See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318250972

Effect of Poroelasticity on Hydraulic Fracture Interactions

Conference Paper · July 2017

DOI: 10.1061/9780784480779.249

CITATIONS READS
16 1,997

4 authors:

Tomoya Usui Saeed Salimzadeh

Cosmo Energy Exploration & Production Co., Ltd The Commonwealth Scientific and Industrial Research Organisation
5 PUBLICATIONS 115 CITATIONS 61 PUBLICATIONS 1,458 CITATIONS

SEE PROFILE SEE PROFILE

Adriana Paluszny R. W. Zimmerman

Imperial College London Imperial College London
62 PUBLICATIONS 4,975 CITATIONS 263 PUBLICATIONS 17,372 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Saeed Salimzadeh on 06 November 2017.

The user has requested enhancement of the downloaded file.

 Effect of Poroelasticity on Hydraulic Fracture Interactions

Tomoya Usui,1 Saeed Salimzadeh,2 Adriana Paluszny3 and Robert W. Zimmerman4

1
Cosmo Energy Exploration and Production Co., Tokyo, Japan; e-mail: tomoya_usui@cosmo-
oil.co.jp
2
Centre for Oil and Gas - DTU, Lyngby, Denmark; e-mail: [email protected]
3
Department of Earth Science & Engineering, Imperial College London, London, United
Kingdom; e-mail: [email protected]
4
Department of Earth Science & Engineering, Imperial College London, London, United
Kingdom; e-mail: [email protected]

ABSTRACT

This study investigates, by performing finite element-based simulations, the influence of fluid
leak-off and poroelasticity on growth of multiple hydraulic fractures that initiate from a single
horizontal well. In this research, poroelastic deformation of the matrix is coupled with fluid flow
in the fractures, and fluid flow in the rock matrix, in three dimensions. Effects of the fluid leakoff
and poroelasticity on the propagation of the neighboring fractures are studied by varying the
matrix permeability, and the Biot coefficient. Simulation results show that the stress induced by
the opening of the fractures, and the stress induced by the fluid leak-off, each have the effect of
locally altering the magnitudes and orientations of the principal stresses, hence altering the
propagation direction of the fractures. The stress induced by the opening of the fractures tends to
propagate both of the fractures away from each other in a curved trajectory, whereas the effects
of fluid leak-off and poroelasticity (i.e., a higher Biot coefficient) tend to straighten the curved
trajectory.

INTRODUCTION

Hydraulic fracturing is widely used in the oil and gas industry to stimulate production from a
reservoir. Despite improvements in diagnostic technology, such as micro-seismic monitoring, it
is still difficult to measure the actual fracture geometry directly; hence, modelling of hydraulic
fracturing propagation remains important. The methods for deriving the fracture geometries can
be divided into two categories, analytical methods and numerical methods.
As analytical methods, three classic fracture models are widely known: the Perkins-Kern-
Nordgren (PKN) (Perkins and Kern 1961; Nordgren 1972), the Khristianovic-Geertsma-de Klerk
(KGD) (Geertsma and De Klerk 1969) and the penny-shaped or radial models (Abe et al. 1976).
These are 2D plane strain or axisymmetric models which assume some constraints in either
fracture height/length. More recently, fracture propagation has been modelled by incorporating
the flow of fluid inside the fracture, and/or the flow of fluid into the pore space of the formation
matrix, which is called “leakoff” (Savitski and Detournay 2002; Bunger et al. 2005; Kovalyshen
2010; Garagash et al. 2011). These studies derive asymptotic solutions of the fracture geometry
and pressure for different regimes (e.g. viscosity/toughness dominated, storage/leakoff regimes)
for penny-shaped fractures. In the cases where the fracturing fluid leaks off into the formation,
the pore pressure of the matrix is increased, resulting in the dilation of the matrix (Biot 1941).
The dilated matrix induces additional compression on the fracture. This additional compression

1
is called the “back-stress” (Cleary 1980; Kovalyshen 2010).
Advancement in computational power has allowed the modelling of evolving fracture
geometries through numerical simulations. To numerically model the hydro-mechanical
evolution of fluid-driven fracturing with leakoff, it is essential to couple the fluid flow in the
matrix with the mechanical deformation of the rock and the fluid flow inside the fracture (Carrier
and Granet 2012; Salimzadeh and Khalili, 2015). Salimzadeh et al. (2016) define separate flow
models for rock matrix and fractures to capture the hydraulic loadings on the fractures surfaces,
as well as the poroelastic deformations of rock matrix on the mesh-independent, finite element-
based framework. The mesh independent, finite element-based method decouples the crack
propagation and the meshing, and performs each process separately to calculate the non-planar
3D fracture propagation in accurately and efficiently (Paluszny and Zimmerman 2011; Paluszny
et al. 2013).
In many field operations of hydraulic fracturing, multiple parallel transverse fractures are
induced from single horizontal wells to maximise the productivity from each well. Many of the
previous studies which model the growth of multiple hydraulic fractures (Olson and Taleghani
2009; Wu and Olson 2015; Peirce and Bunger 2015) only model the mechanical interaction
among the fractures by assuming a formation matrix where fluid leakoff is negligible, and do not
incorporate the poroelastic effect into their simulation models.
The objective of this study is to numerically model the interaction of multiple hydraulic
fractures initiated from penny-shaped cracks in a horizontal well in 3D, and to evaluate the effect
of poroelasticity on the predicted growth such as the propagation paths or the growth speeds of
the fractures, by using the techniques to build three dimensional fully coupled poroelastic models
(Paluszny and Zimmerman 2011; Salimzadeh et al. 2016).

METHODOLOGY

Rock Deformation and Fluid Flow Model. The stress interacting on the fracture is modelled by
coupling the (a) elasticity equation which describes the mechanical deformation of the fracture,
(b) an equation based on lubrication theory describing the fluid flow in the fracture, and (c) an
equation based on Darcy’s law representing the fluid flow in the matrix. Separate flow models
are defined for the rock matrix and the fractures to accurately capture hydraulic loadings on the
fracture surfaces as well as poroelastic deformations of rock matrix (Salimzadeh and Khalili
2015; Salimzadeh et al. 2016). Each of the governing equations are shown below. When deriving
the governing equations, the fractures are represented by discontinuity c in the domain with
boundary 
(a) Mechanical deformation: ∫ [div(Dε − α𝑝𝑚 I) + F]𝑑Ω − ∫ 𝑝𝑓 n𝑐 𝑑Γ = 0 (1)
Ω Γ𝑐
𝑎𝑓3 𝑎𝑓 𝜕𝑝𝑓 𝜕𝑎𝑓 𝑘𝑚 𝜕𝑝
(b) Fracture fluid flow: ∇∙( ∇𝑝𝑓 ) = ∙ + − (2)
12𝜇𝑓 𝐾𝑓 𝜕𝑡 𝜕𝑡 𝜇𝑓 𝜕n𝑐
(c) Matrix fluid flow:
𝑘𝑚 𝜕(div 𝑢) 𝛼 − 𝜙 𝜕𝑝𝑚 𝑘𝑚 𝜕𝑝 (3)
∫ div [ (𝛻𝑝𝑚 + 𝜌g)] 𝑑𝛺 = ∫ [𝛼 + (𝜙𝑐𝑓 + ) ] 𝑑𝛺 + ∫ 𝑑𝛤
𝛺 𝜇𝑓 𝛺 𝜕𝑡 𝐾𝑠 𝜕𝑡 𝛤𝑐 𝜇𝑓 𝜕n𝑐
where 𝐃 is the drained stiffness, 𝛆 is the strain, 𝛼 is the Biot coefficient (Biot 1941), 𝑝𝑚 is the
fluid pressure in the rock matrix pore, 𝐈 is the second-order identity tensor and 𝐅 is the body
force per unit volume. The strain is related to displacement 𝐮 by 𝛆 = 1⁄2 [∇𝐮 + (∇𝐮)𝑇 ]. The

2
definition of Biot coefficient is 𝛼 = 1 − (𝐾 ⁄𝐾𝑠 ) where 𝐾 is the bulk modulus of the entire
porous rock and 𝐾𝑠 is the bulk modulus of the rock matrix grain excluding the pore space. 𝑎𝑓 is
fracture aperture, 𝑝𝑓 is the fracture fluid pressure, 𝐾𝑓 is the fluid bulk modulus, and 𝑐𝐿 is the
leakoff coefficient. The fracture aperture relates to displacement by 𝑎𝑓 = (𝐮+ − 𝐮− ) ∙ 𝐧𝑐 where
𝐮+ and 𝐮− are the displacements of opposing faces of the fracture and 𝐧𝑐 is the vector
perpendicular to fracture plane. 𝐤 𝐦 is the intrinsic permeability of rock matrix, 𝜇 is the fluid
velocity, 𝜙 is the porosity of the rock matrix. 𝑝𝑚 in (2), (3) is the pressure of the matrix cell
adjacent to the fracture. Following assumptions are set for simplification; (i) the rock matrix is
linear elastic, homogeneous and isotropic (ii) fracture fluid is single phase and Newtonian, and
(iii) effect of gravity is negligible. The mechanical deformation is coupled to the fracture flow
through the aperture 𝑎𝑓 and 𝛆 both being related to displacements, while it is also coupled to the
matrix flow by the Biot coefficient 𝛼. When 𝛼 = 0, the mechanical deformation and the matrix
flow are decoupled, in which case the mechanical loading and the matrix pressure lose relation to
each other. The fracture flow and matrix flow are coupled by mass transfer sharing the leakoff
term. Note that in Equation (3), the leakoff term is only applicable to the matrix cells adjacent to
the fracture, and eliminated from the equation when applied to other cells.

Fracture Propagation Model. As the rock matrix deforms due to more fluid being injected into
the fracture, the stress concentration at the tip of the fracture increases. In this work, the stress
intensity factor (SIF) propagation criterion is used to determine the fracture failure at the tip. SIF
represents the amount of energy concentrated around the fracture tip. It can be related to the J-
Integral, which is the amount of energy released to extend the fracture tip for unit length in the
direction of the crack plane (Paluszny and Zimmerman 2011). The three SIFs are mode I (𝐾𝐼 ) for
opening due to tensile loading, mode II (𝐾𝐼𝐼 ) for in-plane shearing due to sliding, and mode III
(𝐾𝐼𝐼𝐼 ) for out-of-plane shearing due to tearing. The crack grows if and when the equivalent stress
intensity factor 𝐾𝐼𝑒𝑞 , overcomes the material toughness (𝑘𝑖𝑐 ). The equivalent SIF in the direction
of propagation (𝜃𝑝 ) is calculated as (Schöllmann et al., 2002)
1 𝜃𝑝 2 + 4𝐾 2 }
𝐾𝐼𝑒𝑞 = 2 cos ( 2 ) {𝐾𝑐𝑠 + √𝐾𝑐𝑠 𝐼𝐼𝐼 (4)
𝜃𝑝 3
where 𝐾𝑐𝑠 = 𝐾𝐼 cos 2 ( 2 ) − 2 𝐾𝐼𝐼 sin(𝜃𝑝 ), 𝜃𝑝 is the propagation angle, and 𝜓𝑝 is the deflection
angle. The propagation angle (𝜃𝑝 ) and deflection angle (𝜓𝑝 ) are determined using a modified
maximum circumferential stress method that takes into account modal stress intensity factors
under mixed loading (Schöllmann et al., 2002). The SIFs and growth computations are
performed at fifty locations along the fracture tip. Once the geometry is updated after each
growth step, the model is outsourced to an octree volumetric mesher and re-meshing is done on
the entire model. The discretized coupled equations and the fracture propagation model are
implemented as part of the Imperial College Geomechanics toolkit (hereinafter, referred to as
“the simulator”).

SIMULATION RESULTS AND ANALYSIS

Figure 1 shows the schematic diagram of the simulation model with two penny-shaped cracks.
We focus on identifying the effect of fluid leakoff and poroelasticity by altering the permeability
and the Biot coefficient through simulations in which multiple fractures are induced
simultaneously. Table 1 shows the fixed parameters that are used for every simulation presented

3
in this section. To identify the effects listed above, simulations are run for five different input
sets for comparison as shown in

Table 2. The input parameters are based on Carrier and Granet (2012) and also on correlation
from previous studies which based their parameters on the Barnett shale (Olson and Taleghani
2009; Wu and Olson 2015). From previous studies (Olson and Taleghani 2009; Wu and Olson
2015), it is known that parallel fractures propagate away from each other. This is because the
stress induced by the opening of the hydraulic fractures locally alter the magnitudes and directions of
the principal stresses, hence altering the propagation direction of the fractures as described in Figure 2
due to the “stress shadow effect” (Fisher et al. 2004).

Minimum horizontal stress 𝜎 min

Maximum horizontal stress 𝜎 max

Fracture 1
Vertical stress 𝜎𝑣

Crack (initial radius = 𝑟0 , spacing = 𝑙)

Fracture 2
Well

Outer boundary of the model

Figure 1. Schematic diagram of the two penny-shaped fracture model. Cracks are located
with a spacing l and perpendicularly to (𝝈𝒉 )𝐦𝐢𝐧 . Fluid is injected into the cracks from the
intersection points with the well. (𝝈𝒉 )𝐦𝐚𝐱 , (𝝈𝒉 )𝐦𝐢𝐧 , 𝝈𝒗 is applied in the x, y, z-direction,
respectively.

Table 1. Parameters and values used in simulations for multiple fracture model.
Parameter Symbol Unit Value Parameter Symbol Unit Value
Young’s modulus 𝐸 GPa 20 Poisson’s ratio 𝜈 - 0.2
-1
Fluid 𝑐𝑓 Pa 110 -9
Initial crack 𝑟0 M 1
compressibility size/radius
Fracture toughness 𝐾I𝑐 MPa m1/2 1 Injection rate 𝑄 m3/s 0.01
Fluid viscosity 𝜇 Pa s 0.1 Porosity 𝜙 - 0.2

Table 2: Varied parameters used in simulation for comparison.

Biot 𝑘𝑥 , 𝑘𝑦 , 𝑘𝑧 (𝜎ℎ )max , 𝜎𝑣 (𝜎ℎ )min
ID Regime Spacing 𝑙(m)
coefficient 𝛼 (m2) (MPa) (MPa)
MF1 viscosity-storage 0 0 20 20 5
MF2 viscosity-leakoff 1 5  10-15 20 20 5
MF3 viscosity-leakoff 1 2  10 -14
20 20 5
MF4 viscosity-leakoff 0 2  10 -14
20 20 5
MF5 viscosity-leakoff 0.4 2  10 -14
20 20 5

4
Effect of fluid leakoff and poroelasticity on interacting hydraulic fractures. To identify the
effect of fluid leakoff on propagation paths of interacting hydraulic fractures, three conditions are
compared as shown in Figure 3. These conditions are: storage (MF1), 3D leakoff with 𝑘 = 5 ×
10−15 m2 (MF2), and 3D leakoff with 𝑘 = 2 × 10−14 m2 (MF3). As the amount of fluid leakoff
increases, the curvature of the propagation path tends to flatten. Three simulation results of
different Biot coefficients 𝛼 = 0 (MF4), 𝛼 = 0.4 (MF5), 𝛼 = 1 (MF3) are presented together
with the storage case (MF1) in Figure 4. The comparison of MF1, MF3, and MF5 reveal that the
propagation path straightens as the Biot coefficient increases. The increase in permeability and
Biot coefficient both contribute together to the increase in fluid leakoff (Salimzadeh et al. 2016)
which causes the aperture of the fracture to decrease, and the compressive stress in the direction
perpendicular to the fracture plane to reduce. As shown in in Figure 5, the reduction of this
compressive stress re-orientates the near-field principal stresses, which make the propagation
paths straighter compared to the storage cases. By comparing MF3 and MF5, when the Biot
coefficient is higher, the propagation path shows a separation from the storage case (MF1) at an
earlier growth step. This may be explained that the higher Biot coefficient in MF3 led to a higher
amount of leakoff compared to MF5, which allowed to shift the propagation from storage regime
to leakoff regime at an earlier time.
When the Biot coefficient 𝛼 is 0 (MF4), the mechanical deformation and the matrix flow
is decoupled, hence the leakoff does not affect the magnitude of the aperture and the propagation
path at the same growth step and propagation paths nearly overlap with the storage case (MF1).
When modelling fracture growth, previous studies (Olson and Taleghani 2009; Bunger et al.
2012; Wu and Olson 2015; Peirce and Bunger 2015) mostly assume that the fluid leakoff is
negligible. The effect of the leakoff on the hydraulic fracturing of a single fracture has been
shown in many studies (Carrier and Granet, 2012; Salimzadeh and Khalili, 2015; Salimzadeh et
al., 2016). In the present study, the poroelastic effect of the leakoff on the interactions between
multiple hydraulic fractures is shown. The present simulations is performed for injection of a
viscous fluid ( = 0.1 Pa s) into a low permeability rock (𝑘 = 5 × 10−15 m2 ). However, in the
field water is commonly used as the fracturing fluid, and considering the in situ temperature of
the reservoirs, the water would have a viscosity in the range 0.0003-0.001 Pa s. Therefore, in
case of using water as the fracking fluid, the present results is valid for rock permeability as low
as 3 × 10−18 m2 to 1 × 10−17 m2 . This range of the rock permeability agrees with many shale
gas reservoirs. In Barnett shale, which contains the largest shale gas reserves in USA,
permeability ranges from 5 × 10−21 m2 to 2 × 10−18 m2 (Vermylen 2011; Bhandari et al.
2015).

CONCLUSIONS

A 3D finite element based modelling approach has been presented to investigate the
interaction of multiple hydraulic fractures initiated from penny-shaped cracks in a horizontal
well, and to evaluate the effect of poroelasticity on the predicted growth. The simulator was
applied to simulate a model with two penny-shaped hydraulic fractures placed within a close
distance in a homogeneous and isotropic matrix. From the simulation results, we conclude that
the stress induced by the opening of the fractures, and the stress induced by the leakoff, each
have an effect on altering the orientations of the in-situ principal stresses. Stresses induced by the

5
opening of the fractures tend to propagate the fractures away from each other in a curved
trajectory. The incorporation of fluid leakoff and back-stress acts to straighten the curved
trajectory by a noticeable amount.

far-field maximum I : Region far from fracture

principal stress

max
max
min
I
Compressional stress
induced by Fracture 1 II
II : Region next to fracture tip
Fracture 1

Fracture 2
near-field maximum
min III min principal stress

near-field minimum
principal stress
far-field minimum
III : Region next to fracture centre
principal stress

Matrix
max
max

min
Figure 2. Far-field and near-field stresses acting on Fracture 2. Left side shows the far-
field stresses and the compressive stress induced by Fracture 1 uncoupled. Right side of the
figure shows the near-field stresses at Region I, II, III.

2
3

Figure 3. Comparison of propagation paths of fractures under storage-dominated

condition and 3D leakoff conditions with matrix permeability 5  10-15 and 2  10-14 m2 in all
directions.

6
 4
5
3

Figure 4. Comparison of propagation paths of fractures under storage-dominated

condition and 3D leakoff conditions with Biot coefficient 0, 0.4, and 1.

Near-field stress Reduced stress due to leakoff Near-field stress

for storage case causing aperture closure for leakoff case

Figure 5. Reorientation of near-field stress from storage case to the leakoff case at Region
II (region next to the tip of Fracture 2).

REFERENCES

Abe, H., Keer, L., and Mura, T. (1976). "Growth rate of a penny-shaped crack in hydraulic
fracturing of rocks, 2." J. Geophys. Res., 81(35), 6292-6298.
Bhandari, A. R., Flemings, P. B., Polito, P. J., Cronin, M. B., and Bryant, S. L. (2015).
"Anisotropy and stress dependence of permeability in the Barnett shale." Transp. Porous
Media, 108(2), 393-411.
Biot, M. A. (1941). "General theory of three‐ dimensional consolidation." J. Appl. Phys., 12(2),
155-164.
Bunger, A. P., Detournay, E., and Garagash, D. I. (2005). "Toughness-dominated hydraulic

7
 fracture with leak-off." Int. J. Fract., 134(2), 175-190.
Bunger, A. P., Zhang, X., and Jeffrey, R. G. (2012). "Parameters affecting the interaction among
closely spaced hydraulic fractures." SPE Journal, 17(01), 292-306. SPE-140426-PA.
Carrier, B., and Granet, S. (2012). "Numerical modeling of hydraulic fracture problem in
permeable medium using cohesive zone model." Eng. Fract. Mech., 79 312-328.
Cleary, M. P. (1980). "Analysis of mechanisms and procedures for producing favourable shapes
of hydraulic fractures." SPE Annual Technical Conference and Exhibition, Society of
Petroleum Engineers, .
Fisher, M., Heinze, J., Harris, C., Davidson, B., Wright, C., and Dunn, K. (2004). "Optimizing
horizontal completion techniques in the Barnett Shale using microseismic fracture
mapping." SPE Annual Technical Conference and Exhibition, Society of Petroleum
Engineers, .
Garagash, D. I., Detournay, E., and Adachi, J. I. (2011). "Multiscale tip asymptotics in hydraulic
fracture with leak-off." J. Fluid Mech., 669 260-297.
Geertsma, J., and De Klerk, F. (1969). "A rapid method of predicting width and extent of
hydraulically induced fractures." J. Pet. Technol., 21(12), 1571-1581. SPE-2458-PA.
Kovalyshen, Y. (2010). "Fluid-driven fracture in poroelastic medium". PhD. University of
Minnesota.
Nordgren, R. (1972). "Propagation of a vertical hydraulic fracture." Soc. Petrol. Eng. J., 12(04),
306-314. SPE-3009-PA.
Olson, J. E., and Taleghani, A. D. (2009). "Modeling simultaneous growth of multiple hydraulic
fractures and their interaction with natural fractures." SPE Hydraulic Fracturing Technology
Conference, Society of Petroleum Engineers.
Paluszny, A., Tang, X., and Zimmerman, R. W. (2013). "Fracture and impulse based finite-
discrete element modeling of fragmentation." Comput. Mech., 52(5), 1071-1084.
Paluszny, A., and Zimmerman, R. W. (2011). "Numerical simulation of multiple 3D fracture
propagation using arbitrary meshes." Comput. Meth. Appl. Mech. Eng., 200(9), 953-966.
Peirce, A., and Bunger, A. (2015). "Interference fracturing: Nonuniform distributions of
perforation clusters that promote simultaneous growth of multiple hydraulic fractures." SPE
J., 20(02), 384-395. SPE-172500-PA.
Perkins, T., and Kern, L. (1961). "Widths of hydraulic fractures." J. Pet. Technol., 13(09), 937-
949. SPE-89-PA.
Salimzadeh, S., and Khalili, N. (2015). "A three-phase XFEM model for hydraulic fracturing
with cohesive crack propagation." Comput. Geotech., 69 82-92.
Salimzadeh, S., Paluszny, A., and Zimmerman, R. W. (2016). "Three-dimensional poroelastic
effects during hydraulic fracturing in permeable rocks." Int. J. Solids Struct., DOI:
10.1016/j.ijsolstr.2016.12.008.
Savitski, A., and Detournay, E. (2002). "Propagation of a penny-shaped fluid-driven fracture in
an impermeable rock: asymptotic solutions." Int. J. Solids Struct., 39(26), 6311-6337.
Schöllmann, M., Richard, H. A., Kullmer, G., and Fulland, M. (2002). "A new criterion for the
prediction of crack development in multiaxially loaded structures." Int. J. Fract., 117(2),
129-141.
Vermylen, J. P. (2011). "Geomechanical Studies of the Barnett Shale, Texas, USA". PhD.
Stanford University, Palo Alto, California.
Wu, K., and Olson, J. E. (2015). "Simultaneous multifracture treatments: fully coupled fluid flow
and fracture mechanics for horizontal wells." SPE J., 20(02), 337-346. SPE-167626-PA.

View publication stats