The role of preexisting fractures and faults during multistage hydraulic fracturing in the Bakken Formation

We performed an integrated study of multistage hydraulic fracture stimulation of two parallel horizontal wells in the Bakken Formation in the Williston Basin, North Dakota. There are three distinct parts of this study: development of a geomechanical model for the study area, interpretation of multiarray downhole recordings of microseismic events, and interpretation of hydraulic fracturing data in a geomechanical context. We estimated the current stress state to be characterized by an NF/SS regime, with SHmax oriented approximately N45°E. The microseismic events were recorded in six vertical observation wells during hydraulic fracturing of parallel wells X and Z with three unusual aspects. First, rather than occurring in proximity to the stages being pressurized, many of the events occurred along the length of well Y, a parallel well located between wells X and Z that had been in production for approximately 2.5 years at the time X and Z were stimulated. Second, relatively few fracturing stages were associated with an elongated cloud of events trending in the direction of SHmax as was commonly observed during hydraulic fracturing. Instead, the microseismic events in a few stages appeared to trend approximately N75°E, approximately 30° from the direction of SHmax. Earthquake focal plane mechanisms confirmed slip on faults with this orientation. Finally, the microseismic events were clustered at two distinct depths: one near the depth of the well being pressurized in the Middle Bakken Formation and the other approximately 800 ft above in the Mission Canyon Formation. We proposed that steeply dipping N75°E striking faults with a combination of normal and strike-slip movement were being stimulated during hydraulic fracturing and provided conduits for pore pressure to be transmitted to the overlaying formations. We tested a simple geomechanical analysis to illustrate how this occurred in the context of the stress field, pore pressure, and depletion in the vicinity of well Y. Introduction The Mississippian-Devonian Bakken Formation is a restricted, shallow-water, mixed carbonate-clastic sequence deposited over most of the deep part of the Williston Basin (Gerhard et al., 1990). Using horizontal drilling and multistage hydraulic fracturing led to the successful development of the Elm Coulee (Montana) and Parshall (North Dakota) fields and demonstrated the great potential of the Bakken Formation. Despite the generally successful exploitation of the Bakken Formation, questions remain about how to optimize hydraulic fracturing and the importance of preexisting fractures and faults as fluid pathways in the reservoir. The study area consists of three horizontal production wells (X, Y, and Z) and six vertical monitoring wells (A-F) (Figure 1). The three approximately 10;000‐ ftlong horizontal wells are located in the Middle Bakken. The well spacing is approximately 500 ft. Our study focuses on stimulation of wells X and Z. The middle well, well Y, had been hydraulically fractured previously and was in production for about 2.5 years prior to stimulation of wells X and Z. Well-log data including gamma ray, electrical resistivity, density, and Pand S-wave sonic velocities are available from geophysical logs in vertical wells A, B, D, E, and F. Well logs from vertical well A are shown in Figure 2. The total thickness of the Bakken Formation is approximately 140 ft in the study area, with the top of the reservoir at a depth of approximately 10,000 ft. Wells X, Y, and Z were drilled in the Middle Bakken. As shown in Figure 2, the Upper, Middle, and Lower Bakken members can be easily distinguished from gamma-ray logs. The high gamma ray and high resistivity indicate the oil-saturated, organicrich shale layer in the Upper and Lower Bakken, and the low gamma ray and low resistivity in the Middle member is an indication of the target formation comprised principally of a dolomitic siltstone. The Upper and Lower Bakken shales are also characterized by lower density (approximately 2.3 g∕cm3) and lower Stanford University, Department of Geophysics, Stanford, California, USA. E-mail: [email protected]; [email protected]. Manuscript received by the Editor 7 October 2013; revised manuscript received 15 December 2013; published online 21 May 2014; corrected version published online 4 June 2014.This paper appears in Interpretation, Vol. 2, No. 3 (August 2014); p. SG25–SG39, 20 FIGS., 3 TABLES. http://dx.doi.org/10.1190/INT-2013-0158.1. © 2014 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved. t Special section: Microseismic monitoring Interpretation / August 2014 SG25 VP and VS values, which can be attributed to lithology (clayand organic-rich shale) and fluid-saturation differences. In the sections below, we first report the development of a geomechanical model that includes knowledge of the magnitude and orientation of principal stresses, the existence and orientation of natural fractures and faults, and the mechanical properties of the formations being produced. Knowledge of the stress field can be used to better understand microseismic events and fracture networks resulting from hydraulic fracture stimulation (Fehler et al., 1987; Phillips et al., 1998; Rutledge et al., 1998; Maxwell et al., 2002). Therefore, linking observations of microseismic data with geologic and geophysical information can help us better understand the geomechanical properties of the Bakken Formation and the role of preexisting fractures and faults on the effectiveness of hydraulic fracturing stimulation in this reservoir. Geomechanics of the study area To build a quantitative geomechanical model, we follow the general procedure outlined in Zoback (2007) to constrain the orientation and magnitudes of the three principal stresses in the reservoir. The magnitude of vertical principal stress was determined from the weight of the overburden, whereas estimates of pore pressure (Pp) were available from diagnostic fracture injection test (DFIT) data, which provide instantaneous shut-in pressure (ISIP) to constrain the magnitude of Shmin. Physical properties were determined by commercial laboratory tests on Bakken core samples from well A. The values are listed in Table 1. Unfortunately, there are no image logs at this site to determine the orientation of the maximum compressive stress from the orientation of compressive and/or drilling-induced tensile wellbore failures. Previous studies close to the study area report the orientation of SHmax to be approximately N45-55° E, based on the DITFs observed in Formation MicroImager (FMI) logs obtained from three horizontal wells (Olsen et al., 2009; Sturm and Gomez, 2009). Using the physical property measurements from the core samples, as well as estimates of SV , Shmin, and Pp from this study, we used the methodology outlined in Peška and Zoback (1995) to analyze the occurrence of tensile fractures in the wells studied by Sturm and Gomez (2009) as a function of SHmax orientation and magnitude (see Appendix A). This modeling indicates that when SHmax is oriented toward N4555°E, transverse tensile fractures that are quite similar to those reported by Sturm and Gomez (2009) will occur. However, the magnitude of SHmax is not well constrained by the modeling. The stress state could either be a normal faulting or a strike-slip faulting regime. Microseismic events during hydraulic stimulation As shown in Figure 1, microseismic events during hydraulic fracturing of wells X and Z were monitored with approximately 1900-ft-long, 40-level, 3C geophone arrays in six vertical observation wells (A-F). The spacing between seismometers in each array was 49.2 ft. Hydraulic fracturing was first performed along well X from toe to heel and Figure 1. Entire microseismic events and the well geometry in this study, including three horizontal wells (X, Y, and Z), and six vertical observation wells (A-F). Events are colored by stages. (a) Map view with the arrows represents stimulation sequence along wells X and Z, (b) north–south cross view with geologic formations and geophone locations along each vertical observation well, and (c) histogram of number of events for each stage.