SPE-182747-MS Enhancing Oil Recovery - Advanced ... - OnePetro

SPE-182747-MS Enhancing Oil Recovery - Advanced Simulations for More Accurate Frac-Stages Placement Bilal Saad, Ardiansyah Negara, Maaruf Hussain, and Mokhtar Elgassier, Baker Hughes; Shuyu Sun, and King Abdullah, University of Science and Technology

Copyright 2016, Society of Petroleum Engineers This paper was prepared for presentation at the SPE Kingdom of Saudi Arabia Annual Technical Symposium and Exhibition held in Dammam, Saudi Arabia, 25–28 April 2016. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract Hydraulic fracture stimulation designs are typically made of multiple stages placed along the lateral section of the well using various well completion technologies. Understanding how multiple hydraulic fractures propagate and interact with each other is essential for an effective stimulation design. The number and placement of stages are important factors for optimizing the performance of the laterals. This in turn depends on accuracy in determining fracture interference. We present advanced simulations for accurate placement of well stages. In this paper, we use a 3-D fully coupled geomechanical-fluid flow simulator which incorporates anisotropic geomechanical properties. Densely complex natural fractures and lamination are built into the model based on available core and log information. Multiple fractures are concurrently imployed to simulate real life scenarios. Fluid pressures are incrementally computed such that stress state changes dynamically with time as it happens in real field situation. Our simulations were run on Cray XC 40 HPC system. The results demonstrate that the stress shadow effects can significantly alter hydraulic fracture propagation behavior, which eventually affects the final fracture geometry. The results show that there are large differences in aperture throughout the stimulation which persists to the end of pumping. Furthermore comparison between cases with and without complex natural fractures (discrete fracture network (DFN)) and lamination was conducted with even and uneven spacing configurations. Fracture interference and spacing analysis conducted based on model with perforation frictions shows that while spacing between fractures is important, the largest impact was observed in the presence of lamination and DFN. The large differences in the way the fracture propagates highly depend on the DFN connectivity. Late-stage connection throughout the model implies later disconnection when the pressure drops. Though the computations are time intensive, we believe this is a valuable tool to use in the planning stages for asset development to increase production potential.

Introduction Hydraulic fracturing is a well-stimulation technique. Pumping fluid into a reservoir at high pressure creates artificial fractures that significantly improve reservoir’s permeability and enhance hydrocarbon recovery. Proppants mixed in the fluid keep the induced fractures open after the pumping is stopped.

2

SPE-182747-MS

Fracture generally propagates few feet to a few hundred feet from the wellbore. It provides a conductive pathway to flow to the wellbore. Hydraulic fracturing stimulation designs are typically made of multiple stages placed along the lateral section of the well using well completion technologies. Understanding how multiple hydraulic fractures propagate, interact with each other and with natural fractures is essential for an effective stimulation design. Furthermore, shale reservoirs geologic systems are highly complex and heterogeneous. Different patterns require different degree of complexity. For example, joints are often orthogonal to bedding and to each other and perhaps be modeled using ⬙gridded⬙ discrete fracture networks (DFN). Laminations and inclined natural fractures are also common that would require more complex models. Especially, where fracture systems are less well organized. Analysis of the performance of hydraulic fracturing stimulation jobs and optimization of the treatment design requires modeling that accounts for all important features of the process. This ideally covers both the treatment and post-stimulation production of the well. Incorporating the complex interactions between fluid, rock matrix, and rock interfaces, as well as the interactions between propagating fractures and existing natural interfaces remains a difficult numerical modeling task. Several hydraulic fracturing simulators are available to simulate hydraulic fractures propagation numerically in these tight reservoirs; however, none of them has been able to accurately predict how these reservoirs are stimulated based on fracturing job design and rock-mechanics parameters. A fully coupled 3-D geomechanical modeling technique that takes into account all the complexities would be required to have better understanding of these complexities. Such technique possesses the ability to model both the deformation of the rock mass (allowing fractures to accept fracturing fluids and proppants), and the fracture propagation process itself. Rock mass deformation is a combination of fracture deformation and the deformation of the rock matrix to accommodate fracturing fluids penetrating natural and induced fractures. Many publications have discussed the hydraulic fracturing model developments and their applications for example to optimize the job design and placement, to improve post-treatment diagnostics, to estimate fracture dimensions, and to validate the laboratory scale measurement, etc (Bai et al. 2006, Ji et al. 2006, Denney 2008, Damjanac et al. 2010, Jacot et al. 2010, Meyer et al. 2011, Rafiee et al.2012). An accurate hydraulic fracture simulation highly depends on the model complexity and heterogeneity for example natural fractures and lamination. Defining natural fractures in the model is critical towards understanding hydraulic fracture propagation in any given reservoir. Dershowitz and Doe (2011) summarizes three significant impacts of natural fractures in developing unconventional shale reservoirs: first, natural fractures are planes of weakness that may control hydraulic fracture propagation; second, high pressures from the hydraulic fracturing treatment may cause slip on natural fractures that increases their conductivity; and third, natural fractures that were conductive prior to stimulation may affect the shape and extent of a wells’ drainage volume. A complete description about fractures in the reservoir involves characterization of their occurrence, size, intensity, variation, orientation, and sensitivity with respect to the stress (Edris et al. 2015). Fractures are commonly described in terms of statistical distributions and correlations to reservoir properties (known as discrete fracture network (DFN)) because their individual description is not easily achieved. According to King (1983) and Davy (1993), the spatial distributions of the fractures and faults usually follow the fractal or multi-fractal scaling laws. Integration of data from image logs, CT scans of core, wireline logs, reservoir model, faults, production logging tools, well test data, and permeability tests is required to have a complete description about natural fractures. From these data, interpretation of image logs data is the most important because it is used to determine the fracture locations and orientations across the reservoirs. Fracture orientations are very important to understand the alignments of the fractures because they control the connectivity and anisotropy of the fractures’ network away from the wells. According to Edris et al. (2015), a conceptual model for DFN should accommodate three main things: (i) the location where the fractures occur, (ii) factors controlling the fracture orientation, and (iii) evidence that fractures will affect the production. The DFN model utilizes parameterization of fracture orientations and fracture intensity relative to 3-D model for lithology and geomechanical

SPE-182747-MS

3

properties. Variations of fracture orientation and intensity across the field can be determined by the fracture characterization and associated to the 3-D geological and geomechanical models (Masri et al. 2015). The fracture locations, orientations, and sizes of individual fractures are usually sampled by a Monte Carlo approach. In this paper, we perform 3-D hydraulic fracturing simulation that takes into account both the geomechanics changes and the fluid pressure changes. We use GEOS, which is a 3-D multiphysics hydraulic fracturing simulator code developed by Lawrence Livermore National Laboratory. We consider various cases of single and multiple clusters with different stress profiles. Complex natural fractures (DFN) and lamination are built into the model. Multiple fractures are concurrently simulated for real field scenarios. Critical stress intensity factor was used as a fracture criterion governing the generation of new fractures or propagation of exisiting fractures and their directions. Our simulations were run on a Cray XC 40 HPC system. The results demonstrate that the impact of stress shadow should be considered when designing hydraulic fracturing stimulation treatment as it affects the final fracture geometry significantly. Large variations in aperture throughout the stimulation were observed. Models were built based on even and uneven spaing configuration and with and without DFN and lamination. Fracture interference and spacing analysis conducted shows that while spacing between fractures is important, the largest impact was observed in the presence of lamination and DFN. The DFN connectivity influences the way the fracture propagates.

Modeling Methodology GEOS includes the following four modules in the numerical model (Fu et al. 2011, 2013): i. Geomechanics – it contains a solid solver that provides non-local mechanical responses of the rock matrix; ii. Hydrodynamics – it contains a flow (fluid) solver that solves the fluid flow in interconnected fracture networks; iii. Geomechanical joint model – to capture the hysteretic behavior of the interfacial tensions and the permeability changes; and iv. Adaptive remeshing module – to generate new meshes for the solid and fluid solvers as the fractures propagate. These four modules are required to model a real hydraulic fracturing process that involves complex interactions between multiple fractures, which include the new fractures and their propagation in the rock matrix, the exisiting isolated fractures that could be connected with the new fractures, and the change of apertures of exisiting fractures due to the change of stresses. Figure 1 depicts the coupling of four modules implemented in GEOS.

4

SPE-182747-MS

Figure 1—Modules used in GEOS hydraulic fracturing simulator and their coupling (redrawn from Fu et al. 2013).

The solid solver uses the linear-strain triangle and assuming linear elasticity and small deformations occur for the intact rock response. This solver implemented a central-difference explicit time integration scheme. The fluid solver assumes laminar flow between two parallel plates to represent the fluid flow in open rock fractures. The flow solver implements an explicit integration scheme. The governing equations used in GEOS for the fluid flow are described as follows: (1) (2) (3) where q is the flow rate in the fracture at a given cross-section, w is the aperture of the fracture, x is the length along the fracture, t is time, k is the permeability of the fracture, P is the fluid pressure inside the fracture, and ␮ is the dynamic viscosity of the fluid. Equations (1)-(3) are solved with a 2-D finite volume method formulated based on a three-dimensional approach (Johnson and Morris 2009). In this study, we consider the discrete fracture network (DFN) and lamination in our numerical models. Defining DFN model is critical towards understanding hydraulic fracture propagation in any given reservoir. For this study, we are constrained to very limited data so we generated a structured DFN realization consists of deterministic wellbore image-based model to develop a simple stochastic model of fractures between and around wellbores based on the conceptual models for fracture orientation, spatial distribution, and the intensity-size scaling. Laminations are horizontal weak planes in the reservoirs and were incorporated into the models as horizontal natural fractures. This study also takes into account the wellbore hydraulics in the model in order to investigate the hydraulic fracture propagation behavior in the presence of perforation and wellbore friction. The perforation friction is given by (Izadi et al. 2015) (4)

(5) where Qtotal is the total flow rate, Po is the reference pressure at bottomhole, Pc,i is the closure pressure

SPE-182747-MS

5

of each frac, ⌬Pn,i is the net pressure of each frac, and ⌬Ppf,i(Qi) is the perforation friction of each frac, and ⌬Pcf,i(Qi) is the casing friction. The perforation friction, ⌬Ppf,i(Qi), is calculated by (6) Where ␳ is the fluid density, Qi is the total flow rate, N is the number of perforations, d is the perforation diameter, and Cd is the discharge coefficient, which captures perforation tunnel effects.

Model Setup This study uses the stresses measured by overcoring and micro-fracking in the Western Canada Sedimentary Basin (Bell et al. 1994). The target shale zone in this area ranges from 2096 m to 2152 m deep. Greater in-situ horizontal stress magnitudes are observed in overlain and underlain intervals compared to the target zone. Therefore, they are likely to act as barriers to contain the vertical fracture propagation during the stimulation process. Figure 2 shows the vertical distribution of the three in situ principal stresses components, namely the minimum horizontal stress (Shmin), the maximum horizontal stress (SHmax),and the vertical stress (Sv), which are used in our models. We considered two different faulting stress regimes in our simulation cases, i.e., normal faulting stress regime (Shmin ⬍ SHmax ⬍ Sv) as depicted in Figure 2(a) and its modification into strike-slip faulting stress regime (Shmin ⬍ Sv ⬍ SHmax) as shown in Figure 2(b) in the target zone. In the second faulting stress regime, there is a change of stress profile, from strike-slip to reverse fault stress regime, which is located at the depth of 2128 m. It is expected that the hydraulic fracture propagation is strongly impacted by the stress profile. In our model, the perforation location is depicted by the black line in Figure 2, which represents the wellbore.

Figure 2—Principal stress profiles used in the model: (a) normal and (b) strike-slip stress profiles in the target zone.

6

SPE-182747-MS

The following is a series of simulation scenarios to establish the hydraulic fracture propagation in the reservoir, fracture interference, and optimum spacing with respect to the stress data complexity.. In order to generate a realistic model and provide a final solution to an optimum spacing, a fully coupled DFN and lamination model are incorporated into the model. The computational domain has the size of 560m in the -direction, 800m in the -direction, and 240m in the -direction. Figure 3 illustrates the finite element mesh that is used in the model. The meshes are finer in the near-field around the fractures with 4 meter mesh resolution and coarsermesh elements are used in the far-field region in order to reduce the computational cost.

Figure 3—Field scale model geometry with high mesh density region in the middle (4 meter resolution).

Base case model development A base case model using the two different faulting stress regimes mentioned above was developed to understand the hydraulic fracture propagation in single cluster. The stress profiles and rock properties are assumed changing with depth. The perforation depth is 2116m and the fluid pumping rate is 20 barrel per minute (bpm). The model was run for 3600 s, which represents one hour of hydraulic fracturing job in the field. Our simulations were run on Cray XC 40 HPC System. Interference and spacing We investigated fracture interference due to the induced stress associated with multiple hydraulic fracturing stimulations. The induced stress, also known as a stress shadow, can easily alter local stress condition that can significantly hamper the fracture development and the effectiveness of stimulation design. Stress shadow can cause hydraulic fracture to rotate, coalescence, and terminate. This will eventually lead to an early screen-out. Hence, the stress shadow impact must be taken into account for any optimum fracturing stimulation design. To investigate the impact of fracture interference and spacing, we considered different multiple clusters scenario spacing. In the first scenario, DFN and lamination were ignored and were incorporated in the second scenario. We considered multi-cluster stimulation for fracture interference using three models: 1. Even spacing between clusters without DFN and lamination using the two stresses profiles; 2. Uneven spacing between clusters without DFN and lamination using the two stresses profiles; and 3. Uneven spacing between clusters with DFN and lamination using the two stresses profiles.

SPE-182747-MS

7

DFN and lamination From the base case model, we gradually increase the complexity of the model by incorporating the DFN and lamination. Complex DFN were defined wherein the natural fractures are perpendicular to each other. The DFN model utilizes parameterization of fracture orientations and fracture intensity relative to 3-D model for lithology and geomechanical properties. Variations of fracture orientation and intensity across the field can be determined by the fracture characterization and associated to the 3-D geological and geomechanical models. DFN are defined such that they have 4 m length connecting to each other orthogonally and cut across the laminations at different zones above and below the target zone. Perforation friction Perforation friction were included in the model to allow for effective stress shadow analysis. Below are the parameters used in the model: ● ● ● ●

Fluid density, ␳ ⫽ 8.347 ppg (⬇ 1000 kg/m3) Number of perforations, N ⫽ 10 Perforation diameter, d ⫽ 0.37 inch (⬇ 0.0094 m) Discharge coefficient, Cd ⫽ 0.85

Results and Discussions Hydraulic fracturing propagation in single cluster In the first part of this section we present the simulation results focusing on the impact of stresses profiles on the hydraulic fracture propagation in single cluster with and without presence of DFN and lamination. The fluid pumping rate is 20 bpm and fluid viscosity is 40 centipoise (cp). The simulation time is 3600 s, which represents one hour of hydraulic fracturing job in the field. Figure 4 depicts the oblique view of hydraulic fracture propagation at the end of simulation for two different stress profiles. The hydraulic fracture in the normal faulting stress regime mainly propagates vertically. Whereas, in the strike-slip faulting regime, a T-shaped fracture developed as thefractures propagate horizontally in addition to the vertical fracture propagation. Figure 5 displays the fracture length and height where in this case the fracture in the normal faulting stress regime is longer and has larger height compared the fracture in the strike-slip faulting regime. The aperture for the fracture in the normal faulting stress regime is also wider than the fracture in the strike-slip faulting stress regime (Figure 4). We also plot the minimum horizontal stress profiles at the perforation at different time as shown in Figure 6, where it can be seen that the minimum horizontal stress in the normal faulting stress regime is 0.5 MPa higher than the one in the strike-slip faulting regime.

Figure 4 —Oblique view of fracture propagation at the end 3600 s simulation time using normal and strike-slip stress profiles.

8

SPE-182747-MS

Figure 5—Side view of fracture propagation at the end of 3600 s simulation time using normal and strike-slip stress profiles.

Figure 6 —Minimum horizontal stress values at the perforation for normal and strike-slip stress profiles.

Next, we increase the complexity of the model by incorporating the natural fracture (DFN) and lamination. The aim is to investigate the impact of natural fractures and lamination on the hydraulic fracture propagation. The outputs illustrated in Figure 7 describe the fractures propagation in the two stress regimes where it is clearly shown that the fractures length is greatly impacted by the DFN and lamination. The fracture length has reduced to about 480 m from 550 m using the normal stress regime and to about 340 m from 500 m using the strike-slip stress regime. Also, the hydraulic fracture propagation in the presence of DFN and lamination is substantially different from the fracture propagation without these complexities especially in the case of strike-slip stress regime (Figure 7 and Figure 8). Thus, any hydraulic fracture design based on the assumption that the rock is homogenous and the fracture propagates symmetrically in a plane perpendicular to the minimum stress would generate erroneous results. Due to natural fractures and lamination interactions with hydraulic fracture, propagation can be asymmetrical or in multiple directions or segmented as shown by the output in this study.

SPE-182747-MS

9

Figure 7—Side view of fracture propagation at the end of 3600 s simulation time with DFN and laminated incorporated into the model for normal and strike-slip stress profiles.

Figure 8 —Oblique view of fractures propagation at the end of 3600 s simulation time with DFN and lamination incorporated into the model for normal and strike-slip stress profiles.

Figure 9 —Minimum horizontal stress values at the perforation for normal and strike-slip stress profiles.

Hydraulic fracturing propagation in multi-cluster The second part of this section presents two cases of hydraulic fracturing simulation in multi-cluster. In the first case, we investigate the impact of stress shadow and spacing between the clusters on the hydraulic

10

SPE-182747-MS

fracture propagation. In the second case, we study the impact of fluid injection rate on hydraulic fracture propagation in multi-cluster. The normal faulting stress regime is used in the first case while we use the strike-slip faulting stress regime in the second case. Case 1 – Impact of stress shadow and spacing on hydraulic fracture propagation In the first case we considered two scenarios of hydraulic fracturing simulations in multi-clusters with five clusters considered in the model. In the first scenario, uneven spacing (15/25/25/15 m) between the clusters is considered as depicted in Figure 10 while the second scenario has even spacing (15/15/15/15 m) between the clusters. The normal faulting stress regime is used in the model. Fluid pumping rate and fluid viscosity for both scenarios are 50 bpm and 40 cp, respectively. We ran the model for 3600 s simulation time. Figure 11 displays the oblique view of fracture propagation at the end of 3600 s simulation time with even and uneven spacing between clusters. Figure 12 shows the side views of all five fractures (fractures 1, 2, 3, 4, and 5, from left to right). From these figures we observed that the fractures growth in the first scenario is greater than the second scenario. This is because the spacing between the fractures in the first scenario, especially the interior spacing, is greater than the spacing in the second scenario such that the impact of stress shadow is less. The two exterior fractures (fractures 1 and 5) and the middle fracture (fracture 3) propagate more than the interior fractures (fractures 2 and 4) with about 30 m difference in the first scenario. Almost similar to the second scenario where the exterior fractures propagate more than the interior fractures, about 50 m difference, and in this scenario the middle fracture grows together with the other interior fractures (fractures 2 and 4) because of the same distance spacing. It is also observed that the multi-cluster fracture apertures are smaller than the fracture in single cluster because of the stress distribution to the five fractures. The first scenario yields less height and wider aperture compared to the second scenario. Table 1 summarizes the fracture geometry for all of the five fractures in the first and second scenarios.

Figure 10 —Clusters configuration for the uneven spacing.

SPE-182747-MS

11

Figure 11—Oblique view of fracture propagation at the end 3600 s simulation time with uneven and even spacing between clusters.

Figure 12—Side view of each fracture in multi-cluster with uneven and even spacing between clusters.

12

SPE-182747-MS

Table 1—Fracture geometries for the two scenarios. Scenario 1 (uneven spacing) Fracture 1 2 3 4 5

Scenario 2 (even spacing)

Length (m)

Height (m)

Aperture (mm)

Length (m)

Height (m)

Aperture (mm)

400 370 390 370 400

77 77 77 77 77

3.6 3.6 3.6 3.6 3.6

410 350 350 350 410

80 80 80 80 80

3.4 3.4 3.4 3.4 3.4

Figure 13 illustrates the flow rate at the perforation points for the first and second scenarios. The plots in this figure show that fractures 1 and 5 have the same flowrate and both are the highest flowrate than the interior fractures. The flowrate in fracture 3 (middle fracture) has the lowest flowrate. Similar behavior is observed in the second scenario where the spacing between the fractures is uniform. We also plot the apertures against time for both scenarios and we see similar behavior in which fractures 1 and 5 have the widest aperture compared to the interior fractures because the exterior fractures have less impact of stress shadow.

Figure 13—Flow rate at perforation versus time with uneven and even spacing between clusters.

Figure 14 —Aperture at the perforation versus time with uneven and even spacing between clusters.

SPE-182747-MS

13

We conducted induced stress analysis to understand its impact on the five fractures for the two scenarios. First, we compare the minimum horizontal stress, Shmin, at the perforation between the multi-cluster with uneven spacing (first scenario) and single cluster (base case). Then, we compare the minimum horizontal stress between multi-cluster with even spacing (second scenario) and single cluster. From the two comparisons, we observe that the minimum horizontal stresses in the multi-cluster cases are higher than the one in single cluster, as depicted in Figure 15 and Figure 16, implying that the stress distribution in multi-cluster is higher than the single cluster. Therefore, a properly planned spacing between perforations clusters would be necessary to avoid impaired fracture development.

Figure 15—Comparison of minimum horizontal stress between multi-cluster with uneven spacing and single cluster.

Figure 16 —Comparison of minimum horizontal stress between multi-cluster with even spacing and single cluster.

Case 2 – Impact of injection rate on hydraulic fracture propagation In the second case we investigate the impact of fluid injection rate on the hydraulic fracture propagation in multi-cluster with uneven spacing. We consider two scenarios with two different injection rates where the first and second scenarios have injection rate of 50 bpm and 40 bpm, respectively. Similar to case 1,

14

SPE-182747-MS

there are five clusters in the model, the fluid viscosity is 40 cp, and the simulation time is 3600 s. The stress profile used in this second case is strike-slip faulting stress regime. The spacing between the clusters is 40 m, 45 m, 45 m, and 40 m as illustrated in Figure 17. Figure 18 shows the fractures propagationafter 3600 s where we can see that the fractures do not only propagate vertically but also horizontally. With the injection rate of 50 bpm, we observe the horizontal fractures on the top of five fractures interact each other. On the contrary, the horizontal fractures do not interact each other in the second scenario due to lower injection rate. Figure 19 summarizes the fracture geometry of the five fractures for both scenarios. As can be seen in this figure, the lengths of the fractures in the first scenario are longer than the second scenario because of higher injection rate. Similarly, the fractures apertures in the first scenario are larger than the fractures apertures in the second scenario. We also plotted the minimum horizontal stresses at the perforation at different times and the plots reveal that there is no large difference of minimum horizontal stresses between the two scenarios.

Figure 17—Clusters configuration with uneven spacing.

Figure 18 —Oblique view of hydraulic fractures propagation with 50 bpm and 40 bpm pumping rates.

SPE-182747-MS

15

Figure 19 —Side view of each fracture in multi-cluster with 50 bpm and 40 bpm pumping rates.

Case 3 – Impact of natural fractures (DFN) and lamination on hydraulic fracturing propagation in multi-clusters We have presented the hydraulic fracture propagation in multi-cluster without the presence of natural fractures (DFN) and lamination. Next, we increase the complexity of the model by introducing the DFN and lamination into the model. We still consider five clusters in the model with uneven spacing between the fractures. Two scenarios are considered with the first scenario uses the normal faulting stress regime and the second scenario uses the strike-slip faulting regime. The spacing configuration in the first scenario is 20/25/25/20 m. The injection rate is 50 bpm and fluid viscosity is 40 cp. Similar to the previous cases, the model was run for 3600 s simulation time. Figure 21 depicts the hydraulic fractures propagation in normal stress regime at the end of the simulation. The magenta color represents the DFN and lamination. Figure 22 illustrates side view of each fracture in which it is obviously seen that the exterior fractures (fractures 1 and 5) grow about 60 m longer than the interior fractures (fractures 2, 3, and 4). We may compare the results with the case without DFN and lamination (Case 1) in order to observe the impact of DFN and lamination on the hydraulic fracture propagation. The first scenario in this section is still comparable with the first scenario in Case 1 above although the spacing between the fractures is not exactly same. The results demonstrate that with larger spacing the lengths of the outer fractures of the first scenario in Case 1 and the first scenario in this section are similar. This indicates that the DFN and lamination retard the growth of the fractures. This is further confirmed by the propagation of the interior fractures with the presence of DFN and lamination is slower, about 25 m to 90 m, than the propagation of the interior fractures without the presence of DFN lamination (Case 1). The aperture, however, is larger than the Case 1 because Case 3 has large spacing so the effect of stress shadow is less (Figure 23). Figure

16

SPE-182747-MS

24 shows the minimum horizontal stress value at the perforation, which is slightly higher than the minimum horizontal stress in Case 1 (Figure 15, left figure).

Figure 20 —Minimum horizontal stress profiles for 50 bpm and 40 bpm pumping rates.

Figure 21—Oblique view of hydraulic fractures propagation in normal stress regime with DFN and lamination incorporated into the model.

SPE-182747-MS

Figure 22—Side view of each fracture in normal stress regime with DFN and lamination incorporated into the model.

Figure 23—Aperture at the perforation versus time with DFN and lamination incorporated into the model.

17

18

SPE-182747-MS

Figure 24 —Minimum horizontal stress values at the perforation for normal stress regime.

Next, we consider the second scenario in which we use the strike-slip stress regime. The injection rate and fluid viscosity are 40 bpm and 40 cp, respectively. The spacing configuration is 40/45/45/40 m. The second scenario in this section is similar with the second scenario in Case 2 except the presence of DFN and lamination. Figure 25 presents the hydraulic fractures propagation at the end of the simulation with DFN and three layers of lamination exist in the reservoir (represented by the magenta color). The fractures propagation is significantly different from the output of the second scenario in Case 2. The lengths of the exterior fractures (Figure 26) are shorter than the exterior fractures of Case 2 (Figure 19). The difference is even more apparent on the interior fractures where the middle fracture is about 100 m shorter than the middle fracture of case without DFN and lamination as depicted in Figure 19. Similarly, the other two interior fractures are about 80 m shorter. This emphasize a significant influence of DFN and lamination on the fracture propagation. Figure 27 illustrates the aperture growth with time during the simulation. We also plot the minimum horizontal stress at the perforation in Figure 28 and if we compare with the minimum horizontal stress in the case of without DFN and lamination, we do not observe a significant difference.

Figure 25—Oblique view of hydraulic fractures propagation in strike-slip stress regime with DFN and lamination incorporated into the model.

SPE-182747-MS

Figure 26 —Side view of each fracture in strike-slip stress profile with DFN and lamination incorporated into the model.

Figure 27—Aperture at the perforation versus time with DFN and lamination incorporated into the model.

19

20

SPE-182747-MS

Figure 28 —Minimum horizontal stress values at the perforation for normal stress regime.

Conclusions In this study, we have performed the fully 3-D hydraulic fracturing simulations that take into account the geomechanics changes and the fluid pressure changes. We use GEOS simulator, which contains geomechanics, hydrodynamics, geomechanical joint model, and adaptive remeshing modules. We considered various cases of single and multiple clusters with different stress profiles. Complex natural fractures (DFN) and lamination were built into the model based. Multiple fractures were concurrently simulated for real field scenarios. Our simulations were ran on Cray XC 40 HPC System. The results demonstrate that the stress shadow effects can significantly alter hydraulic fracture propagation behavior, which eventually affects the final fracture geometry. The results show that there are large differences in aperture throughout the stimulation which persists until the end of pumping. The development of T-shaped horizontal fractures due to high stress contrast between minimum horizontal and overburden stresses were observed. Furthermore comparison between cases with and without DFN and lamination was conducted with even and uneven spacing configurations. Fracture interference and spacing analysis shows that while spacing between fractures is important, the largest impact was observed in the presence of lamination and DFN. The large differences in the way the fracture propagates highly depend on the DFN connectivity. Late-stage connection throughout the model implies later disconnection when the pressure drops. Though the computations are time intensive, we believe this is a valuable tool to use in the planning stages for asset development to increase production potential.

Acknowledgements The authors are grateful to Baker Hughes for the permission to publish this paper and to King Abdullah University of Science and Technology (KAUST) for providing access to Shaheen Cray XC40 HPC system.

References Bai, M., Green, S., Casas, L. et al. 2006. 3-D Simulation of Large Scale Hydraulic Fracturing Tests. Presented at the Golden Rocks 2006, The 41st U.S. Symposium on Rock Mechanics (USRMS) held in Golden, Colorado, USA, 17-21 June. Bell, J.S., Price, P.R. and McLellan, P.J. 1994. In-situ Stress in the Western Canada Sedimentary; in Geological Atlas of the Western Canada Sedimentary Basin, G.D. Mossop and I. Shetsen (comp). Canadian Society of Petroleum Geologists and Alberta Research Council, URL http://www.ags.aer.ca/reports/atlas-of-the-western-canada-sedimentary-basin.html.

SPE-182747-MS

21

Damjanac, B., Gil, I., Pierce, M. et al. 2010. A New Approach to Hydraulic Fracturing Modeling in Naturally Fractured Reservoirs. Presented at the 44th U.S. Rock Mechanics Symposium and 5th U.S.-Canada Rock Mechanics Symposium held in Salt Lake City, Utah, USA, 27-30 June. Davy, P. 1993. On the Frequency-Length Distribution of the San Andreas Fault System. Journal of Geophysical Research: Solid Earth 98: 12141–12151. Denney, D. 2008. Hydraulic Fracturing: Modeling Hydraulic-Fracture Treatments in San Juan Basin Coals: Fully Functional 3D Fracture Model. Journal of Petroleum Technology 60: 62–66. Dershowitz, W. and Doe, T.W. 2011. Modeling Complexities of Natural Fracturing Key in Gas Shales. The American Oil and Gas Reporter. Edris, M.A., Haggag Amin, M., Al Benali, K., et al. 2015. Integrated Characterization and Modeling of Faults and Fractures: Their Impact on Reservoir Performance with Changing In-situ Stresses, Abu Dhabi. Presented at the SPE Reservoir Characterisation and Simulation Conference and Exhibition held in Abu Dhabi, UAE, 14-16 September. Elsaid, M.E., Al-Jafri, G.M., Mercado, G. et al. 2007. Detecting and Analysing Fractures in the Subsurface: The Critical Steps in Modeling Fractured Carbonate Reservoir. Presented at the SPE Middle East Oil and Gas Show and Conference held in Manama, Bahrain, 11-14 March. Izadi, G., Moos, D., Settgast, R. et al. 2015. Fully 3D Hydraulic Fracture Growth within Multi-stage Horizontal Wells. Presented at the International Congress on Rock Mechanics held in Montreal, Quebec, Canada. Ji, L., Settari, A., and Sullivan, R. B. 2006. A New Approach to Hydraulic Fracturing Modeling – Fully Coupled with Geomechanical and Reservoir Simulation. Presented at the SPE Europec/EAGE Annual Conference and Exhibition held in Vienna, Austria, 12-15 June. Fu, P., Johnson, S.M., and Carrigan, C.R. 2011. Simulating Complex Fracture Systems in Geothermal Reservoirs Using an Explicitly Coupled Hydro-Geomechanical Model. Presented at the 45th US Rock Mechanics/Geomechanics Symposium held in San Fransisco, CA, 26-29 June. Fu, P., Johnson, S.M., and Carrigan, C.R. 2013. An Explicitly Coupled Hydro-Geomechanical Model for Simulating Hydraulic Fracturing in Arbitrary Discrete Fracture Networks. International Journal for Numerical and Analytical Methods in Geomechanics 37: 2278 –2300. International Energy Agency. 2011. Are We Entering a Golden Age of Gas? World Energy Output. Special Report. Jacot, R.H., Bazan, L.W., and Meyer, B.R. 2010. Technology Integration – A Methodology to Enhance Production and Maximize Economics in Horizontal Marcellus Shale Wells. Presented at the SPE Annual Technical Conference and Exhibition held in Florence, Italy, 19-22 September. Johnson, S.M. and Morris, J.P. 2009. Hydraulic Fracturing Mechanisms in Carbon Sequestration Applications. Presented at the 43rd US Rock Mechanics Symposium and the 4th US-Canada Rock Mechanics Symposium, Asheville, NC, 28 June-1 July. King, G. 1983. The Accommodation of Large Strains in the Upper Lithosphere of the Earth and Other Solids by Self-Similar Fault Systems: the Geometrical Origin of b-Value. Pure and Applied Geophysics 121: 761–815. Masri, A., Barton, C., Hartley, L. et al. 2015. Structural Permeability Assessment Using Geological Structural Model Integrated with 3D Geomechanical Study and Discrete Fracture Network Model in Wayang Windu Geothermal Field, West Java, Indonesia. Proceedings at the Fourtieth Workshop on Geothermal Reservoir Engineering held in Stanford University, Stanford, California, 26-28 January. Meyer, B.R. and Bazan, L.W. 2011. A Discrete Fracture Network Model for Hydraulically Induced Fractures: Theory, Parametric, and Case Studies. Presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition held in The Woodlands, Texas, USA, 24-26 January. Rafiee, M., Soliman, M.Y., Pirayesh, E. et al. 2012. Geomechanical Considerations in Hydraulic Fracturing Designs. Presented at the SPE Canadian Unconventional Resources Conference held in Calgary, Alberta, Canada, 30 October – 1 November.