Many of the wells drilled in the Montney Formation of the Farrell Creek dry gas ... John. Fig. 1 Location of the Farrell Creek Montney Field. 2.2. Stratigraphy.
DFNE 2014 - 224
Investigation of the Effects of Natural Fractures and Faults on Hydraulic Fracturing in the Montney Formation, Farrell Creek Gas Field, British Columbia Steve Rogers Golder Associates Ltd., Vancouver, British Columbia
Pat McLellan McLellan Energy Advisors Inc., Calgary, Alberta
Gordon Webb Golder Associates Ltd., Vancouver, British Columbia ABSTRACT: The Lower Triassic Montney Formation is an areally extensive unconventional resource in northeastern British Columbia, Canada with significant natural gas and gas liquids in place. Development of this mixed siltstone, sandstone, and shale resource, which can reach thicknesses of up to 400 m, is typically achieved by drilling horizontal wells followed by multistage hydraulic fracture treatments. Many of the wells drilled in the Montney Formation of the Farrell Creek dry gas reservoir showed typical hydraulic fracture geometries as revealed by their microseismic response, i.e., relatively simple SHmax-parallel groupings of events propagating symmetrically away from the well. However, a number of well pads in the field, which are located close to major faults that were delineated with 3D seismic, display significant structural control. These latter wells had a range of microseismic responses including: (1) lineaments evolving at 30° oblique to SHmax; (2) regardless of what interval of the Lower Montney was stimulated, an identical spatial response was observed, (3) a virtual absence of seismicity in the Upper Lower Montney even when directly stimulated; and (4) a temporal pattern of events consistent with the diffusion of pressure around a conductive propped fracture and a connected network of natural fractures. Discrete Fracture Network (DFN) analysis and modelling were used to help quantify and describe both the natural and induced fracture architecture in the Lower Montney and their role in influencing injection fluid migration during the well stimulation process. A comprehensive geomechanical characterization of one pad, including in-situ stress magnitudes and orientations, pore pressures, rock mechanical properties, and log and seismic-derived discontinuities, was the foundation for building the DFN model. Our analysis shows that much of the microseismic response is the result of both the jacking open (dilation) of existing fractures and shearing (slippage on existing discontinuities when a critical stress level is reached) rather than solely tensile parting of new rock which is commonly associated with hydraulic fracturing. This raises interesting challenges for designing the most effective stimulation strategy for structurally controlled parts of the field.
1. INTRODUCTION The Lower Triassic Montney Formation is an areally extensive unconventional resource in northeastern British Columbia with significant natural gas and gas liquids in place. Development of this mixed siltstone, sandstone, and shale resource, which can reach thicknesses of up to 400 m, is typically achieved by drilling an array of horizontal wells followed by multistage hydraulic fracture treatments. Many of the wells drilled in the Montney Formation of the Farrell Creek reservoir showed typical hydraulic fracture geometries as revealed by their microseismic response, i.e., relatively simple SHmax parallel groupings of events propagating symmetrically away from the well. However, a number of well pads in the field, which are located close to major faults, that were delineated with 3D seismic, display significant structural control. These latter wells had a range of microseismic responses including: (1) lineaments evolving at 30° oblique to SHmax; (2) regardless of what interval of the Lower
Montney was stimulated, an identical spatial response was observed, (3) a virtual absence of seismicity in the Upper Lower Montney even when directly stimulated; and (4) a temporal pattern of events consistent with the diffusion of pressure around a conductive propped fracture and a connected network of natural fractures. This paper describes efforts to integrate a range of static and dynamic fracture, discontinuity, geomechanical and microseismic monitoring observations into a coherent conceptual model of significant reservoir discontinuities and subsequent DFN modelling that was used to help understand their influence on hydraulic fracture behavior. The Farrell Creek Montney field is operated by Progress Energy as part of a joint venture with Sasol Canada. Prior to March 2014 the property was previously operated by Talisman Energy as part of a joint venture with Sasol Canada.
2. THE FARRELL CREEK MONTNEY RESERVOIR 2.1. Location This paper is focused on the 13-36 pad location in the eastern part of the Farrell Creek dry gas field located in north-eastern British Columbia, approximately 75km west of Fort St. John. See Figure 1.
British Columbia
paper we are focusing on the results of a DFN investigation undertaken in the vicinity of the 13-36 pad drilling location where the two horizontal wells were drilled in a NW-SE orientation, offset 250m from each other, and targeting the Upper Lower (UL) and Lower Lower (LL) units of the Montney Formation. In this area of the field the UL consists primarily of fine to coarse bituminous siltstone. The LL has a similar lithology but typically possesses more laminations and a greater silt and clay sized fraction. Table 1. Stratigraphic column of the Montney Formation and bounding strata in the study area Middle Triassic
Doig Phosphate Formation
Farrell Creek Field (Sasol-Progress Energy JV)
Upper Upper (UU) Upper Montney
Fort St. John
Lower Triassic
Lower Upper (LU) Upper Lower (UL) Lower
Fig. 1 Location of the Farrell Creek Montney Field
Montney
2.2. Stratigraphy In the Farrell Creek area the basic stratigraphy of the Montney Formation and its bounding strata used by Talisman/Sasol is shown in Table 1. Over 100 horizontal wells have been drilled and stimulated in the field targeting the four basic sub-divisions of the Montney Formation which is largely based on a sequence stratigraphy framework, see [10] For the purposes of this
Permian
Lower Lower (LL)
Belloy Formation
Conventional reservoirs “Basin centre” tight silt and sandstones
Bituminous siltstone reservoirs
A
Tangent
Dawson
Farrell
Fig. 2. Regional west-east geological cross-section, after [10]
A’
Figure 2 shows a regional west-to-east Montney geological cross-section in the Farrell area that distinguishes argillaceous siltstones of the Lower Montney, carbonaceous siltstones of the Upper Montney, and shelfal siltstones and mudstones of the Upper Montney. East of the Farrell Creek area the Montney thins and is comprised of a more silty and sandy facies, that includes a number of conventional low permeability reservoirs. 2.3. Structure The Farrell Creek field sits at the edge of the Dawson Creek Graben Complex. (Figure 3) East-west normal faults formed in the Carboniferous-Permian age were reactivated during the Late Cretaceous Laramide orogeny. Numerous low angle, NW-SE thrust faults from this event are present in the field. Sub-seismic scale faults, associated with these major faults, have been identified throughout the field in ant-tracker images processed from 3D seismic.
Farrell Ck
Boundary Lake
Table 2: Range of geomechanical properties for the Lower Montney in the 13-36 pad area, Farrell Creek Property
Range
Average Value Used
Matrix Permeability
10-200nD
50 nD
Static Young’s Modulus UL
32-51 GPa
40 GPa
Static Young’s Modulus LL
28-40 GPa
32 GPa
Shmin Gradient
14-26 kPa/m
21.1 kPa/m
Reservoir Pressure Grad
10-20 kPa/m
16.6 kPa/m
Sv/Shmin
1.0-1.8
1.2
SHmax/Shmin
1.1-1.4
1.3
SHmax Orientation Discontinuity Coefficient of Friction
o
N40-44 E
N42oE
0.4-0.7
0.5
Pouce Coupe
Fig. 3. Dawson Creek graben complex, after [13]
SW
2.4.
NE BLSK
Geomechanical Characteristics of the Montney CDMN Formation in the Study Area Top TRSC Table 2 is a summary of in-situ stress magnitudes and orientations, and basic rock mechanical propertiesHLFY in the Lower Montney which were determined from laboratory U MNTN L MNTN tests and log analyses. These have been previously BLLY described by [7], [8], and [9]. These inputs wereDBLT used with the DFN model in order to simulate the behavior SHND of natural discontinuities under injection conditions. Figure 4 shows a profile of vertical, minimum and WBMN maximum horizontal stress, and pore pressure from data obtained in the vicinity of Pad 13-36.
Fig. 4. Example profile of pore pressures and in-situ stresses in the 13-36 pad area. Compiled from DFIT stress test data, borehole breakout inversion and bulk density logs, [9].
3. WELLBORE STATIC FRACTURE DATA ANALYSIS 3.1. Introduction Static data sources help characterize some of the geometric properties of the fracture/fault system, e.g., type, intensity and size. Data were available from two processed 3D surveys, allowing the automatic extraction of intermediate and larger scale structures and curvature analysis of the top Belloy horizon. Image log data were available for some parts of the build and horizontal legs of the two wells in the UL and LL, although the data quality was poor over much of the wellbore drilled with oil-based mud.
3.2. Seismic Fault Analysis The seismic cube was processed using Schlumberger’s proprietary Ant Tracking technology [12]. The Ant Tracking analysis reveals a relatively high density of seismically resolvable structures around the area of interest, many in close proximity to the well pad involved in this study. The Ant Tracking analysis allows a level of detail to be extracted from the seismic cube with a detailed fault network being produced, see left inset, Figure 5.
Micro-seismic features- Green
Key interpreted micro-seismic lineaments
Ant tracker features - Red
integration of size observations from a number of different length scales. Length scale data from microseismic lineament interpretation from a pad with a strong structural signature is seen in Figure 5. The results indicate the microseismic lineaments are in fact part of the same size population as the Ant Tracker identified seismic scale lineaments, thought to be faults. This strongly supports the idea that the hydraulic fracture stimulation is stimulating natural “fault” structures.
3.3. Image Log Data Electrical image log data provide the most common form of information on wellbore structures. Across the field two different types of imagers have been utilised: Schlumberger’s FMI tool using a water-based mud and their EI tool used in oil-based mud. By far the best results are obtained with the water-based mud system and this has allowed the calculation of fracture intensity properties for both the Upper Lower and Lower Lower Montney. Whilst originally recorded as a fracture frequency measure (P10, #/m), these results have been converted to the less directionally biased P32 fracture intensity measure (fracture area/volume) using an analytical technique, [19]. The DFN fracture intensity system is described in [3]. The conversion from linear biased measurements (e.g. wellbore fracture frequency) to the less biased P32 measurement is an important part of the intensity analysis. With some wells drilled vertically and others horizontally, it is crucial that the best unbiased estimate of fracture intensities and therefore spacings are determined. Typically either a Terzaghi correction [8] or a Mauldon correction [6] are applied. Both seek to increase the apparent intensity of near-wellbore parallel structures that are generally under sampled. Table 3. Summary of fracture intensity properties of the Lower Montney P32 Intensity (m2/m3)
Fig. 5. Power law analysis of lineament data from (left) Ant Tracker interpreted structures and (right) microseismic lineaments.
A key issue in any Discrete Fracture Network (DFN) analysis is describing the size distribution of structures. Natural fracture size is a critical component of DFN modelling as this strongly controls network connectivity. One of the most commonly used techniques for characterizing the size population of natural structures is power law size analysis of regional fault data sets and this has been carried out on the Ant Tracker interpretation. The power law analysis allows the
Well Inclinat’
Tool
Upper Lower
Lower Lower
Vertical
FMI
0.29
.03
Orientation data from the FMI tool show that the most common structural trends are NW-SE and E-W.
3.4. Curvature Analysis of the Belloy Fm Surface Curvature analysis calculates the rate of change of slope on a surface and is often used in fracture studies for either the prediction of fracture intensity or as a way of identifying more subtle features such as faults with limited throw [14]. The results of 3D seismic derived curvature analysis on the top Belloy Fm surface (not shown) reveal some strikingly different structural
characteristics across the field. Some areas show a pronounced crenulated texture, with similar NW-SE structural trends to that seen in image logs from nearby boreholes. Other areas however show markedly different characteristics, without the crenulated surface, with a smaller number of larger structures. This difference in structural style is also reflected in pad performance across the field with some wells have limited structural response to stimulation whereas the area focused on here is favourably influenced by larger structures in the subsurface.
4. DYNAMIC DATA ANALYSIS
4.1. Introduction Dynamic data sources help characterize the conductive nature of the fracture system in terms of how conductive the fractures are and how they extend and dynamically connect away from the well. Data were available from the analysis of drilling mud losses, a Production Logging Tool (PLT) and from the microseismic response to stimulation.
4.2. Drilling Mud Losses & PLTs Drilling records can be analyzed in order to identify and quantify the nature of significant mud losses to natural fractures or faults. An example compilation of these data from one of the wells on Pad 13-36 is shown in Figure 6.
Fig. 6 Graph showing mud loss locations (red squares), cumulative fracture frequency (brown squares) and relative PLT flow contribution (blue diamond).
During drilling of one NW-oriented well a number of locations were observed where the rate of losses was greater than 2.5 m3/s. These losses are the equivalent to Darcy magnitude permeabilities when using a simple flow calculation based upon the measured loss rate and the calculated pressure difference [11]. There is evidence that these areas of higher loss are associated with localised zones of higher fracture/fault intensity as indicated on the image log. However it was not possible to resolve any features on the image log that might correspond to the mud loss events.
In addition to the drilling losses, a production logging run (PLT) was carried out in one well to investigate the distribution of relative flow contribution for each perf/stage, see Figure 6 (blue diamonds). The PLT could only be run in the last 5 stages of the well (at the heal end) but it does record the distribution of flow over 15 different perf clusters. The distribution of flow contribution shows that the largest number of perfs produced zero flow (8/15), a lesser number produced intermediate flow rates (6/15) and a single perf provides significantly higher flow, almost double the next highest
Regardless of which layer is being stimulated, virtually the same response is seen which can only be explained by the stimulation of existing structures, thought to be faults or fracture corridors; and
contributor. This distribution of flow is consistent with the characteristics of most fractured reservoirs, where a skewed distribution of fracture permeability results in a few wells (or intervals) dominating production, with most of the other wells (intervals) contributing very little, [16]. This strongly suggests a natural fracture signature to the flow behaviour in this Montney well.
From one stage to the next, there is nearly always a re-activation of structures observed from the previous stage, again supporting the claim of the stimulation of existing structures.
4.3. Microseismic response to stimulation Hydraulic fracturing of these two adjacent Montney wells was carried out by pumping slickwater into one of up to 10 stages for approximately 150 – 200 minutes, with injection rates varying between 12-18 m3/min. Surface injection pressures were typically near a maximum of 60 MPa. Microseismic monitoring was conducted on the pad during these fracture stages to track fluid and pressure migration. Examples of the Upper Lower and Lower Lower microseismic responses are shown in Figure 7. There are a number of characteristics that can be seen from these data including:
It might be argued that stress shadowing or poro-elastic responses can explain the observed response. However these processes can't explain these observations. When viewing the microseismic though time the evolution of event clouds follow logical linear connection pathways. It appears therefore that the hydraulically stimulated fractures at this pad are primarily natural fractures and small sub-seismic scale faults connected to the well and extending out into a larger network of faults away from the well.
Regardless of whether the UL or LL is stimulated, almost all of the microseismicity is concentrated in the LL;
Late
Upper Lower Stimulation
Section
Late
Upper Lower Lower Lower
Circle size is relative to SNR Pumping Stage
Early
Early Late
Lower Lower Stimulation
Late
Upper Lower
26
Early
Section
Lower Lower
Early
Fig. 7. Example microseismic responses in plan (left) and section (right). The events are coloured by elapsed time and scaled by Signal to Noise Ratio (SNR) value. A conceptual network has been added to the cloud indicating possible flow paths consistent with the elapse time
5. CONCEPTUAL DISCONTINUITY ARCHITECTURE & DFN MODEL All of the fracture and geomechanical observations were compiled into a conceptual fracture architecture or model that describes the style and distribution of
fracturing within the Lower Montney and how that explains the dynamic response to stimulation (as observed by the microseismic data). The following are the key components of the model as shown in Figure 8:
Fig.8. Conceptual model of the architecture of discontinuities in the Lower Montney, Pad 13-36
The Upper Lower (UL) is comprised of higher intensity shorter fractures that are largely confined within the UL layer (Fig. 8 A). The Lower Lower (LL) is comprised of lower fracture intensity but longer structures which are thought to at least partially penetrate up into the UL layer (Fig. 8 B). Multi-zone fracturing with conventional hydraulic fracture advancement (Fig. 8 C) or by the direct stimulation of major structures at or near the perforations (Fig. 8 D) or a combination of both. Bedding planes are also believed to have hydraulic significance, providing conductive pathways for fluid when the injection pressures approach Smin (Fig. 8 E). Under high injection pressures weak bedding planes are known to have undergone shear displacement.
A view of a single DFN realisation that draws upon the conceptual architecture and uses properties derived from the previously described analysis is shown in Figure 9. Once the DFN model is constructed, some of the geometric characteristics of that network can be investigated. For instance, the DFN model can be sampled with wellbores to check the likelihood that the well-to-network connectivity is achieved in a certain completion (see Figure 9, top right). Additionally the overall connectivity of the DFN can be investigated using FracMan’s cluster analysis tools. By identifying all fractures that are connected to at least 10 other fractures, the architecture of connected fractures can be seen. Comparison of the identified clusters with an example of the microseismic cloud shows the strong similarity between a stimulated series of fractures and a possible network of connected fractures present in the Lower Lower, Figure 10.
Fractures intersecting the well
DFN Model
Image Log
UL LL
Fig. 9. Views of DFN model of components of the Lower Montney fracture architecture. Left image shows the larger and sparser Lower Lower fractures (red and purple) and the smaller more intense Upper Lower fracture (blue). Upper Right image shows the fractures penetrated by a horizontal well and a comparison between the actual and simulated fracture orientations. The lower right image shows fractures indirectly connected to the well.
See comparison between a microseismic cloud and a connected fracture volume
Fracture clusters shown with background fractures removed Fig. 10. FracMan cluster analysis of the Lower Lower identifies all clusters of fractures where there are at least 10 fractures connected (left image). The centre rectangle shows the identified clusters (N>=10 fractures, each cluster coloured differently) with the background turned off and compares them to the microseismic cloud (right) showing a very similar pattern of connected fractures.
appropriate properties (see Fig 12) or deterministically where features are created through the array of microseismic points to represent an approximation of the actual sub-surface network in the Lower Montney. Both of these approaches have been tested and are described below.
6. GEOMECHANICAL ANALYSIS OF HYDRAULIC FRACTURING IN A FRACTURED ENVIRONMENT 6.1. Introduction Microseismic monitoring of the fracture stimulation treatments on NW-SE wells in this pad do not produce classic NE-SW oriented event clouds defining induced hydraulic fractures. A modelling approach is required that allows the injection of fluid into a fractured media, where the structural influence is allowed to play a role. Consequently simulations have been carried in the FracMan code using the methodology described in [2] and has been previously applied regionally [20]. The basic mechanics of the FracMan approach are shown in Figure 11. FracMan allows a hydraulic fracture to propagate away from the well (A) until it reaches natural fractures. If the fluid pressure is greater than the closing stress on that fracture, then fluid will be taken up by that fracture (B). These are called inflated normal fractures and have components of both dilation and shear, therefore behaving as a hybrid between a hydraulic fracture and conductive natural fracture. Fractures that become critically stressed (see section 6.4) can transmit fluid pressure but are not considered to significantly dilate and receive fluid, and by logical extension, proppant (C). Not all critically stressed fractures behave in the same way, with rougher fractures experiencing greater dilation and therefore increased permeability. This can be captured by the dilation angle i. Using this simple geomechanical approach, the evolution of fluid through a combination of a hydraulic fracture and stimulated natural fractures can be rapidly captured.
B.
t, Shear Stress
smax
Plan View
Simulated microseismicity
q smin Rough Shear-Dilated Fractures
C.
Smooth Un-Dilated Fractures
smax
f+i
Inflated Natural Fractures
f
smax A.
Using the DFN model of existing natural fractures and faults as shown in Figure 12, hydraulic fracture stimulations were simulated. Images of these simulations for the Upper Lower are shown in Figure 12 with the extent of the fractures representing the extent of fluid injection (i.e., actual mass movement). What is seen from these simulations is a reduction in the overall extent of fractures generated in the UL as more fluid volume is occupied by the natural fractures, reducing the propagation distance from the well. The more varied orientation trends observed in these stochastic simulations are broadly consistent with those observed in the field.
smax
smax
smax
6.2. Stochastic DFN based hydraulic fracture simulations
tension
Pf
compression
sn, Normal Stress
Section
Hydrofrac
Fig. 11. Basic mechanics of the FracMan hydraulic fracture simulator as shown on a Mohr Circle.
As is seen in Figure 7, the actual microseismic patterns are strongly indicative of being controlled by a largescale connected fracture and fault system. This can be reproduced in a DFN model in two primary ways: either stochastically where the fracture are generated with the
Fig. 12. Simulation of hydraulic fracturing using a stochastically generated DFN model. Injection into multiple stages of the Upper Lower well only. Simulated microseismicity illustrated in the top right inset.
However, the UL stimulations have resulted in considerable simulated microseismicity that is not observed in the field and so the simulation is not capturing all of the actual processes that took place.
6.3. Deterministic DFN based hydraulic fracture simulations
magnitudes and compare to a standard Mohr Coulomb failure criteria assuming a cohesion of zero and a coefficient of sliding friction for faults. In this example for Pad 13-36 a friction coefficient of 0.5 was used based on the average of a set of bedding plane parallel residual strength tests described by [7]. An example critical stress analysis for fractures from a borehole for the pre-injection stress conditions is shown in Figure 14.
In an attempt to improve upon the pure stochastic models, a series of more deterministic models were constructed, using an interpretation of microseismically defined lineaments, extruded into 3D using stochastically assigned dips from image logs. When hydraulic simulations were run through these models, it was consistently found that the extent of the stimulated length was significantly less than observed by the microseismic events, see Figure 13. The FracMan simulations primarily map the extent to which dilatational opening has occurred and this represents a relatively small proportion of the identified structural network. It is believed that a dilation only of existing structures model on its own can’t fully explain the Lower Montney response and an additional/alternative mechanism is needed.
Directly stimulated volume (dark blue)
sHmax
s3
s2
s1
Fig. 14. Critical stress analysis of fractures under natural (noninjection) conditions showing a number of critically stressed fractures (red triangles).
As can be seen from this figure, even at pre-stimulation pore pressure conditions, a considerable number of fractures are critically stressed, with these related to NNE-SSW and ENE-WSW striking structures. When the general form of the induced microseismic clouds are considered particularly in the toe half of the wells, it can be seen that the main orientation trends follow those predicted from critical stress analysis, as shown in Figure 15.
Model enlarged
Fig. 13. Simulation of hydraulic fracture stimulation using a deterministic DFN model. Left, the DFN model based upon microseismic data and right, the part of the model that has actually received fluid (zoomed in).
6.4. Critical Stress Analysis The concept of critically stressed fractures was introduced by [1]. [18] showed that in many settings with stiff, hard crystalline rocks that fractures and faults that experienced high induced shear stresses appeared to be more permeable. These features were responsible for significant wellbore inflows, production, mud loss events and also wellbore instability. Due to the rough nature of fracture surfaces in some lithologies, when fractures experience small amounts of shear, they ride up and over the natural asperities and dilate, resulting in increasing permeability as well as often generating seismicity. The basic test for critically stressed fractures is to calculate the resolved normal and shear stress
Micro-seismic trends Major Critical Stress Trends
Fig. 15. Comparison between the orientations of the critically stressed fractures, coloured red (left) and the major trends of fluid migration as shown by the microseismic cloud (right).
This provides evidence that during injection, the most favourably oriented fractures and faults for shear failure (i.e., those with the highest ratio of shear stress to normal stress) are re-activated, resulting in the
generation of microseismicity along their length. Given that none of the previously described hydraulic fracture simulations could generate fractures as far from the well as the observed microseismicity, this suggests that it is pressure diffusion ahead of the fluid front that is changing the effective stress, resulting in structure reactivation.
7. SUMMARY When all of the different data sets and modelling are considered together, a reasonably coherent conceptual mechanism emerges to describe what is thought to be happening during fracture stimulations in this setting. It is believed that a number of differing geomechanical processes are controlling the stimulation response that explains some of the anomalous observations. Natural fracture dilation occurs where fluid pressure within the fractures is greater than the stresses acting to close the fractures. When both tensile opening and shearing conditions exist both the inflation and shearing of the fractures will occur. This appears to be happening relatively close to the wellbore and represents the distances predicted by hydraulic fracture simulation and modelling (see black dots on Figure 16). Note how this propped volume corresponds to the microseismicity with the strongest signal-to-noise ratio (SNR), as shown by the largest event sizes.
Additionally, as fluid is being injected into a connected fracture system, there is a pressure front moving in advance of the fluid front. This increases the pore pressure, causing a reduction in the effective normal stress on structures, resulting in the shearing of these structures and the generation of additional microseismicity. Note how the microseismic response on the features in Figure 16 outside the near-wellbore propped zone, show much lower SNR. The timing of these more distant events shows that these are the early events (as shown by their colour), consistent with rapid pressure transmission through natural structures rather than fluid movement. It is believed that conventional hydraulic fracturing does occur. However in the presence of conductive fractures prone to shear dilation and given the relatively high permeability of these structures (relative to the intact matrix), it is much more likely that these structures will take fluid in preference to breaking new rock. Under this situation as seen in this part of the Montney, conventional design and modelling of hydraulic fracture stimulation will fail to consider the natural structures that actually impart some of the most significant influence on the stimulation response. Therefore operators are encouraged to recognize early where a significant structural influence is present and to design and evaluate the effectiveness of stimulation with this natural system of fractures in mind.
Fig. 16. Initial conceptual model to explain structurally controlled stimulations. Microseismic data is coloured by elapse time and scaled by SNR. Earliest events seen at distance along existing structure as pressure diffusion causes shearing. Closer to the well propping of natural structures with shear generates a strong microseismic response. DFN model with microseismic data overlaid on the right.
8. ACKNOWLEDGEMENTS The authors would like to thank Talisman Energy and their joint venture partner Sasol Canada who sponsored the DFN characterization and modelling project which is partially described here. We would also like to acknowledge the contribution of numerous co-workers in the Talisman/Sasol Montney Integrated Subsurface Task Force. The authors wish to thank the management of Talisman, Sasol Canada and Progress Energy (the new co-owner and operator of this asset) for permission to publish this paper.
9. REFERENCES Barton, C.A., M.A. Zoback, and D. Moos. 1995. Fluid flow along potentially active faults in crystalline rock. Geology; v. 23; no. 8; p. 683–686. 2. Dershowitz, W.S., M.G. Cottrell., D.H. Lim. and T.W. Doe. 2010. A Discrete Fracture Network Approach For Evaluation of Hydraulic Fracture Stimulation of Naturally Fractured Reservoirs. ARMA-10-475. 44th U.S. Rock Mechanics Symposium and 5th U.S.-Canada Rock Mechanics Symposium, 27 - 30 June 2010, Salt Lake City, Utah. 3. Dershowitz, W.S. and H.H. Herda. 1992, “Interpretation of Fracture Spacing and Intensity”, Proc. 33rd U.S. Rock Mech. Symp., Santa Fe, NM. 4. Golder Associates 2014. FracMan Discrete Fracture Network Analysis – Reservoir Edition. User Documentation, Version 7.4. Golder Associates Inc, Redmond WA. 5. Mauldon, M. and W.S. Dershowitz. 2000. A MultiDimensional System of Fracture Abundance Measures. Geological Society of America Annual Meeting Reno, Nevada. 6. Mauldon, M. and J. Mauldon. 1997. Fracture sampling on a cylinder: From scanlines to boreholes and tunnels. Rock Mechanics and Rock Engineering, July–September, Volume 30, Issue 3, 129-144. 7. McLellan, P. 2012. Direct shear measurements of bedding plane strength and stiffness, Montney & Doig Formations, Farrell Creek Field, Northeast British Columbia. Presented at Canadian Rock Mechanics Association Conference, Edmonton, Alberta. May 2012. 8. McLellan, P., V. Mostifavi, and I. Anderson. 2013. Geomechanical Characterization of an Unconventional Montney reservoir: In-situ Stresses, Rock Properties and Natural Fractures. Presented at the Canadian Society of Petroleum Geologists Gussow Conference, Banff, Alberta, 15 – 17 October 2013. 9. McLellan, P, I. Anderson, J. Wong and V. Mostafavi. 2014. Geomechanical Characterization of the Farrell Creek Montney Reservoir, Northeast British Columbia. CSPG CSEG CWLS GeoConvention 2014, Calgary, Alberta. 10. Moslow, T. Haverslew, B., & Pelletier, H. 2014. Fabric Selective Impacts on Reservoir Quality and Permeability Anisotropy in Sedimentary Facies of the Montney
11. 12.
13.
14.
15.
16.
1.
17.
18. 19.
20.
Formation, Northeast British Columbia, CSPG CSEG CWLS GeoConvention 2014, Calgary, Alberta. Nelson, R. 2001. Geologic Analysis of Naturally Fractured Reservoirs. Gulf Professional Publishing. Pedersen, S. I., T. Randen, L. Sonneland, and O. Steen. 2002, Automatic 3D Fault Interpretation by Artificial Ants: 64th Meeting, EAEG Expanded Abstracts, G037. O’Connell, S., G. Dix, and J Barclay. 1990. The Origin, History, and Regional Structural Development of the Peace River Arch, Western Canada, Bulletin of Canadian Petroleum Geology, Vol. 38A, No. 1, 4-24. Roberts, A. 2001. Curvature attributes and their application to 3D interpreted horizons. First Break, Volume 19. February. Terzaghi, R. 1965. Sources of Error in Joint Surveys. Géotechnique,Volume 15, Issue 3,01 September, 287 – 304. Thiem, G. 1906. Hydrologische Methoden; Gebhardt, Leipzig. Will, R, R.A. Archer, and W.S. Dershowitz. 2005. Integration of Seismic Anisotropy and Reservoir Performance Data for Characterization of Naturally Fractured Reservoirs Using Discrete Feature Network Models. SPE 84412. Zoback, M. 2010. Reservoir Geomechanics. Cambridge University Press. Wang, X. 2006. “Stereological Interpretation of Rock Fracture Traces on Borehole Walls and Other Cylindrical Surfaces”, Ph.D. Thesis, Virginia Polytechnic Institute and State University. Rogers, S, D Elmo, R Dunphy, D Bearinger, Understanding Hydraulic fracture geometry and interactions in the Horn River Basin through DFN and Numerical modeling, SPE 137488, 2010.
