1

Fracture Interpretation in the Barnett Shale, using Macro and Microseismic Data

Murray Roth* (murray@transformsw.com) Transform Software and Services, Littleton, Colorado, USA

and Amanda Thompson

Devon Energy, Oklahoma City, Oklahoma, USA

Summary

The Mississippian Barnett Shale of the Fort Worth Basin (Texas, USA) is potentially the largest gas field in North America. As with most tight-gas fields, successful Barnett well creation requires accurate well path placement in the gas zone: with judicious alignment of the well path relative to existing fractures and current stress orientations, followed by effective hydraulic fracturing. Previous work has established a current stress orientation of approximately 50 to 60 degrees east of north. Microseismic data recorded during hydraulic fracturing in a horizontal well in the study area exhibits a very diffuse spatial pattern and dramatic differences in microseismic activity, in adjacent fracture stages. Natural karst-related collapse chimneys are known to exist in this region, with larger structural collapses visible on auto-tracked seismic horizons. A combination of volume seismic curvature, incoherence and dip/azimuth estimates provided complementary indicators for the location, extent and magnitude of structural collapse. Crossplot analysis of micro- and macroseismic data highlighted the correlation of induced seismicity and natural karst-related fracturing. Volume interpretation of calculated seismic attributes was used to map collapse chimneys in three dimensions. Composite depth displays of seismic horizons and karst bodies provided an ideal environment for horizontal well planning, to avoid tapping into water-bearing formations. Background

The Barnett Shale of the Fort Worth (Texas, USA) Basin has emerged as potentially the largest North American gas field, with an estimated 250 tcf in place and USGS ultimate recovery expectations of 26.2 tcf. First drilled by Mitchell Energy in 1981, the Barnett Shale has become the proving ground for effective and efficient tight-gas production, with over 10,000 wells to date. As with most tight-gas fields, successful Barnett well creation requires accurate well path placement in the gas zone, judicious alignment of the well path relative to existing fractures and current stress orientations, followed by effective hydraulic fracturing, generally using slick water techniques. The Barnett Shale was deposited during the Mississippian period in a marine setting and unconformably overlies Ordovician carbonates. In the area of this study, the underlying Viola limestone has been eroded, juxstaposing the Barnett Shale against the locally porous and water-bearing Ellenberger Limestone. The Barnett Formation ranges in thickness from about 50 feet in the south to nearly 1000 feet to the northeast. From the southwest boundary of the Llano arch, the Barnett Shale dives from a depth of 2500 feet down to 8000 feet near the Muenster Arch, the northeastern field boundary. The top of the Barnett in the study area is located at 6400 feet and has a thickness of 500 feet. Barnett shale porosity ranges from 0.5 to 6% with permeabilities as low as 70 nanodarcies. Low clay content (20-30%) makes the Barnett more "fracture friendly" than typical shales, which is verified by analyzing microseismic magnitudes and density.

2

Figure 1 - Fort Worth Basin - near Dallas Texas (USGS)

Figure 2 - West-East Cross-section showing Barnett Shale (brown) overlying Ellenberger Limestone (Pollastro, USGS)

3

Fractures There is ongoing debate about the impact that natural fractures have upon well production within the Barnett Shale. Gas production near faults and structural flexures are noted to be poorer (Bowker 2007) and in the worst cases scenarios, natural fractures can act as conduits to the water-bearing Ellenberger Limestone. However, reactivation of natural fractures are also seen as essential for creation of complex fracture fairways (Fisher 2006). Natural fractures observed in cores and FMI logs are rarely open, generally being calcite filled and oriented approximately NW to WNW (Gale 2008, Fisher 2006). These estimated orientations support contemporary maximum stress orientations normal to the Ouachita Thrust Belt, which ran approximately NE (Figure 1). Investigation of available stress information (e.g. World Stress Map) and our measurement of published microseismic data all indicate a current maximum stress orientation along a 50-60/230-240 degree azimuth (i.e. NE, normal to original stress orientation). Horizontal wells in the Fort Worth Basin are commonly oriented perpendicular to the current maximum stress orientation, inducing primary fracturing in a NE direction which may reopen natural fracturing along a NW direction (Fisher 2006). The most prominent form of natural fracturing in the Forth Worth Basin is associated with "collapse chimneys" (Kerans 1990). Often initiated by basement faulting, these features are enhanced by karsting in the Ellenberger and shallower carbonates. Individual collapse chimneys can extend vertically over 2000 feet, ranging from beneath the Ellenberger to the Pennsylvanian Caddo level (Sullivan 2006). Both 3D seismic data and derivative volume attributes, including curvature and incoherence, can be used to identify the most significant karsting features (Sullivan 2006) (Figure 3).

Figure 3 - Fracture chimneys identified with incoherence time slice and vertical slices of seismic amplitude (slumping features)

and incoherence (rippled patterns)

4

Macroseismic Analysis A high-quality 3D seismic survey exists over the study area. Automated horizon tracking of major geologic interfaces provides both key boundary information as well as indication of larger collapse chimneys (Figure 4). Seismic inversion, combining well log impedances and spectrally enhanced seismic data, provides subtle lithology information for the Barnett Shale and surrounding carbonates. (Figure 4). A velocity model was constructed by integrating well top and horizon interpretations, creating accurate depth surfaces for the Pennsylvanian Marble Falls and Ellenberger interfaces. This velocity model was also used to convert seismic and attribute data to depth so that all macroseismic analysis could be performed with well and engineering data in the depth domain.

Figure 4 - Ellenberger surface identifies base of the Barnett Shale zone and highlights karsted chimney features. Seismic

inversion contrasts Barnett Shale (reds and greens) from surrounding carbonates (blues). A range of seismic attributes were created from the 3D seismic data including: volume curvature; incoherence; fault moment; and spectral decomposition. Previous literature suggests that the volumetric attributes of maximum and minimum curvature and incoherence are particular useful for identifying karst collapse chimneys (Sullivan 2006). Methodical visualization and crossplot analysis of attributes was used to determine best approaches for highlighting fracture features. Key findings include: 1) minimum curvature responds to subtle collapses, with largest response at the core of collapse chimneys 2) incoherence responds most to the circumference of the collapse chimneys, with greatest vertical displacement 3) dip/azimuth responds most to the greatest amount of chimney collapse, and is best represented with vector arrows 4) corendering of seismic and inversion data with either curvature or incoherence attributes most effectively shows the vertical and horizontal extents of collapse chimneys

5

A composite display was designed that combines minimum curvature, incoherence, dip/azimuth and structure surfaces to easily differentiate high and low-density concentrations of fracturing (Figure 5).

Figure 5: Negative curvature (red), incoherence (black patterns) and dip/azimuth (red arrows) all provide diagnostic indicators of

the location and amount of karst-related collapse chimneys, relative to actual well (purple). Microseismic Analysis For this study, microseismic (micro-earthquake) data were recorded and located for three hydraulic fracture stages of a horizontal well (Figure 6). The orientation of the horizontal well, from heel to toe, is approximately 120 degrees east of north (~NW/SE). Analysis of the resulting microseismic data reveals a very diffuse pattern, which is generally desired as induced fractures tie into approximately orthogonal natural fractures. What is surprising is the high variability in number of microseismic events, ranging from slightly over 500 for the middle and heel stages and only 76 for the stage nearest the toe of the horizontal well. Microseismic data were analyzed in multi-dimensions spanning 1-D well plots, 2-D crossplots, 3-D visualization and 4-D time lapse control (Figure 7). While all three fracture stages exhibit similar treatment characteristics for pressure, slurry flow rates and proppant injection volumes, microseismic measurements are very diverse. The fracture stage nearest the well toe (magenta) was observed to have not just the fewest microseismic events, but also the most diffuse and most latent occurrence in time. In contrast, the middle fracture stage (red) has approximately 7 times the microseismic activity that begins shortly after pressuring-up of the well completion. Similarly, the heel fracture stage (blue) has a multitude of microseismic events that are concentrated very early in the hydraulic fracture.

6

Figure 6: Diffuse microseismic patterns (blue=heel stage, red=middle stage, magenta=toe stage) relative to karst chimneys

(redder zones)

Figure 7: Extracted curvature and incoherence seismic attributes (upper right crossplot) show strong correlation with heel (blue) and middle (red) fracture stage microseimic events (upper left view). Heel stage microseismic events (magenta) are much fewer

and occur much later than either other fracture stages (lower time lapse view)/

7

Incoherence seismic attributes were converted to depth and a depth slice was visualized just below the horizontal well bore (Figure 7 - upper left). Microseismic event density appears to correlate with incoherence attribute anomalies. Moving beyond visual correlation, both maximum curvature and incoherence were extracted from seismic depth volumes around each microseismic event. Maximum curvature and incoherence attributes were crossplotted (Figure 7 - upper right) and largest values of each attribute were highlighted with a polygon selection. The corresponding microseismic events were in turn highlighted in the plan view (Figure 7 - upper left), validating the correlation of microseismic attributes with key volume seismic attributes. Fracture Interpretation Microseismic data are valuable in situ measurements of rock response to hydraulic fracture treatment. In this study three fracture treatments provided a means to calibrate microseismic activity with macroseismic attributes. The toe fracture stage is interpreted to initiate in a largely fracture-free area, resulting in a limited, but diffuse microseismic pattern. Eventually, however, the introduction of proppant stimulates a latent burst in microseismic activity, centered approximately 1400 feet on either side off the wellbore. It is inferred that this microseismic response corresponds to a delayed breakout of hydraulic treatment to adjacent fractured collapse chimneys. As the stages move to the middle and heel of the well, it appears that the hydraulic fracturing directly overlies two separate collapse chimneys (Figure 7 - upper left). In both cases, microseismic intensity and location appear to correspond to a reactivation of fracturing within these collapse chimneys. Examination of the vertical extents of microseismic events reveal that none of the stages significantly extrude into the water-bearing Ellenberger limestone formation. Gas production data for this horizontal well supports a conclusion that hydraulic fracture treatment is contained within the Barnett Shale zone. However, comparison with adjacent wells indicates a significant degradation in gas production as fracture energy is "wasted" inside adjacent collapse chimneys. Optimized well placements avoid these fracture features and intersect the center of the Barnett Shale zone.

Figure 8: Volume interpretation of collapse chimneys (red) displayed with other depth data, including interpreted seismic

horizons, provides a backdrop for horizontal well planning and completions design.

8

Macroseismic attributes of maximum curvature and incoherence are effective at identifying karst-related collapse chimneys within the Fort Worth Basin. Crossplotting of minimum curvature and incoherence seismic data volumes provides a means to delimit features and visually identify potential water-tapping collapse chimneys. Volume interpretation of calculated seismic attributes is useful for mapping collapse chimneys in three dimensions (Figure 8). These feature not only dominate the Barnett Shale zone, but extend well above the Marble Falls and below the Ellenberger and down to the basement, in some cases. Conclusions Hydraulic fracturing in the Barnett Shale, as monitored with microseismic data, can be shown to interact with naturally fractured collapse chimneys, with the potential for tapping into water from the underlying Ellenberger limestone. Volumetric seismic attributes, including minimum curvature and incoherence, are useful for identify collapse chimneys and mapping them using volume interpretation tools. A composite depth scene that combines seismic surfaces, seismic attributes and interpreted karst collapse bodies is useful for optimizing well trajectories and completions strategies (Figure 8). Acknowledgements

We thank Devon Energy for access to data and permission to publish this work. References Bowker, K.A., 2003, Recent developments of the Barnett Shale play, Fort Worth Basin:West Texas Geol. Society Bulletin, v. 42, no. 6, p. 4-11 Bowker, K.A., 2007, Barnett Shale gas production, Fort Worth Basin: Issues and discussion, AAPG Bulletin; April 2007; v. 91; no. 4; p. 523-533; DOI: 10.1306/06190606018 Fisher, M. K., 2006, Barnett Shale fracture fairways aid E&P, World Oil, v. 227, no. 8 Fisher, M. K., and J. R. Heinze, C. D. Harris, B. M. Davidson, C. A. Wright, and K. P. Dunn, 2004, Optimizing horizontal completion techniques in the Barnett Shale using microseismic fracture mapping, SPE 90051, presented at the SPE ATCE, Houston, Sept. 26 - 29, 2004. Gale, J. F. W., 2008, Natural fractures in the Barnett Shale in the Delaware Basin, Pecos Co., West Texas: comparison with the Barnett Shale in the Fort Worth Basin (abs.), in West Texas Geological Society Fall 2008 Symposium, WTGS digital publication #08-120, p. 13. Givens, N. and Zhao, H., 2005, The Barnett Shale: Not so simple after all (http://www.republicenergy.com/Articles/Barnett_Shale/Barnett.aspx)

Heidbach, O., Tingay, M., Barth, A., Reinecker, J., Kurfeß, D., and Müller, B. (2008): The release 2008 of the World Stress Map (available online at www.world-stress-map.org)

Kerans, C., 1990, Depositional systems and karst geology of the Ellenburger Group _Lower Ordovician_, subsurface West Texas: The University of Texas at Austin, Bureau of Economic Geology Report of Investigations, 193, 63.

Montgomery, S. L., D. M. Jarvie, K. A. Bowker, and R. M. Pollastro, 2005, Mississippian Barnett Shale, FortWorth basin, north-central Texas: Gasshale play with multi-trillion cubic foot potential: AAPG Bulletin, 89, 155–175. Pickering Energy Partners, Inc., 2005, The Barnett Shale: Visitor Guide to the Hottest Gas Play in the US (available online at http://www.tudorpickering.com/pdfs/TheBarnettShaleReport.pdf)

Roberts, A., 2001, Curvature attributes and application to 3D interpreted horizons: First Break, 19, 85–99. Sullivan , E.C., Marfurt, K.J., Lacazette, A. and Ammerman, Mike, 2006, Application of new seismic attributes to collapse chimneys in the Fort Worth Basin, Geophysics,Vol. 71, No. 4 July-August 2006; P.B111–B119