SPE 128715-PP

Hydraulic Fracture Propagation in a Naturally Fractured Reservoir

D. A. Chuprakov, SPE, A. V. Akulich, Schlumberger Moscow Research, E. Siebrits, SPE, TerraTek, A Schlumberger
Company, and M. Thiercelin, SPE, Schlumberger Regional Technology Center, Dallas

Copyright 2010, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Oil and Gas India Conference and Exhibition held in Mumbai, India, 2022 January 2010.
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
We present the results of numerical modeling that quantify the physical mechanisms of mechanical activation of a natural fault due
to contact with a pressurized hydraulic fracture. We focus on three stages of interactions: hydraulic fracture approaching, contact,
and subsequent infiltration of the fault. Fracture interaction at the contact is shown to depend on four dimensionless parameters:
net pressure in the hydraulic fracture, in-situ differential stress, relative angle between the natural fault and the hydraulic fracture,
and friction angle of the natural fault. Numerical model based on the Displacement Discontinuity Method allowing for fracture
closure and Mohr-Coulomb friction was used to calculate the displacements and stresses along the natural fracture as a result of the
interaction with the pressurized hydraulic fracture. The analysis of the total stress state along the fault during the hydraulic fracture
coalescence stage makes it possible to define a criterion for reinitiation of a secondary tensile crack from the natural fault. We
show that the most probable location for tensile crack initiation is the end of the open zone of the fault, where the highest tension
peak is generated by the hydraulic fracture contact. In our numerical analysis we study the magnitude of maximum tensile stress
and its position along the fault for a wide range of key dimensionless parameters. Given real reservoir properties these
measurements can be used to detect the possible fracturing scenarios in naturally fractured reservoirs. Using simplified uncoupled
modeling of fluid penetration into the fault after the contact with the hydraulic fracture, we demonstrate that either an increase or a
decrease of the tensile stress at the opposite side of the fault can be realized depending on the ratio of increments of net pressure
and the fluid front as it penetrates the natural fault.
Introduction
Determining the trajectory of a propagating hydraulic fracture is an important challenge in reservoirs with pronounced
heterogeneities. Stratum bedding interfaces can arrest or deflect hydraulic fractures, propagating from one reservoir rock layer into
another (Cook and Gordon 1964; Keer and Chen 1981; Blanton 1982; Lam and Cleary 1984; Renshaw and Pollard 1985; He and
Hutchinson 1989; Polturi 2005). Pre-existing natural discontinuities in the rock, such as natural joints or faults (NF) also affect the
propagation path of a hydraulic fracture (Beugelsdijk et al. 2000; Cooke and Underwood 2001; Jeffrey 1987). For the majority of
gas shale reservoirs, the number of discontinuities is large (Gale et al. 2007), and, therefore, the fracturing path can become quite
complex and ambiguous (Warpinski and Teufel 1987; Thiercelin 2009). However, for some conditions, the fracture may deviate
by small amounts, having a relatively direct propagation path even through many pre-existing fractures (Zhang and Jeffrey 2008;
Jeffrey et al. 2009).
The problem of crack deviation at natural cracks or faults has been widely investigated both experimentally (Blanton 1982;
Renshaw and Pollard 1985; Warpinski and Teufel 1987; Beugelsdijk et al. 2000; Zhou et al. 2008; Chalivendra and Rosakis 2008)
and numerically (Thiercelin 1987, He and Hutchinson 1989; Zhang and Jeffrey 2006, 2007ab, 2008; Thiercelin and Makkhyu
2007; Akulich 2008). There are even several analytical approaches to describe the crossing condition based on the various
simplifications (Blanton 1986; Renshaw and Pollard 1985; Warpinski and Teufel 1987; Polturi 2005). However, a comprehensive
analysis of how different parameters influence the fracture behavior has not been fully investigated to date. An understanding of
the main physical criteria during the interaction of a hydraulic fracture with pre-existing discontinuities plays an important role in
prediction of the propagation scenario. In this work, we study the process of fault activation by an approaching hydraulic fracture
before it reaches the fault, at its coalescence, and during fluid penetration into the fault. We present here approximated fluiduncoupled modeling due to a prescribed fracture pressure. Investigation of the total stress state induced along the frictional fault

SPE 128715-PP

allows us to predict possible locations and reveal the conditions for a new tensile crack to be initiated on the opposite side of the
fault.
Problem statement and governing equations
For simplicity, consider a hydraulic fracture (denoted by HF in Fig. 1) in two dimensional space, that propagates in an
impermeable elastic medium, and has a uniform profile of inner pressure p . This represents the limiting case of a HF containing a
zero viscosity fluid. Its right tip moves quasi-statically following a predefined path along the x axis towards the inclined preexisting natural fracture (fault) of effectively infinite length. The straight natural fault (denoted by NF in Fig. 1) has an inclination
with respect to the approaching hydraulic fracture. The NF is initially closed (zero normal displacement discontinuity) and
static (zero shear displacement discontinuity) under the given far-field confining stresses, characterized by the far-field stress
components 1 and 3 , where we have assumed compression to be positive. If we choose a reference frame for the axis OX to
coincide with the initial (unperturbed) direction of HF propagation, and its origin to be at the point of fracture crossing, then the
left tip of the HF is supposed to be fixed at x = LHF , while the right tip is at some distance from the NF along the axis
direction, or x = . In this case, LHF is the maximum length of the pressurized HF at the moment of contact (Fig. 1).
For quasi-static conditions, where inertial terms can be
neglected, the elastic state of the medium is governed by the
following relationship between displacement discontinuities Dk
(k = 1, 2) and elastic stresses

ij (Crouch and Starfield 1983):

r r

ij (r ) ij( ) = ij ,k (r r ' ) Dk (r ' ) d ,

(1)

where

ij( )

are the far-field stress tensor components,

r r
ij ,k ( r r' ) is the unitary elastic influence function (Greens
r
function), D k ( r ' ) is the magnitude of fracture sliding ( k = 1 )
Figure 1. Geometry of the problem

or opening ( k = 2 ) with summation implied over the repeated

indices, denotes the boundary of the complete system of two
r
r
fractures under consideration, and r and r ' are the spatial

locations along boundary . To resolve the integral equation for two unknown distributions of fracture sliding and opening, the
boundary conditions for the normal and shear tractions at the fracture face are prescribed. For the open fracture parts, the normal
fracture traction is set equal to the inner pressure, and shear traction is set to zero. For the closed fracture parts, the opening is set
to zero, and frictional sliding is computed from the condition for the shear traction, which is set according to a Coulomb-Navier
failure criterion.
r
The solution for Dk ( r ) can be found numerically using the Displacement Discontinuity Method (DDM) (Crouch and
Starfield 1983) by introducing a finite mesh consisting of boundary segments of a finite length along the fractures. The solution for
the stresses and displacements is then obtained with a necessary degree of mesh resolution.
To reduce the number of independent variables and parameters, the following dimensionless stress tensor components are
introduced:

ij = ij
m

( )
ij

ij( )
=
m

(2)

m = ( 1 + 3 ) / 2 is the mean far-field stress. The space coordinates, and, consequently, any spatial distances, such as
fracture lengths, are normalized by the total hydraulic fracture length (after the contact) LHF as
where

x=

x
y
r

, y=
, r =
, =
.
LHF
LHF
LHF
LHF

(3)

SPE 128715-PP

In the normalized units, the maximum HF length is unity at the fracture contact. Additionally, the displacement discontinuities

D k are replaced by dimensionless analogs

Dk =

Dk
G
.
LHF (1 ) m

(4)

The final set of independent parameters of the fracture interaction problem is listed as follows: = ( p 3 ) / m is the

0 < u , where the upper limit u

depends on frictional rock properties, d = ( 1 3 ) / 2 , the NF slope angle range is 0 < 90 ; and rock frictional
coefficient range is 0 < < . If the HF fracture tip has not yet coalesced with the NF, one more independent parameter is the
dimensionless gap between the HF and NF with range 0 < 1 .
normalized HF net pressure; = d / m is the differential far-field stress with range

In what follows, all stresses, spatial coordinates, and displacement discontinuities are treated according to the normalization
provided by (2-4). In order to reverse the dimensionless variables to their dimensional analogs, substitution of actual reservoir
properties into (2-4) is required in order to obtain the physical values.
Finally, the condition for initial steady state of the NF must be taken into account. In compressed rock the NF is always closed,
but sufficient differential far-field stress is able to cause NF slippage. We will assume that the NF is initially in a steady nonslipped state, and neither pore pressure nor inner cohesion exists within the NF. Then

( ) < nn( ) ,

(5)

where n is a local coordinate orthogonal to the NF boundary, and the upper admissible limit for differential far-field stress u is
thus

u =

sin
=
,
sin 2 + cos 2 sin (2 )

(6)

where is the frictional angle ( = arctan ). In order to model and study the process of NF activation with different values of
model parameters, the constraint for differential far-field stress (6) has to be taken into account.
To find the conditions for hydraulic fracture re-initation at the NF, the induced stress state along the natural fracture has to be
analyzed under various reservoir circumstances, covered by an admissible range of parameters. Zones along the opposite side of
the NF to the HF with sufficient tension may serve as potential locations for new tensile crack onset. Such localized tension zones
can become the sites of further hydraulic fracture propagation. For a given tensile strength of the rock, T0 , the criterion for the
tensile crack initiation is:

r
T
min ( r j ) > 0

(7)

where

min ( r j ) is the minimal principal (maximal tensile) dimensionless stress component, generated at some point r j . The

determination of the value of tensile stress peak

min

as well as its position r j along the NF for a wide range of parameters can

thus provide a quantitative preliminary answer to the question of hydraulic fracture propagation through natural frictional
discontinuities.
Evolution of jog position before the coalescence of the HF with the NF
Numerical investigation of the secondary fracture initiation from the NF and its subsequent propagation should embrace all stages
of the fracture interaction: HF tip approaching, fracture coalescence, and fluid penetration into the NF. One of the approaches to
predict the initiation of a new tensile crack at the NF is to keep track of the tensile stresses generated along the opposite side of the
NF by the moving HF tip. Once the criterion (7) is satisfied along the NF, a new tensile crack can simultaneously appear at the
new location. Obviously, the HF-induced stress change and additional rock deformation due to slippage and opening along the NF
develop at any of these stages, and, therefore, have to be separately analyzed.
First, consider the initial stage of the interaction HF tip approaching, where the nucleation of a secondary HF is observed to
take place in many modeling examples. Initially, when the NF stays closed and has not yet slipped (i.e., non-activated), the new
position of the hydraulic fracture can be defined by inspecting the known analytical solution for stresses generated at the NF from
a uniformly pressurized crack (Jaeger and Cook 1979). However, the tensile stress peaks may not satisfy criterion (7) in the case of

SPE 128715-PP

a non-activated NF because the NF activation can trigger earlier and make an analytic approach (solution) inappropriate. After the
beginning of NF slippage and/or opening, the position of the new fracture must be inspected numerically. In the current study, a
linear elastic DDM with frictional contact logic has been used for that purpose (Crouch and Starfield 1983; Tuhkuri 1997;
Christensen et al. 1998; Elvin and Leung 1999; Phan et al. 2003; Wang and Crouch 2008). This numerical technique allows us to
obtain an accurate distribution of the total stresses along each side of the NF (Zhang and Jeffrey 2006; Akulich 2008).
The two fractures are modeled using constant e DD elements subjected to a constant internal pressure loading if they are within
the HF and are subjected to zero internal pressure in the non-pressurized parts of the NF. When the NF is opened during the
approaching stage, it is assumed to have no fluid pressurization yet, because the effects of leak-off into the reservoir rock are not
taken into account in this study. In the case when the HF contacts the NF, the constant DDs at the junction form a T-shaped
configuration with an angled top, consisting of 3 touching DD elements. At first contact, the NF is unpressurized, and at later
incremental steps, the DDs along the opened part of the NF are subjected to internal pressure.
Once the NF is activated, the tensile stress peaks on the opposite side of the NF start to grow dramatically. Using (7), we can
establish a new tensile crack onset at the NF after that stage is reached. However, the induced tensile crack may not have a
sufficient capability to grow further without fluid infiltration. Therefore to study its growth, the subsequent stages of fracture
coalescence and fluid flow into the NF have to be taken into account.
The new crack onset is generally due to the tangential stress component (where n - normal to the fault, - tangent to the
fault), which becomes tensile in the opening NF zone. For the majority of modeling situations, the maximum tension peak
coincides with and tracks the moving ends of the opening zone, as is evident in Fig. 2. The normal nn and shear n stress
components are zero along the open section of the NF, and the minimal principal stress (maximum tension) coincides with the
tangential stress in this region, including the peak values. Thus, the new crack, which is more likely to appear at the tension
maximums, is expected to be generated at the tips of the open zone, and will have an opening parallel to the NF, and the direction
of initial growth perpendicular to it.

Figure 2. Profiles of sliding and opening (upper graphs), and back-side stresses (i.e., along the opposite side of the NF, lower graphs)
for a gap between the HF tip and the NF = 0.1 (left) and full contact (right). = 40 , = 1 ,
below scales in graphs for sliding and opening (upper graphs) are different according to (3) and (4).

= 0 .1 , = 0 .5 . Note that here and

SPE 128715-PP

Once appeared, the magnitude of the tensile stress peak increases with a decreasing gap between the HF tip and the NF, until
fracture coalescence, when these peaks achieve their maximal value. Fig. 3 shows the distribution of the principal tensile stress
along the opposite side of the NF for various gaps between the HF tip and the NF. The tension peak propagates along the NF and
grows in magnitude (see Fig. 3). Its maximal value is reached at the moment of full contact.
It is of interest that, at the very moment of fracture contact, an additional tensile peak is generated at the junction of the NF and
the HF (Fig. 2, right). This peak is the result of a sudden jump in the NF opening and sliding profiles at this coalescence point.
Once the fracture junction is created, a finite shear stress induced by the pressurized HF is applied at the wall of the compressed
left NF branch (y < 0, in Fig. 1) where sliding is inhibited. The step-like nature of the shear stress (see grey curve in Fig. 2, lower
right) along the NF yields a logarithmic singularity for the tangential stress, according to the Cerutti surface problem (Jaeger and
Cook 1979). Numerical modeling shows that, for sufficiently large differential far-field stress, the shear stress at the left NF branch
can be either cancelled, or its sign inverted. In the latter case, at this point the tangential stress peak becomes compressive rather
than tensile.
It is interesting to compare the induced shear and tangential stress distributions along the NF with other published works. For
example, Blanton (1986) used a linear variation of the shear and normal tractions along the NF. However, our direct simulations
show that the normal and shear stress profiles along the NF during interaction are more parabolic than linear (see Fig. 2).
Moreover, in Blantons theory the position of the peak is within the slippage zone, at some finite distance from its beginning. In
Renshaw and Pollard (1985), the re-initiation point corresponds to the non-activated part of the NF. But in contrast to these
previous works, our results demonstrate that the tensile peak, which should initiate the tensile crack, appears at the end of the open
zone and beginning of slippage.
In the presented modeling results the NF is unpressurized. Therefore, there is an additional jump of the pressure profile at the
fracture contact point, which even more favors the generation of the singular tension peak. For realistic reservoir conditions, once
fluid enters the open part of the NF, this tension peak significantly diminishes. Even if a tensile fracture initiates at the junction on
the opposite side of the NF, it is expected to be short-lived, because with continued fluid injection along the NF, a second stronger
tensile peak will appear further along the NF. We will discuss the case of a fluid infiltrated NF later.
A plot of the maximum tension magnitude of the second tensile peak and its offset with respect to the coalescence point are
demonstrated in Fig. 4. It is possible to define the position (jog) of the first appearance of the tensile crack using the criterion of the
tensile strength of the rock T0 (7). If the rock is sufficiently strong to sustain quite large tensile stresses, the new crack will be
jogged with respect to the HF into the right NF branch (i.e., y > 0 section of NF). Otherwise, if the rock is weak, the beginning of
the tension can occur in the other, left NF branch (i.e., y < 0 section of NF), and the created crack cannot propagate because, after
fracture coalescence, this region will be under compression.
0.6

Peak position / value

0.4
0.2
0

-0.2 0.33

0.15

0.07

0.03

0.01

-0.4
-0.6

jog
max stress

-0.8
-1

-1.2
-1.4
Figure 3. Profiles of the principal tensile stress along the opposite
side of the NF for various gaps between HF and NF.

= 1 , = 0 . 1 , = 0 .5 .

= 40 ,

Figure 4. The offset of the tensile stress peak (jog) (blue) and the
peak value (magenta) along the opposite NF side as a function of
the gap between HF and NF. Both plots are in dimensionless values
according to (2), (3).

The numerical results shown in Fig. 2 and 3 are obtained without including any prehistory of the sliding process, as a series of
statically located fractures at different distances from each other. It is important to realize that the frictional sliding of the NF is a
non-linear process and its historical behavior must be properly taken into account during iterative modeling of the approaching HF

SPE 128715-PP

tip. Evolution of this stress component with account for sliding prehistory is found to generate additional fault sliding along the left
NF branch. As a result, it favors the scenario where a tension peak at the fracture contact point is always generated independently
of the value of the differential stress in the rock. Nevertheless, because, as stated above, this tensile peak is expected to be rapidly
diminished in real fluid penetration conditions, we will neglect the first tension peak in further analysis.
Parameter study of unstable tension zone at the NF at the moment of coalescence
The most probable opportunity for the creation of a tensile crack is at the moment of fracture coalescence because in this case, the
tangential stress peaks reach their maximum (see Fig. 2) as a result of maximum deformation of the fault zone at the moment of
coalescence. It is intentionally supposed in this section that even if the tensile cracks appear before the coalescence during HF tip
approaching, the condition for their subsequent growth is not satisfied. Moreover these pre-existing cracks are small enough such
that their contribution into the stress state at the NF can be neglected.
Consider now the special case of fracture coalescence. The tensile stress distribution has the qualitative form depicted in Fig. 2
for = 0 . As noted earlier, the first tension peak, which has a singular nature, will be omitted here. The second tension peak is
always shifted from the coalescence point along the NF. It can be characterized by the position and magnitude of maximal tension.
The maximum tension (minimum principal) stress coincides with the tangential stress studied above. Thus, the direction of the
maximum principal stress at this point, which corresponds to the initial direction of the tensile crack generated here, is
perpendicular to the NF. The latter takes place for a majority of the fracture interaction cases, where the peak is located at the end
of the open zone, where there is no shear stress component. If that peak appears in a sliding or sticking part of the fault, which
mostly refers to the non-activated fault regime, the orientation of the secondary crack opening might be different.
In the following numerical experiments, the fractures touch each other but do not cross. For this configuration, the following
four dimensionless parameters are independent: HF net pressure , far-field differential stress , relative angle of the fault
inclination , and frictional coefficient . In the following analysis of jogging, these parameters are used for the investigation.
The aim of this parametric study is to find out the corresponding dependencies of the position and magnitude of the second tensile
stress peak, where the fracture reinitiation is primarily expected.
First, consider the effect of HF net pressure. As it is known from the fracture mechanics of a uniformly pressurized crack, a
stress field perturbation near its tip is proportional to the magnitude of the applied pressure (Jaeger and Cook 1979). Thus, if the
HF net pressure is increased, then the stress perturbation along the NF is also expected to increase. As a result, the open NF zone
becomes larger, and the position of the tensile stress peak at the end of the open zone moves farther from the point of fracture
contact. The graphs in Fig. 5 indicate the shift of the tensile stress peak as a function of HF net pressure for the fault angle 40.
Principle tensile stress peak magnitude vs HF pressure for different ,
=0.5, =40 deg

Peak magnitude

-0.4
-0.6 0.1

0.3

0.5

0.7

0.9

-0.8
-1
-1.2
-1.4

=0

=0.1

=0.2

-1.6
-1.8

Normalized HF net pressure

Figure 5. Offset of the second tensile stress peak along the NF
with respect to the fracture contact point as a function of HF net
pres sure in the case of 40 NF angle. Plots are given for
various values of far-field differential stress.

Figure 6. Tensile stress peak magnitude as a function of HF net

pressure in the case of 40 NF angle. Plots are given for various
values of far-field differential stress.

As the stress perturbation field is larger in magnitude for larger net pressure, the tension peak at the end of the open zone also
grows. However, the growth of the maximum tensile principal stress magnitude is not as steep as for the offset. A slower rate of
growth of the stress maximum occurs because of the increasing distance between the peak and HF tip. Fig. 6 shows the change of
tensile stress peak versus HF net pressure. A detailed investigation of the principal tensile stress peak behavior for a wide range of
the chosen problem parameters indicates that the peak increases with the net pressure in the HF for any angle of fault inclination.
The dependency of the generated tensile stress peaks on the differential far-field stress appears more ambiguous. For small
values of NF inclination angle the differential stress makes the open zone larger, and the tensile stress peak shifts further along
the NF. A smaller inclination angle causes a steeper dependency of peak offset along the NF. The opposite situation occurs with

SPE 128715-PP

large angles of NF angle: a higher differential stress causes a shorter offset. The open zone length along the NF is governed mainly
by the confining stress acting perpendicular to the NF. As the mean far-field stress in the dimensionless units is equal to 1, the
normal far-field stress component for the NF is 1 cos 2 . Therefore, changing the dimensionless differential stress
implies that the normal stress along the NF decreases proportionally for

< 45 , and increases for > 45 .

Principal tensile stress peak magnitude vs. for different , =1, =0.5

Principal tensile stress peak offset vs differential stress , =1, =0.5

0.25

=10
=30
=40
=50
=60
=80
=90

0.15
0.1
0.05

-1

Peak magnitude

Offset

0.2

0.2

0.4

0.6

0.8

-2
-3
-4
-5
-6

0
-0.1

=10

0.1

0.3

0.5

0.7

0.9

point as a function of differential stress

for various angles of the NF inclination

Examples are shown

=40

=50

=60

=80

=90

Normalized differential stress

Normalized differential stress

Figure 7. Tensile stress peak offset with respect to the HF contact

=30

-7

Figure 8. Tensile stress peak magnitude as a function of

differential stress
the NF inclination.

. Examples are shown for various angles of

The influence of the far-field shear stress along the NF, proportional to , has a secondary effect on the length of the open
zone. Fig. 7 demonstrates the results of measurements of the offset of the second tensile stress peak with respect to the coalescence
point. The non-linear behavior of the curves is the additional effect due to changing far-field shear stress along the NF. Fig. 8
shows the dependency of the maximal principal tensile stress on the opposite side of the NF as a function of dimensionless
differential stress for different angles of the NF, provided that all other dimensionless parameters are fixed. The linear
dependencies for the nearly parallel and nearly perpendicular NF cases are the result of the proportional variation of the normal
stress along the NF. The far-field shear stress is negligible in these cases. For the 40 and 60 cases, the far-field shear stress along
the NF influences the tensile stress leading to its complex non-linear behavior.
Next, we investigate the effect of the
angle of the NF relative to the HF.
Different NF inclinations create different
zones of opening and sliding induced by
the near-tip HF stress field. The angle of
the NF slope is thus one of the major
parameters for the fracture interaction
outcome. The distribution of the principal
tensile stress along the NF is plotted in
Fig. 9 for different NF inclinations. One
can see how significantly the tensile stress
peak generated along the NF depends on
the inclination. For small angles, the peak
is located farther from the fracture contact
point as the open zone is large. As the angle
tends to a perpendicular orientation with
respect to the HF, the length of the open
zone, having a tensile region, decreases.
For the case of perpendicular interaction,
the peak is still shifted but this offset is the
smallest. The tensile stress magnitude is
higher for the nearly perpendicular NFs,
and hence the probability to crack and cross
them without penetration into the NF is
increased.
Figure 9. Principal tensile stress distribution along the NF for various angles of the
NF inclination. HF net pressure

= 1 , frictional coefficient = 0.5 .

SPE 128715-PP

Influence of high friction in approaching stage

More complex angular dependency for the hydraulic fracture jogging can be seen in a special case of high friction along the NF
before the fracture coalescence. In the next series of our experiments, where dimensional units are used instead of normalized, the
fractures have a finite gap equal to 0.2 m. The total length of the HF is 6 m, horizontal and vertical far-field stresses are 2 MPa and
1 MPa, respectively, HF net pressure is 0.65 MPa, and the frictional coefficient along the NF is 5.6. This particular modeling
example was taken from (Thiercelin and Makkhyu 2007).
As the HF tip approaches, the activation of the NF starts earlier for more acute angles of interaction, but for large angles close
to the perpendicular case the fault activation may begin only when the fracture tip is very close to the fault. As a result, there are
situations where, for small angles, the NF is activated, but for a large enough angle, it is not.

Figure 10. A case of high frictional coefficient: the sliding and opening displacements (upper graphs), and the stress profiles along the NF
(lower graphs) for NF inclination 10 (left), 40 (middle), and 60 (right). HF net pressure 0.65 MPa, far-field stresses are 2 and 1 MPa, and

= 5.6 .

Fig. 10 shows the sliding and

opening profiles, three stress
components, and the maximal
principal tensile stress generated
along the NF for three modeling
examples with different values of
the NF inclination. One can see
that, for a fracture inclination of
10, the activated NF zone is the
largest. The two peaks of the
maximum tensile stress are placed
in this case at the largest distance
from the projected coalescence
point (corresponding to the zero
coordinate). The offset of the
secondary tensile cracks is therefore
maximal in the case of such nearparallel fractures. Our previous
analysis of fracture interaction
showed that the magnitude of
tension is smallest for small angles
of interaction. For large friction
angles, as in these particular

Fault angle effect

90
80

1st s3 peak offset

70

fault angle (deg)

frictional coefficient

60

2nd s3 peak offset

50

1st stt peak offset

40

2nd stt peak offset

30
20
10
0
-0.4

-0.3

-0.2

-0.1

0.1

0.2

0.3

0.4

jog
Figure 11. The offset of the principal tensile stress peak, and that of the tangential stress
along the opposite side of the NF, for different values of NF inclination. HF net pressure is
0.65 MPa, far-field stresses are 2 and 1 MPa, and frictional coefficient
ellipsoids denote the maximum peaks.

= 5.6 . Dotted

SPE 128715-PP

examples, the tension peaks along the NF are increased in magnitude, and the tensile stress peaks are quite high even for 10 of
interaction.
The second case shown in Fig. 10 has 40 for the NF inclination. The tensile stress peaks become closer to each other and their
magnitude increases with respect to the smaller NF inclination. In the third modeling example, the angle between fractures is 60,
and the NF activation is absent for this sufficiently large angle. The maximum of the principal tensile stress profile is not so large
in this case, because sliding does not help to concentrate the tension zone.
Fig. 11 summarizes the jog prediction results for the whole spectrum of NF angles. There are two branches of the jog position
expected to correspond to the two peaks in the tensile stress. In the plot, both the tangential and the principal stress tensile peak
offsets are indicated. In the open NF zone, the normal and shear stresses are zero, and the principal tensile stress peak coincides
with the maximum of tangential stress. For the cases where the peaks are out of the open zone, these peaks are different in
magnitude and even in position along the NF boundary (see the case of 60 fault angle in Fig. 10).
The areas denoted by the dotted zones in Fig. 11 show the dominant peaks, where the tension is higher and the jogging is
expected to occur. This complex angular behavior of the jogging along the NF is actually due to the switching between the regimes
of NF activation and non-activation. If the distance between the HF tip and the NF would be smaller, the activation would persist
for all NF angles and the jogging characteristic would be monotonic.
The complex angular dependency of the most probable jogging position at the NF, demonstrated in Fig. 11, suggests that
fracturing scenarios in practice can be much more complicated than those investigated for the given friction coefficient above.
Friction, which adds a strong non-linearity to this problem, is one of very interesting and special parameters that deserves a special
attention to.
Fluid flow into the NF
In this section, we perform some approximate (uncoupled) modeling of the fluid inflow into the NF after the coalescence with the
HF. This modeling strives to imitate the fully coupled process of fluid injection into a NF after fracture coalescence without
resorting to fully coupled simulations. This analysis may help us to understand the key physical parameters controlling the
hydraulic fracture reinitiation at the NF after fracture coalescence during fluid injection into the NF. Solving the problem of a
partially pressurized NF by gradually extending the pressurized part along it, the profile of the principal tensile stress along the NF
can shift together with the fluid-filled zone. As a result, the maximum of the tension can change in magnitude and in position. The
analysis of the tension maximum can allow us to conclude about the most probable point of HF reinitiation for different
mechanical characteristics of the injection process, and, thus, determine which properties of the fluid injection mainly control the
jogging process.
In the modeling presented here,
we use the dimensionless variables
which were introduced above. The
NF has an inclination of 40 with
respect to the HF, and the two
orthogonal far-field stresses are both
set to unity. The friction coefficient
is set to 0.5. At the start of the
penetration process we assume that
the fractures have coalesced, and
injected fluid propagates along the
open NF section on the upper part of
the NF for acute angles (Fig. 1). Let
us choose the initially pressurized
NF zone b0 = 0.1 , and initial net

Figure 12. Maximal principal tensile stress profiles along the NF line for different pressurized
zone lengths,

p0

= 1 MPa,

b0 =

0.1 m. Solid curves case of permanent pressure and

changing pressurized zone length, dashed curves case of changing fracture pressure and
permanent pressurization length.

pressure to be = 1.
First, consider the following two
limiting scenarios: HF fluid injection
with free fluid leakage along the NF
and fixed fluid pressure within the
fractures (Case 1), and the opposite
one, when fluid injection along the
NF is inhibited, and additional fluid
injection inevitably leads to a
pressure increase along the entire HF
(Case 2).

10

SPE 128715-PP

In Case 1, the constant net pressure along the HF and partly along the NF does not change. The pressurized NF segment in the
vicinity of the fracture contact point monotonically grows with a predefined step b0 = 0.1 . For the 5 subsequent evolution steps
used in the simulation, the fluid path along the NF regularly extends from b0 = 0.1 up to 0.6. Tracking the evolution of the
maximal principal tensile stress profile along the NF allows us to judge the tendency of the tensile crack generation at the fault. In
Fig. 12 several snapshots of the maximal principal tensile stress along the NF during the fluid penetration process are shown (see
solid curves). In the current modeling example, the pressure profile was prescribed uniformly along the HF and the pressurized
zone of the NF. Net pressure used was rather high (equal to the fracture confining stress). As a result, the fluid pressure front never
coincides with the end of the open zone along the NF, allowing some finite lag zone (empty open zone without fluid) to be created
behind the pressurized zone along the NF. The zone of tension is created primarily along this lag zone, as one can see in Fig. 12.
The maximum of tension is still located at the end of the NF opened zone, as was shown previously in the case of a nonpressurized NF (also shown in Fig. 12 by the solid black line).
It is interesting to note that the first tension maximum located at the HF-NF contact point becomes much smaller in magnitude
with respect to the non-pressurized NF case (compare black curve with other solid curves in Fig. 12). It is believed that this is the
result of a smoothed continuous pressure profile at the point of fracture coalescence, which helps to reduce additional stress spikes.
It also can be shown that the partial pressurization of the open fault zone decreases the shear stress jump created by the HF tip at
the closed NF branch (i.e., lower one in Fig. 1), which generates the singular tangential stress component at the point of fracture
contact.
The second principal tensile stress peak serves as the main maximum of the tensile stress profile and thus the most probable
location for fracture reinitiation. As the open zone grows during fluid penetration into the NF, the position of the principal tensile
stress peak moves away from the fracture junction. Because the pressure profile remains constant and its front moves far from the
HF tip, the induced stress at the end of the open zone weakens, and the tensile peak monotonically decreases. Evolution of the
tensile maximum due to the fluid penetration along the NF is plotted in Fig. 13 (blue curve).
The next set of experiments (Case 2, Fig.12, dashed curves), simulating the opposite asymptotical case of a slowly penetrating
fluid along the NF ( b0 0 ) demonstrate qualitatively different behavior. The length of the pressurized segment along the NF is
fixed, b0 = 0.2 m , while the magnitude of pressure within the HF and NF is varied with a constant step = 0.5. As a result of
the 5 evolution steps in this series of experiments, the pressure value in the whole fracture system has been increased from 1 up to
3. Similar to the case of a non-pressurized NF, the open zone length and the tension maximum increase as the HF net pressure
increases. However, the rate of the tensile stress peak growth with the HF pressure should be different between these two cases
because of the finite pressurized zone along the NF. Nevertheless, the qualitative behavior in the cases of pressurized and nonpressurized NF, such as tension peak growth, moving peak position, looks similar. The corresponding principal tensile stress
profiles for the Case 2 are plotted in Fig. 12 by dashed curves. In Fig. 13 the tendency of the tensile stress peak growth is plotted
by the solid brown curve.
Principal tensile stress peak value vs the NF pressurized zone length,
=0, =40 deg, =0.5

max

1.3

Principal tensile stress peak offset vs the fault pressurized zone length,
=0, =40 deg, =0.5

1.6
dp=0, db=0.1m

dp=0, db=0.1

1.4

dp=0.1, db=0.1
dp=0.5, db=0.1

1.2

Offset

dp=0.1, db=0.1
dp=0.5, db=0.1

1.2

dp=0.5, db=0.05

dp=0.5, db=0.05

dp=0.5, db=0

dp=0.5, db=0

1.1

0.8

0.6
0.9
0.4
0.8

0.2
1

step number
Figure 13. Principal tensile stress peak magnitude versus the NF
pressurization step.

step number
Figure 14. The offset of the tension maximum along the NF
with respect to the fracture contact point versus the NF
pressurization step.

SPE 128715-PP

11

Strictly speaking, the above two processes of a pressure increase (Case 2) and fluid propagation along the NF (Case 1) are
combined in a coupled model. The uncoupled modeling approach, chosen here as a simplified approach, cannot give an answer as
to how exactly or in what proportion these quantities change during fluid propagation. The ratio between the steps must depend on
the differential stress, NF angle, and injecting fluid properties (fluid viscosity, injection rate). As the fluid flow equations are not
solved here, we approximated the real fluid penetration behavior by setting up different steps for the pressure and pressurized NF
segment zone increase. It has been found that for relatively low ratios of the steps ( / b0 1 , fast penetration) the tension peak
decreases as a function of time. For sufficiently large values of that ratio ( / b0 > 5 , slow penetration) the tension peak
increases with time as fluid is injected (Fig. 13). The asymptotic cases given above represent the extreme cases of an increase or a
decrease in the tension maximum, respectively. Some intermediate values of that ratio are found to change insignificantly
compared to the initial value (e.g., green curve in Fig. 13).
As for the offset of the tension maximum, it is proportional to increasing both pressure and pressurized zone length. The
maximal growth of the open zone and the peak corresponds to the maximal rates for these two scenarios (Fig. 14).
Discussion
The fluid-uncoupled simulations of fluid penetration into the NF can be seen as being far from reality. However, this is a first
attempt to understand the independent influence of the two synchronous processes, which are naturally coupled pressure increase
after the contact and fluid front advance along the NF. The ratio between increments of fluid pressure and advance of the fluid
front depends on the fluid viscosity, injection rate, relative angle of the NF and HF, and in-situ differential stresses. Investigation
of their interplay in a fluid-coupled model is the subject for separate study. Based on our simplified numerical tests with step-wise
fluid infiltration, we find that the ratio of their increments have significant influence on the jog size.
Other aspects to consider are the natural roughness of NFs, pre-existing flaws or asperities, which are potential stress
concentrators affecting the jog size, and the possibility of a fluid lag zone ahead of the fluid front. Zhang and Jeffrey 2007ab, for
example, have studied secondary fracture growth from weak points. Our numerical modeling does, however, indicate dominant
parameters that can influence jog size under otherwise homogeneous conditions.
Conclusions
In this work the problem of elastic interaction between a pressurized HF and a pre-existing NF in an impermeable rock has been
investigated numerically under the simplified assumption of a uniform pressure distribution along the fractures (zero fluid
viscosity). To our knowledge, this is the first time that an analysis of the total stress state induced along the NF is fulfilled
numerically for different stages of hydraulic fracturing (HF tip approaching, coalescence, fluid infiltration along the NF). It is
found that a tensile stress is generated on the opposite side of the NF (with respect to the initial HF location) tangentially, creating
the possibility for generation of a new tensile fracture perpendicular to the NF. The pronounced maximums of the tensile stress are
found at the very ends of the open zone along the NF, which move as the HF tip approaches the NF. These peaks are the expected
location of the new tensile crack onsets. Initially these peaks appear along the lower NF branch, but the resulting fracture offsets
will not develop due to a change in the stress state along this NF section to compressive after fracture coalescence.
The points of expected HF reinitiation are determined along the upper NF branch for a wide range of problem parameters (net
pressure, angle of NF, differential stress). This numerical study shows qualitative agreement with experimental results (Blanton
1982; Renshaw and Pollard 1995) and modeling results (Thiercelin and Makkhyu 2007) where a high differential stress and large
angles between the fractures both favor the crossing scenario for HF propagation. This is supported numerically here by growth of
the tension peaks for increasing NF angles and differential far-field stresses. Influence of other parameters such as net fracture
pressure and rock frictional coefficient is more evident and proved by our analysis as well. Both higher net pressure and rock
friction lead to larger tension peaks along the NF.
The case of high friction allows more complex angular behavior for the maximum tension peak because for acute angles the
NF can be opened and may slip, whereas for sufficiently larger angles it may not. This can cause a sudden change between the
positions of the maximum tensile stress in the angular jog dependency.
The net fracture pressure can be an important parameter to control the crossing behavior. In a real HF treatment, it is controlled
by the injection rate and fluid viscosity. Higher viscosity fluids injected at higher rates will tend to result in smaller jog sizes when
a HF propagates in a naturally fractured reservoir, as was found experimentally by Beugelsdijk et al. 2000 in laboratory
experiments, assuming that the reservoir rock can be hydraulically fractured. If not, then the fluid invasion into the NF network
will become more dendritic. Thus, since the pressure is proportional to the generated tension peaks, it can be a real-time trigger
between the different HF propagation outcomes (arrest versus direct or jogged crossing).
Acknowledgments
The authors thank Schlumberger for permission to publish this work.

12

SPE 128715-PP

Nomenclature
HF acronym for Hydraulic Fracture
NF acronym for Natural Fault
= dimensionless differential stress
= dimensionless HF net pressure
= NF slope angle (with respect to the HF)

= coefficient of rock friction

= frictional angle

u = upper limit for dimensionless differential stress

b0 = initially pressurized zone at the NF (with the same pressure as in the HF)
b0 = step of the NF pressurized zone increase in numerical experiments
= step of the HF net pressure increase in numerical experiments
= distance between the HF tip and the NF
r
r = radius-vector in space
= comprehensive trajectory of all fractures
Dk = displacement discontinuities (shear or normal)
p = inner fracture pressure
ij = stress tensor components

min

= minimal compressional (max tensile) principal stress

T0 = rock tensile strength

m = dimensional mean far-field stress

d = dimensional differential far-field stress
LHF = full dimensional length of the HF
1 = maximum principal far-field stress (parallel to the HF)
3 = minimum principal far-field stress (perpendicular to the HF)
References
Akulich, A.V. 2008. Numerical Simulation of Hydraulic Fracture Crack Propagation. Moscow University Mechanics Bulletin, 63 (1): 612.
Beugelsdijk, L.J.L., De Pater, C.J. and Sato, K. 2000. Experimental hydraulic fracture propagation in a multi-fractured medium. Paper SPE
59419 presented at the SPE Asia Pacific Conference on Integrated Modelling for Asset Management: 177-184.
Blanton, T.L. 1982. An experimental study of interaction between hydraulically induced and pre-existing fractures. Paper SPE 10847 presented at
the SPE Unconventional Gas Recovery Symposium, 16-18 May 1982, Pittsburgh, Pennsylvania. DOI: 10.2118/10847-MS.
Blanton, T.L. 1986. Propagation of hydraulically and dynamically induced fractures in naturally fractured reservoirs. Paper SPE 15261 presented
at the Unconventional Gas Technology Symposium of the Society of Petroleum Engineers: 613-627.
Chalivendra, V.B. and Rosakis, A.J. 2008. Interaction of dynamic mode-I cracks with inclined interfaces. Engineering Fracture Mechanics, 75
(8): 2385-2397.
Christensen, P.W., Klarbring, A., Pang, J.S. and Strmberg, N. 1998. Formulation and comparison of algorithms for frictional contact problems.
International Journal for Numerical Methods in Engineering, 42 (1): 145-173.
Cook, J. and Gordon, J.E. 1964. A mechanism for the control of crack propagation in all brittle systems. Proc. Roy. Soc. 282 A: 508-520.
Cooke, M.L. and Underwood, C.A. 2001. Fracture termination and step-over at bedding interfaces due to frictional slip and interface opening.
Journal of Structural Geology, 23 (2-3): 223-238.
Crouch, S.L. and Starfield, A.M. 1983. Boundary Element Methods in Solid Mechanics. George Allen and Unwin.
De Bremaecker, J.-C., Ferris, M.C. and Ralph, D. 2000. Compressional fractures considered as contact problems and mixed complementarity
problems. Engineering Fracture Mechanics, 66 (3): 287-303.
Elvin, N. and Leung, C. 1999. A fast iterative boundary element method for solving closed crack problems. Engineering Fracture Mechanics, 63
(5): 631-648.
Gale, J.F.W., Reed, R.M. and Holder, J. 2007. Natural fractures in the Barnett Shale and their importance for hydraulic fracture treatments.
American Association of Petroleum Geologists Bulletin, 91 (4): 603-622.
He, M.Y. and Hutchinson, J.W. 1989. Crack deflection at an interface between dissimilar materials. International Journal of Solids and
Structures, 25: 1953-1967.

SPE 128715-PP

13

Jaeger, J.C. and Cook, N.G.W. 1979. Fundamentals of Rock Mechanics. London: Chapman and Hall.
Jeffrey, R.G., Vandamme, L. and Roegiers, J.-C. 1987. Mechanical interactions in branched or subparallel hydraulic fractures. Paper SPE 16422
presented at SPE/DOE Low Permeability Reservoirs Symposium held in Denver, Colorado, May 18-19, 1987: 333-341.
Jeffrey, R.J., Zhang, X. and Thiercelin, M. 2009. Hydraulic fracture offsetting in naturally fractured reservoirs: Quantifying a long-recognized
process. Paper SPE 119351 presented at the 2009 SPE Hydraulic Fracturing Technology Conference held in the Woodlands, Texas, USA,
19-21 January 2009.
Keer, L.M. and Chen, S.H. 1981. The intersection of a pressurized crack with a joint. J. Geophys. Res. 86 (B2): 10321038.
Lam, K.Y. and Cleary, M.P. 1984. Slippage and re-initiation of (hydraulic) fractures at frictional interfaces. International Journal for Numerical
& Analytical Methods in Geomechanics, 8 (6): 589-604.
Phan, A.V., Napier, J.A.L., Gray, L.J. and Kaplan, T. 2003. Symmetric-Galerkin BEM simulation of fracture with frictional contact.
International Journal for Numerical Methods in Engineering, 57 (6): 835-851.
Potluri, N., Zhu, D. and Hill, A.D. 2005. Effect of natural fractures on hydraulic fracture propagation. Paper SPE 94568 presented at the SPE
European Formation Damage Conference held in Netherlands, 25-27 May 2005.
Renshaw, C.E. and Pollard, D.D. 1995. An experimentally verified criterion for propagation across unbounded frictional interfaces in brittle,
linear elastic materials. International Journal of Rock Mechanics & Mining Sciences, 32 (3): 237-249.
Thiercelin, M., Roegiers, J.C., Boone, T.J. and Ingraffea, A.R. 1987. An investigation of the material parameters that govern the behavior of
fractures approaching rock interfaces. Proceedings of 6th International Congress on Rock Mechanics: 263-269.
Thiercelin, M. and Makkhyu, E. 2007. Stress field in the vicinity of a natural fault activated by the propagation of an induced hydraulic fracture.
Proc., First Canada-US Rock Mechanics Symposium Rock Mechanics Meeting Society's Challenges and Demands, 2: 1617-1624.
Thiercelin, M. 2009. Hydraulic fracture propagation in discontinuous media. Proceedings of the Conference on Rock Joints and Jointed Rock
Masses, Tucson, Arizona, January 7-9, 2009.
Tuhkuri, J. 1997. Dual boundary element analysis of closed cracks. International Journal for Numerical Methods in Engineering, 40 (16): 29953014.
Wang, J. and Crouch, S.L. 2008. An iterative algorithm for modeling crack closure and sliding. Engineering Fracture Mechanics, 75 (1): 128135.
Warpinski, N.R. and Teufel, L.W. 1987. Influence of geologic discontinuities on hydraulic fracture propagation. Journal of Petroleum
Technology, 39 (2): 209-220.
Zhang, X. and Jeffrey, R.G. 2006. The role of friction and secondary flaws on deflection and reinitiation of hydraulic fractures at orthogonal preexisting fractures. Geophysical Journal International, 166 (3): 1454-1465.
Zhang, X., Jeffrey, R.G. and Thiercelin, M. 2007. Deflection and propagation of fluid-driven fractures at frictional bedding interfaces: A
numerical investigation. Journal of Structural Geology, 29 (3): 396-410.
Zhang, X., Jeffrey, R.G. and Thiercelin, M. 2007. Effects of frictional geological discontinuities on hydraulic fracture propagation. Paper SPE
106111 presented at Hydraulic Fracturing Technology Conference held in College Station, Texas, USA, 29-31 Jnauary 2007: 268-278.
Zhang, X. and Jeffrey, R.G. 2008. Reinitiation or termination of fluid-driven fractures at frictional bedding interfaces. Journal of Geophysical
Research, 113 B08416. DOI:10.1029/2007JB005327.
Zhou, J., Chen, M., Jin, Y. and Zhang, G.-Q. 2008. Analysis of fracture propagation behavior and fracture geometry using a tri-axial fracturing
system in naturally fractured reservoirs. International Journal of Rock Mechanics and Mining Sciences, 45 (7): 1143-1152.