Identification of slowly slipping faults in the Barnett ... · At the Barnett 2 site, two horizontal treatment wells, Treatment Well 1 (TW1) and Treatment Well 2 (TW2) were fractured

$Page 1: Identification of slowly slipping faults in the Barnett ... · At the Barnett 2 site, two horizontal treatment wells, Treatment Well 1 (TW1) and Treatment Well 2 (TW2) were fractured$
Identification of Slowly Slipping Faults in the Barnett Shale Utilizing Ant-tracking Noha S. Farghal* and Mark D. Zoback, Stanford University, Stanford, CA, USA Summary In this study, we present the results of using Ant-tracking to identify slowly-slipping faults in the vicinity of hydraulically-fractured horizontal wells at two sites in the Barnett shale, referred to as Barnett 1 and 2. At both sites, Long Period Long Duration (LPLD) events were observed. These events are believed to be caused by relatively large faults slipping slowly during hydraulic stimulation. At the Barnett 1 site, Ant-tracking revealed a steep fault which intersects the middle of five wells at a place where a high density of fractures was found in FMI images. Moreover, the local orientation of the fault has the same strike as the observed fractures. This fault appears to be located in the source of the LPLD events produced during hydraulic fracturing. It is also coincident with a distinct gap in microseismicity. It appears that slip on this relatively large scale fault is responsible for the observed LPLD events. At the Barnett 2 site, two horizontal wells were hydraulically fractured with seismic monitoring in two vertical observation wells. Ant-tracking identified several faults which explain both lineations in the microseismicity as well as the locations of LPLD events between the two treatment wells. Introduction Long Period Long Duration (LPLD) events are low frequency seismic events observed during multi-stage hydraulic fracturing (Das and Zoback, 2013a,b) which persist for tens of seconds. As the events are similar to tremors occurring due to slow slips of large tectonic plates, LPLD events are thought to be caused by relatively large faults slipping slowly and stably (Zoback et al., 2012). The importance of LPLD events lies in that they signify a process (stable slip of large faults) that could account for a substantial amount of the increase in permeability associated with hydraulic fracturing stimulation. Das and Zoback identified LPLD events on recordings of micro-earthquakes induced by hydraulic fracturing in the Barnett Shale at sites referred to as Barnett 1 and Barnett 2. At the Barnett 1 site, five sub-horizontal wells A, B, C, D and E (shown in Figure 1) with well separations of 150m were fractured sequentially from toe to heel. Wells A and B were fractured simultaneously with a horizontal recording array of nine three-component (3C) geophones placed in well C (geophone separation of about 12m). Wells D and E were fractured with zipperfrac technique (one fracture stage carried out in a well with the following fracture stage carried out in the neighboring well). During hydraulic fracturing of wells A, B, D and E, microseismicity was

monitored in well C with ten 3C geophones which was moved along the well several times to keep the array close to the intervals being hydraulically fractured. Well C was fractured last, with a recording array of twelve 3C geophones placed in the vertical section of well B. Around 4500 microseismic events were located by the contractor with magnitudes ranging between -1.5 and -3 (Das and Zoback, 2013a). A pre-fracturing (baseline) 3D reflection seismic survey was acquired to identify important subsurface features for well and process design purposes. The acquisition geometry used in this 3D seismic survey principally involved 3C geophones deployed in lines parallel to the wells shown in Figure 1. A sampling interval of 2 ms was used, and the record length was 4s. The Figure shows hypocenters of micro-earthquakes displayed as dots, with dot size representing magnitude and color representing fracture stage. FMI fracture strikes from well C are also displayed, as well as the range of angles from which LPLD events emanated (shown as a yellow wedge), which is discussed below.

Figure 1: A plot of the locations of all recorded microseismic hypocenters, FMI fracture strikes from well C, and the range of angles from which LPLD events emanated displayed as a yellow wedge (Das and Zoback, 2013b). At the Barnett 2 site, two horizontal treatment wells, Treatment Well 1 (TW1) and Treatment Well 2 (TW2) were fractured in 8 and 10 stages, respectively (shown in Figure 2). Microseismicity was monitored with 40 3C geophone arrays in two vertical wells (Monitoring Well

SEG New Orleans Annual Meeting Page 4919

DOI http://dx.doi.org/10.1190/segam2015-5811224.1© 2015 SEG

Slowly Slipping Faults in the Barnett Shale

(MW) 1 and 2) with geophone separation of 15m. The deepest geophone in MW1 was 450m above the treatment wells, while the deepest in MW2 was 300m above the wells. Das and Zoback showed that there seems to be a NE lineation in recorded microseismic event hypocenter locations as shown in Figure 2, in addition to a few (compared to Barnett 1) recorded and located LPLD events shown on the same figure. The microseismic hypocenter lineations are consistent with natural fracture strikes found on FMI logs in TW1. Das and Zoback also found evidence of fluid communication the treatment between wells, as some microseismic events did not seem to emanate from regions close to the stimulated stages, but rather from areas where stimulation was not being carried out. For example, during fracturing of stage 5 of TW1, microseismic events started emanating near TW2 accompanied by a pressure drop of 300psi in TW1, suggesting a fluid connection between the two wells, which may be due to a fault.

Figure 2: A plot of the locations of all recorded microseismic hypocenters (dot size indicative of magnitude) and locations of large LPLD events (ellipses) colored by stage for Barnett 2 dataset. The first number on LPLD ellipses indicates treatment well number while the second indicates stage number in which the LPLD event emanated (Das and Zoback, 2013b (modified)). Fault Identification

For Barnett 1, 3D reflection seismic data was provided to us in the form of a post-stack migrated volume. To locate fault(s) in this area, we calculated the variance attribute of the seismic data to detect discontinuities and followed it by two passes of Ant-tracking for discontinuity/edge enhancement. The variance of each voxel in the data is a measure of local amplitude dissimilarity which indicates discontinuities such as faults. Discontinuities are enhanced when the variance volume is used as input to Ant-tracking

which improves edge definition in the data by tracking consistent discontinuities that are likely to be faults. Figure 3 shows the result of the second Ant-tracking pass, showing a large apparent fault intersecting well C where the highest density of natural fractures was observed on the FMI log, and having their same strike orientation of approximately N100oE. The figure also shows the constraints on the location of LPLD events calculated by Das and Zoback. Since it is difficult to identify P and S waves on LPLD events, Das and Zoback were able to find the general angular range from which LPLD events originated by analyzing event move-outs from well B and C recording arrays. With the monitoring array deployed in the vertical section of well B, LPLD events located during fracturing of well C were found in the range of angles between the broken line green circles in Figure 3.

Figure 3: Ant-tracking time slice showing the slowly slipping fault intersecting well C where the highest pre-existing fractures were found as well as the ranges of angles from which the LPLD events were emitted during the fracturing of well C (between broken line green circles) and wells A and B (within the wedge between orange broken lines).

The wedge located within the orange broken lines indicates the angular range from which LPLD events were emitted during fracturing of wells A and B and recorded with a geophone array in well C. The fault we identified is clearly located in the range of angles from which LPLD events emanated. The fault seems to be the most likely source of LPLD events, as it matches the dense natural cluster of fracture orientations found in well C in both location and




strike direction and it is located within the range of angles from which LPLD events were emitted.

(a)

(b)

(c) Figure 4: (a) Ant-tracking time slice of the post-frac survey showing identified faults A, B and C, (b) the agreement between fault locations and orientations and microseismic lineations and (c) red ellipse showing approximate locations of some LPLD events located by Das and Zoback in Figure 2. In the Barnett 2 site, raw shot gathers of four 3D seismic surveys (time lapse) in the Barnett Shale were available. The surveys were very repeatable, with geophones remaining in the same locations from the time of their deployment to the last recording in the final survey. The surveys were acquired to help understand the effects of hydraulic fracturing and production on reservoir properties. The survey under consideration in this work was acquired after all hydraulic fracturing treatments took place and

before flowback. We will refer to this survey as the post-frac survey. After data processing and migration, the variance attribute of the post-frac survey was computed and two passes of Ant-tracking were carried out on the volume. Figure 4 (a) is the second pass Ant-tracking time slice of the post-frac survey showing identified faults A, B and C. Figure 4 (b) shows the agreement between fault locations and orientations and microseismic lineations. Finally, panel (c) of the figure shows approximate locations of the red ellipse in Figure 2 encompassing most of the large LPLD events located by Das and Zoback. Microseismic Gap along a Slowly Slipping Fault in Barnett 1 To visualize the locations of the microseismic events relative to the fault discussed above, we used the Automatic Fault Extraction feature of Petrel (courtesy of Schlumberger) to create a three dimensional representation of the fault identified through Ant-tracking. To display only the fault under consideration, we limited the azimuth and dips of the extracted patches to that of the slowly slipping fault. We then plotted the hypocenters of the microseismic events in the same 3D window as the fault. By examining the locations of the microseismic hypocenters, we identified a region near the fault in which microseismic events from stages 7-9 of the simulfrac treatment of wells A and B are almost nonexistent, as shown in Figure 5.

Figure 5: A 3D plot of hypocenters of micro-earthquakes from stages 7-9 of wells A and B simulfrac and the extracted slowly slipping fault showing a significant drop in microseismicity near the fault.




Discussion In the Barnett 1 dataset, we were able to identify a large fault using the variance attribute of the 3D reflection seismic data and enhance it with Ant-tracking. The fault matched the strike direction and position of the highest density of natural fractures seen on the FMI log of well C. The fault also lies in the angular range from which LPLD events were emitted, suggesting that slow stable slip of this fault is the cause of the LPLD events recorded during the hydraulic fracturing treatments. A microseismic gap (a region where microseismic events are absent) was also found for events recorded during the hydraulic fracturing stages of wells A, B and D close to the fault. The microseismic events seem to drop significantly in number close to the area of the slowly slipping fault from which LPLD events emanated. It seems that as the activated fault slips stably emitting seismic energy in the form of LPLD events, microseismic events on the fault become almost nonexistent. Post-frac survey variance attribute analysis in Barnett 2 data showed three faults connecting the two treatment wells, providing an explanation for pressure communication between the wells during hydraulic fracturing. Fault C which connects the two wells caused the fluid pressure drop during fracturing of TW1 and microseismic events appearing near TW2. We also see in Figure 4(c) that the LPLD events located between the two wells by Das and Zoback (Figure 2) are close to fault B, or possibly the more prominent fault seen in the same area. Acknowledgement We are grateful to ConocoPhillips for providing the data used in this work.



EDITED REFERENCES Note: This reference list is a copyedited version of the reference list submitted by the author. Reference lists for the 2015 SEG Technical Program Expanded Abstracts have been copyedited so that references provided with the online metadata for each paper will achieve a high degree of linking to cited sources that appear on the Web. REFERENCES

Das, I., and M. Zoback, 2013a, Long-period, long-duration seismic events during hydraulic stimulation of shale and tight-gas reservoirs — Part 1: Waveform characteristics: Geophysics, 78, no. 6, KS97–KS108.

Das, I., and M. Zoback, 2013b, Long-period long-duration seismic events during hydraulic stimulation of shale and tight-gas reservoirs — Part 2: Location and mechanisms: Geophysics, 78, no. 6, KS109–KS117.



Documents

Identification of slowly slipping faults in the Barnett ... · At the Barnett 2 site, two horizontal treatment wells, Treatment Well 1 (TW1) and Treatment Well 2 (TW2) were fractured