies that abruptly occur in shale gas sections, similar to those revealed in the previous discussion about the lithofacies of the Barnett Shale. An example of how lithofacies can affect empirical methods such as those discussed above is seen in Fig.1. Uranium has been proposed to exhibit a positive linear relationship with the total organic carbon for some marine source rocks (Fertl and Rieke, 1980). However, the uranium concentrated in organogenic and authigenic calcium phosphate, such as apatite, which is a common mineral that contributes to the phosphatic content of organic marine sections such as the Barnett, must be compensated for before the empirical relationship yields favorable results (Kochenov and Baturin, 2002), (Fig. 1).
SPE 114925
3
8
8
7
7
6
6
5
% t w , C O 4 T e r o C 3
5
% t w , C O 4 T e r o C
y = 0.3x + 2.2 R2 = 0.4
3
2
2
1
1
0
0 0
5
10
15
Core Uranium, ppm
20
25
y = 1.5x - 2.0 R2 = 0.9
0
1
2
3
4
5
6
Core Uranium, ppm
Fig. 1: Correlation between weight percent of TOC and ppm uranium content for core data from a Barnett well with all cores (left), and excluding cores with significant amounts of apatite (right).
The influence of apatite can be seen in Fig. 1. The graph on the left represents the TOC versus the uranium recorded from cores whose mineralogy includes varying amounts of apatite, while the graph on the right shows the relationship after samples containing apatite have been removed. Clearly, the correlation (R 2 = 0.9 as opposed to R 2 = 0.4) between TOC and the uranium is improved when samples containing elevated amounts of apatite are excluded. Phosphatic mudstone lithofacies found in the Barnett are reported to contain the highest TOC and uranium within the section (Hickey and Henk, 2007; Jarvie et al., 2007). In addition to the mineral apatite, both pyrite and siderite found in many lithofacies can elevate rock, grain densities that could also present a challenge when using bulk density as a method for determining the TOC of shale gas intervals. Schmoker (1979) and Schmoker and Hester (1983) recognized large density contrasts between organic and inorganic formations and suggested the use of bulk density logs for TOC evaluation. The authors recognized, however, that density logs alone can only be used if porosity, fluid and TOC density as well as matrix density do not vary significantly over the interval of interest, and empirical corrections are made to eliminate density anomalies caused by the presence of pyrite. The TOC recorded from core data for some shale gas reservoirs has been suggested as a possible predictor of the probable total gas content (Jarvie et al., 2005), (Fig. 2). To infer the TOC empirically, without accounting for facies and mineralogical factors, could cause greater uncertainty with estimates of the total gas value using the Jarvie et al. (2005) method. Because of the potential effect of these mineralogical and lithofacies variances, measuring in situ carbon from the formation using a geochemical logging sonde should be preferred for deriving a TOC estimate. The inability to make a direct measurement of the TOC from the wellbore, and thus the reliance on empirical approaches, is one of the reasons why core analysis is used more readily for determining TOC and estimating the maturation of source rocks (Jarvie 2005, Jarvie et al.,2007; Montgomery et al., 2005). The benefit of making an in situ carbon measurement of the formation, using a geochemical logging sonde (Pemper et al., 2006), will be demonstrated with the discussion of the shale gas facies expert system.
4
SPE 114925
Fig. 2: Correlation between total gas yield and amount of total organic carbon (TOC) for Chattanooga shale (from Jarvie et al., 2005).
Lithofacies Influence on Completion Strategies Lithofacies of shale gas reservoirs can also influence the successful completion and subsequent recovery of gas from wells. Because of the low matrix permeability, gas production from many shale gas reservoirs is dependent on the ability to produce an extensive network of induced fractures into the matrix (Cardot, 2007; Fisher et al., 2005; Johnston, 2004). Hydraulic fracturing, using low-proppant, high-flow-rate water-based methods are usually the stimulus method of choice that is employed to recover the free gas associated with the rock matrix and gas trapped in the kerogen (Gale et al., 2007; Jarvie et al., 2007; Johnston, 2004; Montgomery et al., 2005). Gas stored in shale source rocks is reported to exist in two principal ways: (1) as gas adsorbed (chemical) and absorbed (physical) to or within organic matter and (2) as free gas in pore space and/or fractures (Jarvie et al., 2007). Organic richness, kerogen type, and thermal maturity also influence the amount of absorbed gas in kerogen (Jarvie et al., 2007). The recovery of gas from shales using hydraulic fracture techniques is significantly dependent on the mineralogical homogeneity of the targeted zone, as the petrophysical and geomechanical properties can vary abruptly both vertically and laterally (Boyer et al., 2006). The preferred zones in the Barnett, for example, have been recognized by Jarvie et al. (2005) as very silica-rich intervals with high organic content found both in the upper and lower Barnett. The dominance of these are found in the lower Barnett section, but still prevalent in the upper Barnett in certain regions of the basin. Bowker (2002) in turn has suggested that these preferred lithofacies of the Barnett interval are generally composed of the following mineralogy and TOC, by volume: 45% quartz (mainly as siliceous radiolarian tests and sponge spicules); 27% illite with very minor smectite; 8% calcite + dolomite; 7% feldspar; 5% organic matter; 5% pyrite; 3% siderite; and trace amounts of copper and phosphatic materials. Geomechanically, these silica-rich, low-clay zones are usually the most brittle lithofacies within the interval (Johnston, 2004). Stimulation strategies that target the more brittle zones are reported to improve overall well deliverability (Boyer et al., 2006). Mainly, because the fractures produced in these lithofacies tend to remain open after stimulation, and also aid to create a fracture fairway throughout the strata conducive for gas recovery (Gale et al., 2007; Johnston, 2004). Targeting the more brittle intervals for stimulation is also supported by the strong relationship reported between the thermal maturity of the organic-rich kerogen and the corresponding brittleness of the rock. Both are reported factors that can be used to predict the possible gas flow rate from a given well and subsequently the expected gas production (Jarvie et al., 2007), (Fig. 3). In addition to induced fractures, locating subvertical natural fractures within shale sections are also considered to be advantageous for the recovery of reserves (Johnston, 2004; Gale et al., 2007). In many studies of core from the Barnett, natural fractures are found annealed and filled with calcite (Gale et al., 2007; Papazis, 2005). Other phases of cement mineralization are also found in addition to the calcite including pyrite, albite, quartz, barite and dolomite (Gale et al., 2007). The cements within the fractures are generally not bonded to the grains of the wall rock. Because of this, the mineralization is thought to provide planes of weakness that when intersected by induced hydraulic fractures are postulated to reactivate, thus opening, and optimizing further recovery of reserves (Gale et al. 2007). In other cases, these fractures as well as open fractures detected from core studies are considered by some as very rare, and when present are not factors that contribute to well productivity, and in some instances may be detrimental to well performance (Bowker, 2007; Montgomery et al., 2005). Still, others suggest that open fractures are predicted to occur in the Barnett, based on stress analysis, usually as localized clusters, and must be considered in addition to the closed fractures as possibly enhancing gas permeability locally, but overall could affect the efficiency of hydraulic fracture treatments (Gale et al., 2007). Whatever the agreement or
SPE 114925
5
disagreement about their existence or contribution to well productivity, evidence of their occurring as both cemented and open fractures is interpreted from image logs; fractures both natural and those drilling-induced can be detected using image logs (Johnston, 2004), (Fig. 4).
Fig. 3: Schematic of the gas flow rate as a function of organic richness (TOC), thermal maturity, GOR, and fractures (from Jarvie et al., 2007).
Image logs could be used to correlate fractures to certain lithofacies. For example, Gale et al. (2007) has reported that carbonate concretionary zones found in the Barnett typically have more complex annealed fracture networks than those found in mudstone intervals. However, the fractures tend to terminate abruptly within the concretions. This was not the case for the hydrocarbon-productive mudstones. Thus, this occurence has the possibility of being detected by image logs in correlation with other petrophysical data, provided the thicknesses of these zones are compatible with the resolution of the instrument. Just as stimulation strategy plans are designed to account for certain intervals that enhance hydraulic fracturing, they must also account for other lithofacies that can severly reduce or foul fracture propagation (Johnston, 2004; Gale et al., 2007). For example, fracture propagation, in the more clay-rich zones surrounding the siliceous target intervals, are reported to be less advantageous for producing optimal fracture networks. As a result, the location of these clay facies within the section, by volume and number, are crucial to assess their effect on hydraulic fracturing strategies (Boyer et al., 2006; Johnston, 2004; Cardott, 2006). Furthermore, the location of other lithofacies or structural features that conduct stimulation energy away from the target zones is crucial for developing hydraulic fracture plans. Carbonate lithofacies, with anisotropies due to natural fractures, karsting, vuggy and cavernous porosity, can destroy the efficiency of the hydraulic fracturing process (Johnston, 2004; Jarvie, 2007). Moreover, faults are also considered to attenuate fracture stimulus as the energy is diverted along fault planes away from the targeted zones (Bowker, 2007). When both of these diagenetic and structural features are intersected by induced hydraulic fractures, water can enter the reservoir and curtail gas production generated from the kerogen in the low permeability matrix; this is the primary reason that fracture plans are often designed to avoid intersecting the underlying Viola and Ellenberger carbonate lithofacies at the base of the Barnett section (Bowker, 2007; Johnston, 2004; Gale et al., 2007; Montgomery et al., 2005).
6
SPE 114925
Fig. 4: Example of an image log from a horizontal Barnett well. In addition to the natural fractures (labeled), a large drilling-induced fracture can be seen running along the axis of the borehole (dashed line).
Lithofacies and Porosities The porosity and permeability of shale gas facies is typically very low, below that commonly encountered in conventional reservoirs, yet they are parameters deemed important in determining whether a shale gas reservoir is commercially viable for production (Bowker, 2007; Johnston, 2004; Jarvie, 2007). In the Barnett Shale, in situ porosity measurements derived from conventional logging methods such as neutron-density can be challenged, because of the varied lithological and organic facies of the intervals. As a result of these occurrences, core analysis is heavily relied upon to obtain reservoir porosity and permeability properties (Montgomery et al., 2005). Core analysis results of the productive organic siliceous mudstones of the Barnett often show porosities ranging from 3-6% and permeabilities in the microdarcy to nanodarcy range (Montgomery et al., 2005; Zhao et al., 2007). Unlike neutron, density, and acoustic porosity data, which are affected by all components of the reservoir rock, nuclear magnetic resonance (NMR), which has been used in tight gas formations, has a signal that contains no contribution from the rock matrix and only responds to hydrogen associated with pore-filling fluids. The NMR porosity does not need to be calibrated for lithology or lithofacies changes, eliminating one of the most significant problems of conventional porosity logs, which, if not calibrated properly, may provide inaccurate results. This is particularly beneficial in complex reservoirs such as the Barnett Shale where matrix calibration is difficult due to a high degree of heterogeneity as has been discussed. Also, NMR porosity is immune to the propagation of errors from calculating the fluid volumes. This is especially important in tight reservoirs where the fluid contribution is insignificant compared to that of the rock matrix. As a result of these factors, the NMR measurements applied in the Barnett have shown comparable and reasonable estimates of porosity ranging from 4-6%, compared to that commonly derived from core analysis of the productive siliceous Barnett zones (Fig. 5). Components of the Expert System We have demonstrated in our discussion that the complexity of shale gas lithofacies, similar to that found within the Barnett requires the acquisition of a number of petrophysical measurements for complete reservoir characterization. In order for empirical methods to be effective for estimating TOC, maturation, and gas production from kerogen in the matrix, knowledge of the mineralogy is essential. Moreover, the reliance on these methods can be circumvented by using a more direct geochemical measurement of the formation using geochemical logging sondes. Shale gas lithofacies location, fractures, and porosity within the reservoir are also part of the information that is needed for developing reservoir stimulation strategies and for predicting reservoir productivity. This too can be enhanced by information gathered by a combination of the geochemical
SPE 114925
7
logging sondes, acoustic sonde, density, neutron, NMR instruments and resistivity image devices. The integration of these measurements and their relationships is the basis of the expert system which was designed for deriving shale gas facies to enhance recovery and completion of wells. In the proceeding sections we will be discussing the philosophy and methodology behind the development of the model.
Fig. 5: A log showing NMR, Neutron and Bulk density through a Barnett section. Note the NMR porosity estimation from the section covered, usually determined from core anlaysis ranging from the lowest of 2% to the highest at 4-6%. Compare this trend to that of the neutron porosity which overestimates the porosity. Below X100 where the three curves coincide is the contact of the Ellenberger limestone.
Th/U Ratio The Th/U ratio can be used as a chemostratigraphic method for determining whether ancient marine or continental regimes, or both influenced the depositional cycles associated with lithological strata (Adams and Weaver, 1958). Under oxidizing conditions, such as those that are prevalent in continental depositional environments, uranium will often occur as the uranyl ion (U6+ often as UO 22+). The uranyl ion is soluble and readily forms stable complexes with other anions such as CO 32- and PO43- , which are common polyatomic anions that can increase uranium’s mobility in the sedimentary environment (Langmuir, 1997). Under reducing conditions, which typically occurr in a marine environment, bottom and sediment pore waters can be depleted in oxygen due to the decomposition of elevated concentrations of organic particulate, thus uranium is found concentrated as U4+ (Fig. 6).
8
SPE 114925
Fig. 6: Diagram showing change in uranium concentrations in sediment versus pore water. The marked decreases in pore water 6+ concentrations with depth are paralleled with increases in solid phase uranium. Soluble complexed U at near surface oxidizing 4+ conditions is converted quickly to insoluble U sorbed to clays because organic particulate in the pore w aters increases with depth and promotes reducing conditions below the sediment-water interface (from Calvert and Pederson, 1993).
Uranium as U4+ , which is its most stable valence under reducing conditions, causes uranium’s mobility to be significantly sequestered as it forms insoluble compounds sorbed to clay surfaces, organic matter, organogenic phosphatic particulate, and other colloids found in pore waters such as smectite and iron-oxyhydroxyl complexes (Calvert and Pederson, 1993; Kochenov and Baturin, 2002). Similar, reducing, anoxic conditions could have been prevalent during the Missisipian that led to the accumulation of the organic matter in the Barnett Shale; oxygen-depleted bottom waters became concentrated with organic matter, constantly fostered and supplied by elevated bioproductivity that was promoted by deep upwelling of nutrients into aerated surface waters (Kochenov and Baturin, 2002; Loucks and Ruppel, 2007). The uranium was then concentrated by diffusion from seawater into the euxinic, organic-rich, interstitial sediment, over a prolonged period. (Paxton et al., 2006). It is suggested that increased uranium concentrations in sediments deposited under these conditions can be used as an indicator of the degree of anoxia in a basin (Kochenov and Baturin, 2002; Olsen, 1982). Unlike uranium, no such redox sensitivity is associated with Th in the sedimentary environment, thus its concentration can be directly attributed to the provenance (source area) of the accumulated sediment. The relative immobility of Th, as a stable, conserved, trace element in the marine environment compared to the transient mobility of U due to fluctuations in oxidation-reduction potential is a relationship that can be used to identify the boundaries of the Barnett marine section, and also delineate the possible sequence stratigraphy in the Barnett shale. The Barnett section is reported to have accumulated in a restricted foreland basin whose structural architecture was created as a result of the Ouachita orogeny (Loucks and Ruppel, 2007; Pollastro et al., 2007). The Barnett lithofacies are reported to represent deep water deposition that was influenced by a sediment-starved platform and deposited below the storm-wave base at water depths of between 400-700 ft. (Hickey and Henk, 2007; Loucks, and Ruppel, 2007; Papazis, 2005). Typically, the bottom waters at these depths are not subject to storm activity during lowstands in sea level. Moreover, the Fort.Worth basin’s projected paleogeographic location during the Mississipian indicates that its interconnection with the open oceans was limited (Loucks and Ruppel, 2007; Pollastro et al., 2007)). Thus, the Barnett intervals are postulated to not have been reworked or otherwise significantly influenced by changes in sea level, as is evident by the lack of bioturbation in the beds and the dominance of thinly laminated lithologies, the succession of which is only disturbed by periodic debris and sedimentation associated with mud flows (Hickey and Henk, 2007; Loucks and Ruppel, 2007). However, recent reports of wavy to flaser-bedded deposits discovered within the Barnett from core taken in the Newark East field of the Fort Worth basin, suggest that the intervals–upon closer inspection–could contain sedimentation patterns attributed to both low energy and high energy environments that could be traced to sea level cycles (Singh et al., 2007). The study which also relied upon whole core similar to those used in other studies, suggests that the Barnett may not completely represent a continuous succession of deposition as has been proposed, but there is also evidence that it may be comprised of stacked sequences bounded by intervals signifying abrupt changes in the energy of the depositional environment (Singh et al., 2007). Periodically, the anoxia supported by the deep water column in the basin could have been disturbed by sudden turbidity currents, mud plumes and debris flows (Loucks and Ruppel, 2007) as is seen in the lithofacies recorded from many core studies, but the sedimentation may also have been influenced on a larger scale by eustatic changes in sea level, partly
SPE 114925
9
influenced by the crustal tectonics associated with the Ouchita orogeny. Ross and Ross (1987), in their compilation of sea level curves associated with the Late Paleozoic, suggest there were cyclic drops in sea level of 150 ft. during the Mississipian period. These sea level changes would have occurred during the deposition of the Barnett, and may have been substantial enough to have changed the mobility of uranium, an occurrence that would also record variations in the oxidation-reduction potential in the anoxic bottom waters of the Fort Worth basin. This in turn could also be correlated to changes in lithofacies and the organic content accumulated within the sediments as well. The chemostratigraphic relationship of Th/U compared to the U across the Barnett section suggest there were possible sea level cyclic influences upon the deposition and the accumulation of organic matter within the basin. The chemostratigraphic section of the lower Barnett presented in Fig. 7, shows the possible major transgression and regression periods associated with the section. For example, the sudden spikes marked by sudden increases in U at 652 ft. relative to decreases in Th/U ratio below 2, a value that Adams and Weaver (1958) suggests is characteristic of marine origin, could mark the position of condensed flooding surfaces due to transgressive events. Under these conditions, quiescent, stable anoxic bottom water would prevail, thus supporting accumulations of organic matter in the sediments (Kochenov and Baturin, 2002). These condensed intervals often contain elevated phosphatic content, clay, and the highest TOC related to what are called “hard grounds”.
Fig. 7: Plot of Th/U versus U for the Barnett shale from a well in depicting the use of the data and the architecture of the curve for delineating sequence stratigraphy.
This sudden spike in uranium previously described is then followed by a gradual decrease in uranium, up the section, to 580 ft., which could be related to a gradual regression of sea levels. The Th/U ratio gradually moving above 2 coupled with the drop in the uranium through the section could mark basinal changes in anoxia due to low stands in sea level. Thus, more oxidizing conditions are introduced by continental influences due to the fall in sea level. During such events, increased erosion would occur and detrital deposition could have migrated seaward introducing more detrital silt or quartz and feldspar and in some cases carbonate debris into the deep water column during the accumulation of the organic-rich sediment, an occurrence that could increase the matrix porosity. Together, the trangressive and regressive episodes would constitute one cycle of sea level change. This is then followed by yet another cycle and so on according to Fig. 7. Such an approach can be used provided that there is an understanding that slight variations in U that occur throughout these trends could be related to depositional events such as slope failures or mudflows, autocyclic events that introduced carbonate debris or mud which contribute to slight disturbances in the redox conditions in the deep water column. However, the main signal where spikes and gradual overall declines are seen is what is needed to assess the possible eustasy changes that occurred in the basin.
10
SPE 114925
The Shale-Gas Facies Expert System The chemostratigraphic trace of U and Th/U across the Barnett shows how various trends between the two could be indicative of eustatic changes within the Ft. Worth basin. Using this relationship, along with the addition of geochemical data (chemistry and mineralogy), we contend that it is possible to identify various depositional facies described in the sections previously discussed. Most importantly, it is possible to identify silty intervals (quartz-rich, brittle zones favorable for hydraulic fracturing), organic-rich black shale (high kerogen content and source of gas), calcareous zones, and phosphatic facies (both as possible zones of fracture energy attenuation). Based on these criteria, and information from the literature and logging data from four wells in the Barnett Shale, an algorithm was developed for identifying depositional facies in the Barnett Shale. The algorithm is a rule-based expert system. The expert system uses geochemical data from a new geochemical logging sonde called the Formation Lithology Explorer (FLEX sm )and Spectra Log II (SLIIsm ), which, used in tandem, measures and collects Si, Ca, Fe, S, Ti, Gd, Cl, Mg, Al, and C, and naturally occurring radioactive K, U, and Th data from the formation (Pemper et al., 2006). This chemical data is then used to determine the general lithology, then the specific lithology and finally the mineralogy and the TOC for each interval measured (Jacobi et al., 2007). The following minerals can be readily computed: quartz, K-feldspar, plagioclase, calcite, dolomite, siderite, pyrite, anhydrite, smectite, illite, chlorite, kaolinite, glauconite, apatite and organic carbon, either as kerogen, coal, or oil. The results from these computations and measurements are then used as input into the shale gas facies expert system to distinguish between seven different lithofacies in the Barnett Shale: •
Organic-rich shale (Si-rich, high TOC)
•
Non-siliceous organic-rich shale
•
Low organic shale
•
Siliceous mudstone
•
Calcareous mudstone
•
Phosphatic mudstone
•
Phosphatic Zone
Pyritic zone The input data includes U, and Th; elemental data (Ca, C and Si); and the computed mineralogy from the measured elemental data (calcite, dolomite, and pyrite), (Fig. 8). The lithofacies of low-organic shale (grey) and non-siliceous organic-rich shale (orange) are rare in the Barnett and signify that the interval is either classified as a poor source rock or lies outside of the typical Barnett organic-rich facies. The most important facies is siliceous mudstone (yellow), the deposition of which could have been influenced by gradual declines in sea level. We contend this facies is optimal for initiating hydraulic fracturing, due to its siliceous composition, possible elevated detrital content, and therefore higher porosity than might be expected from this zone. This zone is considered a favorable facies for the entrapment of free gas expelled from the kerogen during the process of hydrocarbon cracking in the organic-rich intervals shown by the organic black shale facies (black). Moreover the phosphatic (green) and calcareous facies (blue) as well as the pyritic facies (red) are considered non-essential to the stimulation of production in the Barnett, and in some cases may serve as fracture energy attenuators. The degree to which the above lithofacies is designated as favorable or non-favorable for hydraulic fracturing is based on the geomechanical properties computed from mineralogy, the TOC and the facies of each formation. Using geomechanical software, acoustic data, (acoustic sonde (XMACsm)), coupled with mineralogy, TOC, density and porosity is used to compute the following static elastic rock properties for each interval: Poisson’s ratio, Young’s Modulus, and unconfined compressive strength (UCS). The porosity value, in the absence of neutron-density, can be provided by nuclear magnetic resonance data from the NMR sonde (MREX sm). •
SPE 114925
11
Compute Apatite Weight Fraction
Apatite
Phosphatic zone
Calcareous mudstone
Total carbonate
Pyrite concretion
Pyrite
Siliceous mudstone
Black shale Si
Uranium and Th/U
Low-organic shale
Si Non-siliceous black shale
Fig. 8: A simplified version of the flow diagram for the expert system used to identify the facies within the Barnett. These various facies are coded according to color and their existence is presented in a single logging track as an interval in which they occur across the succession (Fig. 10).
Elevated values of UCS above a certain threshold associated with some intervals indicate more indurated rock integrity, while lower values correspond to less obdurate and more brittle intervals. Since the character of Barnett Shale, its composition, mineralogy, and geomechanical properties are very complicated, we use the UCS curve only in a qualitative sense. Namely, we compute the maximal value of UCS for the whole Barnett shale section and then further compare calculated UCS for each depth with this maximal value, taking into account the facies, established earlier. If the UCS value at a given depth is below a certain threshold relative to the maximal UCS value, which is connected with the rock integrity, and the percentage of TOC is considered productive (organic-rich zone), and the interval is identified as siliceous mudstone facies, then the zone can be selected as a preferred hydraulic fracturing interval (green). If, on the other hand, the UCS value is greater than a threshhold value and the interval is identified as a calcareous mudstone facies or phosphatic zone, then the interval can be designated as an interval for possible fracture energy attenuation (red), again all based on the UCS value and TOC for the interval (Fig 9).
If in siliceous m mudstone UCS and TOC
Preferred fracturing zone
If in calcar eous m mudstone or phosphatic z zone – UCS and TOC
Zone of fracture energy attenuation
Fig. 9: A flow diagram depicting the identification of preferred zones for hydraulic fracturing, versus those that are considered as fracture energy attenuators. Input data are TOC, UCS and lithofacies. These various intervals are coded according to their color and their existence is presented as a single logging track beside the facies interval in which they occur.
12
SPE 114925
Results From the Expert System Applied to the Barnett Shale The demonstration of the shale gas expert system for characterization of the Barnett shale according to the facies and those zones designated as both favorable and non-favorable for hydraulic fracturing is presented for three logs from the Fort Worth basin. The intervals depicted correspond to the lower Barnett shale down to the underlying Ellenberger carbonate. We will refer to the logging header in Fig.10 and the bulleted list below for further interpretation of the various logging measurements and computations used in the expert system. The tracks from left to right can be used to interpret the data of the logs (Figs. 11-13) presented on succeeding pages: Track 1 presents Gamma Ray (green), Caliper (red), and SP (black dashed line) • •
Track 2 – depth
•
Track 3 – resistivity
•
Track 4 – bulk density (blue), neutron porosity (red), and PE (pink)
•
Track 5 – K (wt %) (red) and Th (ppm) (blue)
•
Track 6 – Si (wt%) (yellow) and Ca (wt%) (light blue)
•
Track 7 – Mg (wt%) (dark blue) and Fe (wt%) (brown)
•
Track 8 – Al (wt%)(grey) and S (wt%) (pink)
•
Track 9 – C (wt%) (green) and Ti (wt %) (magenta)
•
Track 10 – UCS (unconfined compressive strength, MPa)
•
Track 11 – curves for TOC (total organic carbon) (wt%)
•
Track 12 – U (red) (in ppm) and Th/U ratio (blue)
•
Track 13 – facies (colors correspond to the facies color shown in Fig. 8)
•
Track 14 – frac intervals (colors correspond to the intervals colors in Fig. 9)
•
Tracks 15, 16, and 17 – general lithology, specific lithology, and mineralogy as weight percent, respectively.
Discussion The results presented from the three logs in Figs. 11, 12, and 13 were computed using the shale gas facies expert system and show how integration of petrophysical measurements along with geochemical and mineralogical data measured from the Barnett formation can produce a more selective method for characterizing the complex facies that constitute shale gas plays. In Track 17, one can trace the complex mineralogy of the Barnett, the differences in the amounts of clay and clay types, illite (dark gray – ranging from 20-30%) and smectite (light gray – 2-5%) as well as quartz (yellow – 40-60%) and carbonate minerals, dolomite (dark blue) and calcite (light blue), between all three Barnett successions. Plagioclase (red cross hatched) and K-feldspar (green cross hatched) and minor amounts of pyrite (red), siderite (orange) and apatite (light green) are also detected in varying amounts. Moreover, differences in the TOC between wells can also be seen from Track 11 and is also included as part of the matrix as weight percent in Track 17. All of the mineralogy and TOC is derived from the chemistry (Tracks 5 -9) supplied by the geochemical logging sondes. The TOC for each interval is computed using the following relationship: Organic C = [ C
total –
C calcite – C Dolomite – C Siderite ]
C total is the C measured from geochemical logging sonde, and CCalcite, CDolomite and C amount of these minerals computed from geochemical data from the intervals.
(1) siderite is
the carbon derived from the
SPE 114925
13
Fig. 10: Log headers for reference to interpreting the three succeeding logs (Figs. 11-13) presented for evaluation of the shale gas facies expert system.
14
SPE 114925
1
2
3
4
5
6
Fig. 11: Well example #1 of the Barnett Shale.
7
8
9
10
11
12
13 14 15 16
17
SPE 114925
1
15
2
3
4
5
6
Fig. 12: Well example # 2 of the Barnett Shale.
7
8
9
10
11
12
13 14 15 16
17
16
SPE 114925
1
2
3
4
5
6
Fig. 13: Well example # 3 of the Barnett Shale.
7
8
9
10
11
12
13 14 15 16
17
SPE 114925
17
The mineralogy and chemistry is key not only for the shale gas facies model, but is also needed for interpreting the sequences of deposition using the Th/U ratio versus U. When comparing the facies (Track 13) between all of the wells, three main facies are prevalent throughout the section: the siliceous mudstone (yellow zones), organic rich mudstones (black zones) and calcareous mudstone intervals (blue). Two phospatic facies (green) occur based on the computed apatite in Example 3 (Fig. 13). One is located just above the 150-ft. mark and in Example 2 at the base of the section near the contact of the Ellenberger. The organic-rich shales correlate with increased uranium, (Track 12) which is indicative of the possible deposition of the strata during highstands in sea level. The siliceous mudstones are not marked by these increases in uranium, a sign of uranium’s mobility during sea level declines and changing oxidation-reduction potential. One major spike in uranium occurs just above the Ellenberger carbonate which is exhibited in all three wells and could represent a major trangressive flooding surface. This serves as a datum for segregating and interpreting the position of the remaining depositional sequences, moving up the section. The variation in the Th/U ratio with U has been used to suggest the existence of four major depositional sequences as shown from the black bracketed intervals in the three examples (Figs. 11, 12, and 13). The beginning of these sequences is marked by a sudden spike in uranium with a sudden decrease in the Th/U ratio well below 2. This is followed by a gradual decline in uranium down to a minimum marked by an increase in Th/U above 2. The next major depositional cycle is then initiated by a sudden increase in uranium with yet another corresponding decline in the Th/U ratio. As one can see from the shale gas facies (Track 13), these cycles coincide with, first, a series of organic-rich mudstones and then followed and capped by siliceous mudstone intervals. The start and cessation of these proposed depositional cycles are rather distinct on log wells # 1 (Fig.11) and # 3 (Fig.13) and less so on #2 (Fig. 12). The facies intervals found within each cycle also differ in thickness and frequency from well to well, an occurrence that could be attributed to the the localized basin architecture, source area for the sediments and or depositional conditions as the Barnett has been reported to vary both vertically and horizontally throughout the basin. Also, smaller depositional sequences are detected within the major sequences which could be due to less notable changes in sea level. Furthermore, while uranium and the Th/U ratio has been demonstrated to have trends associated with the deposition of the Barnett, one must take into account that the sudden declines in the signal can be attributed to periodic mudflows containing calcareous fossil debris and other carbonate lithofacies and diagenetic lithofacies such as concretionary zones (Loucks and Ruppel, 2007; Papazis, 2005). Typically, the calcite and dolomite composing carbonate lithologies do not contain uranium, thus in those intervals where the mineralogy or facies is dominated by carbonate, one must conclude that part of the decline in uranium can be related to this factor. As a result, the facies and the mineralogy can be used to discern when these could have influenced the signal. These calcareous facies are determined to occur throughout the sections in all three wells exhibited as is evident in the facies track 13 (blue intervals). Along with the identification of the facies present in all three wells, the expert system also selected those intervals favorable for hydraulic fracturing and those considered fracture energy attenuators. The results are found in Track 14. In Example 1 (Fig. 11), the favorable zones (green) for fracturing within the siliceous mudstone facies are considered small in both number and volume throughout the interval. This suggests that although these are siliceous mudstones, they do not possess the unconfined compressive strengths needed for propagation of fractures. The potential productivity can be determined from their average TOC content (3-5%) as well. TOC can often be correlated to the potential gas production from a given well. As a result of both the TOC and the computed geomechanical properties, the expert system does not select many zones as favorable. The organic mudstone facies appears as the dominant lithofacies compared to the siliceous mudstone facies which are somewhat thin, thus the propagation of fractures through these organic mudstones, which are considered less brittle than the siliceous mudstone, would not create extensive open fracture fairways for recovery of gas reserves. The facies considered detrimental to hydraulic fracturing within this well, is the underlying Ellenberger limestone, so designated by the red beside the carbonate facies interval (Track 14). Compared to Well 1, Well 3 contains thicker and more siliceous mudstone facies. Many of these are selected as favorable for hydraulic fracturing due in part to the combination of their TOC (3-8%) and computed UCS values. The volume and number of the siliceous mudstone facies present in Well 2 are comparable to those in Example 1, however, because of the geomechanical and TOC (3-5%) properties these are considered to be acceptable for fracturing for recovery of the reserves contained in the section. Again, similar to Well 1, those facies designated as stimulus barriers are designated in red, in this case the Ellenberger in Example 2 and the limestone section in Well 3. The facies results for the Ellenberger in Example 3 are not presented, otherwise it to would be considered a fracture stimulus barrier.
Confirmation of the Expert Sytem Using Production Data The shale gas facies model thus introduced and presented has been applied to several wells in the Barnett for general evaluation, but to date, whole core data has not been available to test the results. The model has primarily relied on confirmation from cuttings analysis and reference literature. However, recent production data provided by a customer from these three different wells located in different parts of the Fort Worth basin indicate that the model shows promise for providing useful information for recovery plans. This is especially true, when comparisons are made between the numbers of favorable productive fracture zones predicted by the facies model, and the volume of gas produced from each of the three wells. The well in Eexample 1, the least productive in gas production compared to Wells 2 and 3, was predicted by the shale gas facies model to have the least number of organic siliceous facies which are considered ideal for fracturing and productive.
18
SPE 114925
In contrast, the volume of these types of facies along with the organic rich mudstones in both Wells 2 and 3 are more common in comparison. Table 1 shows the accumulative production rates for these wells from February to July 2007, and the corresponding logs in Fig.11, 12, and 13 . .
Well Name
Gas MCF
Water BBLS
#1 #2 #3
185,463 249,618 240,539
102,312 25,301 35,322
Table 1. Gas production recorded for each well (Examples 1-3 (Figs. 11-13)) from February to July 2007.
A distribution of lithologies and organic content differs between the wells shown in Figs. 11-13. Along the entire log of Well 1, one can see substantially more clay and shale zones and less siliceous intervals as well as slightly lower TOC levels (average 3-5%) than those in Wells 2 and 3, where the siliceous facies and organic rich intervals seem to dominate (TOC- 38%). Because the shale gas facies model relies on the mineralogy and TOC for computation of the geomechanical properties of the intervals, the brittleness of the lithofacies in Well 1 is significantly less than in the other two wells. One can see that most of the green zones (Track 14) in both Wells 2 and 3 coincide with the more siliceous silty intervals established earlier by the mineralogy and the shale gas facies model using the geochemical logs. In addition, the organic-rich intervals (intervals in black) in Well 2 are slightly greater in volume than that from Well 3, which also correlates with slightly larger gas production found with Well 2. Incidentally, Well 1, the poorest in gas production, was completed first and is the oldest by a year, followed by Well 3, and finally Well 2, which is the youngest prospect, but the highest in production.
Confirmation of the Expert System Using Image Logs The use of both acoustic and resistivity image logs for shale gas characterization is considered necessary for noting differences in strata that could provide further evidence of productive zones based on facies changes. When the expert system’s results from an interval in Well 1 (Fig.11) are compared to resistivity image logs, a reasonable correlation is found with the derived facies (Fig. 14). Using the static resistivity image (second track from the left), one can see starting from the bottom that there are three distinct changes in resistivity contrasts. The static normalized image is preferred because the increased contrast of the image provides a better indication of the bulk electro-facies than the dynamic image because of the complexity of the layering in the lithofacies. The first zone near the bottom exhibits a high resistivity (lighter banding) that corresponds to an organic-rich mud facies computed by the expert sytem. This facies is also coincident with an increase in TOC as shown by the organic carbon (black) computed in the mineralogy found in the last track. Without the additional information provided by the facies model and the mineralogy, the high resistivity contrasts using image data alone could be misinterpreted as a carbonate concretion or other high resistivity facies. Instead the mineralogy and lithology and the facies expert system analysis indicate that the zone is organic-rich black shale. Also, this zone contains minor amounts of apatite (light green) which is often associated with these organic-rich facies and can contribute further to the high resistivity. The occurrence of this mineral also coincides with an increase in uranium as well. Above the organic-rich mudstone facies is a transition to a siliceous mudstone, which is marked by a corresponding decrease in resistivity. In this zone, the organic matter, as part of the matrix, is seen to decrease in concentration moving up the section. This facies is then followed by yet another increase in resistivity from the image log, which coincides with the occurrence of a carbonate interval that is confirmed by the facies expert system, the lithology and also the mineralogy computed for the section. The general lithology is a carbonate, and the specific lithology is a shaley carbonate whose mineralogy is mainly calcite, quartz and illite. Note that the image log shows resistivity contrasts that are indicative of laminated sequences that are common to shaley carbonates. NMR, Density, Neutron and Geochemistry Used for Estimation of TOC The description of the organic contents of source rocks remains a challenging task for the geoscience community to this date. With no readily available direct measurements, the majority of the practical methods are based on indirect correlations with conventional log responses and/or core analysis deliverables as was discussed previously. In the unconventional Barnett shale section, kerogen is a substantial part of the matrix, however, since the density of the kerogen changes with the degree of maturation and the amount of gas contained therein, computing an accurate grain density with mineralogy including carbon as part of the matrix can be challenging. The computation requires carbon to be treated as part of the matrix, which is in sharp contrast to conventional reservoirs where hydrocarbon associated with the rock is contained in the pores, thus it does not influence the matrix density. This occurrence has provided another opportunity for NMR to provide an independent TOC estimate from that computed using the geochemical elemental data. We have found that the NMR porosity measurement along with the density measurement can be used to compute a grain density for comparison with that derived from the inorganic grain density computed from the mineralogy using geochemical data. The NMR measurement responds to the
SPE 114925
19
volume of movable fluids in the pore spaces, with the assumption that the measurement is independent of the volume of kerogen comprising shale gas plays. In contrast, the bulk density measurement responds to the whole rock, including fluids and kerogen contained in the matrix. Thus the difference between the two grain densities can be used to assess the TOC of the formation. This in turn can be used to further verify the TOC derived using the carbon obtained from the downhole geochemical sonde.
U / h T
s e i c a F
y y g g o o l l o o h h t t i i l l L L l c a i r f i e c n e e p G S
y g o l a r e n i M
sm
Fig. 14: Resistivity image log (STAR ) compared to the shale gas facies model results for an interval in Well Example 1. Lithology and mineralogy and the facies derived are computed from geochemichal logs.
Figure 15 shows a log from a 150-ft. section with gamma-ray, density, and mineral grain density (computed from the mineralogy) derived from the geochemical logging data and NMR logging data acquired from a well drilled through the Barnett Shale. The gamma-ray track (Track 1) indicates a sharp boundary between the Barnett Shale and the underlying carbonate formation. Schmocker’s (1979) observation of anomalously low density within the organic-rich zone is clearly verified by the bulk density log, shown as a solid curve in Track 2. With a typical matrix density estimate of 2.71 g/cm 3 for the Barnett Shale, density porosity often exceeds 10 p.u. while NMR total porosity, presented as a dashed curve in Track 2,
20
SPE 114925
remains at or below 6 p.u. If fluid density is known, the matrix density can be directly calculated from the bulk density and NMR porosity logs. Track 5 shows that the pore fluids responding to NMR are almost exlclusively clay-bound and capillary bound water in the Barnett Shale, suggesting a fluid density of 1 g/cm3 for the calculation of total matrix density, displayed in Track 3 as a red curve. Note that the matrix density is slightly above 2.7 in the carbonate while it drops to 2.5-2.6 g/cm 3 in the organic shale. The density of the inorganic matrix, shown as the blue curve in Track 3, can be determined based on the detailed mineralogic analysis provided by geochemical logs. As a means of quality-control in zones without organic matrix constituents, the inorganic and total matrix densities should match as seen in the carbonate zone below (X100 - X150) the Barnett Shale. The shaded area in Track 3 highlights the difference in the total matrix density with respect to the inorganic matrix density. These differences can be converted into a TOC weight fraction if a good estimate of the organic material density is available from core studies or other geologic information. The TOC weight fraction is presented in Track 4 as the green shaded area. If no organic density data is available, we propose a log-based procedure to estimate TOC density, by comparing the density-NMR based TOC with the TOC provided by the geochemical sonde. The gray shaded area in Track 4 displays the TOC weight fraction directly obtained from mineralogy analysis. If the organic material density estimate is correct, these two curves should provide similar results. The cross-plot between the density-NMR based TOC assuming ρTOC = 1.2 g/cm3 and the geochemical TOC is shown in Fig. 16. Points above the diagonal suggest a TOC density below 1.2 g/cm 3 while entries below the diagonal line indicate a TOC density above 1.2 g/cm 3. Despite the relatively large scatter mainly due to different curve resolutions, it is apparent that most points fall below the identity line, indicating that the density of TOC exceeds 1.2 g/cm3 (Fig 16 ). We adjust the TOC density estimate until the number of points above and below the diagonal is equal or similar. With this procedure we determined an organic material density of 1.44 g/cm 3 (Fig. 17). The density-NMR based TOC curve shown in Track 4 is computed using ρTOC = 1.44 g/cm3.
Fig. 15: A log from a Barnett well depicting the relationship for establishing TOC from inorganic grain densities computed from geochemical sonde-derived mineralogy versus that computed from NMR and density.
SPE 114925
21
10
) T N E C R E P T H G I E W ( C O T Y T I S N E D R M N
ρ
est =1.20g/cm 3 TOC
8 ρ
est <1.20 g/cm3 TOC
6
4
2 ρ
0
0
est >1.20 g/cm3 TOC
2 4 6 8 ROCKVIEW TOC (WEIGHT PERCENT)
10
3
Fig. 16: Comparison between density-NMR derived TOC (assuming TOC density of 1.2 g/cm ) weight fraction and TOC computed from geochemical logging data. 10
) T N E C R E P T H G I E W ( C O T Y T I S N E D R M N
ρ
est =1.44g/cm 3 TOC
8 ρ
est <1.44 g/cm3 TOC
6
4
2 ρ
0
0
est >1.44 g/cm3 TOC
2 4 6 8 ROCKVIEW TOC (WEIGHT PERCENT)
10
Fig. 17: Comparison between density-NMR derived TOC weight fraction and TOC computed from geochemical logging data. TOC 3 density is adjusted to 1.44 g/cm , so that number of points above and below 1:1 line is equal.
Conclusion The shale gas facies expert system was developed to identify the complex facies of the Barnett Shale and also select those zones best suited for hydraulic fracturing. The model and algorithms rely on the integration of a number of downhole measurements using acoustic, density, neutron, NMR and geochemical sondes. All these tools, when used together, show promise to minimize the number of intervals selected for hydraulic fracturing within a play and also to avoid those zones considered detrimental to fracture network propagation. We contend the model can be used to minimize the costs involved in stimulating production in shale gas reservoirs and enhance recovery and improve completion strategies. We have demonstrated this claim indirectly from the well production data for the three wells discussed. Qualitatively, the expert system model predicted which of the three wells would be the most productive based on the number of favorable siliceous mudstone facies intervals selected based on the geomechanical properties (UCS) versus the TOC content computed for each well.
22
SPE 114925
We have also demonstrated that the application of NMR for estimation of porosity in shale gas sections provides, relative to the published core analysis values a more accurate porosity as opposed to that derived from neutron-density methods. The interpretation of resistivity image logs is improved with correlation to the shale gas facies, lithology and mineralogy computed using our methods. Finally, we have found using Th/U versus U that four major depositional sequences related to eustasy changes are possibly present within the Barnett section. Moreover, the use of the Th/U versus U for describing the possible sequence stratigraphy of the Barnett needs validating with petrographic studies. These might be able to confirm differences in grain sizes such as coarsening upward or fining upward in the sequences compared to the trend of Th/U and U as indicating a fall or rise in sea level as has been proposed by this paper; remembering that some of the variation in lithofacies in the Barnett has already been proposed as attributed to high energy and low energy environments (Singh et al., 2007). The Th/U and U need to be considered further to investigate whether there is a relationship. We look forward to further validation of the expert system as we continue to test the facies model on other wells with available core data. This is a subject of future research.
References 1. 2. 3.
4. 5. 6. 7. 8. 9. 10. 11. 12.
13.
14. 15. 16. 17. 18. 19.
20. 21. 22. 23. 24. 25.
Adams, J.S. and Weaver, C.E., 1958, “Thorium-uranium ratios as indicators of sedimentary processes: example of concept of geochemical facies”, AAPG Bulletin, 42 (2), p. 387-430. Boyer, C., Kieschnick, J., Suarez-Rivera, R., Lewis, R., and Waters, G., 2006, “Producing gas from its source”, Oilfield Review, Autumn 2006, p. 36-49. Bowker, K.A., 2002, “ Recent developments of the Barnett Shale play, Fort Worth basin”, in B.E. Law and M Wilson, eds., Innovative Gas Exploration Concepts Symposium: Rocky Mountain Association of Geologists and Petroleum Technology Transfer Council, p. 16. Bowker, K.A., 2007, “Barnett shale gas production, Fort Worth basin: issues and discussion”, AAPG Bulletin, 91, p. 523-533. Calvert, S.E. and Pederson T.F., 1993, “Geochemistry of recent oxic and anoxic marine sediments: implications for the geological record”, Marine Geology, 113, p. 67-88. Cardott, B.J. 2006, “Gas shales tricky to understand: energy mineral division”, AAPG Explorer, www.aapg.org/explorer/divisions/2006. Fisher, M.K., Wright, C.A., Davidson, B.M., Goodwin, A.K., Fielder, E.O., Buckler, W.S., and Steinsberger, N.P., 2005, “Integrating fracture-mapping technologies to improve stimulations in the Barnett Shale”, SPE paper 77441. Gale, J.F.W., Reed, R.M. and Holder, J., 2007, “Natural fractures in the Barnett Shale and their importance for hydraulic fracture treatments, AAPG Bulletin, 91, p. 603-622. Hickey, J.J, and Henk, B., 2007, “Lithofacies summary of the Mississippian Barnett Shale, Mitchell 2 T.P. Sims Well, Wise County, Texas”, AAPG Bulletin, 91, p. 437-443. Hill, D.G and Nelson C.R., 2000, “ Gas productive fractured shales: an overview and update”, Gas TIPS , Summer, p. 4-13. Jacobi, D., Longo, J. M., Sommer, A., Pemper, R., 2007, “A chemistry-based expert system for mineral quantification of sandstones”, presented at Petrotech 2007 Oil and Gas Conference and Exhibition. Jarvie, D.M., Hill, R.J., Pollastro, R.M., Wavrek, D.A., Bowker, K.A., Claxton, B.M., and Tobey, M.H., 2003, “Evaluation of unconventional natural gas prospects: the Barnett Shale fractured natural gas model”, 21st International Meeting on Organic Geochemistry, September 8-12, Krakow, Poland. Jarvie, D.M., Hill, and R.J., Pollastro, 2005, “Assessment of the gas potential and yields from shales: the Barnett Shale model”, in B. Cardott, ed., Oklahoma Geological Survey Circular 110: Unconventional Gas of the Southern Mid-Continent Symposium, March 9-10, Oklahoma City, OK, p. 37-50. Jarvie, D.M., Hill, R.J., Ruble, T.E., and Pollastro, R.M., 2007, “Unconventional shale-gas systems: the Mississippian Barnett shale of North-Central Texas as one model for thermogenic shale-gas assessment”, AAPG Bulletin, 9, p. 475-499. Johnston D., 2004, “Technological advances expand potential pay”, Oil and Gas Journal , p. 51-59. Kochenov and Baturin, 2002, “The paragenesis of organic matter, phosphorus, and uranium in marine sediments”, Lithology and Mineral Resources, 37(2). Lancaster, D. E., McKetta, S.F., Hill, R.E., Guidry, F.K., and Jochen, J.E., 1992, “Reservoir evaluation, completion techniques, and recent results from the Barnett Shale development in the Fort Worth basin”, paper SPE 24884. Langmuir D., 1997, “Aqueous environmental geochemistry”, New Jersey, Prentice Hall. Le Calvez, J.H., Tanner, K.V., Glenn, S., Kaufman, P., Sarver, D.S., Bennett, L., Panse, R., and Palacio, J.C., 2006, “Using induced microseismicity to monitor hydraulic fracture treatment: a tool to improve completion techniques and reservoir management”, SPE paper 104570. Loucks, R.G., and Ruppel, S.C., 2007, “Mississippian Barnett Shale: lithofacies and depositional setting of a deep-water shalegas succession in the Fort Worth basin, Texas”, AAPG Bulletin, 91, p. 579-601. Mayerhofer, M.J., Lolon, E.P., Youngblood, J.E., and Heinze, J.R., 2006, “Integration of microseismic fracture mapping results with numerical fracture network production modeling in the Barnett Shale”, SPE paper 102103. Milliken, K., 2007 personal communication. Montgomery, S.L., Jarvie, D.M., Bowker, K.A., and Pollastro, R.M., 2005, “Mississippian Barnett Shale, Fort Worth basin, North-Central Texas: gas-shale play with multi-trillion cubic foot potential”, AAPG Bulletin, 89, p. 155-175. Moore, L.P., and Ramakrishnan, H., 2006, “Restimulation: candidate selection methodologies and treatment optimization”, SPE paper 102681. Olsen, R.K., 1982, “Factors controlling uranium distribution in Upper Devonian-Lower Mississipian black shales of Oklahoma and Arkansas”, Dissertation, University of Tulsa, Tulsa, Oklahoma.
SPE 114925
26.
27. 28. 29.
30. 31. 32.
33.
23
Papazis, P., 2005, “Petrographic characterization of the Barnett shale, Fort Worth Basin, Texas”, Master’s Thesis, University of Texas at Austin, Austin, TX, 142 p., CD-ROM (SW0015), available from Bureau of Economic Geology, University of Texas at Austin. Paxton, S.T., Cruse A.M., and Krystyniak A.M., 2006, “Fingerprints of global sea-level change revealed in hydrocarbon source rock?”, AAPG presentation, Houston, TX. Pemper, R., Sommer, A., Guo, P., Jacobi, D., Longo, J., Bliven, S., Rodriguez, E., Mendez, F., Han, X., 2006, “A new pulsed neutron sonde for derivation of formation lithology and mineralogy”, SPE paper 102770. Pollastro, R.M., Hill, R.J., Jarvie, D.M., and Henry, M.E., 2003, “Assessing undiscovered resources of the Barnett-Paleozoic total petroleum system, Bend Arch – Fort Worth Basin Province, Texas”, AAPG Southwest Section Convention, Fort Worth, Texas, March 1-5. Pollastro, R.M., Jarvie, D.M., Hill, R.J. and Adams, C.W., 2007, “Geologic framework of the Mississipian Barnett Shale, Barnett-Paleozoic total petroleum system, Bend arch-Fort Worth Basin, Texas”, AAPG Bulletin, 91, p. 405-436. Pollastro, R.M., 2007, “Total petroleum system assessment of undiscovered resources in the giant Barnett Shale continuous (unconventional) gas accumulation, Fort Worth Basin”, AAPG Bulletin, 91, p. 551-578. Ross, C.A., and J.R. P. Ross, 1987, “Late Paleozoic sea levels and depositional sequences” in, C.A. Ross and D. Haman, eds., Timing and deposition of eustatic sequences: Contrains on seismic stratigraphy: Cushman Foundation for Foraminiferal Research Special Publication, 24, p. 137-149. Singh, P., Slatt, R., Rai, C., Sond ergeld, C., and Stamm, C., 2007, “Stratigraphy of the Barnett Shale, Newark East Field, Texas”, AAPG Southwest Section Convention, Wichita Falls, TX, April.
