SPE-184847-MS

A Fully-Coupled, Poro-Elasto-Plastic, 3-D Model for Frac-Pack Treatments in

Poorly Consolidated Sands

Dongkeun Lee and Mukul M. Sharma, The University of Texas at Austin

Copyright 2017, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Hydraulic Fracturing Technology Conference and Exhibition held in The Woodlands, Texas, USA, 24-26 January
2017.

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
Frac-pack completions have been used as a technique for well stimulation and sand control. Conventional
hydraulic fracturing models based on linear elastic fracture mechanics often lead to inaccurate predictions of
fracture geometry and fracturing pressure response due to large inelastic deformations and strong fluid-solid
coupling. We present a fully-coupled, three-dimensional hydraulic fracture model in poro-elasto-plastic
materials using a finite volume cohesive zone model for frac-pack applications.
Our model is capable of capturing the high net fracturing pressure commonly observed during frac-
packing operations. The 3-D model simulates both shear and tensile failure around the fracture and
computes the stresses and displacements around the fracture. We observe that plasticity causes lower stress
concentration around the fracture tip which shields the tip of the propagating fracture from the fracturing
pressure. High leak-off can also lead to shear failure around the fracture and ahead of the tip due to pore
pressure diffusion. The low cohesion sands tend to fail in shear first then in tension if sufficient pore
pressure and poroelastic backstress build up. We use a three-dimensional numerical computation of fluid
leak-off using a fracture-reservoir domain coupling in this model to overcome the limitation of linear, one-
dimensional leak-off models. Higher pressure gradients due to lower permeability, higher viscosity and
injection rate results in a faster fracture propagation rate.
LEFM models inherently predict lower net fracturing pressure, smaller fracture widths and longer lengths
in soft formations than observed in the field. In many instances, such models are used to fit the net pressure
data by manipulating input parameters beyond physically reasonable values. The model presented here
provides a much more physically realistic approach to model fracture growth in unconsolidated reservoirs
with lab measured mechanical properties. The model has allowed us to design and analyze hydraulic
fracturing stimulations much more accurately than LEFM models used in the past.

Introduction
Hydraulic fracturing is a widely applied well stimulation technique that has been utilized for more than
60 years in the oil and gas industry (King, 2012). Modeling hydraulic fractures is a complicated process
because it involves several coupled phenomena (Adachi, Siebrits, Peirce, & Desroches, 2007; Carrier &
Granet, 2012; Chen, 2012):
2 SPE-184847-MS

Poro-elastic or poro-elasto-plastic deformation coupled with pore pressure diffusion in porous

reservoirs
Preferential injected fluid flow in the fracture

Proppant transport coupled with fluid flow in the fracture

Tightly coupling between the fracture pressure and the reservoir deformation

Fluid leak-off from the fracture into the surrounding porous medium

Propagation of the fracture

Solving these coupled phenomena in a fully three-dimensional model at a reservoir scale is a challenging
problem due to the lack of computational resources (Searles et al., 2016). Furthermore, for the design and
analysis of frac-packs, i.e. hydraulic fracture operations in soft formations, the exact mechanisms of failure
remain an unresolved issue, although a large number of hydraulic fracturing operations have been performed
in soft formations.
There are several experimental studies and theoretical modeling approaches, but a robust field-tested
theory and model to understand the physics of fracturing in soft rocks is not available. Thus, most hydraulic
fracturing projects carried out in poorly consolidated reservoirs are designed using commercial simulators
based on the theory of Linear Elastic Fracture Mechanics (LEFM). LEFM theory and the assumption that
brittle failure occurs with small and linear-elastic deformation are not applicable to soft rocks. Also, it is
important to simulate fluid leak-off accurately. Significant amounts (~ 60 to 90%) of the injected fracturing
fluid can leak off from the fracture into the high permeability porous reservoir during the stimulation
treatment (Adachi et al., 2007).
Users who tried to design and analyze the frac-pack treatment using hard rock models need to calibrate
and often manipulate input parameters such as fracture toughness, Young's modulus, leak-off coefficient,
and minimum horizontal stress. In many instances, such models are used to fit the net pressure data by
manipulating input parameters beyond physically reasonable values. In spite of this procedure, hard rock
models do not explain the very high net fracturing pressure reported in the field and laboratory experiments
(Papanastasiou 1997, van Dam & Papanastasiou 2002). LEFM models consistently predict lower net
fracturing pressure, smaller fracture widths and longer lengths in soft formations than observed in the field.
To address this challenge, we developed a fully-coupled, three-dimensional hydraulic fracture model
using a finite volume cohesive zone model. Details of the model are provided in a separate paper (Lee et
al., 2015). We have extended this work by developing the model further in three-dimensions and applying
it to a field case. We present comparisons of our results with field data and results from an LEFM hard
rock model. Then, we investigate factors that determine the growth of a fracture in soft sands by a series
of simulations and a parametric sensitivity study.

Model Formulation
Two separate domains, one for the fracture and one for the reservoir as shown in Figure 1, are discretized to
model coupled phenomena involved in the hydraulic fracturing process. Separate equations are defined and
solved in the two domains. Displacement and pore pressure are primary unknowns in the reservoir domain.
Fluid flow with proppant transport is solved in the fracture domain. The two domains are coupled through
appropriate boundary conditions. Fluid leak-off from the fracture to the reservoir is numerically calculated
by iterating between the two domains until the continuity of the pressure and fluid leak-off rate is achieved.
We present the formulation of the displacement vector in detail here.
SPE-184847-MS 3

Figure 12-D conceptual image of the two domains: reservoir and fracture.

Displacement Equation in the Poro-Elastic Reservoir

We formulate a mathematical model for deformation of the reservoir by combining the equilibrium equation,
the constitutive equation which relates stress and strain relations, and the kinematic equation (strain-
displacement equations). The equilibrium equation is written as:

(1)

Where S is a total stress tensor and f is the density of body force. In this work, we assume the
gravitational body force is negligible, and a tension-positive sign convention is used. To have a governing
partial differential equation for the displacement vector, we need to define the stress-strain law and the
strain-displacement equations. For the stress-strain relation, we adopt the constitutive equation of linearized
poroelasticity (Jaeger, Cook, & Zimmerman, 2007).
(2)
(3)

(4)

(5)

(6)

By combining the constitutive equation and the small strain-displacement relation with the equilibrium
equation, we can write an equation for the displacement vector, u, as:

(7)

Incremental Displacement Equation in a Poro-Elasto-Plastic Reservoir

For poro-elasto-plastic material behavior, the nonlinear stress-strain relation is approximated by a Mohr-
Coulomb perfect plasticity model. The Mohr-Coulomb plasticity model is a classical model used to capture
shear failure in frictional materials such as soils and rocks (Vermeer & de Borst, 1984). The plastic flow
formulation rests on the assumption that the total strain increment is decomposed into elastic and plastic
components, where only the elastic strain contributes to the stress increment.
4 SPE-184847-MS

(8)
The following equations summarize the mathematical model for incremental displacement, du. They
represent incremental forms of the governing equation, a non-linear perfect elastoplastic constitutive
relation, and strain-displacement relations.

(9)

(10)

(11)

(12)

Plastic strains are generated when the yield surface f = 0 is reached. A tensile cut-off Mohr-Coulomb
criterion is written as:

(13)

(14)

(15)

In the plastic flow rule, g specifies the direction and magnitude of the incremental plastic strain
(Papanastasiou, 1997; Vermeer & de Borst, 1984). The flow potential function is described by two yield
surfaces, gs and gt, to define shear and tensile plastic flow.

(16)

(17)

(18)

(19)

Where dplastic is the plastic multiplier of a scalar function. By combining the above equations into a
governing equation, a mathematical model for the incremental displacement is derived:
SPE-184847-MS 5

(20)

Model Implementation
Recent work using finite volume method (FVM) for hydraulic fracture modeling, reservoir simulation,
and sanding prediction have made FVM a viable approach for solving such problems (Bryant, Hwang, &
Sharma, 2015; Hwang, Bryant, & Sharma, 2015; Lee et al., 2015; Manchanda, Bryant, Bhardwaj, & Sharma,
2016; Bhardwaj, Hwang, Manchanda, & Sharma, 2016; Wang & Sharma, 2016). We implement our model
in the finite volume discretization framework by building and using C++ libraries within OpenFOAM-3.2-
ext and our geomechanics library, FROGG (FRamework for Operations in General Geomechanics) (Philip
Cardiff et al., 2015; Lee et al., 2015). All codes are designed to run in parallel using dynamic mesh
refinement. This enables us to handle complex problems with greater accuracy and less computation time.
A cell-centered FV domain discretization and segregated solution procedures are employed, where each
primary unknown is solved separately, and outer Picard iterations provide coupling between variables and
domains. Implementation of the cohesive zone model, plasticity return algorithm, numerical leak-off using
domain coupling, and solution algorithm were explained in a previous paper (Lee et al., 2015). We present
the discretization and boundary conditions employed in detail here.

Discretization
The solution domain is divided into a number of cells by a number of faces. Figure 2 shows two control
volumes (cells) with the computational point P (center of current owner cell) and N (center of the neighbor
cell) in its center. Every internal face is shared by two cells. Cells are contiguous and don't overlap each
other to fill the complete spatial domain. Dependent variables and properties are usually stored at the center
of each cell such as P and N. Some of variables and properties are defined and stored at the center of the
face, or at a point. As shown in Figure 2, the distance vector between the center of cells is not necessarily
parallel to face area vector . This non-orthogonal, unstructured mesh utilization offers more freedom
to handle the complex geometry of the domain (Jasak & Weller, 2000).
6 SPE-184847-MS

Figure 2Neighboring cells (current cell, P, and neighbor, N) and an internal face (OpenFOAM Foundation, 2013).

We use the integral over cell volume form of a mathematical model for each primary unknown. Most
spatial integral terms are converted to the integral over the cell faces by the divergence theorem. Then the
equation is discretized into the explicit and implicit term. In this study, we employ the implicit-explicit split
strategy with a segregated solution procedure. For example, each component of the displacement vector
is solved separately with fixed point iteration to achieve convergence. Linear and uncoupled terms are
treated implicitly, and non-linear and coupling terms are evaluated explicitly from the previous time step or
iteration value. This method is extremely memory efficient as it allows us to construct diagonally dominant
sparse matrices. Instead of one large matrix handling all the components of dependent variables, we have
segregated small matrices for each component of the unknown and solve them consecutively and iteratively.
This segregated approach with fixed point iteration can suffer from poor convergence and a number of
iterations over the inter-coupling term. Therefore, following careful equation discretization is necessary to
ensure fast and reliable convergence (P. Cardiff, Tukovi, Jasak, & Ivankovi, 2016; Jasak & Weller, 2000;
Tang, Hededal, & Cardiff, 2015).
Discretization of the displacement equation is applied on a term-by-term basis. The formulated
mathematical model for the displacement vector in Eq. (7) is integrated over the control volume (cell) as
shown in Eq. (21).

(21)

The second time derivative term integrated over a volume (L.H.S of Eq. (7) is linearized as follows:

(22)

Where P is the current owner cell as shown in Figure 2. The R.H.S of Eq. (7) is converted to the sum of
surface integral over the cell faces with implicit and explicit discretization as follows:
SPE-184847-MS 7

(23)

By using the implicit discretization for the Laplacian term ( ), we have a linear relation which
depends only on the displacement vector on the current cell (up) and the nearest neighbor cells (uN).

(24)

(25)

Coefficients aP and aN make up the diagonal and off-diagonal terms of the sparse matrices. If the surface
area vector (Sf) and distance vector (d) between two cells are not parallel, which is the case of non-orthogonal
unstructured meshes, an additional treatment for non-orthogonal correction is introduced.
Inter-component coupling terms ((u)T +tr(u)I), a constant term (0), an inter-variable coupling term
(bpI) are discretized explicitly:

(26)

(27)

(28)

We need further discretization for the displacement equation by adding a (+)u term into the implicit
discretization and subtracting the equivalent term in the explicit discretization as shown in Eq. (29). Based
on Jasak and Weller, 2000, this helps to have balanced implicit and explicit split and constructs an equal
coefficient matrix for each component of the displacement vector.

(29)

Boundary Conditions
Boundary conditions for each dependent variable can be either Dirichlet or Neumann in most of the cases.
For proppant concentration, we need to specify the concentration of proppant at the inlet as time-varying
8 SPE-184847-MS

fixed value (Dirichlet BC). For the fracture face boundary, zero gradient is specified as a Neumann BC.
A traction boundary condition for the displacement is usually applied on the far-field boundary due to
the far-field in-situ stresses as shown in Figure 3 and fracture-reservoir interface due to the injected fluid
pressure as shown in Figure 4. Traction boundary condition is implemented as a Neumann condition where
the surface normal gradient is specified on the boundary face. Traction is calculated at the boundary face
using Cauchy's stress theorem, and total stress is decomposed to the effective stress and the pore pressure
with Biot's coefficient. Then, as introduced in Eq. (29), we discretize the total stress term to have a balanced
implicit and explicit split. Moving the implicit term to the L.H.S and the other terms to the R.H.S yield
Eq. (31).

(30)

(31)

Figure 32-D conceptual image of boundary conditions.

SPE-184847-MS 9

Figure 4Traction applied on the fracture-reservoir interface. The calculated traction

from the pressure in the fracture domain (which is varying along the fracture) is applied.

Model Verification
We have verified the essential components of the model by comparing our results with several well-known
analytical solutions:

Poro-elasticity: 1D Biot's consolidation problem (Bryant et al., 2015)

Elasto-plasticity: Shear failure around the circular hole in a 2-D infinite plate (Lee et al., 2015)

Fracture propagation: KGD model in a 2-D elastic domain (Lee et al., 2015)

In this paper, we extend the verification of fracture propagation by conducting a new benchmark test for
KGD model propagation in storage-toughness dominated regime. We also investigate the effect of the mesh
density on the rate of fracture propagation.

KGD Verification for the Storage-Toughness Dominated Regime

We benchmark a case of KGD fracture propagation in storage-toughness dominated regime from work
by Carrier & Granet, 2012. We simulate the problem of hydraulic fracture propagation in a homogeneous
domain with leak-off under plane strain conditions, due to the injection of an incompressible Newtonian
fluid into the fracture as shown in Figure 5. We model only half of the domain (defined by the symmetry
plane). Constant injection rate per unit height is specified, and the material parameters used for the numerical
simulation are given in Table 1. Approximate analytical solutions in the form of regular asymptotic
expansions were obtained by assuming an infinite domain with linear elastic materials with Carter's leakoff
model (Bunger, Detournay, & Garagash, 2005; Carrier & Granet, 2012; Zielonka, Searles, Ning, & Buechler,
2014).
10 SPE-184847-MS

Figure 52D Computational domain for KGD verification in storage-

toughness dominated regime with traction boundary conditions.

Table 1Simulation parameters for KGD fracture propagation in storage-toughness dominated regime (after (Carrier & Granet, 2012)).

Parameters Symbol Value Unit

Dimension of the 2D domain in X direction dx 45 m

Dimension of the 2D domain in Y direction dy 60 M
Young's modulus of the domain E 17 GPa
Poisson's ration 0.2 -
Mode I energy release rate GIc 120 Pam

Tensile strength tnc

1.25 MPa

Medium permeability k 0.1 / 1 mD

Biot coefficient b 0.75 -
Biot modulus M 68.7 MPa

Mass injection rate 0.5Q 0.0005 m3/sec

Fracturing fluid viscosity 0.0001 Pasec
Porosity 0.2 -

We perform numerical simulations for two permeabilities, 0.1md and 1md, with 14sec of injection to
ensure that the fracture propagation stays in the storage regime (Carrier & Granet, 2012). The numerical
results of effective stress in the y-direction and close-up around the fracture tip are displayed in Figure 6. Due
to the low permeability of the matrix and short injection duration, pore pressure diffusion and poroelastic
backstress buildup are not dominant, as intended. Figure 6 clearly shows the extent of the cohesive zone
elements around the tip where the face is inward convex due to the cohesive traction.
SPE-184847-MS 11

Figure 6Normal effective stress distribution in the y-direction (Left) and close-up view around the fracture tip
(Right) at the end of the simulation for the k=0.1 md case. Note that the displacement is magnified 3000 times.

The results obtained from simulations are compared to analytical solutions for the storage-toughness
dominated regime (Bunger et al., 2005) in Figure 7. The total numerical fracture length which includes fully
broken and cohesive zone elements and fracture length only with fully broken elements are compared to
the analytical solution of each permeabilities. The cohesive zone length is almost constant throughout the
simulation for the two cases and is approximately 0.5~0.6m. Very good agreement between our model and
analytical solution are obtained as shown in Figure 7 in the two different permeability cases. We further
investigate the effect of the mesh fineness on the propagation rate by running cases with different levels of
mesh refinement. By increasing the level of the refinement around the fracture faces and tips, we have a
converged fracture propagation rate as shown in Figure 7 (Right).

Figure 7Fracture half-length results for two permeability cases: 0.1 md (in black)
and 1md (in red) (Left); The effect of mesh refinement on the propagation rate (Right).

Results and Discussion

We apply our model to simulate and analyze a recent frac-pack treatment conducted in a field. In a frac-
packing operation, a porous reservoir is fractured with short and wide biwing propped fractures and the
annulus is packed with proppant. The fracture half-length usually is less than 50 ft, and the propped width is
approximately 1 to 2 in (Ghalambor, Ali, & Norman, 2009). Reservoir dimensions are 10010060 meters
in our simulations. A fully three-dimensional domain is created and discretized as shown in Figure 8. Finite
volume meshes are generated for only one-quarter of the reservoir and the fracture to take advantage of
12 SPE-184847-MS

symmetry. We adaptively apply mesh refinements around the fracture face to compute accurately within
reasonable computation time. The model consists of 45,000 finite volume cells with the smallest cell having
a dimension of 0.5m. We construct the Base Case with reservoir properties given from the field data (log
analysis for layer properties, step rate test and analysis, pumping schedule, and injected fluid analysis). Our
simulation results are compared with the bottomhole pressure response recorded for 60 min of injection in
the field. Then, we provide sensitivity studies by varying fluid flow parameters and mechanical properties.

Figure 8Numerical domain and mesh discretization for field application cases:
Note that mesh is warped by the displacement vector to visualize the fracture.

Base Case
The model properties for the Base Case are given in Table 2. Our model provides a variety results in both
domains:

Displacement distribution

Stress (total and effective) distribution

Strain (elastic and plastic) distribution

Failure state (shear and tensile) distribution

Pore pressure distribution

Induced fracture geometry (length, width, and height)

Fracture (bottomhole and net) pressure distribution

Proppant concentration distribution

Table 2Simulation parameters for the Base Case

Fluid Flow Parameter Symbol Value Unit

Reservoir permeability k 1250 mD

Biot coefficient b 1 -
Injection rate qinj 40 bpm
Injection duration tinj 60 min
Fracturing fluid viscosity 100 cP
Continued on next page
SPE-184847-MS 13

Fluid Flow Parameter Symbol Value Unit

Porosity p 0.28 -

Initial Value Symbol Value Unit

Initial pore pressure po 70 MPa

Vertical in-situ stress (total) Sv 100 MPa
Max. horizontal stress (total) SH 90 MPa
Min. horizontal stress (total) Sh 85 MPa
Porosity 0.2 -

Mechanical Parameter Symbol Value Unit

Young's modulus E 1 GPa

Poisson's ratio 0.3 -
Cohesion c 5 MPa
Friction angle 30
Tensile strength t 0.5 MPa

Figure 9 shows the distribution of the displacement in the y-direction, pore pressure, and total and
effective stress of normal component in the y direction at the end of the simulation. Fluid leak-off is
numerically computed in this model depends on the fracture propagation and reservoir permeability tensor.
So, three-dimensional radial leak-off is captured in the Figure 9-(B). In all cases, the model is run for a
constant injection rate of 40 bpm for a 60 min injection duration without pumping proppant.

Figure 9Displacement and pore pressure distribution at the end of the simulation.
14 SPE-184847-MS

Figure 10 shows the pressure response during the one-hour fracturing job. Simulation result shows that
calculated fracturing pressure is higher than the maximum horizontal stress. Net fracturing pressure is
also very high which is consistent with field observations. In this simulation, we have a homogeneous
payzone, so a penny-shaped radial fracture is expected to propagate steadily as shown in Figure 11.
Therefore, fracturing pressure and consequently the maximum fracture at the wellbore declines (mainly due
to increasing volume of the fracture). Proppant is not simulated in this case, so we don't have an increased
pressure response at the end of the injection, indicating the tip screen-out. Base Case results are summarized
in Table 3, and it is compared to results from a commercial hard rock model with the same input. We compare
the ratio of the Base Case simulation result to the hard rock model in the last column. Figure 12 compares
the pressure response from the field data, and our model result (FracPack) of the Base Case, and a hard rock
model result. It is observed that our model is capable of simulating the high net pressure value as well as
the overall pressure response without manipulating input parameters. Also, the induced fracture geometry
predicted shows shorter lengths and wider widths compared to the hard rock model.

Figure 10Pressure response during the simulation. Left: Simulated bottomhole fracturing
pressure at the injection point, the injection rate, and proppant concentration (solid line) compared
to the corresponding field data (dot), Right: Net fracturing pressure (line) with field data for the
net pressure (dot) in log-log scale. ppa: pounds of proppant added per gallon of clean fluid.

Figure 11Induced fracture geometry. Left: induced fracture area, Right: maximum fracture width at the wellbore.
SPE-184847-MS 15

Figure 12Net pressure response from the Base Case, the field data, and a hard rock model result.

Table 3Fracture geometry of the Base Case compared to a hard rock model.

Base Case result Hard rock Model Ratio

Fracture Half-Length 15m (50ft) 55m (180ft) 0.27

Max. Fracture width 41.6mm (1.6in) 26mm (1.02in) 1.60
Height 40m (130ft) 47m (150ft) 0.85
Net Pressure 8MPa (1160psi) 4.6MPa (670psi) 1.74

Effect of the pressure gradient

The pressure gradient is mainly determined by the permeability of the reservoir, the injected fluid viscosity,
and the injection rate. The effect of the pressure gradient is evaluated by running sensitivity cases with only
the corresponding parameter changed as shown in Table 4. Then, results from the Base Case and sensitivity
cases are compared for the induced fracture geometry and the pressure response.

Table 4Sensitivity study 1: the effect of the pressure gradient

Base Case Sensitivity I Sensitivity II

Reservoir permeability 1250md 750md 2000md

Injected fluid viscosity 100cp 75cp 125cp
Injection rate 40bpm 30bpm 50bpm

The higher pressure gradient from the lower matrix permeability, higher injected fluid viscosity, and
higher injection rate cases result in a longer and taller induced fracture. Fast propagation of the fracture
leads to the increase of the volume of the fracture. Then, the pressure inside the fracture decreases as the
fracture propagates. Note that even you pump with the higher injection rate of fracturing fluid, you will get
the lower fracturing pressure at the end of the simulation because of the increased volume of the longer
and taller fracture.
16 SPE-184847-MS

Figure 13The effect of the reservoir permeability. Induced fracture area (Left)
and fracturing pressure response (Right) during an hour of injection simulation.

Figure 14The effect of the fluid viscosity. Induced fracture area (Left) and
fracturing pressure response (Right) during an hour of injection simulation.

Figure 15The effect of the reservoir permeability. Induced fracture area (Left)
and fracturing pressure response (Right) during an hour of injection simulation.
SPE-184847-MS 17

Effect of the mechanical properties of the reservoir

We investigate the effect of the mechanical properties by running sensitivity cases with only the
corresponding parameter changed as shown in Table 5.

Table 5Sensitivity study 2: the effect of the mechanical properties of the reservoir

Base Case Sensitivity I Sensitivity II

Young's modulus 1GPa 0.5GPa 5GPa

Cohesion 5MPa 3MPa 7.5MPa
Internal friction angle 30 20 45
Tensile strength 0.5MPa 1MPa 5MPa

ncreasing Young's modulus makes the reservoir harder and the stress more concentrated at the tip. The
fracture initiates earlier, and the propagation rate is higher for the higher Young's modulus case. Therefore,
the higher volume of the induced fracture results in a declining fracture pressure response as shown in
Figure 16.

Figure 16The effect of Young's modulus. Induced fracture area (Left) and fracturing pressure response (Right).

The plastic behavior of the reservoir is determined by the cohesion and internal friction angle. By
changing a value of the cohesion and internal friction angle, we investigate the effect of inelastic deformation
of the reservoir. From Figure 17 and Figure 18, we observe that the time at the fracture starts growing and
the slope of the fracturing pressure increased before it propagates are identical in all cases. This is because
these are controlled by the parameters such as elasticity and fluid flow parameters which are the same.
By lowering the cohesion and internal friction angle, more plastic deformation occurs in the poro-elasto-
plastic reservoir. In these cases, shear failure occurs around the fracture and ahead of the tip due to the pore
pressure diffusion. The sand fails in shear first then in tension if sufficient pore pressure and poroelastic
backstress build up. The plasticity around and ahead of the fracture causes a lower stress concentration at
the tip which shields the tip of a propagating fracture from the fracturing pressure. This results in a higher
fracturing pressure being required to propagate a fracture in a poro-elasto-plastic material.
18 SPE-184847-MS

Figure 17The effect of the cohesion. Induced fracture area (Left) and fracturing pressure response (Right).

Figure 18The effect of the internal friction angle. Induced fracture area (Left) and fracturing pressure response (Right).

Increasing the tensile strength changes the criteria of fracture initiation of the cohesive zone model.
Higher tensile strength requires higher stress concentration at the fracture tip to make it propagate.
Therefore, higher fracturing pressure and a lower rate of fracture propagation are observed from the higher
tensile strength case.
SPE-184847-MS 19

Figure 19The effect of the tensile strength. Induced fracture area (Left) and fracturing pressure response (Right).

Conclusions
We have formulated and implemented a new frac-pack model using the finite volume cohesive zone model.
The model includes the following unique features:

Non-planar fully 3-D fracture growth in poro-elasto-plastic materials

Both tensile and shear failure accounted for

Newtonian or power-law fluid flow coupled with proppant transport in the fracture

Fluid leak-off from the fracture to the porous reservoir computed numerically

3-D fracture propagation using Cohesive Zone Model and dynamic meshing

Parallel computing, adaptive mesh refinement

The model presented here provides a new approach to model fracture growth and to understand the
mechanisms of fracture growth in unconsolidated reservoirs. The model allows us to design and analyze
hydraulic fracturing stimulations in soft formation much more accurately than LEFM models used in the
past. Some important observations from the model development and application are:

Fracture propagation is verified by a benchmarking KGD fracture propagation problem in the

storage-toughness dominated regime. Our model provides good agreement with the asymptotic
analytical solution without mesh fineness dependency.
Our model is capable of capturing the high net fracturing pressure during frac-pack operations.
Plasticity caused lower stress concentration around the fracture tip which shields tips of a
propagating fracture from the fracturing pressure. We observe shear failure around the fracture and
ahead of the tip due to pore pressure diffusion. Low cohesion sand tends to fail in shear first then
in tension if sufficient pore pressure and poroelastic backstress build up.
We investigate factors affect the pressure gradient such as the matrix permeability, injected fluid
viscosity, and injection rate. Higher pressure gradient due to lower permeability, higher viscosity
and injection rate results in a longer and taller induced fracture and a steep declining fracturing
pressure response.
Our frac-pack model predicts shorter lengths and higher widths compared to a hard rock model
This is consistent with field and lab observations. The net pressure response from the simulation
agrees well with field data.
20 SPE-184847-MS

Acknowledgements
This work is supported by companies participating in the Joint Industry Project on Hydraulic Fracturing
and Sand Control at the University of Texas at Austin.

Nomenclature
b : Biot coefficient [-]
c : Cohesion [N/m2]
d : Incremental strain tensor [-]
de : Incremental elastic component of strain tensor [-]
dp : Incremental plastic component of strain tensor [-]
dplastic : Plastic multiplier of a scalar function
du : Incremental displacement tensor [m]
: Strain tensor [-]
v : Volumetric Strain, scalar [-]
fs : Shear failure envelope
ft : Tensile failure envelope
g : Plastic potential function
G : Lame's second constant, Shear modulus [N/m2]
I : Identity matrix
p : Pore pressure field [N/m2]
P : Current owner cell
S : Total stress tensor [N/m2]
tr : Trace of a matrix
u : Displacement vector [m]
u old : Displacement vector at the previous time step [m]
un : Displacement vector at the current time step, un =u(t + t)
u : Displacement vector at the previous time step, u =U(t)
u : u =u(tt)
: Lame's first constant [N/m2]
: Mass density [kg/m3]
f : Body force density [N/m3]
: Internal friction angle []
: Dilation angle []
: Effective stress tensor [N/m2]
0 : Initial effective stress tensor [N/m2]
1 : Effective maximum principal stress [N/m2]
3 : Effective minimum principal stress [N/m2]

References
Adachi, J., Siebrits, E., Peirce, A., & Desroches, J. (2007). Computer simulation of hydraulic fractures. International
Journal of Rock Mechanics and Mining Sciences, 44(5), 739757. http://dx.doi.org/10.1016/j.ijrmms.2006.11.006
Bhardwaj, P., Hwang, J., Manchanda, R., & Sharma, M. M. (2016). Injection Induced Fracture Propagation and Stress
Reorientation in Waterflooded Reservoirs. Presented at the SPE Annual Technical Conference and Exhibition, Dubai,
UAE, 26-28 September. SPE-181883-MS. http://dx.doi.org/10.2118/181883-MS
Bryant, E. C., Hwang, J., & Sharma, M. M. (2015). Arbitrary Fracture Propagation in Heterogeneous Poroelastic
Formations Using a Finite Volume-Based Cohesive Zone Model. Presented at the SPE Hydraulic Fracturing
SPE-184847-MS 21

Technology Conference, The Woodlands, Texas, 3-5 February. SPE-173374-MS. http://dx.doi.org/10.2118/173374-

MS
Bunger, A. P., Detournay, E., & Garagash, D. I. (2005). Toughness-dominated Hydraulic Fracture with Leak-off.
International Journal of Fracture, 134(2), 175190. http://dx.doi.org/10.1007/s10704-005-0154-0
Cardiff, P., Manchanda, R., Bryant, E. C., Lee, D., Ivankovi, A., & Sharma, M. M. (2015). Simulation of Fractures in
OpenFOAM: From Adhesive Joints to Hydraulic Fractures. In 10th OpenFOAM Workshop.
Cardiff, P., Tukovi, ., Jasak, H., & Ivankovi, A. (2016). A block-coupled Finite Volume methodology for
linear elasticity and unstructured meshes. Computers & Structures, 175, 100122. http://dx.doi.org/10.1016/
j.compstruc.2016.07.004
Carrier, B., & Granet, S. (2012). Numerical modeling of hydraulic fracture problem in permeable medium using cohesive
zone model. Engineering Fracture Mechanics, 79, 312328. http://dx.doi.org/10.1016/j.engfracmech.2011.11.012
Chen, Z. (2012). Finite element modelling of viscosity-dominated hydraulic fractures. Journal of Petroleum Science and
Engineering, 8889, 136144. http://dx.doi.org/10.1016/j.petrol.2011.12.021
Dam, D. B., Papanastasiou, P., & Pater, C. J. (2002). Impact of Rock Plasticity on Hydraulic Fracture Propagation and
Closure. SPE Production & Facilities, 17(August), 149159. http://dx.doi.org/10.2118/78812-PA
Ghalambor, A., Ali, S. A., & Norman, W. D. (2009). Frac Packing Handbook. Richardson, TX: Society of Petroleum
Engineers.
Hwang, J., Bryant, E. C., & Sharma, M. M. (2015). Stress Reorientation in Waterflooded Reservoirs. Presented at the
SPE Reservoir Simulation Symposium, Houston, Texas, 23-25 February. SPE-173220-MS (pp. 2325). Society of
Petroleum Engineers. http://dx.doi.org/10.2118/173220-MS
Jaeger, J. C., Cook, N. G. W., & Zimmerman, R. W. (2007). Fundamentals of Rock Mechanics.
Jasak, H., & Weller, H. G. (2000). Application of the Finite Volume Method and Unstructured Meshes to Linear Elasticity.
International Journal for Numerical Methods in Engineering, 48, 267287. http://dx.doi.org/10.1002
King, G. E. (2012). Hydraulic Fracturing 101: What Every Representative , Environmentalist , Regulator , Reporter ,
Investor , University Researcher , Neighbor and Engineer Should Know About Estimating Frac Risk and Improving
Frac Performance in Unconventional Gas and Oil W. Presented at the SPE Hydraulic Fracturing Technology
Conference, Woodlands, Texas, 6-8 February. SPE 152596.
Lee, D., Cardiff, P., Bryant, E. C., Manchanda, R., Wang, H., & Sharma, M. M. (2015). A New Model for Hydraulic
Fracture Growth in Unconsolidated Sands with Plasticity and Leak-Off. Presented at the SPE Annual Technical
Conference and Exhibition, Houston, Texas 28-30 September. SPE-174818-MS. http://dx.doi.org/10.2118/174818-MS
Manchanda, R., Bryant, E. C., Bhardwaj, P., & Sharma, M. M. (2016). Strategies for Effective Stimulation of Multiple
Perforation Clusters in Horizontal Wells. Presented at the SPE Hydraulic Fracturing Technology Conference,
Woodlands, Texas, 9-11 February. SPE-179126-MS. http://dx.doi.org/10.2118/179126-MS
OpenFOAM Foundation. (2013). OpenFOAM Programmer's Guide.
Papanastasiou, P. (1997). The Influence of Plasticity in Hydraulic Fracturing. International Journal of Fracture, 84, 6179.
Searles, K. H., Zielonka, M. G., Ning, J., Garzon, J. L., Kostov, N. M., Sanz, P. F., & Biediger, E. (2016). Fully-
Coupled 3D Hydraulic Fracture Models: Development, Validation, and Application to O&G Problems. Presented at
the SPE Hydraulic Fracturing Technology Conference, Woodlands, Texas, 9-11 February. SPE-179121-MS. http://
dx.doi.org/10.2118/179121-MS
Tang, T., Hededal, O., & Cardiff, P. (2015). On finite volume method implementation of poro-elasto-plasticity soil model.
International Journal for Numerical and Analytical Methods in Geomechanics, 32, http://dx.doi.org/10.1002/nag.2361
Vermeer, P. a., & de Borst, R. (1984). Non-Associated Plasticity for Soils, Concrete and Rock. Heron, 29(3), 164.
Retrieved from http://dx.doi.org/http://www.narcis.nl/publication/RecordID/oai:tudelft.nl:uuid:4ee188ab-8ce0-4df3-
adf5-9010ebfaabf0
Vermeer, P. A., & de Borst, R. (1984). Non-Associated Plasticity for Soils, Concrete and Rock. Heron, 29(3), 164. http://
doi.org/10.1007/978-94-017-2653-5_10
Wang, H., & Sharma, M. M. (2016). A Fully 3-D, Multi-Phase, Poro-Elasto-Plastic Model for Sand Production. In SPE
Annual Technical Conference and Exhibition. Society of Petroleum Engineers. http://dx.doi.org/10.2118/181566-MS
Zielonka, M. G., Searles, K. H., Ning, J., & Buechler, S. R. (2014). Development and Validation of Fully-Coupled
Hydraulic Fracturing Simulation Capabilities. SIMULIA Community Conference, SCC2014, 131.