Geomechanics in Shale Gas Development

Geotechnical Synergy in Buenos Aires 2015 A.O. Sfriso et al. (Eds.) IOS Press, 2015 © 2015 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-61499-599-9-53

53

Geomechanics in Shale Gas Development Maurice B. DUSSEAULT Department of Earth and and Environmental Environmental Sciences Sciences University of Waterloo, Waterloo, ON Canada

Abstract. Hydraulic fracturing in a stiff naturally fractured rock mass such as a shale gas reservoir or a geothermal site in an igneous rock involves a complex interaction between the stress and flow effects of the induced fracture process with a geometrically complex, strongly anisotropic, and undoubtedly heterogeneous medium. Deformation mechanisms include the opening of the primary hydraulic fracture, opening of near-by joints, and small-scale shearing of near-field and more distant joints that are not fully wedged open by the high hydraulic pressures. Aspects of initial state (joint system, stresses, mechanical properties…) are complex and must be better understood. The changes to the system during hydraulic fracturing include large scale stress and pressure changes, as well as irreversible and geometrically complex changes to the joint systems. Because of the different scales involved, large-scale modeling must involve some type of upscaling, but the best approach to this in different cases and for different processes remains obscure. The article attempts to give a physical portrait of the process to help guide model development, but also to help understand what happens in real cases. Some examples of simple 2-D modeling efforts in simulated jointed media are given to show how the physical understanding can, qualitatively, be supported and extended by careful analysis, albeit of a preliminary nature. Keywords. Hydraulic fracturing, naturally fractured rocks, up-scaling, shale gas, shale oil, modeling

1. Introd Introduct uction ion

Some applications of high pressure injection and hydraulic fracturing in naturally fractured rock (NFR) masses include jointed rock mass grouting to reduce permeability and perhaps increase rock mass strength; jointed igneous rock fracturing at depth to aid in the extraction of geothermal energy; deep slurried solid waste injection; and, development development of oil and gas trapped trapped in stiff jointed jointed rock masses. masses. The main driver driver of research at present is the rapid advance of technology in the hydrocarbon industry to exploit shale gas, gas, shale oil and tight gas deposits. deposits. The following sections sections are general discussions, and detail is unreferenced, although many sources over the years could be detailed (Curtis 2002, Harris 2015). 1.1. General Disposition Disposition of Shale Shale Oil and Shale Gas Strata The low-permeability strata in which shale gas and shale oil deposits are currently being exploited in Canada and the United States are called shales, but range in texture and composition from true shales (Haynesville Shale, Louisiana) to mudstones and

54

M.B. Dusseault / Geomechanics Geomechanics in Shale Gas Gas Development

silty shales (Barnett Shale, Texas; Horn River Shale, British Columbia; Utica Formation, Quebec), to siltstones (Montney Formation, Alberta and British Columbia; Bakken Formation, Formation, North North Dakota). Formation mineralogy mineralogy may may be entirely siliceous siliceous (quartz-illite mineralogy - Utica Formation) and range to fine-grained arkosic strata that are carbonated cemented and with secondary calcite cementing one or more joint sets, either partially partially or fully, and even even carbonate-rick laminated laminated mudstones. mudstones. In most cases the hydrocarbon in the shale was generated therein (the “source” rock), although in some shale oil strata (Bakken and Three Forks Formations in North Dakota and Saskatchewan) the light oil and associated gas has migrated into the fine-grained silty strata from adjacent source source rocks. Similarly, in fine-grained fine-grained silts (tight gas) such as in the Ordos Basin in Shanxi, Ningxia and Nei Mongol, China, the gas has migrated substantial distances into the rock, and these reservoirs generally are called “tight gas” strata, and also classified classified as “unconventional” “unconventional” resources. They are all being developed developed with similar technologies: horizontal wells and large-scale l arge-scale hydraulic fracturing. In all these cases, and many other similar formations around the world (Vaca Muerta Formation, Neuquen Basin, Argentina; Longmaxi Formation, Sichuan Basin, China; Frederic Brook Formation, Maritimes Basin, Canada; Canol River Formation, Upper Mackenzie Basin, Canada), the rock mass matrix is of such low permeability (k) that Hydraulic Fracturing (HF) is necessary to generate economically viable production rates. HF may take place place in vertical wells when the the strata thickness thickness is greater than perhaps 200 m (Fig 1), but more and more, HF is implemented in horizontal wellbores from 1 to 3 km long, so as to generate a large volume of enhanced-k rock mass in contact with the well. well. In particular, the thickness thickness of shale gas reservoirs can vary from 20-60 m (sections of the Barnett and Marcellus Shales) to several hundred meters (Montney Formation) Formation) and up to 600-700 m exceptionally, exceptionally, perhaps in cases where thrust faulting has resulted in formation stacking (Frederic Brook Formation). 1.2. NFR – Naturally Fractured Rocks All low-k strata are naturally fractured to some degree; the abbreviation NFR – Naturally Fractured R ock ock - will be used, and these fractures will be called joints, to avoid confusion with the term “fracture”, reserved for the induced HF process as much as possible. One set of discontinuities discontinuities (not actually actually joints), an important important one because because they may serve as general fluid diffusion paths to induced or naturally open fractures, are the sedimentary bedding-planes. bedding-planes. These are generally closed because because of a horizontal disposition and the vertical stress ( v) in the rock rock mass. mass. Bedding-planes Bedding-planes represent represent planes of weakness that are easy to part or shear during during HF. Commonly, there are two or three additional joint sets in a NFR mass that are oriented normal to the bedding planes, formed in the geological geological past when the effective stress () for some reason reason became zero. The condition   0 is the HF condition, condition, and in general one may assume the tensile strength of the rock to have been modest compared to the total stresses stresses when these these joints were were generated. More precisely, the resistance to Mode I (extensional) fracturing was small when joints were generated because granular siliceous sediments sediments have little little true tensile resistance resistance at the grain scale, and because the sharpness of the fractures that were being formed led to local crack-tip extensional stress concentrations concentrations that easily overcame the low tensile strength.. The condition   0 could have been reached when an increasing pore pressure in situ (po) exceeded the least total compressive stress – p o  3. In fine-grained source rocks, this likely occurred during hydrocarbon generation as kerogen was

M.B. Dusseault / Geomechanics in Shale Gas Development

55

catagenetically converted to liquids and gases so that p o rose until HF occurred. This generated a HF network with the dominant joint (fracture) set at 90° to the orientation of  3, which was almost certainly the least horizontal stress ( hmin) at that time. Many shale gas reservoirs, despite being hundreds of millions of years old, are overpressured to this day, especially the Jurassic Haynesville Shale, which has a pore pressure ratio of about 1.8, compared to the hydrostat of fresh water. Others, such as the Horn River Formation, are mildly overpressured (1.3), and in some cases the strata are normally pressured (Utica). Note that the only way that fluid hydrocarbons can escape from argillaceous source rocks is through joints (existing or generated): the water-wet condition of many silica surfaces and the restrictive pore throat sizes (sub-micron) mean that capillary forces as so large that significant non-wetting phase efflux cannot happen through an intact matrix. Thus, internal pressures can be trapped for long times, especially if the shales are clay-rich (Haynesville Shale). The orientation of  3 may have been imposed by distant tectonic forces (extensional or compressional), or lateral stress conditions may have been close to isotropic (hmin   HMAX   3 ( 2)) if tectonism was absent. Logically, the larger the in situ lateral stress anisotropy, the more likely it is that the joint set at 90° to  3 became strongly dominant, whereas in cases of weak  h anisotropy at the time of joint formation, two or even three (e.g. @ 60º orientations to each other) joint sets roughly similar in intensity could form. Mathematical modeling (e.g. Dusseault and Simmons 1982) confirms that when an induced fracture is generated and grows in one orientation, the stress normal to the fracture plane increases somewhat as the aperture expands along with fracture length growth, and the stress parallel to the fracture drops slightly so that locally the direction of 3 rotates. In strata that are horizontally isotropic mechanically, this process must also occur, and favors the generation of a joint set of secondary importance but oriented at 90º to the primary joint set that initiates first. This is shown on Figure 1.

Figure 1: Bedding surface of the Brown Shale, Central Sumatra Basin, Indonesia

The photo shows an exposed bedding surface of the Brown Shale, the oil source rock for the Central Sumatra Basin in Indonesia. In this exposure, the kerogenous


56

Brown Shale is at the surface, but in the central part of the basin it is 2.5-3 km deep, just deep enough to have entered the oil-generation window, and the oil in fields such as Duri (the largest producing heavy oil field in Asia for 30 years) was sourced from this formation. Where the Brown Shale has generated oil deep in the basin, it is thought to be a potentially rich shale oil prospect. The dominant joint set strikes SW, pointing at the mountains, indicating that  3 = hmin was parallel to the mountains, a common condition in foreland compressional basins. The secondary orthogonal joint set (white solid lines) is non-penetrating, with joint terminations against the primary joint set, showing that they formed after the primary set. Another non-tectonic reason for joints to form and generate a NFR mass is loss of volume during diagenesis, which has three primary sources:   

Continued consolidation; i.e., expulsion of water because of an excess pore pressure arising from loading as burial continues, with delayed drainage. Loss of bulk volume as oil and gas are expelled from the source rock Mineral diagenesis, in particular a smectite-to-illite transition, accompanied by a significant -V and generation of quartz and additional free water.

When general volumetric shrinkage occurs,  v remains unchanged because the weight of the overburden must be carried, but  h (both hmin and HMAX) must drop because of the no-lateral-strain condition that exists in laterally extensive flat-lying strata. Thus, 3 (hmin) drops until the HF condition is reached (p o  3) and vertical joint sets are formed. This volumetric shrinkage processes are often taking place at the same time that overpressures are being generated through oil and gas formation. 1.3. Nature of the Discussion The response of NFR masses to HF stimulation is the subject of this discussion. A descriptive mechanics-based approach is taken, focusing on the nature of the physical processes involved, which then serves as the foundation of mathematical model development. It is also important to understand the large-scale general behavior of NFRs because of significant constraints on what can be achieved in multi-physics mathematical modeling, requiring some scale of homogenization (up-scaling) while still capturing the physical evolution of the system (e.g. induced flow anisotropy). Some challenges to mathematical modeling presented by NFRs are the following: 







The joint fabric at depth (frequency, persistence, orientations…) is uncertain, and can only be approximately stipulated near boreholes where limited penetration depth high resolution seismic logging, borehole wall imagery, and cores are available. The initial spatial distribution of open and closed joints, their connectivity and aperture are uncertain; well test methods (pre- and post-fracturing), though valuable, are not as strongly predictive as in conventional reservoirs. The flow properties of the interconnected NFR network created by HF change sharply and in a non-linear manner as effective stress changes ( ij) alter fracture aperture, and the constitutive behaviors (Barton et al ., 1985) of the different joint sets is difficult or impossible to stipulate rigorously. The shear behavior of joints, particularly during HF stimulation, is important in generating interconnected flow paths, but the ductility of the rock, the joint smoothness and their shear stiffnesses and effective normal stress


57

compressibilities within the deep constraining rock mass are hard to specify in a consistent, rigorous manner. These challenges, combined with the efforts needed to mathematically couple different processes (thermal, hydrodynamic, mechanical), lead to a vexing but exciting problem that requires analysis at different scales (Geers et al . 2010): 





The small scale (millimeters  meters) processes include joint-matrix coupling, mechanical behavior of individual joints (compressibility, hysteresis…), and multiphase fluids flow (gas-oil-water) in heterogeneous joint systems. The intermediate scale (meters  10’s of meters) processes involve blocky rock mass behavior, local stress redistributions, shear slip of joints subjected to high deviatoric stresses, and changes in mechanical properties because of the changes in joint apertures that are taking place. The large scale (> 50 m) involves interactions with stress field changes among different HF zones and adjacent wellbores (~100 to 300 m apart), the effect of far-field boundary conditions, the presence of through-going features such as lithology changes or faults, the mechanical properties of the bounding strata, and large-scale pore pressure changes with stress redistribution at the interwellbore scale (>100 m).

It is not yet clear what approaches will be optimum for the mathematical modeling of NRF masses subjected to HF and then to years of pressure decline from production, perhaps followed by one or more episodes of restimulation (re-entry and additional HF). What is clear at the present time is that these models must be sensitive to different lithostratigraphic conditions, different virgin stress and pressure fields, different rock fabric (at the large scale), different phase behavior (especially in the case of oil-gas systems), and different assemblages of flowing fluids (gas, oil, perhaps some water) in an interconnected induced fracture/joint system that is experiencing an evolution of properties. Because of this complexity, there is merit in identifying the first-order issues in particular cases, and focusing the modeling effort on those issues, rather than trying to include all of the physical processes. To achieve this goal and make reasonable predictions, the physics and modeling of the processes must be understood (Jing 2003; Lanru and Feng, 2003); then, parametric analysis are undertaken to identify the most important processes (Sarkarfarshi et al. 2014).

2. General Large-Scale HF Behavior

In a geological environment, the dominant control on HF behavior is the stress field (Hubbett & Willis 1957). Stress distribution details have a major effect on lateral and vertical extension, as lateral stresses may differ in different strata because of burialerosion history, diagenesis and tectonic loading or unloading. HF is a typical work minimization process, and there are several sources of work in a generalized HF propagation process: 

The work of opening a fracture against the resisting force, which in a simple planar case can be taken to be approximately 3 (hmin). o For HF in a NFR mass, this will involve block rotations.


58

o

 



Elastic energy is also stored by increases in compressive forces, mainly  to the fracture plane, evidenced as + 3 in practice. The viscous energy dissipation arising from flow along the induced and natural fractures and flow into the porous medium bounding the fractures. The work of fracture, the process of rupturing the intact rock or opening a somewhat cemented natural joint, which is mainly a largely scale-independent crack-tip process of overcoming inherent cohesive forces. The work of shear, associated with shear displacement along joint surfaces having frictional resistance and some cohesion.

These processes interact in a complex manner in a NFR mass, but some generalities can be noted. Once the fractured zone length is well beyond the wellbore scale in a NFR mass, the proportion of the work from fracture tip dissipation processes is overall negligible. The largest work component by far is the expansion and extension of the zone, the first in the list above. Viscous flow work is a strong function of the viscosity, and if high viscosity fluids and high rates are used, the induced fracture zone is shorter and fatter, but higher injection pressures are needed. Conversely, if slickwater fracturing is employed, based on adding friction reducers to reduce viscous resistance of water flow in narrow-aperture fractures, greater stimulation distances and volumes can be achieved, providing that fluid leak-off rates are not excessive. High leak-off rates lead to much energy loss through flow into the flanks of the fracture, suppressing propagation and contributing to a larger poroelastic back-stress because of the slight expansion of the rock mass and small-scale sjoint hearing with some dilation as the high pressures slowly propagate into the rock matrix (a slow process in true shale, far easier in tight sandstone). Finally, although the work done in shearing of natural joints is only a small but steady contribution to the work involved in HF, far less than the work of general expansion, its role cannot be ignored in enhancing natural joint conductivity.

Figure 2: HF development in a shale gas or shale oil reservoir

For the first HF in a horizontal well (Figure 2), generally at the toe of the well, one set or several sets of perforations are isolated, and a fluid pumped in at high pressure. In this case, the virgin stress field controls fracture behavior and an approximately ellipsoidal shape is generated for the stimulated zone. For the second and subsequent


59

HF along the horizontal axis of the well, stresses have been altered somewhat as the result of the volumetric strain imposed by the first sets of HF, and induced fractures can be expected to respond to the altered stress field by propagation direction changes. The presence of adjacent wells that have been subjected to HF stimulation, and the stress changes associated with depletion as production develops, will also affect HF propagation behavior. Complexities of HF operations are beyond the discussion scope, but much trial-and-error work is done to find optimized approaches, which will certainly be different in the various types and depths of reservoirs being developed. 2.1. Effects of the Initial Stress Field on Fracture Behavior In the great majority of shale gas HF operations,  3 = hmin, therefore induced fractures or fracture networks will be vertical and oriented macroscopically at 90° to the  hmin direction. In cases of geological uplift and large-scale erosion in an otherwise nontectonic basin, such as the Michigan Basin or the Western Canada Sedimentary Basin 100 or 200 km from the disturbed belt, the condition  3 = v may extend from the surface to depths of perhaps 1 km. Thus, deep HF will be vertical because 3 = hmin. The stress contrasts in the horizontal plane will affect how fractures propagate. A high contrast (e.g. hmin > 1.2·HMAX at 2-3 km depth) will result in a much stronger and consistent HF orientation with a smaller tendency to generate ancillary fracturing in the orthogonal direction. Smaller stress contrasts make it easier to have additional smaller fractures in the direction parallel to  hmin, and this should lead to less length extension of a fracture network  to  hmin, and a concomitant increase in the width of the fracture network. Furthermore, the weaker the stress contrast, the more likely it is that stress changes induced from one fracture stage will affect the fracture propagation in subsequent stages, so that the fracture networks become affected and adopt complex shapes, dependent on the local stress field alterations. Spanish Peaks dike patterns show these effects: the primary HF (dike) propagated  to hmin (which is regionally  to the mountain front, but subsequent injection episodes were more and more affected by the induced stress changes from repeated magma injections (Figure 3).

Figure 3: Previous HF injection episodes (dikes) affect subsequent propagation behavior


60

2.2. Stress Changes from Depletion Consider a laterally extensive flat-lying reservoir that is being depleted uniformly. p means that there is a small V (increase in  ij). The induced vertical displacement component occurs in a straightforward manner, but the lateral displacement boundary condition is  = 0, so stress change must take place. For a homogeneous elastic porous rock, the following happens:  v = p, and h  p·(1 – 2)/(1 – ), v remains constant, and  is Poisson’s ratio. So, using continuum mechanics understanding, we can extend this simple model to more realistic reservoir development practices (nonuniform in time and space), and some major points can be made: 



 



Reservoir depletion reduces the total lateral stress in the p region, therefore it also reduces the HF initiation and propagation pressures. ij effects propagate farther in the reservoir than p effects. The h “lost” in the depleted zone must be redistributed above and below the reservoir, into the bounding strata, therefore creating a “stress barrier” to upward propagation (see next Section). The spatial distribution of p will not be uniform, therefore the values of  h (and indeed ij) will also be non-uniform. During later well life refracturing, a practice being carried out more and more in shale gas and shale oil wells, HF behavior will not be the same as during the primary well completion using HF in the virgin rock mass. o Preferred HF propagation orientations will be different from place to place, depending on the spatial distribution of p. o HFs will not be of constant orientation as they propagate because the hmin orientations are no longer consistent in one direction, as petroleum engineers would normally expect. Because stresses are transmitted through the solid matrix but pore pressures are local and scalar in nature, poroelastic analysis is needed to calculate the new values of ij at all locations in the region of interest.

The last conclusion, in particular, presents the petroleum geomechanics practitioner a major challenge: 



 

To rationally optimize HF refracture treatments and even primary HF stimulation, it is best to have an understanding of the distribution of  hmin (3), and secondarily, of the magnitudes of the other principal stresses. After some depletion, this requires knowledge at some reasonable level of the spatial distribution of p, as well as a good geological model, so that  can be calculated through use of a hydromechanically coupled calculations in a 3D mathematical model - often called poroelastic modeling from Biot (1941) & Coussy (2004) - despite the plasticity aspects that may be involved. It is not possible to install many monitor wells, and the depleted regions’ spatial extent cannot be determined by testing of the production wells, thus: Perhaps the only realistic path forward to improve models is high-sensitivity (high-frequency) periodic 3-D seismic surveys that can delineate p and  ij distribution through small changes in wave velocities and attenuation. A resolution scale of even 100×100 m in the reservoir would be a great boon.






61

Geomechanical models that are more detailed must be developed because heterogeneities in geomechanical properties exist a priori, and additional heterogeneities are generated by the HF network creation, including effects of fracture opening, shearing and dilation of joints, proppant placement, and related effects that take place during HF and depletion. Lab and field data are needed (Jeffrey et al . 2009, Van Der Baan et al . 2014).

2.3. Fluid Buoyancy Effects Large-volume HF is generally executed with water-based fluids; smaller scale HF in reservoirs with water-sensitivity strata (containing swelling clays, sensitive to capillary blocking…) may be executed, at greater expense, using liquid CO 2, liquefied propane, even N2 or foams. These fluids have different densities, and this affects fracture propagation, especially the tendency of fractures to rise. Figure 4 shows a sketch of the concept. In the simplest case of a static fracture, the excess pressure at the top is a function of the difference between the fluid gradient in the fracture and the lateral stress gradient – hmin/z. For typical water-based HF fluids this is about 11 kPa/m, whereas lateral hmin gradients in virgin conditions are in the range 18-22 kPa/m in typical shale gas conditions at 2-3 km. If less dense fluids such as gelled propane or liquid CO2 are used, this density contrast is increased, but in those cases, injection volumes per fracture stage are far less. At this stage in the evolution of the HF industry, non-aqueous HF fluids are costly and still in the experimental or very early development phase.

Figure 4: Induces HFs tend to rise because of pressure and stress gradient differences

This rising tendency is often (but not quite accurately) called the fluid buoyancy effect (it is actually a mismatched gradient effect), and it leads to a higher HF driving pressure at the top of the fracture than at the bottom, enough to enhance upward and suppress downward propagation. Under normal conditions, this would lead to placement of horizontal wellbores at about the 0.25 level of the reservoir, depending on fluid viscosity, density and injection rate, unless unusual stress conditions exist. In some cases (Barnett Shale), there is a basal aquifer that must be avoided, so HF well placement is higher in the zone to restrict catastrophic communication with free water.

62


3. Wedging, Propping and Shear Dilation

3.1. Wedging During HF The primary mechanism for HF in a geomaterial without strong planes of weakness (e.g. a relatively homogeneous sandstone), is the propagation of a Mode I (extensional) fracture. For a rock mass with strong discontinuities, the author refers to this as “wedging”, shown schematically in Figure 5. Indeed, in a homogeneous weak rock without significant discontinuities, parting or wedging fully dominates the induced fracture growth, and the fracture is, at least initially, a single plane  to hmin. In a NFR, it is the natural planes of weakness that are wedged open by the fluid pressure. Proppant may be added to the fluid, but because proppant can only enter a crack at least 5-6 grain diameters in width, the proppant lags behind the crack propagation, and the black wedge in the figure is intended to represent this. In a high modulus (rigid) rock mass, the crack can be open some distance (several meters?) in advance of the proppant penetration front.

Figure 5: Wedging open and propping a natural fracture plane with injection pressure

An important aspect of the HF process in a stiff NFR is that the wedging involves some small rotations, and these rotations must be accommodated by the rock mass, the importance of which will be discussed further. In particular, these rotations are transmitted through the rock mass, and before the displacements become small with distance from the induced wedging fracture, there may be important displacements happing in adjacent joints (Figure 6).

Figure 6: Rigid block rotation easily opens some adjacent unbonded fractures

Nevertheless, the general shape of the fracture network, including opened adjacent joints, must approximate an ellipsoid, with some shape modifications because of buoyancy effects, non-uniform distribution of stresses among strata, non-uniform mechanical rock properties (stiffness, Poisson’s ratio) and the possible presence of strong natural fabric oriented at a low angle to the principal stresses.


63

Continuum mechanics shows that an ellipsoidal shape  to 3 is the minimum work geometry to accommodate the injected volume. If there is little tip cohesion (uncemented natural fractures), this shape is slightly modified at the fracture tips. The shape of the zone must be macroscopically smooth, narrow, and long. In a homogeneous case without buoyancy, the bulk normal aperture is greatest at the midpoint. Thus, the shape of an induced fracture zone in terms of aperture distribution is less like the shape of the root system of a tree, but more like that shown in Figure 7, with bulk V in the tip region very small, and increasing toward the point of maximum aperture (at the wellbore in this representation). The ellipsoidal zone is the region most affected by displacements, dominated by wedging.

Figure 7: Generation of a wedged and partially propped zone in a NFR

The representation is greatly exaggerated in thickness in the direction of  3; the L/a ratio is much greater than shown. The proppant cannot enter in to the induced wedged joints all the way to the tip, many of the small aperture lateral responses are devoid of proppant, but still possess greatly enhanced conductivity that will aid in gas flow, and that increases the open surface area within the volume, accelerating diffusional gas escape from the matrix blocks in the NFR. The diagram shows the primary HF, in darker blue, as a stepped shape. The next Section describes the mechanics of this shape. 3.2. Small-Scale Fracture Propagation in a NFR Mass Assume a joint intersects or is close enough to the initiation point so that it presents a plane of weakness to the fluid being injected. This is of high probability if perforations are used or if a substantial length of borehole (e.g. 25 m) is simultaneously subjected to the high HF pressure. Given a high matrix cohesive strength, it is highly probable, indeed almost certain, that the fracture will initiate in some direction other than  to 3. This also means that the fracture initiation pressure will reflect a different tangential stress than one might surmise from an elastic borehole stress analysis. Indeed, the classical concept of breakdown pressure must be recast for HF in NFRs, and values from one point to another in the rock mass are unlikely to be consistent because of


64

different intersecting joint angles to the borehole. The wellbore scale mechanisms involved are shown in Figure 8.

Figure 8: Fracture departure along an intersecting natural fracture

Because the matrix has substantial fracture resistance compared to the joints, and because the joints in opening slightly as the pressure increases generate to a high Mode I tip stress concentration, propagation of the induced fracture opening is little impaired by cohesive resistance, and as the fracture grows longer, the cohesive strength of a largely unbonded joint contributes almost negligibly to the overall fracture zone growth. However, fracture tip processes remain important at the local scale. Figure 9 shows the mechanism of induced fracture orientation changes. A more detailed explanation is warranted. 

 



  

When a fracture is opened in a joint-controlled direction that is not coincident with the principal stress direction, the shear stress must be relieved because the fluid in the fracture cannot sustain a shear force (upper sub-figure). As the shear force is relieved, there is a small shear displacement relative to the faces of the joint, in the example shown, leading to left-lateral movement. This creates a loading a distortion of the rock mass in the region of the open crack, and this causes loading and unloading of the faces of ancillary joints, an effect shown in the crack tip figure inset in the upper left. When the internal fracture pressure “feels” a sufficiently large local stress redistribution in the vicinity of the tip, it will follow the path of least compressive stress. This causes the fracture to turn in a specific direction predicated by the natural fracture system fabric. The re-oriented fracture continues to propagate in the new direction, only now the induced shear displacement is a right-lateral movement in the figure. And, the process continues as the overall fracture length grows, keeping the macroscopic fracture (scale much larger than the characteristic joint length) oriented approximately  to 3, in the most energetically favored global orientation (work minimization).




65

This leads to various distortions and small rotations of the adjacent rock blocks, leading to the effect in generating a fracture network shown in Figure 7, leading to an approximately ellipsoidal shape, shown in Figure 2.

Figure 9: Induced local fracture direction changes as a function of orientation and scale

The propagation distance at which the fracture will reorient is a function of the stress difference magnitude, the orientation of the joint fabric with respect to the principal stress field orientation, the cohesive resistance of the joints sets, and the mechanical properties of the matrix blocks and of the rock mass. For example, if the major joint fabric angle is close to 90° from  3, the induced fracture will propagate a long distance from the borehole before changing direction. On the other hand, if the initial propagation is along a joint oriented at a large angle to the primary preferred HF direction, such as 60-75° from  3, the rectification forces build up more quickly with length because the shear stresses that are relieved by the opening of the joint are larger. Note that the fracture tip processes are scale-independent: they occur at the scale of the rock fabric, which is predicated by the geological history and lithostratigraphic factors. However, because the induced fracture zone grows in size, the tip energy becomes less important as the large-scale fracture length extends.

3.3. Joint Slip and Shear Dilation A final major point for HF processes in NFRs is that the entire process is accompanied by small-scale shear slip, sub-millimetric scale generally, along limited lengths of the affected joint. These are detectable as stick-slip microseismic events (Warpinski 2014) which can be recorded within the propped pod region, but as pressures propagate, it

66


also occurs in the region beyond the propped pod (Pirayehgar & Dusseault 2015 simulation attempts). The joint slip and shear dilation effect occurs because of the reduction in  n across a favorably oriented joint as pore pressure increases. Figure 10 shows the process.

Figure 10: Shear displacement of a joint in a favorable orientation to stresses

When the slip criterion is satisfied for the joint, a small shear displacement occurs to relieve the shear stresses, and this causes a local (scale no larger than a few meters) stick-slip event. However, the joint surface is rough at the small scale, and some dilation takes place, leaving remnant conductivity along the joint plane. This is a classical Mohr-Coulomb shear mechanism (Figure 11); irreversible, controlled by joint friction and cohesion, as well as by the stress orientations and the degree of pore pressure penetration along the joint.

Figure 11: Favorably oriented joint shear as a Mohr-Coulomb process (simplified)

When the joint slips, the distortion may be mainly ductile if the rock is clay-rich and less cemented; smearing and plastic deformation may lead to little or even no remnant conductivity. In brittle rocks, the small aperture is more easily maintained, but as depletion occurs,  n increases, and the joint will compress, reducing the drainage effectiveness (a function of asperity ductile deformation and the joint compressibility). This helps explain why, in clay-rich shale gas strata such as the Haynesville shale in


67

Louisiana, HF treatments do not lead to as long a well life as in other, more brittle strata, such as the Marcellus quartz-illite (and some CaCO3) shale. The importance of small-scale fabric at the joint surface level is substantial, as rough joints will be far more difficult to mobilize, but will sustain conductivity, whereas smooth joints will shear more easily, but with far less dilation. Of course, the proper choice of joint compressibilities and mechanical properties remains one of the challenges of mathematical simulation of the HF in NFRs process. Now, the picture is more complete, and an additional component of deformation, the shear dilation, has been described. This contributes to the overall flow enhancement of the HF zone, and leads to the concept of the “stimulated volume”, shown in Figure 12, where there is a central “propped pod” surrounded by a region where shear and dilation (evidenced by microseismic events) has enhanced conductivity, albeit far less than within the central zone. This has led many to use an analogy of a road network, where the superhighways are the propped, wide-aperture joints, the secondary highways are the unpropped joints opened that were opened by rigid block rotations near to the propped joints (Figure 7), a nd the distant region mostly contains the network of country roads arising from shear of joints. The latter roads are rough and small; they can only sustain limited, slow traffic, yet nevertheless contribute substantially to the overall flow toward the wellbore because the stimulated volume is much larger than one would surmise only from the wedging and propping processes.

Figure 12: The stimulated volume from HF in an NCR

4. Induced Permeability Anisotropy in NFR Masses

NFRs not only are anisotropic in their bulk permeability, the nature on the anisotropy is that of a high-order tensor because of the presence of different joint sets at different orientations (i.e. a second order tensor has an insufficient number of independent parameters and assumes orthogonality). Each joint has a conductivity, and averaging over a REV (L>>LS), where LS is a characteristic mean spacing, yields a bulk directional permeability value. In a NFR with two orthogonal joint sets at 90° to bedding, one might reasonably expect to define three principal permeability values co-

68


axial with the major discontinuity directions, and this might be, at a large scale (L>>L S), approximated by a second order permeability tensor. However, in a NFR with three joint sets orthogonal to bedding, for example at 60° orientations, additional terms are needed to describe the bulk up-scaled permeability. More challenging than defining initial up-scaled permeability is tracking kchanges at an appropriate scale during HF, altering joint apertures, and thereby altering individual joint flow capacity in a strongly non-linear manner, often taken to be cubic in nature (i.e. if all else is constant, q  a3). Furthermore, these changes take place at scale less than the REV scale necessary for up-scaling. It is not clear how to resolve this conundrum to allow more efficient simulation of HF processes. The discussion below does not lead to a strong recommendation, but it suggests that simulations will have to be based on the stipulation of a spatiotemporally evolving “permeability” akin to a higher-order damage mechanics formulation, linked to aperture increases and shear displacements. How to formulate this in a useful manner remains an important task. 4.1. The Effect of an Induced Fracture The impact of an induced fracture in a NFR is to introduce a strong additional anisotropy upon an already complex permeability. Figure 13 shows the approach to a 2D DEM simulation of the pressure development in a NFR mass, developed by ITASCA and others (e.g. Thallak et al . 1991). Further details of the DEM approach may be found at the ITASCA website and widely-available professional articles.

Figure 13. The flow computation scheme in DEM modeling – from the ITASCA Corp. web site

The first major effect is the opening of the discrete fracture system by the HF process, shown in a DEM simulation. The model simulation parameters for fluid are not so relevant as the clear imposition of flow anisotropy governed by the differential stress field that acts upon the NFR. The DEM example shown below (Figure 14) is statistically homogeneous at the large scale, formulated with random Voronoi polygonization that gives an initially isotropic up-scaled permeability. All joints and contact stiffnesses are identical, and in principle it is straightforward to introduce spatial or statistical variations in these


69

material parameters. The induced propagating HF discontinuity provides one dominant flow path roughly normal to  3, other effects such as shear are probably less impactive.

Figure 14. Induced anisotropy on the flow regime as the result of a HF (Pirayehgar, ongoing)

The effect of stresses and the primary fracture propagation direction is clear, even in this statistically random simulated joint system that does not have strong joint sets. 







The black lines represent those joints wedged open more than a small defined limit, and represents the HF network with the ancillary, close-by joints that are also wedged open. The system is subjected to a differential boundary stress, and injection takes place at the central site. The strong fabric (the rock blocks cannot be fractured in this simulation causes a stepped shape in the directly affected zone as the fracture propagates approximately to  to the far-field 3 orientation. Local non-randomness of fabric at a small scale leads to some departures in the HF zone orientation. The colors are linked to different values of pore pressure in the joint system which possesses a small initial joint conductivity value, and is also linked to aperture changes in deformed joints. Pressures are highest near the black lines representing the wedged region of the rock mass, giving an overall anisotropy of pressure field response, a roughly elliptical area. This process is affected by joint fabric, aperture-conductivity relationships linked in part to joint compressibility, and the magnitude of the differential stress boundary conditions ( MAX – min).

Figure 15 shows how some of the effects of a fault and of shear of joints can be evaluated in a DEM approach. The figure (a different tessellation than before) has an included fault, and it is interesting to note several major features resulting from the simulation. In examining these figures, remember that the ultimate goal is not only to understand the various effects, but potentially to find a way to up-scale such results (D’Addetta et al . 2004).


70

Figure 15. Introduction of a fault plane and shear effects in the DEM simulation (Pirayehgar ongoing)

 





Overall, the pressure pattern is little affected by the presence of the fault, the flow patterns remain dominated by the black-lined central HF zones. On the right-hand side, the planes of slip are shown, with colors scaled to the magnitude of slip. The fault has slipped a large amount because of the differential stress field, the favorable fault inclination, and the fracturing. Outside of the central region are locations where the joints have exhibited shear above a small, pre-set value. The low level of interconnectivity of these slip surfaces is an artefact mainly of the choice of frictional coefficients of the joint surfaces and the displacement limit chose for graphical representation. As stated above, specific results are affected by fabric, dilation functions, and the magnitude of MAX – min.

4.2. The Effect of Stress Anisotropy on Flow Patterns in a NFR Mass To more clearly show the effects of different magnitudes of differential stress, several figures are presented. Figure 16 is not a DEM model; it is a model formulated with displacement discontinuities for the joints combined with a finite-difference scheme for the flow both within the joints and in somewhat permeable rock matrix blocks. This scheme is more suitable for investigation of such cases as well as for thermal coupling issues (T causes aperture changes and therefore affects p and ij). Figure 16 does not contain a HF plane (HF is somewhat easier to study in a DEM approach). In other words, all of the stress-induced flow pattern changes in this model are pre-HF, to emphasize that even in cases without the generation of open HF fractures (wedging) the interaction between the strong fabric elements and the principal stress directions is important. Figure 17 is a DEM simulation of several HF cases in a random tessellation similar to Figures 15 and 16, but with different stress ratios ( 1 is the same in all cases). There is a clear change in the HF opening zone shape, linked to the stress difference. The right-hand figure shows that in an isotropic stress field there is a roughly circular branching of the zone in which joint apertures are wedged open. Also, the propagation distance is less than in the other examples. Though not shown, flow patterns are


71

similar, with more pressure anisotropy in the examples with higher deviatoric stresses. Also not shown are the joints that experienced shear slip, although it is clear that this joint shear-slip behavior is stronger in cases where the deviatoric stresses are the largest.

Figure 16. Effect of stress rotation on fluid flow in joints in simulation without hydraulic fracturing,  is the CW rotation angle of the 3 direction applied to the model (e.g. Jalali 2014, Jalali et al . 2015)

Figure 17. Effect of stress difference on the HF zone shape in a random DE tessellation (Pirayehgar 2015)

The major factor on flow pattern anisotropy in these simulations remains the opening of the highly conductive HF planes, but there is an effect of the stress differential, mostly in that the shape of the zone of fracturing becomes less strongly anisotropic as an isotropic stress condition is reached. For simplicity, only the “wedged” primary cracks were shown.

72


Figure 18 is a composite of the effect of stress differences on a DEM pore pressure simulation, showing the pressures across the horizontal midpoint of the models. Many further observations can be made, only a few are given here.

Figure 18. Effect of stress difference on area of the zone and pressure distributions (Yetisir 2015)

The pressure distribution control changes sharply throughout the affected zone because of the induced aperture changes and the tip of the HF opened zone can be seen as a sharp corner in the pressure curve across the midpoint. Beyond the HF opened zone, there is still a pressure decline because all joints are permitted some flow. As discussed earlier, it is feasible to alter and statistically distribute such joint parameters, but the major effects can be seen in the figures.


73

In an attempt to “quantify” the effects of fabric and stress ratio on the affected zone area and the pore pressure distribution, values are given on the plots. It is well to remember that these are simulated values, and although we believe that the general physical picture is clarified, scaling of DEM simulations to real cases (D’Addetta et al . 2004), introducing stochastic factors, and altering fabric patterns represents a great deal of computation effort, in addition to trying different ways to display the results and determine the dominant (first-order) effects in a relative manner. 4.3. Fabric Effects and Flow Patterns The final computational results shown for this general introductory article are two figures showing the effects of strong fabric on the HF response. The effect of strong fabric on pressure distribution in a non-HF case was already shown in Figure 16; Figures 19 and 20 are only examples of the effect of several of the myriad joint fabric patterns that may be postulated. Figure 19 shows some results on both normal stress-controlled wedging of joints and shear displacements of joints for the same fabric model, but subjected to the different differential stresses in the plane. The magnitudes of the simulated shear displacements is larger than what takes place in situ (less than a millimeter based on microseismic event inversion), but these displacements are functions of the joint properties chosen for the simulation. Because a high limit for plotting was chosen, 2.5 mm, only those joints very close to the wedged regions on the left are shown. Had a lower displacement limit been chosen, the region would be shown to be populated with far more slip planes, decreasing in displacement outward from the wedged planes. In Figure 20, the beginnings of studies on the effects of various scales are represented by a preliminary investigation of the size of the model with respect to the scale of the DEM blocks. The question was initially to understand if it was possible to generate reasonable simulations without worrying if the rock mass discretization scale was too coarse with respect to the dimensions of the model. The results show that, for these two scales, the results are similar. Although the larger block model gave a larger affected area for the pore pressure changes, and there is, as expected, more fine detail in the smaller block simulations, this gives some confidence that first-order effects can be reasonably captured by relatively coarsely discretized models. However, such tests must be repeated when strongly different fabrics are investigates to assure that artefacts arising from model ill-design are kept to a minimum. These results should also give some insights into the issues associated with up-scaling so that more effective, reservoir-scale modeling can be reasonably achieved with current computational capabilities.

74


Figure 19. Showing the impact of stress ratio on both wedging and shear (Pirahehgar & Yetisir 2015)


75

Figure 20. Exploring the scale effects of simulation results (Pirahehgar & Yetisir 2015)

5. Final Comments

Extended discussions about many ancillary factors, including approaches to up-scaling, the effect of injection rate, fluid viscosity, and so on, are the subjects for more detailed articles, and the students mentioned in the acknowledgements are working on some of these issues. Figure 21 is a physical representation of the major mechanisms of wedging, propping, and shear during fracturing. Each aspect is non-linear, the system is heterogeneous at many scales, and the fabric, joint conductivity and stresses are altered as HF goes on. Post-HF, the depletion behavior is also non-linear as the system parameters (e.g. joint conductivities and compressibilities) change with the effective stresses, and large-scale stress change effects take place. All of our simulations so far are carried out in models with homogeneous stress conditions, despite clear evidence in practice that stresses can be different in different strata, that there is a regional gradient in the lateral stress, and that the lateral stress gradient with depth is also changing. In summary, a few points: 

The HF process in NFR masses is a complex interplay of the induced HF process with a zone that has an important joint fabric. o Fracture propagation is dominated locally by the fabric o Globally, the fractured zone development is intimately related to the differential stress field, dominantly the minimum stresses and the deviatoric value. o Large-scale stress changes and pore pressure changes take place during injection and depletion, affecting HF behavior during the initial HF stages,


76





 

the production history, and the rock mass behavior during subsequent refracturing episodes. o Changes in primary orientations are likely because stresses are affected at an appreciable scale by the volumetric changes associated with HF o A stimulated volume is developed, dominated by wedging in the central part, and by shear displacement and dilation outside of the central zone, which is partially propped by co-injection of a granular agent. The processes involved lead to nonlinear slip and wedging behavior of joints, and also of the rock mass itself, in that large-scale properties (such as rock mass compressibility) are altered by creation of wedged and sheared zones. o Shear slip is a major effect, but is dominated by wedging in terms of energy dissipation, stress changes, and impact on permeability changes. o The magnitude, geometric scale (extent) and production impact of the shearing and wedging are governed by joint behavior and the lithology and stiffness of the rock mass, as well as the joint geometry, and the magnitude and continued existence of the dilation effect as stresses and pressures change. Geoscience aspects of heterogeneity in rock properties and stress distributions are primary factors. Designing HF operations and optimizing treatments must be based on a detailed geomechanics earth model, well populated with reasonable parameters measured or inferred using strong inference methods. Petrophysics contributions to an understanding of mechanical response cannot be underestimated (Bust et al . 2014): geoscience underpins engineering. Upscaling approaches must be developed to allow physically robust modeling of the processes involved; this is a challenging area.

Figure 21. The “child’s blocks” representation helps to understand the mechanisms

References [1] Barton, N., Bandis, S., Bakhtar, K. (1985). Strength, Deformation and Conductivity Coupling of Rock Joints. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts. 22 (3), 121-140. [2] Biot, M. (1941) General Theory of Three Dimensional Consolidation. Journal of Applied Physics. 12(2): 155-164. [3] Bust, V.K., Majid, A.A., Oletu, J.U., Worthington, P.F. (2013) The petrophysics of shale gas reservoirs: Technical challenges and pragmatic solutions. Petroleum Geosciences, 19, 91–103. [4] Coussy, O. (2004) Poromechanics. Wiley, London. [5] Curtis, J.B. (2002) Fractured shale-gas systems. AAPG Bulletin, 11(11), 1921–1938.


77

[6] D’Addetta, G.A., Ramm, E., Diebels, S., Ehlers, W. (2004) A particle center based homogenization strategy for granular assemblies. Engineering Computations 21(2/3/4), 360-383. [7] Dusseault, M.B., Simmons, J.V. (1982). Injection-induced stress and fracture orientation changes, Canadian Geotechnical J ., 19, 4, 483 493. [8] Geers, M.G.D., Kouznetsova V.G., Brekelmans W.A.M. (2010) Multi-scale computational homogenization: Trends and challenges. Journal of Computational and Applied Mathematics 234, 2175-2182. [9] Harris, N. (2014, 2015) Personal communication and presentations. Univ. of Alberta, Edmonton AB [10] Hubbert, M.K., Willis, D.G. (1957) Mechanics of hydraulic fracturing. Transactions of the American Institute of Mining Engineers. 210, 153-168. [11] Jalali, M.R. (2014) Thermo-Hydro-Mechanical Behavior of Conductive Fractures using a Hybrid Finite Difference – Displacement Discontinuity Method , PhD Thesis, University of Waterloo, Ontario. [12] Jalali, M.R., Cameron, R. and Dusseault, M.B. (2012). Hydro-Mechanical Response of Hydraulic Fractures. Proc. 21 st Canadian Rock Mechanics Symposium. Canadian Rock Mechanics Association (CARMA) Edmonton, AB, 8p. [13] Jalali, M.R., Evans, K.F., Valley, B.C., Dusseault, M.B. 2015. Relative Importance of THM Effects during Non-isothermal Fluid Injection in Fractured Media. Proc. Amer Rock Mech Assoc Conf. ARMA 15-0175, 8p. [14] Jeffrey, R.G., Bunger A.P., Lecampion, B., Zhang, X., Chen, Z.R., Van As, A., Allison, D.P., De Beer, W., Dudley, J.W., Siebrits, E., Thiercelin, M., Mainguy, M. 2009. Measuring Hydraulic Fracture Growth in Naturally Fractured Rock, In: Proc. of the 2009 SPE Annual Tech. Conf. and Exhib ., New Orleans, LA, 4-7 October, SPE 124919. [15] Jing, L. (2003). A Review of Techniques, Advances and Outstanding Issues in Numerical Modelling for Rock Mechanics and Rock Engineering. International Journal of Rock Mechanics & Mining Sciences. 40, 283-353. [16] Lanru, J., Feng, X.T. (2003) Numerical Modeling for Coupled Thermo-Hydro-Mechanical and Chemical Processes (THMC) of Geological Media – International and Chinese Experiences. Chinese Journal of Rock Mechanics and Engineering . 22(10): 1704-1715. [17] Li Ruiqiang (2013-2015) Work in progress toward a PhD thesis, University of Waterloo, Waterloo, Ontario. [18] Pirayehgar A, Dusseault MB 2014. The stress ratio effect on hydraulic fracturing in the presence of natural fractures. Proc. 48th U.S. Rock Mechanics Symp ., Minneapolis, Paper ARMA 14-137, 7 p. [19] Pirayehgar, A. (2013-2015) Work in progress toward a PhD thesis, University of Waterloo, Waterloo, Ontario. [20] Pirayehgar, A., Dusseault, M.B. (2015). Numerical Investigation of Seismic Events Associated with Hydraulic Fracturing. Proc ISRM Congress, Montreal, 15 p. [21] Pirayehgar, A., Yetisir, M. (2015) Work in progress toward PhD and MSc theses, University of Waterloo, Waterloo, Ontario. [22] Sarkarfarshi, A., Malekzadeh, F.A., Gracie, R., Dusseault, M.B. 2014. Parametric sensitivity analysis for CO2 geosequestration. Int. J. of Greenhouse Gas Control , 23 (2014), 61-71. [23] Thallak, S., Rothenburg, L., Dusseault, M.B. 1991. Hydraulic fracture simulation in granular assemblies using the discrete element method. AOSTRA J. of Research, 141-153. [24] Van Der Baan, M., Eaton, D. & Dusseault, M.B. (2013) Microseismic Monitoring Developments in Hydraulic Fracture Stimulation. Proc. Conf. on Effective and Sustainable Hydraulic Fracturing , Dr. Rob Jeffrey (Ed.), ISBN: 978-953-51-1137-5, InTech, DOI: 10.5772/56444. Available from: http://www.intechopen.com/books/effective-and-sustainable-hydraulic-fracturing/microseismicmonitoring-developments-in-hydraulic-fracture-stimulation [25] Warpinski, N.R. (2014) A Review of Hydraulic-Fracture Induced Microseismicity. In Proc. 48 th U.S. Rock Mechanics Symp., American Rock Mechanics Assoc., Minneapolis, 1-4 June 2014.

Acknowledgements Simulation results provided by Reza Jalali, PhD, Atena Pirayehgar, MSc, Mike Yetisir, BSc, Li Ruiqiang, MSc. ITASCA Corp. provided research access to software. Canada’s Natural Sciences and Engineering Research Council awarded a four-year grant to Robert Gracie (Civil Engineering, Waterloo) and me. I have learned a lot about shale gas geosciences form Professor N. Harris of the University of Alberta.

78


Appendix 1: Summary of Geomechanical Properties of Some Shale Oil and Shale Gas Plays, North America


79

Notes:  The assembling of geomechanical data for this table is part of a PhD project by Li Ruiqiang (2015) at the University of Waterloo, Canada



The value ranges are set based on data from various studies or testing results on various specimens within one study. Extreme values are eliminated to ensure a representative range, so that better comparative results can be expected.



Missing of certain types of data for some formations occurs due to the lack of robust research results in the open literature.



Those Young’s modulus and Poisson’s ratio ranges with confining/differential pressure numbers specified are static values, whereas the ranges given without specifications are dynamic values.



Comprehensive research on the geomechanics of shale gas/oil formations is still sparse at this stage; many geomechanical properties are not included in this table but are considered important for comparisons. Further research and literature review are needed for a more sophisticated table.

References for Appendix 1: [1] Sone, H., Mechanical Properties of Shale Gas Reservoir Rocks and its Relation to the In-situ Stress Variation Observed in Shale Gas Reservoirs, PhD Thesis, Stanford University, 2012 [2] Chatellier, J. et al ., Multidisciplinary Integration and Tools to Better Address the Utica Shale Stratigraphy, Rock Properties and Fraccability, geoConvention 2013: Integration, 2013 [3] Molgat, M., Shale Gas in the Quebec Lowlands; High Potential in a New Frontiers Area, PowerPoint Presentation from Talisman Energy [4] Heidbach, O., Tinguay, M., Barth, A., Reinecker, J., Kurfeß, D., and Müller, B., The World Stress Map database release 2008 doi:10.1594/GFZ.WSM.Rel2008, 2008 [5] Call, T., Geomechanical Properties of Marcellus Shale Core Samples with a Sequence Stratigraphic Framework, Master’s Thesis, The Pennsylvania State University, 2012 [6] Izadi, G., Junca, J.P., Cade, R., and Rowan, T., Multidisciplinary Study of Hydraulic Fracturing in the Marcellus Shale, ARMA 14-6975, 2014 [7] Chou, Q. and Gao, J., Analysis of Geomechanical Data for Horn River Basin Gas Shales, NE British Columbia, Canada, SPE 142498, 2011 [8] Yang, S., Harris, N., Dong, T., Wu, W., and Chen, Z., Mechanical Properties and Natural Fractures in a Horn River Shale Core from Well Logs and Hardness Measurements, SPE 174287, 2015 [9] Bellman, L.M. and Leslie-Panek, J., Reading Between the Lines: A NEBC Shale Gas Quantitative Interpretation Case Study, GeoConvention 2014, 2014 [10] National Energy Board, Energy Briefing Note: A Primer for Understanding Canadian Shale Gas, 2009 [11] Vermylen, J.P., Geomechanical Studies of the Barnett Shale, Texas, USA, PhD Thesis, Stanford University, 2011 [12] Davey, H., Geomechanical Characterization of the Montney Shale, Northwest Alberta and Northeast British Columbia, Canada, Master’s Thesis, The Colorado School of Mines, 2013 [13] Burnie Sr., S.W., Montney and Doig Rock Mechanical Properties Study of the Pouce Coupe Area, T.76 – T.80, R.11 – R.13 W6M, Canadian Discovery Ltd., 2008 [14] Rivard, C., Lavoie, D., Lefebvre, R., Sejourne, S., Lamontagne, C., and Duchesne, M., An Overview of Canadian Shale Gas Production and Environmental Concerns, International Journal of Coal Geology 126(2014) 64 – 76, 2014 [15] Wang, C. and Zeng, Z., Overview of Geomechanical Properties of Bakken Formation in Williston Basin, North Dakota, ARMA 11-199, 2011 [16] Narr, W. and Burruss, R. C., Origin of reservoir fractures in Little Knife Field, North Dakota, The American Association of Petroleum Geologists Bulletin, Vol. 68 (9): 1087-1100. 1984 [17] Havens, J., Mechanical Properties of The Bakken Formation, Master’s Thesis, The Colorado School of Mines, 2012 [18] Jin, H. and Sonnenberg, S.A., Source Rock Potential of the Bakken Shales in the Williston Basin, North Dakota and Montana, AAPG Search and Discovery Article #20156, 2012 [19] McLennan, J.D., Roegiers, J.C., Marcinew, R.P., and Erickson, D.J., Rock Mechanics Evaluation of the Cardium Formation, Petroleum Society of CIM, Paper No. 83-34-38, 1983 [20] Creaney, S., Allan, J., Cole, K.S., Fowler, M.G., Brooks, P.W., Osadetz, K.G., Macqueen, R.W., Snowdon, L.R., and Riediger, C.L., Geological Atlas of the Western Canada Sedimentary Basin: Chapter 31 – Petroleum Generation and Migration in the Western Canada Sedimentary Basin, 2012 [21] Soltanzadeh, M., Fox, A., and Rahim, N., Application of an Integrated Approach for the Characterization of Mechanical Rock Properties in the Duvernay Formation, GeoConvention 2015 [22] Low, W.S., Duvernay Shale: The New Millennium Gold Rush, BMO Capital Markets, 2012

Geomechanics in Shale Gas Development

Recommend Documents