Sie sind auf Seite 1von 17

Available online at www.sciencedirect.

com

ScienceDirect
Underground Space xxx (xxxx) xxx
www.elsevier.com/locate/undsp

Modeling hydraulic fracture in heterogeneous rock materials


using permeability-based hydraulic fracture model
Ming Li a, Peijun Guo b,⇑, Dieter F.E. Stolle b, Li Liang a, Yitao Shi a
a
Department of Civil Engineering, Northeastern University, Box 265 N, 3-11, Wenhua Road, Heping District, Shenyang 110819, PR China
b
Department of Civil Engineering, McMaster University, 1280 Main Street West, Hamilton, Ontario L8S 4L8, Canada

Received 26 February 2018; accepted 24 December 2018

Abstract

Hydraulic fracturing is one of the most important techniques for enhancing oil/gas production. The permeability-based hydraulic
fracture (PHF) model, which is based on the smeared-crack method and considers the interaction between the pore pressure and solid
phase, is adopted in the present study for a fully-coupled simulation of the hydraulic fracture in a heterogeneous rock formation. The
level set method (LSM), which is used to describe the distribution of material properties of heterogeneous rocks, is coupled with the PHF
model. Using the coupled PHF–LSM model, a series of finite-element method (FEM) simulations are carried out to investigate the char-
acteristics of a hydraulic fracture (e.g., the breakdown pressure and fracture propagation) in heterogeneous rocks. Three types of hetero-
geneous rocks are examined: layered rock, rock with distributed inclusions, and rock with random spatial variations in the material
properties. The results of the numerical simulations show that the coupled PHF–LSM model can describe the material interface without
changing the FEM mesh used to discretize the physical domain. Further, the model effectively simulates hydraulic-fracturing problems
for various heterogeneous rocks.
Ó 2019 Tongji University and Tongji University Press. Production and hosting by Elsevier B.V. on behalf of Owner. This is an open access article
under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Hydraulic fracture; Heterogeneous rock material; Smeared-crack method; Level set method

1 Introduction wellbore. Assuming plane-strain conditions, Perkins and


Kern (1961) proposed the Perkins–Kern–Nordgren
The prediction of the geometry of a hydraulic-fracture (PKN) fracture model that permits the estimation of frac-
zone and the variation in the injection pressure during an ture widths under a variety of flow conditions for both hor-
entire fracture process is important to evaluate the effec- izontal and vertical fractures. Nordgren (1972) further
tiveness of a hydraulic fracture. Following the pioneer modified the PKN model to simulate the propagation of
work of Khristianovitch and Zheltov (1955) to model a hydraulically induced fractures of limited vertical extent
fracture induced by a viscous fluid in the horizontal plane and elliptic cross-sections considering the effect of fluid
under plane-strain conditions, Geertsma and de Klerk loss. Similar work was conducted by Economides and
(1969) developed a rapid method—the Khristianovic– Nolte (1989) and Valko and Economides (1995). A repre-
Geertsma–de Klerk (KGD) model—to predict the width sentative model for a three-dimensional (3D) analysis is
and extent of hydraulically induced fractures around a the P-3D-C model (Adachi, Sie brits, Peirce, &
Desroches, 2007; Rahman & Rahman, 2010), which is a
⇑ Corresponding author. pseudo-3D (P-3D) model incorporating Carter’s solution
E-mail addresses: liming-neu@163.com (M. Li), guop@mcmaster.ca of material balance (Simonson, Abou-Sayed, & Clifton,
(P. Guo), stolle@mcmaster.ca (D.F.E. Stolle), liangli@mail.neu.edu.cn 1978; Settari & Cleary, 1986; Warpinski & Smith, 1989).
(L. Liang).

https://doi.org/10.1016/j.undsp.2018.12.005
2467-9674/Ó 2019 Tongji University and Tongji University Press. Production and hosting by Elsevier B.V. on behalf of Owner.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
2 M. Li et al. / Underground Space xxx (xxxx) xxx

Developed within the framework of the classical elasticity they need more computational resources owing to the need
theory, these analytical models all assume a linear elastic for finer meshes.
behavior of the reservoir rock and neglect surface energy Regarding heterogeneous rocks with inclusions, depend-
considerations for the crack tip or plastic deformation ing on the contrast between the properties of the con-
effects. The leak-off is treated as an independent process stituents, a heterogeneous rock can be treated as a
and merged with the crack propagation problem via itera- homogeneous continuum (matrix) with different types of
tive methods that assume self-consistency. inclusions, which can be hard inclusions, weak inclusions,
Numerical methods are more effective in simulating or voids. The geometry of an inclusion, which usually has
hydraulic fracturing problems; particularly, when the non- a finite volume and complex shape, can be simplified to a
linear material behavior, hydro-mechanical coupling, circle (Wu & Wong, 2013), sphere, or an ellipse (Markov,
heterogeneity of the rock formation, and complex bound- Jarillo, & Markov, 2014). Wu and Wong (2013) reported
ary conditions are considered. Various numerical modeling the test results of hydrocal specimens containing a single
techniques have been developed to simulate hydraulic frac- inclusion and showed that cracks can propagate along
tures: the finite-element method (FEM) (Wangens, 2011), the hard inclusion and pass through the soft inclusion.
distinct-element method (Fatahi, Hossain, Fallahzadeh, & Li, Guo, Stolle, and Li (2017) carried out FEM simulations
Mostofi, 2016; Zhang, Damjanac, & Huang, 2013), and to investigate the influence of regularly or randomly dis-
meshfree (meshless) method (Zhuang, Augarde, & tributed hard/soft inclusions with different shapes. In addi-
Mathisen, 2012). Based on the FEM, the smeared-crack tion, numerical algorithms have been developed in XFEM
approach (Li, Guo, Stolle, & Liang, 2016), extended to consider the effect of inclusions, such as the enriched
FEM (XFEM) (Lecampion, 2009; Mohammadnejad & numerical manifold method (Wu & Wong, 2013) and level
Khoei, 2013), and cohesive-zone approach (Sarris & set method (LSM) (Osher & Sethian, 1988).
Papanastasiou, 2011) have been successfully used to simu- In this study, a permeability-based hydraulic fracture
late hydraulic-fracture problems. Different from the FEM (PHF) model developed by Li et al. (2016) for two-
in which a real crack must be physically defined, the stan- dimensional (2D) rock material with layered features is
dard XFEM allows the presence of discontinuities in an used with FEM to simulate hydraulic fracturing in different
element by enriching degrees of freedom (DOF) with spe- types of heterogeneous rock materials. These include lay-
cial displacement functions. The major drawback of ered rock systems, rock material with inclusions, and
XFEM is that the solution corresponding to additional heterogeneous rock with random spatial distributions of
DOF requires additional system resources (either more material properties following the Weibull distribution
computer power or higher computing costs); in particular, (Weibull, 1939). To improve the modeling efficiency and
for 3D, large-scale field problems. Even though the cohe- to control the mesh sensitivity of the FEM simulations,
sive zone model was originally proposed to investigate the same mesh is used for different types of rocks with a
the behavior of perfectly brittle or ductile materials exhibit- proper control of the aspect ratio of the FEM elements
ing a nonlinear zone ahead of the crack tip due to plasticity (Li, Li, Guo, & Li, 2015). The LSM is introduced to track
or micro-cracking (Dugdale, 1960; Barenblatt, 1962), it has the evolution of the material interface without changing the
been used to model the process of hydraulic fractures in FEM mesh (Li et al., 2017). Different from the study in Li
rocks. Different from the discrete-crack model aiming at et al. (2017), the LSM is used to track inclusions, layered
simulating the initiation and propagation of dominant and spatial distributions of material properties in the
cracks, the smeared-crack approach assumes that the heterogeneous rock materials.
cracked element remains continuous by smearing out the The remainder of the paper is structured as follows: Sec-
cracks over the continuum. Thus, the cracking effect is tion 2 introduces the modeling method for heterogeneous
translated to a deterioration of the material properties rocks, followed by an outline of the PHF model in Sec-
(e.g., the stiffness, permeability, and strength). Originally tion 3. Next, the characteristics of a hydraulic fracture in
used to analyze the crack propagation in concrete struc- homogeneous rock, layered rock, and rock with distributed
tures (Bažant and Oh, 1983; de Borst, 1987), methods inclusions and Weibull distributions of the material proper-
based on the smeared-crack approach have been developed ties are discussed in Sections 4 and 5. Section 6 provides the
to simulate hydraulic fracturing in rock mass (Hu, Chen, conclusions.
Cheng, & Yang, 2013). However, the classical smeared-
crack model exhibits some disadvantages; particularly, 2 Modeling of heterogeneity in rock sample
the propagation of a crack along the direction of the mesh
lines and a spurious sensitivity with incorrect convergence Three types of heterogeneous rock materials are studied
when the elements are refined to a vanishing size (Bažant in the present work: rock with layered features, rock with
and Oh, 1983; Remmers, de Borst, & Needleman, 2003). distributed hard inclusions, and rock with random spatial
Various methods, such as an introduction of a characteris- variations of the material properties described via a Wei-
tic length and the non-local smeared-crack method, have bull distribution (Weibull, 1939). To model the heterogene-
been adopted to reduce the mesh sensitivity (Bažant and ity of a rock, we extend the LSM proposed by Li et al.
Oh, 1983; Bažant and Lin, 1988; Hu et al., 2013). However, (2017) to a general form. The extended LSM model can

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
M. Li et al. / Underground Space xxx (xxxx) xxx 3

be used to describe the boundaries of different constituent 2.2 Rock material with distributed hard inclusions
materials in a heterogeneous material system, which can
then be used to control the quality of the FEM mesh. Explicit expressions for particular shapes (e.g., circles,
In a heterogeneous rock sample, the material interface L ellipses, polygons) via the level set function have been stud-
(zero-level set) between the matrix and different material ied by Sukumar, Chopp, Moës, and Belytschko (2001). An
features is assumed independent of time. By ignoring the arbitrary elliptical inclusion located at a random position
influence of the time t, the static material interface can be ðxi0 Þ in the n-dimensional solution domain can be
simplified to: expressed by introducing a random number:
L ¼ fx 2 Rn : ui ðxÞ ¼ 0g ði ¼ 1; 2; 3; :::; mÞ; ð1Þ Xn  2
x  xi0
ui ðxÞ ¼  1; ð4Þ
in which x is the coordinate in the n-dimensional space Rn j¼1
aja aj
and m the total number of different material regions. The
union of the material regions is implicitly represented as in which aja 2 ð0; 1 are random numbers that control the
follows: semi-major length aj of the elliptical inclusions. If the rota-
tion of inclusions is included, the coordinates should be
uðxÞ ¼ minðu1 ðxÞ; u2 ðxÞ; :::; um ðxÞÞ: ð2Þ
rotated by random angles h 2 ½0; pÞ and u 2 ½0; pÞ (u ¼ 0
for 2D conditions) before using Eq. (4). Depending on
2.1 Rock material with layered features ui , the point ðxÞ is in the ith inclusion domain for ui < 0,
in the matrix domain for ui > 0, and on the interface for
To model a rock system with layered features, the y-axis ui ¼ 0 (Duddu, Chopp, Voorhees, & Moran, 2011). Vari-
is chosen to be in vertical direction parallel to gravity. The ous mathematical functions can be used to describe the
material properties are assumed to vary randomly along transition between two different materials, as discussed by
the y-direction. The level set function for the layered rock Li et al. (2017).
material is: Figure 2 shows six cubic rock samples with random dis-
ui ðxÞ ¼ pprop i ðy Þ; y bottom < y < y top ; ð3Þ tributions of hard inclusions. The number of inclusions m
varies in the range of 4 300. According to the figure, when
where pprop i ðy Þ stands for the properties of the ith material LSM is used to control the mesh generation, the meshes
at the y-location. Various material properties can be con- can be kept equal for different numbers of inclusions.
sidered such as the elastic modulus ðEÞ, Poisson’s ratio
ðvÞ, permeability ðjÞ, and tensile strength ðP t Þ. The varia- 2.3 Rock material with random variation in properties
tions in pprop i ðy Þ can be obtained from in situ tests or following a Weibull distribution
numerically generated according to the expectation and
chosen probability distribution (e.g., a uniform distribu- Regarding a heterogeneous rock formation with ran-
tion, normal distribution, or Weibull distribution). dom spatial variations in the material properties, when
We now choose the 3D rock material as an example. the Weibull distribution is used to describe the distribution
Figure 1 shows the distribution of the material properties of the material properties (Wong, Wong, Chau, & Tang,
along the depth in y-direction in the FEM. The values of 2006), the level set function can be expressed as:
the elastic modulus E are extracted from field test results. 1=x
Moreover, the mesh quality is controlled using the LSM. ui ðxÞ ¼ Kf1n½uðxÞg ; 0 < u 6 1; ð5Þ
where uðxÞ is a random number following the uniform dis-
tribution with the variable being the location vector x; K is
a scale parameter, i.e., the mean (or the expectation) of the
corresponding material property, and x is the shape
parameter termed as ‘‘homogeneity index”. When
x ! þ1, the rock material becomes homogeneous. In a
FEM modeling procedure, the random number u is gener-
ated at and the corresponding material properties are
assigned to different integration points of the elements.
Consider the distribution of the elastic modulus E as an
example. Regarding a 2D plain strain problem with expec-
tation K ¼ 17 GPa, for variations in the homogeneity index
x from 1.5 to þ1, the variation in E in a square rock sam-
ple with lengths Lx ¼ Ly ¼ 100 mm is illustrated in Figs. 3
(a)–(i). The degree of the material heterogeneity decreases
with increasing shape parameter x. Figure. 4 further pre-
sents the upper and lower bounds of E for different cases.
Fig. 1. 3D rock sample with layered features. For a reduced value of x, the difference between the upper

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
4 M. Li et al. / Underground Space xxx (xxxx) xxx

Fig. 2. 3D rock samples with randomly distributed hard inclusions.

Fig. 3. Variations in rock properties following Weibull distribution (2D problem).

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
M. Li et al. / Underground Space xxx (xxxx) xxx 5

Based on the smeared-crack method, the PHF model


was developed for fully-coupled simulations of hydraulic
fractures considering the pore pressure–solid phase interac-
tion. To describe the fluid flow in a cracked rock, the per-
meability is expressed as a function of the mean effective
stress r0m , which increases with the crack propagation.
The governing equations of the PHF model are summa-
rized as follows:
rij;j þ qf i ¼ 0 ði; j ¼ 1; 2; 3Þ; ð6Þ
j 2 1 @p @ev
rp¼ a ; ð7Þ
l Q @t @t
1 
eij ¼ ui;j þ uj;i ; ð8Þ
2
0
Fig. 4. Upper and lower bounds of material properties in 2D rock sample rij ¼ kekk dij þ 2Geij ; ð9Þ
with material property following the Weibull distribution.
rij ¼ r0ij þ pdij ; ð10Þ
  
and lower bounds of E increases, which implies a higher jðr; pÞ ¼ bj0 exp n r0m  ap ; ð11Þ
level of heterogeneity. When the homogeneity index is very jq g
large (e.g., x ¼ 9 999), E approaches its mean of K¼ w ; ð12Þ
l
K ¼ 17 GPa.   
Figures 5 and 6 show the distributions of different mate- h ¼ tanh n r0m  P t ; ð13Þ
rial properties generated by the ABAQUS software for 2D 1 þ h UB 1  h
and 3D conditions, respectively. The modeling procedure K¼ K þ K 0; ð14Þ
2 2
with the ABAQUS software will be presented in the next
section. in which k and G are the Lamé constants of the materials,
dij the Kronecker delta, q the density of the rock mixture,
3 PHF model coupled with LSM and ui and qf i are displacement and unit body force in
direction xi , respectively; rij , r0ij , and p are the total stress
This section provides a brief outline of the coupled tensor, effective stress tensor, and fluid pressure, respec-
PHF–LSM model for the simulation of hydraulic fractur- tively; eij and ev ¼ ui;i are the strain tensor and volumetric
ing in rocks. More details can be found in Li et al. (2017). strain, respectively. The variables a and Q1 are the Biot’s

Fig. 5. Distribution of elastic modulus in 2D rock samples following the Weibull distribution with K ¼ 17 GPa and x varying from 1.5 to 100.

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
6 M. Li et al. / Underground Space xxx (xxxx) xxx

Fig. 6. 3D rock samples with material properties following the Weibull distribution in ABAQUS.

coefficient and storage coefficient, respectively. The where S wi is the irreducible (minimal) degree of water satu-
hydraulic conductivity K is correlated with the intrinsic ration, S orw the residual oil saturation after flooding with
permeability j via the viscosity l and density qw of the pore water, k r0 the end point of the water relative permeability
fluid, with g being the acceleration of gravity [Equation. (considered as a material constant), and N w a constant.
(12)]. Eq. (14) is introduced to describe the dependency In addition, the degree of the water saturation is expressed
of the hydraulic conductivity on the mean stress and the as a function of the porosity U:
damage coefficient n, and with K 0 and K UB being the U0 ð1  UÞS w0 þ U  U0
hydraulic conductivity of the intact and fully-damaged Sw ¼ ; ð16Þ
Uð1  UÞ
rock, respectively. h is a measure that describes the mean
effective stress r0m relative to the tensile strength of the rock. where S w0 is the initial degree of the water saturation and
The post-cracking behavior and failure criterion of the U0 the initial porosity. In this study, we assume
solid phase are described by the Mohr–Coulomb plasticity N w ¼ 3:66, S wi ¼ S orw ¼ 0:15, and k r0 ¼ 0:5.
and piece-wise linear hardening law. Regarding a two-phase flow in porous rock, the relative
The relation between the relative permeability k r and permeability k ri for phase i is defined as k ri ¼ ji =j, in which
degree of water saturation S w is described by the Corey cor- ji stands for the intrinsic permeability of phase i, and j
relation (Corey, 1954; Brooks & Corey, 1964): denotes the permeability of the rock with a single-phase
 N w flow. Equation. (15) is used to evaluate the equivalent rel-
S w  S wi ative permeability of the fluid mixture of the original fluid
k r ¼ k r0 ; ð15Þ
1  S wi  S orw and injection fluid. In particular, the hydraulic conductivity

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
M. Li et al. / Underground Space xxx (xxxx) xxx 7

K calculated from Eqs.  (13) and (14) is updated  to According to the elasticity theory and under plane-
k r K=lmix , where lmix ¼ l0 S w0 U0 þ linj ðS w U  S w0 U0 Þ =S w U strain conditions, if no initial pore pressure exists in the
is the mixture viscosity of the fluid in the fracture zone with rock, the breakdown pressure pb is (Hubbert & Willis,
l0 and linj being the viscosity of the original and injection 1957):
fluids, respectively. In the engineering practice, the values pb ¼ 3rmin  rmax þ rT ; ð17Þ
of linj =l0 can vary in a certain range, depending on the
where rT is the tensile strength of the rock, and rmin and
property of the injection fluid. In this study, linj =l0 ¼ 100
rmax are the minimum and maximum horizontal in situ
is adopted in the simulations. stresses, respectively. If there is an initial pore pressure pp
The coupled PHF-LSM model is implemented into the
in an impermeable elastic rock, the breakdown pressure
commercial software package ABAQUS. Figure. 7 presents
is determined as:
the flow chart for the coupled PHF–LSM model. Within
the ABAQUS framework, the material property data that pb ¼ 3rmin  rmax þ rT  /pp ; ð18Þ
are transformed from the LSM to the PHF model are trea-
with / being the porosity of the rock. For a porous and
ted as state variables. The following sections will demon-
permeable elastic rock, the breakdown pressure is affected
strate the performances of the coupled PHF–LSM model
by the elastic properties of the rock (Schmitt & Zoback,
in numerical simulations of hydraulic fracturing in different
1989):
rock types.
3rmin  rmax þ rT  app ð1  2vÞ=ð1  vÞ
pb ¼ ; ð19Þ
4 Breakdown pressure of hydraulic fracture in homogeneous 1 þ /  að1  2vÞ=ð1  vÞ
rock
in which v is Poisson’s ratio for a dry rock and a is Biot’s
poroelastic parameter (a 2 ½1; 0). The above analytical
We first present a verification of the PHF model in
solutions indicate that the breakdown pressure is primarily
which the characteristics of a hydraulic fracture in a homo-
influenced by the stress state, pore pressure, Poisson’s ratio,
geneous rock sample are examined and the breakdown
Biot’s poroelastic parameter, and porosity of the rock.
pressure is compared with the theoretical solutions.
When determining the material properties, Biot’s poroe-
In general, the difference between the initiation and
lastic parameter a is usually related to the porosity. Various
breakdown pressures is affected by the in situ stresses, rock
relations have been proposed, for example, by Geertsma
properties, borehole size, pump rate, and viscosity of the
and Smit (1961), Krief, Garat, Stellingwerff, and Ventre
injection fluid (Yew & Weng, 2014). Regarding hydraulic
(1990), Nur (1992), and Ma (2008) via the following
fractures in large boreholes with a fluid of low viscosity,
functions:
the two pressures are approximately equal (Yew & Weng,
2014). a ¼ 1  1=ð1 þ 50/Þ; ð20Þ

Fig. 7. Flow chart of coupled PHF–LSM model.

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
8 M. Li et al. / Underground Space xxx (xxxx) xxx

3=ð1/Þ
a ¼ 1  ð1  /Þ ; ð21Þ

/=/c ; / 6 /c
a¼ ; ð22Þ
1 ; / > /c
 n
ð/=/c Þ ; n ¼ 2=3 ; / 6 /c
a¼ ; ð23Þ
1 ; / > /c
where /c is the critical porosity of rock, and its typical val-
ues can be found in Nur, Mavko, Dvorkin, and Galmudi
(1998). Equations. (20)–(22) are suitable for sandstone,
and Eq. (23) is applicable to various rock types. Figure. 8
shows the variation in a for /c ¼ 0:4 when the porosity /
varies from 0 to 0.3 according to Eqs. (20)–(23).
To demonstrate the influence of the porosity on the
breakdown pressure, we examine the hydraulic fracture in
a square rock sample subjected to initial stresses of
rx ¼ 60 MPa and ry ¼ 70 MPa, and a pore pressure of
pp ¼ 50 MPa (Fig. 9). The tensile strength of the rock is Fig. 9. Homogeneous rock sample with one point injection.
rT ¼ 7:59 MPa, and Poisson’s ratio is 0.228. The porosity
is assumed to vary in the range of 0–0.3, and Eq. (23) is
used to determine Biot’s poroelastic parameter with critical
porosities of /c ¼ 0.3, 0.4, 0.6, and 0.8. Figure. 10 presents
the variations in the theoretical values of the breakdown
pressure with the porosity / according to Eqs. (17) and
(19). As one might expect, the breakdown pressure is
affected by both porosity and critical porosity of the rock.
For a given rock porosity, the higher the critical porosity,
the lower the Biot’s poroelastic parameter, which in turn
results in a higher breakdown pressure. An impermeable
rock has the lowest breakdown pressure.
As an example for the use of the PHF model, a FEM
simulation is performed to simulate the hydraulic fractur-
ing process of the homogeneous rock sample in Fig. 9.
The initial porosity of the sample is set to / ¼ 0:113, and
the fracturing fluid is injected at a rate of 0.005 mm/s into Fig. 10. Dependency of breakdown pressure pb on porosity / and critical
the injection line. A breakdown pressure of pb ¼ 100 MPa porosity /c of rock.
is obtained from the FEM simulation using the PHF
model. The calculated breakdown pressures are 116 MPa,
111 MPa, and 125 MPa for Eqs. (17), (18), and (19),
respectively.
Although the breakdown pressure obtained with the
PHF model approximates the analytical results, the frac-

Fig. 11. Characteristics of fracture in layered rock.

ture mechanisms and influencing factors are different in


the PHF and theoretical models. In the theoretical model,
the breakdown pressure is influenced by the stress state,
Biot’s poroelastic parameter, porosity, and critical porosity
of the rock. The PHF method considers the effect of the
rock porosity, stress state, stress level, and dependency of
Fig. 8. Relation between Biot’s poroelastic parameter and porosity. the permeability on the stress states and fracture propaga-

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
M. Li et al. / Underground Space xxx (xxxx) xxx 9

Fig. 12. Heterogeneous rock with layered features.

Table 1
Material properties of layered rock.
Properties Injection layer Hard layer
Elastic modulus (MPa) 17 733.3 80 000.0
Density (kg/m3) 2 432.1 2 432.1
Poisson’s ratio 0.228 0.200
Hydraulic conductivity (mm/s) 1:33  108 1  1018
Tensile strength (MPa) 3.964 150
Porosity 0.05 0.05
Friction angle (°) 30 30
Dilation angle (°) 10 10

Table 2
Case study for layered rock.
Features Values
hi (mm) 2, 4, 6, 8, 12, 90
hi (°) 0, 15, 30
Hard-layer tensile strength ptmax (MPa) 10, 15, 150

tion. Please note that the PHF model does not adopt the
concept of the critical porosity.

5 Characteristics of hydraulic fracture in heterogeneous rock


Fig. 13. FEM meshes in neighborhood of injection area for homogeneous
5.1 Propagation of fracture induced by fracking in layered and layered rocks.
rock

Regarding a hydraulic fracture in a layered rock system,


previous studies (e.g., Li et al., 2016) have shown that the
fracture propagation is affected by its incident height (hi )
and incident angle (hi ), which are defined in Fig. 11. A ser-
ies of FEM simulations using the PHF model is conducted
to investigate the effect of hi and hi by modeling the fracture
propagation in a layered rock sample taken from a bore-
hole at a depth of 3 060 m. The water pressure pw , vertical
stress rv , and horizontal stress rh at this depth are deter-
mined as 49 MPa, 71 MPa, and 60 MPa, respectively.
Figure. 12 shows the dimension and layer thickness of
the sample. To capture the details of the fracture propaga-
tion, a fine mesh zone is used in the range from 50 mm to
50 mm in x-direction (horizontal) and from 200 mm to
200 mm in y-direction (vertical). The Mohr–Coulomb Fig. 14. Injection pressure versus injection time for hi ¼ 0 , hi ¼ 12 mm,
model is used as criterion for the shearing-induced failure. and ptmax ¼ 10 MPa.

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
10 M. Li et al. / Underground Space xxx (xxxx) xxx

Fig. 15. Characteristics of injection pressure and its dependency on incident height, incident angle, and tensile strength of a hard rock layer. (a) hi = 0,
ptmax = 10 Mpa. (b) hi = 15, ptmax = 10 Mpa. (c) hi = 30, ptmax = 10 Mpa. (d) hi = 0, ptmax = 15 Mpa. (e) hi = 0, ptmax = 150 Mpa. (f) Propagation pore
pressure versus hi .

The location of the fracturing fluid injection is at the mid- neous and layered rock samples in the neighborhood of the
height of the middle layer (Layer 2). The injection rate of injection area. The FEM mesh consists of 278 481 nodes
the fluid is 0:001 mm=s in all simulations. The material and 92 400 elements.
properties and values of hi and hi used in the simulations In general, the hydraulic-fracturing process can be
are listed in Tables 1 and 2. divided into three stages: fracture opening, fracture propa-
Using the modeling method for the rock heterogeneity gation, and fracture closure (Wang, Sun, Au, Yang, &
outlined in Section 2, the rock sample is discretized into Tang, 2009). The initiation and progressive opening pro-
FEM meshes depending on the distribution of the material cesses of a fracture highly depend on the leak-off pressure
properties. Figure. 13 shows the FEM meshes for homoge- and breakdown pressure, which are affected by various fac-

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
M. Li et al. / Underground Space xxx (xxxx) xxx 11

Table 3
Material properties of rock sample with hard inclusions.
Material property Matrix Hard inclusion
Elastic modulus (MPa) 20 000 80 000
Poisson’s ratio 0.26 0.15
Hydraulic conductivity (m/s) 5  109 2  1012
Tensile strength (MPa) 5 150
Friction angle (°) 30 30
Dilation angle (°) 10 10
Porosity 0.05 0

Figure 15 presents the time history of the injection pres-


sure during 300 s of the fluid injection for different incident
angles hi , incident heights hi , and tensile strengths of the
hard rock layer ðptmax Þ. Based on Figs. 15(a)–(e), the injec-
tion pressure–time history curves are compared for the dif-
ferent layered parameters and material properties in
Table 1. With increasing incident height hi , the injection
pressure necessary to maintain the fracture propagation
decreases. However, the variations in the incident angle hi
Fig. 16. Characteristics of inclusions affecting propagation of hydrauli-
cally induced fracture.
and tensile strength ptmax of the hard rock layer only have
a minor effect on the injection pressure during the fracture
tors including the stress state, porosity, viscosity of the propagation. The propagation pressure increases only for a
fracturing fluid, and injection rate (Yew & Weng, 2014). small hi and very high ptmax [Fig. 15(f)]. The breakdown
Figure. 14 presents the variation in the injection pressure pressure remains approximately constant in all cases, which
with the injection time obtained from the FEM simulations implies that the breakdown pressure primarily depends on
using the PHF model for hi ¼ 0, hi ¼ 12 mm, and the material properties of the layer, into which the fractur-
ptmax ¼ 10 MPa. Herein, ptmax is referred to as the tensile ing fluid is injected.
strength of the hard rock layer. Based on the time history
of the injection pressure, the breakdown pressure and pres- 5.2 Propagation of hydraulic fracture in rock with distributed
sure required to maintain the fracture propagation and the hard inclusions
fracture opening and fracture propagation stages can be
explicitly identified. More specifically, the peak injection This section summarizes the findings obtained from the
pressure on the curve indicates the breakdown pressure, FEM simulations regarding the breakdown pressure and
whereas the fracture propagation stage corresponds to propagation pressure as well as their dependencies on the
the steady state of the injection pressure. inclusion characteristics. These include the incident height

Fig. 17. Rock sample with two inclusions near the injection point.

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
12 M. Li et al. / Underground Space xxx (xxxx) xxx

Fig. 18. Breakdown pressure in rock sample with two inclusions (h ¼ 0 ).

Fig. 19. Variations in incident angle and breakdown pressure with orientation angle a of elliptical inclusions.

Table 4
Weibull distribution parameters for different material properties of heterogeneous rock.
Property K (2D) x (2D) K (3D) x(3D)
Elastic modulus (GPa) 10 1:5  100 30 4
Poisson’s ratio 0.33 1:5  100 0.3 15
Hydraulic conductivity (m/s) 5  108 1:5  100 6  109 0.8
Tensile strength (MPa) 13.8 1:5  100 5.0 5

(h), dimension of inclusions (a and b are the minor and tal stresses are 40 MPa , 70 MPa, and 50 MPa, respec-
major principal radii of the inclusion, respectively), inci- tively. The material properties of the rock matrix and
dent angle h, and orientation of inclusions (a) (Fig. 16). hard inclusions are listed in Table 3.
Figure 17 illustrates the rock sample with two inclusions Figure 18 shows the variation in the breakdown pressure
located on both sides of the injection area and the corre- with the aspect ratio of the elliptical inclusions when the
sponding FEM mesh. The inclusions are assumed to be incident angle is h ¼ 0 for different injection heights h (dis-
elliptical with one principal axis fixed to 10 mm and the tance from the injection point to the inclusion). In this sim-
other varying from 1 mm to 30 mm , which corresponds ulation series, the principal axes of the inclusions are either
to an aspect ratio of b/a = 1–3. The orientation angle a of horizontal or vertical, with the vertical axis being set to
the inclusion varies from 0 to 90 , with h varying from ry ¼ 10 mm . When the inclusions are close to the injection
1 mm to 10 mm . The rock sample is an 1 m  1 m area. position (e.g., h ¼ 3 mm), the breakdown pressure
The injection rate is 0:1 m=s , and the injection area with a increases significantly with increasing particle size
width of w ¼ 1 mm is located at the center of the rock (expressed by b=a ratio). As the value of h increases, the
sample. The initial pore pressure and vertical and horizon- breakdown pressure gradually decreases. However, the

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
M. Li et al. / Underground Space xxx (xxxx) xxx 13

Fig. 20. Fracture zone propagation in 2D heterogeneous rock sample with shape parameter of x ¼ 10.

influence of h on the breakdown pressure tends to vanish dramatically for a ! 90 . The influence of the inclusion
for high b=a ratios, as can be observed in the curves corre- orientation on the breakdown pressure is related to the dif-
sponding to b=a ¼ 7 and 10 in Fig. 18(a). The influence of a ferent local stress distributions in front of the fracture near
hard inclusion size and incident height on the breakdown the inclusion for different a values.
pressure can also be observed in Fig. 18(b), which shows
the variation in the breakdown pressure with the incident 5.3 Propagation of hydraulic fracture in heterogeneous rock
height for circular inclusions of different sizes. The results with random variations in material properties
in Fig. 17 confirm that the influences of both inclusion
dimension and incident height tend to vanish when the inci- In this section, we discuss the characteristics of a
dent height exceeds a certain limit. hydraulic fracture in a heterogeneous rock formation with
In general, the orientation of the inclusions varies ran- random spatial variations in the material properties based
domly with a (see Fig. 16) in the range of ½0 ; 90  and on the FEM simulations and coupled PHF–LSM model.
the incident angle h being a function of a, b, and a. Figure. The material properties that follow the Weibull distribu-
19 shows the relation between h and a for varying particle tion (Weibull, 1939) are generated using the method pre-
shapes when a ¼ 1  9 mm and the longer axis of the sented in Section 2. Table 4 lists the scale parameter (i.e.,
inclusion is set to b ¼ 10 mm. Consider a reference state the expectation) K and shape parameter (i.e., the homo-
in which the longer axis of the inclusion is horizontal geneity index) x of various material properties used in
(i.e., a ¼ 0 ). For a given incident height and gradually the FEM simulations for both 2D and 3D.
rotating inclusion, the rotation of the inclusion does not In the simulation of the hydraulic-fracturing processes
have a significant effect on the breakdown pressure for in the rock samples of Figs. 1 and 2, the fracturing fluid
a < 60 . When the longer axis of the inclusion approaches is added at a rate of 0:001 mm=s to the injection area
the vertical position, the breakdown pressure increases. For for 1 h in the 2D simulations. To obtain the representative
a chosen incident height, the breakdown pressure increases characteristics of the hydraulic-fracture propagation with

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
14 M. Li et al. / Underground Space xxx (xxxx) xxx

Fig. 21. Geometry of fracture zone for T ¼ 600 s in 2D heterogeneous rock samples when x varies from 1.5 to 100.

T ¼ 525–3 600 s for x ¼ 10. The fracture zone has irregu-


lar outer boundaries, which are affected by the spatial vari-
ations in the material properties. The direction of the
fracture propagation is generally perpendicular to the
direction of the minor principle stress, as one might expect.
Figure 21 compares the shapes of the crack zones for an
injection time of T ¼ 600 s when the shape parameter x
varies from 1.5 to 100. Since x characterizes the hetero-
geneity of the material properties, one might conclude that
the degree of the material heterogeneity has only a minor
effect on the evolution of the fracture zone, its height,
and propagation direction. However, the outer boundary
of the fracture zone tends to become coarser with decreas-
ing x. Please note that a lower x implies a more severe
heterogeneity in the rock.
Fig. 22. Variation in breakdown pressure for different heterogeneity Figure 22 presents the variations in the breakdown pres-
levels. sure with shape parameter x. For each heterogeneity level
corresponding to a specific x, three to eight FEM simula-
initial stresses and a pore pressure of rh ¼ 20 MPa, tions are performed using randomly generated distribu-
rv ¼ 30 MPa, and pw ¼ 10 MPa, respectively, more than tions of material properties. The mean values of each
30 simulations are carried out with different spatial distri- material property are summarized in Table 4 for all simu-
butions of the material properties with the parameters in lations. The lower the shape parameter, the more scattered
Table 4. the calculated breakdown pressure. When the heterogene-
Figure 20 shows the propagation of the fracture zones ity index x exceeds a certain level, the breakdown pressure
during the hydraulic fracturing for injection times of approaches a constant. Because the rock becomes less

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
M. Li et al. / Underground Space xxx (xxxx) xxx 15

Fig. 23. Propagation of fracture zone in heterogeneous rock sample: 3D simulations.

heterogeneous with increasing x, the results in Fig. 22 including layered rock, rock with distributed inclusions,
reveal that the heterogeneity level does not have a signifi- and rock with random distributions of its material proper-
cant effect on the breakdown pressure. These observations ties. Moreover, the PHF model is coupled with LSM and
indirectly indicate that the coupled PHF–LSM model can implemented into a FEM software package to simulate
capture the salient features of hydraulic fracturing in a the hydraulic-fracturing process in rock formations. With
heterogeneous rock. a constant mesh used to discretize the physical domain of
Regarding the 3D hydraulic-fracture simulations, the rock formation, it is demonstrated that the coupled
Fig. 23 shows the evolution of the fracture zones for differ- PHF–LSM model can simulate hydraulic-fracturing prob-
ent injection times and an initial stress state of lems for various heterogeneous rocks. The main results of
rv ¼ 70:0 MPa, rhmax ¼ 65:0 MPa, and rhmin ¼ the FEM simulations are summarized as follows:
58:8 MPa. The initial pore pressure is pw ¼ 49:0 MPa.
The spatial variation in the material properties is described (1) Regarding layered rocks, the characteristics of a
by the Weibull distributions and parameters listed in hydraulic fracture are influenced by the incident
Table 4. The fracturing fluid is injected at a rate of height and incident angle of the crack. An increasing
0:005 mm=s into the injection area for 5 400 s. Similar incident height tends to decrease the injection pres-
to the 2D cases, the direction of the fracture propagation sure required to maintain the fracture propagation.
is perpendicular to the minor principle stress, regardless The variations in the incident angle and tensile
of the heterogeneity of the material properties. For both strength of the non-fracking layers have a minor
2D and 3D conditions, the development of a fracture zone influence on the propagation pressure. The injection
is mainly controlled by the stress state, whereas the hetero- pressure during the fracture propagation increases
geneity of the rock material causes the outer boundary of only when the injection point is very close to the
the fracture zone to be more irregular (Fig. 23). hard rock layer (i.e., very small incident height) or
when the fluid-injection layer has a high tensile
6 Conclusions strength.
(2) Regarding a rock with distributed hard inclusions,
This paper presents an approach to simulate hydraulic large inclusions can cause changes in the breakdown
fractures in heterogeneous rocks. An extended LSM is pressure under certain conditions. The breakdown
introduced to model the spatial variations in the material pressure increases significantly if the inclusion is close
properties for different types of heterogeneous rocks, to the injection position and the size of the inclusion

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
16 M. Li et al. / Underground Space xxx (xxxx) xxx

is sufficiently large. However, inclusions far from the Geertsma, J., & de Klerk, F. (1969). A rapid method of predicting width
and extent of hydraulically induced fractures. Journal of Petroleum
injection point do not affect the breakdown pressure. Technology, 21(12), 1571–1581.
(3) Regarding a heterogeneous rock with random spatial Geertsma, J., & Smit, D. C. (1961). Some aspects of elastic wave
variations in the material properties with a Weibull propagation in fluid saturated porous solids. Geophysics, 26(2),
169–181.
distribution, the shape parameter (i.e., the hetero- Hu, Y., Chen, G., Cheng, W., & Yang, Z. (2013). Simulation of hydraulic
geneity index) dominates the discreteness of the fracturing in rock mass using a smeared crack model. Computers and
FEM simulation results. A higher shape parameter Structures, 137(6), 72–77.
Hubbert, K. M., & Willis, D. G. (1957). Mechanics of hydraulic
leads to a lower level of the material heterogeneity fracturing. Transactions of Society of Petroleum Engineers of AIME,
and hence smaller discreteness of the modeling AIME, 210, 153–166.
results. The direction of the fracture propagation is Khristianovitch, S. A., & Zheltov, Y. P. (1955). Formation of vertical
fractures by means of highly viscous fluids. In Proc. 4th world
controlled by the stress state and is perpendicular to petroleum congress. Vol. 2. Rome, Italy, pp. 579–586.
the minor principle stress, regardless of the material Krief, M., Garat, J., Stellingwerff, J., & Ventre, J. (1990). A petrophysical
heterogeneity. interpretation using the velocities of p and s wave s (full-waveform
sonic). The Log Analyst, 31, 355–369.
Lecampion, B. (2009). An extended finite element method for hydraulic
fracture problems. Communications in Numerical Methods in Engineer-
Acknowledgments ing, 25(2), 121–133.
Li, M., Guo, P., Stolle, D., & Li, L. (2017). Modeling method for a rock
matrix with inclusions distributed and hydraulic fracturing character-
Drs. Peijun Guo and Dieter Stolle gratefully acknowl- istics. Journal of Petroleum Science and Engineering, 157, 409–421.
edge the partial funding provide by the Natural Sciences Li, M., Guo, P., Stolle, D. F. E., & Liang, L. (2016). Development of
hydraulic fracture zone in heterogeneous material based on smeared
and Engineering Research Council of Canada (NSERC). crack method. Journal of Natural Gas Science and Engineering, 35,
Drs. Ming Li and Li Liang are financially supported by 761–774.
the Fundamental Research Funds for the Central Univer- Li, M., Li, L., Guo, P., & Li, X. (2015). Mesh sensitivity analysis of the
solution to hydraulic fracture problems based on a smeared crack
sities (Grant No. N170104024) and General Program of model. Journal of Northeastern University: Natural Science, 36(9),
National Natural Science Foundation of China (Grant 1337–1341.
No. 51474048). Ma, G. (2008). Experimental investigation into Biot’s coefficient and rock
elastic moduli. Oil & Gas Geology, 29(1), 135–140 (In Chinese).
Markov, A., Jarillo, G. R., & Markov, M. (2014). Elastic properties of
Conflict of interest rocks containing oriented systems of ellipsoidal inclusions. Journal of
Applied Geophysics, 103(1), 114–120.
Mohammadnejad, T., & Khoei, A. R. (2013). An extended finite element
The authors declare that there are no conflicts of interest method for hydraulic fracture propagation in deformable porous
related to the subject matter or materials discussed in this media with the cohesive crack model. Finite Elements in Analysis and
paper. Design, 73, 77–95.
Nordgren, R. P. (1972). Propagation of a vertical hydraulic fracture.
Society of Petroleum Engineers Journal, 12, 306–314.
References Nur, A. (1992). Critical porosity and the seismic velocities in rocks. Eos
Transactions American Geophysical Union, 73(1), 43–66.
Adachi, J., Sie brits, E., Peirce, A., & Desroches, J. (2007). Computer Nur, A., Mavko, G., Dvorkin, J., & Galmudi, D. (1998). Critical porosity:
simulation of hydraulic fractures. International Journal of Rock A key to relating physical properties to porosity in rock. The Leading
Mechanics and Mining Sciences, 44(5), 739–757. Edge, 17(3), 357–362.
Barenblatt, G. I. (1962). The mathematical theory of equilibrium cracks in Osher, S., & Sethian, J. A. (1988). Fronts propagating with curvature-
brittle fracture. Advances in Applied Mechanics, 7, 55–129. dependent speed: Algorithms based on Hamilton-Jacobi formulations.
Bažant, Z. P., & Lin, F. B. (1988). Nonlocal smeared cracking model for Journal of Computational Physics, 79, 1249.
concrete fracture. Journal of Structural Engineering, 114(11), Perkins, T. K., & Kern, L. R. (1961). Widths of hydraulic fractures.
2493–2510. Journal of Petroleum Technology, 13(9), 937–949.
Bažant, Z. P., & Oh, B. H. (1983). Crack band theory for fracture of Rahman, M. M., & Rahman, M. K. (2010). A review of hydraulic fracture
concrete. Matériaux et Constructions, 16, 155–177. models and development of an improved pseudo-3d model for
de Borst, R. (1987). Smeared cracking, plasticity, creep, and thermal stimulating tight oil/gas sand. Energy Sources, Part A: Recovery,
loading – A unified approach. Computer Methods in Applied Mechanics Utilization, and Environmental Effects, 32(15), 1416–1436.
and Engineering, 62(1), 89–110. Remmers, J. J. C., de Borst, R., & Needleman, A. (2003). A cohesive
Brooks, R. H., & Corey, A. T. (1964). Hydraulic properties of porous segments method for the simulation of crack growth. Computational
media. Hydrology Papers 3. Fort Collins: Colorado State University Mechanics, 31(1–2), 69–77.
(pp. 27). . Sarris, E., & Papanastasiou, P. (2011). The influence of the cohesive
Corey, A. T. (1954). The interrelation between gas and oil relative process zone in hydraulic fracturing modelling. International Journal of
permeabilities. Producers, 19(1), 38–41. Fracture, 167(1), 33–45.
Duddu, R., Chopp, D. L., Voorhees, P., & Moran, B. (2011). Diffusional Schmitt, D. R., & Zoback, M. D. (1989). Poroelastic effects in the
evolution of precipitates in elastic media using the extended finite determination of the maximum horizontal principal stress in hydraulic
element and the level set methods. Journal of Computational Physics, fracturing tests: A proposed breakdown equation employing a mod-
230(4), 1249–1264. ified effective stress relation for tensile failure. International Journal of
Dugdale, D. S. (1960). Yielding of steel sheets containing slits. Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 26(6),
the Mechanics and Physics of Solids, 8(2), 100–104. 499–506.
Economides, M. J., & Nolte, K. G. (1989). Reservoir stimulation (2nd ed.) Settari, A., & Cleary, M. P. (1986). Development and testing of a pseudo-
. Houston, TX: Schlumberger Educational Services. three-dimensional model of hydraulic fracture geometry. SPE Produc-
Fatahi, H., Hossain, M. M., Fallahzadeh, S. H., & Mostofi, M. (2016). tion Engineering, 1(6), 449–466.
Numerical simulation for the determination of hydraulic fracture Simonson, E. R., Abou-Sayed, A. S., & Clifton, R. J. (1978). Containment
initiation and breakdown pressure using distinct element method. of massive hydraulic fractures. Society of Petroleum Engineers Journal,
Journal of Petroleum Technology, 33(1), 1219–1232. 18, 27–32.

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005
M. Li et al. / Underground Space xxx (xxxx) xxx 17

Sukumar, N., Chopp, D. L., Moës, N., & Belytschko, T. (2001). Modeling Wong, T. F., Wong, R. H. C., Chau, K. T., & Tang, C. A. (2006).
holes and inclusions by level sets in the extended finite-element method. Microcrack statistics, Weibull distribution and micromechanical
Computer Methods in Applied Mechanics and Engineering, 190(46–47), modeling of compressive failure in rock. Mechanics of Materials, 38
6183–6200. (7), 664–681.
Valko, P., & Economides, M. J. (1995). Hydraulic fracture mechanics. Wu, Z., & Wong, L. N. Y. (2013). Modeling cracking behavior of rock
Chichester, England: John Wiley & Sons. mass containing inclusions using the enriched numerical manifold
Wang, S., Sun, L., Au, A., Yang, T., & Tang, C. (2009). 2D-numerical method. Engineering Geology, 162(1), 1–13.
analysis of hydraulic fracturing in heterogeneous geo-materials. Yew, C. H., & Weng, X. (2014). Mechanics of hydraulic fracturing (2nd
Construction and Building Materials, 23(1), 2196–2206. ed.). Gulf Professional Publishing.
Wangens, M. (2011). Finite element modeling of hydraulic fracturing on a Zhuang, X., Augarde, C. E., & Mathisen, K. M. (2012). Fracture
reservoir scale in 2D. Journal of Petroleum Science and Engineering, 77 modeling using meshless methods and level sets in 3D: Framework and
(3), 274–285. modeling. International Journal for Numerical Methods in Engineering,
Warpinski, N. R., & Smith, M. B. (1989). Rock mechanics and fracture 92, 969–998.
geometry. Recent advance s in hydraulic fracturing. Richardson, TX: Zhang, F. S., Damjanac, B., & Huang, H. Y. (2013). Coupled discrete
SPE. element modeling of fluid injection into dense granular media. Journal
Weibull, W. (1939). A statistical theory of the strength of materials. of Geophysical Research: Solid Earth, 118(6), 2703–2722.
Proceedings, The Royal Swedish of Institute for Engineering Research,
151(15), 1–45.

Please cite this article as: M. Li, P. Guo, D. F. E. Stolle et al., Modeling hydraulic fracture in heterogeneous rock materials using permeability-based
hydraulic fracture model, Underground Space, https://doi.org/10.1016/j.undsp.2018.12.005

Das könnte Ihnen auch gefallen