SPE-184871-MS

Effect of Geological Layer Properties on Hydraulic Fracture Initiation and

Propagation: An Experimental Study

Murtadha J. AlTammar and Mukul M. Sharma, The University of Texas at Austin

Copyright 2017, Society of Petroleum Engineers

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

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

Abstract
Hydraulic fracture initiation and propagation in the presence of multiple layers with different mechanical and
flow properties are investigated experimentally utilizing a novel fracturing cell. Mixtures of plaster, clay and
other materials are used to cast sheet-like test specimens in layers with different mechanical properties. The
specimen is placed between two transparent plates that allow us to take videos of the speckled rock sample
surfaces to facilitate digital image processing of rock displacement. In order to control fracture orientation, a
far-field differential stress is applied to the specimen via pneumatic jacks. Hydraulic fracture growth during
the experiment is recorded using a high resolution digital camera. Key frames are subsequently analyzed
using Digital Image Correlation (DIC) to reveal micro-cracks, strains and other features that are difficult to
detect with the naked eye. Experiments are conducted for fracture propagation across layered porous media
in different stress states.
It is clearly shown that a fracture initiating in the vicinity of a layer interface may exhibit a range of
behaviors including propagation straight through the layer interface, forming secondary fractures, fracture
kinking, and fracture propagation along or parallel to the interface. Straight/plane fracture propagation is
observed across an interface when going from a hard to a soft layer. Conversely, crossing from a soft to
a hard layer, a fracture tends to kink at a slight angle as it approaches the interface. The angle at which
the fracture turns increases as the contrast in Young's modulus and fracture toughness between the layers
increases and/or the differential stress decreases. Tensile stresses on the intact (hard) side of the interface
can also develop a secondary fracture at an offset to the parent fracture with localized interface opening
resulting in a fracture step-over. Given a choice of multiple layers, fracture propagation into a harder layer is
favored over a softer layer. The effect of hard inclusions that can act as fracture barriers is also investigated.
Fractures tend to initiate and propagate parallel to the bounding interfaces when the fracture initiation layer
is relatively thin or in the case of an isotropic stress state and propagate along the interfaces when the bonding
between the layers is weak. A fracture approaching a dipping, harder layer tends to kink towards the low
side of the interface. The experiments, for the first time, clearly demonstrate and quantify experimentally
the factors that control fracture propagation across geological layers. The observed turning and branching
of the fractures can have a profound effect on fluid leakoff and proppant transport.
2 SPE-184871-MS

Introduction
Rock heterogeneity at various length scales abounds in many oil and gas bearing formations. Understanding
fracture growth across layers of different materials is of great interest in hydraulic fracturing. Much of the
earlier published work in this area was geared towards elucidation of the influence of various parameters
on hydraulic fracture containment and vertical fracture growth.
Teufel and Clark (1984) performed laboratory experiments in which three-layer rock specimens are
hydraulically fractured under a uniaxial compression stress state. The unbonded, unconfined rectangular
rock blocks were pressed together by applying a vertical stress normal to layer interfaces. The two outer
layers in each specimen were identical and had different mechanical properties than the middle layer. Under
the applied vertical stress, the composite specimen expands laterally giving rise to a non-uniform horizontal
stress field across layer interfaces due to contrasting mechanical properties of the layeres. Across a given
interface, the induced horizontal stress is tensile in the high shear modulus layer and compressive in the
low shear modulus layer.
The authors performed two categories of experiments. In the first category, the outer layers have higher
shear modulus compared to the middle layer while in the second category, the outer layers have lower shear
modulus than the middle layer. In both categories, initiated fractures are contained in the middle layer when
the applied normal stress is low. Fractures propagate vertically to the interfaces and then propagate along
the interfaces. As the normal stress is increased, the shear strength of the interfaces increases to a critical
threshold above which fractures propagate across the interfaces. It was observed that in the experiments in
the second category, fractures are totally contained (no crossing or fracture propagation along interfaces)
in the middle layer when the applied normal stress is sufficient to induce high compressive stresses in the
low modulus bounding layers.
From linear elastic fracture mechanics, the stress intensity factor of a fracture tip approaching an interface
varies depending on the relative material properties on either side of the interface. Simonson et al. (1978)
showed that for a fracture in a low shear modulus material, the stress intensity factor, and hence the energy
release rate, decreases as it approaches an interface with a high modulus material. In contrast, the stress
intensity factor increases when a fracture in a high shear modulus material approaches an interface with a
low modulus material. As a result, fracture propagation theoretically becomes increasingly harder as the
fracture gets closer to a stiffer material and increasingly easier closer to a softer material.
Stress and energy analysis reported by Wu et al. (2004) suggests plane/straight fracture propagation when
a fracture grows from a stiff to a soft material. Depending on the specifics of a given problem, a number
of fracture behaviors can arise when a fracture propagates from a soft to a stiff material. Such behaviors
include crack arrest at the interface, plane fracture growth, fracture kinking, interface delamination, and
formation of secondary fractures. Some experimental evidence of these fracture behaviors is reported in
their study as well.
Fisher and Warpinski (2012) show some observations from mineback tests. One of the observations is
of a hydraulic fracture growing from a low modulus material to a high modulus material in a fairly straight
path. Another observation is of a fracture that propagated to a weak interface, resulting in termination and
blunting of the fracture. In addition, it was shown that fracture offsetting, kinking and turning, creation of
additional fractures, and fracture termination can result as a fracture propagates through multiple layers and
interfaces.
A laboratory fracturing test was reported by Athavale and Miskimins (2008) in a laminated artificial block
with layers of contrasting mechanical properties under triaxial stress. Post-test block inspection showed
multiple and complex fracture propagation.
Many theoretical studies have been published on hydraulic fracture growth and containment in layered
media (van Eekelen 1982; Thiercelin et al. 1989; Gu and Siebrits 2008; Zhang and Jeffrey 2008; Garcia
SPE-184871-MS 3

et al. 2013; Ouchi 2016). Results from some of these models are used to compare with the experiments
presented in this paper.
In this work, we performed an experimental study in layered specimens that enables direct observation of
fracture growth near material interfaces. The results provide fundamental insights on fracture propagation
behavior in heterogeneous porous media.

Experimental Method
A fracture cell was used to perform the tests as depicted in Figure 1. A test specimen is placed between
two thick, transparent plates. The specimen is a 6-inch by 6-inch cast sheet of 0.2-inch thickness with a
0. 125-inch diameter hole in the center. The fracturing fluid is glycerin with a viscosity of around 942 cp
(Huang et al. 2012). Fluid is injected in the center of the specimen to initiate and propagate a fracture.
Pneumatic jacks on the sides of the specimen apply uniaxial compressive stress inducing a differential far-
field stress and hence, a preferential direction for fracture propagation. Thin, flexible layers of polymer-
based clear adhesive are used on top and bottom of the specimen to prevent fluid leakage. Fracture growth in
the experiments is recorded using a high resolution digital camera at 30 frames per second. The monitored
field of view is approximately 4 inches by 2 inches with the long side being parallel to the applied far-field
stress as shown in Figure 2.

Figure 1Schematic of fracture cell.

Figure 2Specimen configuration as used in the fracturing tests.

4 SPE-184871-MS

Test Specimen
Plaster is the base material for making test specimens. Talc is mixed with plaster to make softer materials
while hydrostone is mixed with plaster to make harder materials. We measured the mechanical properties of
various plaster-talc mixtures and plaster-hydrostone mixtures (Figure 3). Fracture toughness was measured
using semi-circular bending tests described by Park et al. (2004). Uniaxial compression tests on cylindrical
plugs were performed to determine Young's modulus. Attempts to quantify Poisson's ratio showed similar
values for all tested mixtures; however, the values were unreasonably low. Ganesan (2000) reports that
mixtures of plaster and diatomaceous earth can have a wide variation in Young's modulus with varying
proportions but all mixtures have Poisson's ratios in the range of 0.22 to 0.25. Diatomaceous earth, similar
to talc (clay) used in our test specimens, does not chemically react with plaster. Blanton (1982) reports a
Poisson's ratio of 0.22 for hydrostone. Therefore, it is expected that all specimen mixtures used in our tests
would have a fairly constant Poisson's ratio between 0.22 and 0.25.

Figure 3Mechanical properties of mixtures of plaster and talc (left) and mixtures of plaster and hydrostone (right).

Experimental Procedure
Specimens are cast in rubber molds and allowed to dry for 4 days. In a layered specimen, casting adjacent
layers and allowing them to cure at the same time make well bonded interfaces. Then, a random speckle
pattern is painted on the dry specimen using white and black spray paint. The clear adhesive, with a paste
consistency, is applied next on top and bottom of the speckled specimen and the fracture cell is assembled
as illustrated in Figure 1. The fracturing test is performed after 24 hours to allow the adhesive to completely
cure and dry. Pneumatic jacks are inflated to a certain pressure to apply a far-field differential stress. The
base differential stress used in the tests is 65 psi unless the effect of differential stress is being investigated.
Glycerin is then injected into the injection port and the pressure is ramped up until the specimen fractures.
This typically takes between 2.5 to 5 minutes. The fracture cell is disassembled and cleaned after the test.
The process is recorded with a high resolution digital video camera. Finally, we obtain still frames from the
recorded video to collect key frames of fracture propagation during the test. These key frames are analyzed
using a Digital Image Correlation (DIC) software (Blaber et al. 2015) to resolve full-field displacement and
strain of the deforming specimen.
SPE-184871-MS 5

Digital Image Correlation (DIC)

Displacement and strain measurements in DIC algorithms are basically accomplished by comparing a
reference image of a specimen before loading or deformation and a target or deformed image after
deformation. Let us consider the reference image of the semi-circular specimen shown in Figure 4. In order
to resolve specimen deformation using DIC after loading, the region of interest is divided into virtual subsets
as marked by the red squares in Figure 4. Each subset in the reference image is tracked to its new location
in the deformed image (see Figure 5; Pan et al. 2009). The displacement vector can then be computed as
the position difference between the center of the reference subset (point P in Figure 5) and the center of the
target subset (point P'). Points Q and P are neighboring points in the reference subset. These two points are
still neighbors in the target subset (points P' and Q') but their positions changed. The change in spacing of
neighboring points along with the displacement vector are used to find normal and shear strains.

Figure 4Virtual subsets, shown in red squares, superimposed on a reference (undeformed)

image to resolve specimen deformation in Digital Image Correlation (DIC) analysis.

Figure 5A vitural subset before and after deformation (Pan et al. 2009).

Results
We present here a collection of fracturing tests that illustrate the influence of some important factors on
fracture propagation behavior in layered reservoirs. The factors investigated include contrasts in Young's
modulus and fracture toughness between the layers, layer size, layer dip angle, interface strength, and far-
field differential stress.
Fluid injection pressure in all tests shows an increasing pressure trend from the start of injection until
fracture breakthrough at specimen boundaries. Figure 6 shows an example of an injection pressure history
that was recorded during one of the tests. The fracture initiation point marked in the Figure is determined
using the strain field resolved by DIC analysis. As explained earlier, the maximum far-field stress in all
tests is always applied parallel to the long side of the image in the monitored field of view (Figure 2).
6 SPE-184871-MS

Figure 6An example of fluid injection pressure history recorded during a fracture test.

Test 1
This is a base test in which a fluid driven fracture grows in a homogeneous poro-elastic material. Figure
7 shows the raw images of fracture propagation starting with the reference (undeformed) image. The
last image in the Figure shows the final fracture trajectory after the two faces of the fracture separated
completely. Fracture orientation is parallel to the applied far-field stress, as expected. Figure 8 shows the
evolution of the displacement field in the y-direction during the test. Fracture propagation can also be
monitored using the measured normal strain in the y-direction as shown in Figure 9. Since fracture opening
is mainly in the y-direction, we only show the displacement and strain in the y-direction.

Figure 7Raw images of fracture growth in Test 1. Top left is the reference image labeled with Young's
modulus and fracture toughness of specimen material. Bottom right is the final fracture shape.
SPE-184871-MS 7

Figure 8Incremental displacement fields (in y-direction) in Test 1.

Displacement is superimposed on the reference raw image. Scale is in mm.
8 SPE-184871-MS

Figure 9Incremental normal strain fields (in y-direction) in Test 1. Strain is superimposed on the reference raw image.

Test 2
In this test, a fracture propagates in a three-layer specimen from a hard to a soft material as shown in the left
image of Figure 10. Young's modulus and fracture toughness ratios of the outer layers to the middle layer are
0.46 and 0.67, respectively. The specimen is subjected to unconfined uniaxial compression perpendicular to
layer interfaces. In other words, far-field stress is applied to all the layers in one direction (E-W; x-direction)
with zero far-field stress in the other (up-down; y-direction). This arrangement is similar to the loading
setup described previously by Teufel and Clark (1984) in their fracturing experiments. Thus, as far-field
stress is applied to the specimen in the x-direction, we expect to have induced compressive stress in the low
modulus layers and induced tensile stress in the high modulus layers across interfaces in the y-direction.
Accordingly, in this particular test, the induced stresses in y- direction are tensile in the middle layer and
compressive in the outer layers near the interfaces.
SPE-184871-MS 9

Figure 10Reference image of Test 2 on the left and final fracture trajectory on the right.

Figure 11 shows a sequence of images of normal strain in the y-direction as the fracture propagates
through the specimen. The images show clearly that the fracture reached the right interface first and the
left interface second. After reaching both interfaces, the fracture started propagating into the outer layers
in both directions. While overall fracture growth is very gradual, it was observed that when the fracture
became close to the interface, noticable fracture tip jumps occurred until the fracture reached both layer
interfaces and started growing slowly once again. The fracture tip jumps can be seen in the four images from
the second to the fifth image in Figure 11. These four images are separated by 0.033 second only while, for
example, the first and second image in Figure 11 are separated by 4 seconds.

Figure 11Normal strain field evolution, in y-direction, in Test 2.

10 SPE-184871-MS

The general trend in this test is that the fracture propagated in a fairly straight path from the stiff middle
layer to the soft outer layers as shown in Figure 11 and the right image in Figure 10.

Test 3
The same specimen configuration in Test 2 is used here with the same middle layer. The only difference in
this test is that the outer layers are stiffer than the middle layer (left image in Figure 12). Young's modulus
and fracture toughness ratios of the outer layers to the middle layer are 4.75 and 2.27, respectively. The
induced stresses across the interfaces in y-direction are compressive in the middle layer and tensile in the
outer layers.

Figure 12Raw images of Test 3.

The evolution of the normal strain field in Figure 13 shows that the fracture propagated towards the
interfaces in a markedly slow fashion. In Figure 13, the time between images from the second to the seventh
image is 5 seconds and beween the seventh and eighth image is 2 seconds. Also, the fracture had a slight kink
in its path as it approached the right interface. After reaching both interfaces, the fracture started propagating
into the right layer initially and then into both outer layers. The fracture reoriented to the applied far-field
stress as it propagated through the right layer. The raw image of fracture trajectory is shown in the right
image of Figure 12.
SPE-184871-MS 11

Figure 13Incremental images of y-normal strain field in Test 3.

12 SPE-184871-MS

From the results of Tests 2 and 3, it is evident that fracture growth towards an interface becomes easier
when going from a stiff to a soft layer and harder when going from a soft to a stiff layer. This is consistent
with theoretical predictions from linear elastic fracture mechanics (Simonson et al. 1978). Moreover, the
fracture, in both tests, propagated all the way to the right and left interface before crossing either interface.
This suggests that crossing an interface whether from a hard to a soft layer or from a soft to a hard layer is
energetically more difficult than propagation within the layer.

Test 4
A fracture initiates from the same layer as the previous test but with a lower contrast in Young's modulus
and fracture toughness between the middle and outer layers relative to Test 3. Young's modulus and fracture
toughness ratios of the outer layers to the middle layer in this test are 3.28 and 1.80, respectively (Figure
14). As shown in the fractured specimen image of Figure 14, the fracture kinked as it approached the right
interface and continued along the direction of the applied far-field stress after crossing the interface.

Figure 14Reference and fractured specimen in Test 4.

Test 5
This is a similar test to the previous two tests with the outer layers having a higher contrast in Young's
modulus and fracture toughness than the middle layer relative to Test 3 (Figure 15). Young's modulus and
fracture toughness ratios of the outer layers to the middle layer are 7.05 and 2.63, respectively. The fracture
path in this test as shown in Figure 15 is more curved and tilted compared to the previous, lower contrast
cases. The fracture approached the right interface almost orthogonally and propagated into the right layer
while it approached the left interface at an angle and propagated along the interface.

Figure 15Reference and fractured specimen in Test 5.

Collectively, Tests 3, 4, and 5 suggest that fracture kinking increases as the contrast in Young's modulus
and fracture toughness between the bounding stiffer layers and the middle layer becomes larger. In addition,
SPE-184871-MS 13

it is observed that the fracture links away from the high Young's modulus layer, as if trying to avoid it (see
Figure 15).

Test 6
In the previous three tests, a fracture propagated from a soft to a hard layer with varying contrasts in
mechanical properties. The middle layer in these tests was the same with a Young's modulus of 84 ksi
and fracture toughness of 103 psi.in05. In this test (Test 6), the middle layer is weaker having a Young's
modulus of 39 ksi and fracture toughness of 68 psi.in05 (Figure 16). The bounding layers have a higher
Young's modulus and fracture toughness by factors of 7.11 and 2.70, respectively, relative to the middle
layer. Incremental images of the normal strain field in the y-direction are shown in Figure 17. The fracture
crossed the right layer with a slight deflection first. Then, tensile stresses in the left layer induced a secondary
fracture at an offset to the main fracture. The normal strain in the x-direction in the last image of Figure
17 shows a localized fracture or interface opening that occurred momentarily prior to the formation of the
secondary fracture. The resulting fracture trajectory can be realized by the final y- displacement field in
Figure 18 that shows the step-over fracture behavior on the left interface and fairly plane growth on the
right interface.

Figure 16Initial raw image of specimen in Test 6.

Figure 17Incremental y-normal strain fields of fracture growth in

Test 6 and x-normal strain field around the left interface (bottom right).
14 SPE-184871-MS

Figure 18Final displacement field, in y-direction, of the fracture in Test 6.

Test 7
A fracture grows in a layer bounded by stiffer layers in this test under relatively high differential far-field
stress compared to the previous tests (Figure 19). The applied differential stress in this test is 80 psi compared
to 65 psi in all previous tests. As shown in Figure 19, the fracture propagates and crosses both outer layers
in a fairly straight path. Test 7 suggests that fracture kinking or formation of seconday fractures maybe
suppressed or minimized under high differential far-field stresses.

Figure 19Incremental raw images of Test 7 starting with reference

in top left and ending with fully developed fracture in bottom right.

Similar straight fracture growth from low modulus material to high modulus material was also observed
in a mineback test as mentioned earlier (Fisher and Warpinski 2012).

Test 8
. The effect of a relatively lower far-field differential stress is examined in this test. The applied differential
stress here is 50 psi. The left image of Figure 20 shows the properties of the layered specimen which is
identical to the one used in Test 4. Figure 21 shows incremental images of y-displacement field during
the test. Fracture kinking and branching occurred as the fracture approached and crossed the left and right
SPE-184871-MS 15

interfaces. The raw image of fracture trajectory is shown on the right in Figure 20. It was observed that tests
performed under intermediate far-field differential stress had generally more complex fractures than tests
performed under high differential stress.

Figure 20Reference and fractured specimen in Test 8.

Figure 21Evolution of y-displacement field around growing fracture in Test 8.

Test 9
A fracture is initiated in a layer bounded by a harder layer on one side and a softer layer on the other.
Young's modulus and fracture toughness ratios of the left softer layer to the middle layer are 0.51 and 0.68,
respectively, while those ratios of the right harder layer to the middle layer are 2.41 and 1.55, respectively.
The reference raw image labeled with material properties is shown in the top left of Figure 22. The applied
far-field stress is 65 psi. As a result of compressing the specimen in the x-direction, the induced stresses
in the y-direction are compressive in the softer layer across the left interface and tensile in the harder layer
across the right interface. This is expected to occur due to the differences in mechanical properties between
the adjacent layers as explained earlier.
16 SPE-184871-MS

Figure 22Raw images of fracture growth in Test 9.

Figure 23 shows the evolution of the normal strain field in the y-direction with fracture growth in the test.
Initially, the fracture propagated towards the hard layer with a slight kink. The fracture started propagating
into the hard layer first before growing into the soft layer. The raw fracture growth images in Figure 22 also
show that fracture width developed in the hard layer first.

Figure 23Incremental y-normal strain fields of fracture growth in Test 9.

SPE-184871-MS 17

Test 10
In this test, a fracture grows in a bounded layer under an isotropic far-field stress state with zero far-field
stresses. The outer layers are softer on the left and stiffer on the right compared to the middle layer as shown
in the left image of Figure 24. The right image of Figure 24 shows that the fracture propagated parallel to
the interfaces.

Figure 24Raw images of Test 10 before and after fracturing.

Test 11
Using a similar specimen as in the previous test, a fracture is propagated here under very low far- field
differential stress (15 psi) compared to zero differential stress in Test 10. Figure 25 shows the reference
specimen on the left and fractured specimen on the right. The fracture orientation was subparallel to the
layer interfaces.

Figure 25Raw images of Test 11 before and after fracturing.

Test 12
All the previous tests had fairly strong interfaces. Strong interfaces are made by casting adjacent layers
roughly at the same time. By delaying the casting of adjacent layers by a certain time, markedly weaker
interfaces can be made. A specimen with weakly bonded interfaces are used in Test 12 as shown in the
first image of Figure 26. The middle layer is bounded by harder layers in a slightly different configuration
than that of the previous tests. The bounding layers have higher Young's modulus and fracture toughness by
factors of 5.58 and 2.40, respectively, compared to the middle layer. The base far- field differential stress
of 65 psi is applied in the test.
18 SPE-184871-MS

Figure 26Raw images of fracture growth in Test 12.

The fracture initiated and propagated towards the harder layers in a sub-parallel manner to the applied
stress field (Figure 26). Shear failure and slippage along the weak interfaces can be clearly seen in the
evolution of the displacement field in the y-direction as shown in Figure 27. Finally, the fracturing fluid
flowed along the left and right interfaces as can be seen in the last image of Figure 26.
SPE-184871-MS 19

Figure 27Evolution of y-displacement field during fracture growth in Test 12.

Test 13
Here, we examine the effect of relatively smaller layer size on fracture behavior. The width of the middle
layer in which fracture initiates has been about 2 inches in all the tests so far. In this test, a relatively thin
middle layer with about 1-inch width is used in a similar specimen to the one in Test 12 as shown in the
first image of Figure 28. Young's modulus and fracture toughness ratios of the stiffer bounding layers to
the middle layer are 5.58 and 2.40, respectively. The applied far-field differential stress is 65 psi. Figure 28
shows that the fracture initiated and propagated mainly parallel to layer interfaces in a slightly tortuous path
rather than growing in the applied stress direction towards the interfaces.
20 SPE-184871-MS

Figure 28Raw images of fracture growth in Test 13.

Test 14
This is a repeat for the last test where a fracture is induced in the same layered specimen using a higher
differential stress of 80 psi. Similar to Test 13, the fracture propagated parallel to the interfaces with a
relatively more tortuous trajectory as shown in Figure 29.

Figure 29Raw images of fracture growth in Test 14.

SPE-184871-MS 21

Test 15
In this test, a specimen with a similar layer configuration is used with a lower contrast in mechanical
properties between the middle and bounding layers (first image of Figure 30). Young's modulus and fracture
toughness ratios of the bounding layers to the middle layer are 3.58 and 1.80, respectively. A relatively high
differential stress of 80 psi is used. As shown by the raw fracture growth images in Figure 30, one wing
of the fracture grew parallel to layer interfaces while the other wing propagated obliquely towards the left
interface and became an interfacial fracture. The overall fracture trajectory appeared to be markedly less
sinuous than the previous two tests.

Figure 30Raw images of fracture growth in Test 15.

Test 16
A fracture is initiated in a layer bounded by harder, dipping layers as given in the first image of Figure 31. The
base far-field differential stress of 65 psi is applied. The induced fracture propagated towards layer interfaces
noticeably kinking to the low sides of the dipping layers. The fracture reoriented near the interfaces and
intersected them perpendicularly (Figure 31).
22 SPE-184871-MS

Figure 31Raw images of fracture growth in Test 16.

Test 17
In this test, we observe the behavior of a fracture as it approaches a hard barrier embedded in a weaker
matrix. Figure 32 shows the specimen configuration used in this test. The specimen has two hard barriers
that are oriented normal to the applied stress direction. Young's modulus and fracture toughness ratios of
the barriers to to the matrix are 5.58 and 2.40, respectively. The applied far-field stress is 65 psi.

Figure 32Reference raw image of specimen in Test 17.

SPE-184871-MS 23

Figure 33 shows how the displacement field in the y-direction evolves with time during the test. After
reaching the barriers, the fracture branched off to the side of the left barrier and continued propagating along
the applied stress direction. A tensile secondary fracture formed on the side of the right barrier that was not
connected to the parent fracture. Another secondary fracture, connected to the parent fracture, formed next
through the barrier between the first secondary fracture and the parent fracture. During fracture growth in
the test, segments of the fracture were closing gradually as others were forming as can be seen in Figure
33. Fracture growth is also shown by the evolution of the normal strain field in the y- direction in Figure
34. The last image of Figure 34 shows the normal strain field in the x-direction of the last y-normal strain
field image to illustrate the connections between fracture segments. The raw image of the final fracture
trajectory is given in Figure 35.

Figure 33Evolution of y-displacement field during fracture propagation in Test 17.

24 SPE-184871-MS

Figure 34Evolution of y-normal strain field during fracture propagation in Test 17. Bottom
image is the corresponding x-normal strain field of the last y-strain field image above.

Figure 35Raw image of fractured specimen in Test 17.

SPE-184871-MS 25

Test 18
The previous test is repeated here under a higher far-field differential stress of 80 psi. Also, the bonding
between the barriers and the matrix is relatively stronger in this test compared to Test 17. Raw fracture
growth images in Figure 36 show plane fracture propagation through the left barrier and fracture kinking
towards the right barrier. The fracture reoriented to the direction of the applied far-field stress after crossing
the right barrier.

Figure 36Raw images of fracture growth in Test 18.

26 SPE-184871-MS

Conclusions
We have developed a novel experimental capability that allows for direct monitoring of hydraulic fracture
growth in layered porous media. In addition to raw images of fracture evolution, Digital Image Correlation
(DIC) analysis is used to resolve full-field displacements and strains during the process. In this work, a
series of tests were performed in layered specimens with various properties and configurations.
Based on the test results presented here, we observed the following:
1. A propagating fracture approaching a layer interface that is orthogonal to it crosses the layer interface
from a stiff to a soft material in a straight/plane path (Experiment 2).
2. Approaching a layer interface from a soft to a stiff material, a fracture tends to reorient away from
the stiff material, as if trying to avoid it, and it reorients along the applied maximum stress direction
after crossing the interface (Experiments 3, 4 and 5).
3. As a fracture approaches a harder material, the rate of fracture propagation slows down indicating
a higher resistance to fracture growth. Conversely, going from a hard (high Young's modulus) to a
soft material, fracture propagation speeds up towards the softer layer (Experiments 2 and 3). This is
consistent with theoretical predictions from linear elastic fracture mechanics.
4. When a fracture propagates in a layer bounded by different layers, the fracture reaches both layer
interfaces before crossing either interface. This indicates that more energy is required for a fracture to
cross an interface between two materials as compared to fracture propagation in a single material. This
occurs whether going from a hard to a soft material or the other way around (Experiments 2 and 3).
5. The extent of fracture turning/reorientation/kinking increases as the contrast in Young's modulus and
fracture toughness between the layers increases (Experiments 3, 4 and 5).
6. When a fracture propagates from a soft to a hard layer, tensile stresses in the hard side of the interface
can induce a secondary fracture that connects to the main fracture via a localized interface opening
leading to a step-over behavior (Experiment 6).
7. As the far-field differential stress is increased, a fracture will have more of a tendency to propagate
across a layer interface in a straight path without kinking or forming secondary fractures (Experiment
7).
8. If the far-field differential stress is lowered, fractures propagating from softer to harder layers tend
to be have more complex trajectories such as branching, kinking and out of plane fracture growth
(Experiment 8).
9. For zero far-field differential stress (no preferred fracture direction), a fracture tends to initiate and
propagate mainly parallel to layer interfaces (Experiment 10).
10. When the fracture initiates in a relatively thin layer, it tends to propagate mainly parallel to the layer
interfaces or away from the layer interfaces (Experiments 13 and 14).
11. Fracture propagation can be halted by a weak interface due to shear failure and slippage along the
interface (Experiment 12).
12. A fracture initiating in a layer bounded by a stiff layer and a soft layer favors propagation into the
stiff layer (Experiment 9).
13. Fractures tend to kink toward the low side of an interface with a dipping, harder layer (Experiment 16).
14. A fracture approaching a hard barrier embedded in a weak matrix may exhibit a range of behaviors
including plane fracture growth across the barrier, forming secondary fractures, fracture kinking, and
fracture branching (Experiments 17 and 18).

Acknowledgement
The authors would like to thank the members of the Hydraulic Fracturing and Sand Control joint industry
project at UT Austin for supporting the program.
SPE-184871-MS 27

References
Athavale, A. S., Miskimins, J. L. 2008. Laboratory hydraulic fracturing tests on small homogeneous and laminated blocks.
ARMA 08-067. The 42nd U.S. Rock Mechanics Symposium and 2nd U.S.-Canada Rock Mechanics Symposium, San
Francisco, CA, June 29-July 2.
Blaber, J., Adair, B., Antoniou, A. 2015. Ncorr: Open-source 2D digital image correlation Matlab software. Experimental
Mechanics (55, 2015): 1105-1122.
Blanton, T. L. 1982. An experimental study of interaction between hydraulically induced and pre-existing fractures. SPE/
DOE 10847. SPE/DOE Unconventional Gas Recovery Symposium, Pittsburgh, PA, May 16-18.
Fisher, K., Warpinski, N. 2012. Hydraulic-fracture-height growth: Real data. SPE Production & Operations (February
2012): 8-19.
Ganesan, T. P. 2000. Model analysis of structures. Universities Press (India) Limited (2000): p. 100.
Garcia, X., Nagel, N., Zhang, F., Sanchez-Nagel, M., Lee, B. 2013. Revisiting vertical hydraulic fracture propagation
through layered formations - A numerical evaluation. ARMA 13-203. The 47th U.S. Rock Mechanics/Geomechanics
Symposium, San Francisco, CA, June 23-26.
Gu, H., Siebrits, E. 2008. Effect of formation modulus contrast on hydraulic fracture height containment. SPE Production
& Operations (May 2008): 170-176.
Huang, H., Zhang, F., Callahan, P., Ayoub, J. 2012. Fluid injection experiments in 2D porous media. September 2012
SPE Journal: 903-911.
Ouchi, H. 2016. Development of a peridynamics-based hydraulic fracturing model for fracture growth in heterogeneous
reservoirs. PhD Dissertation, The University of Texas at Austin.
Pan, B., Qian, K., Xie, H., Asundi, A. 2009. Two-dimensional digital image correlation for in-plane displacement and
strain measurement: a review. Measurement Science and Technology (20-062001, 2009).
Park, N., Holder, J., Olson, J. E. 2004. Discrete element modeling of fracture toughness tests in weakly cemented
sandstone. ARMA/NARMS 04-553. The 6th North America Rock Mechanics Symposium (NARMS): Rock
Mechanics Across Borders and Disciplines, Houston, Texas, June 5-9.
Simonson, E. R., Abou-Sayed, A. S., Clifton, R. J. 1978. Containment of massive hydraulic fractures. SPE 6089. Society
of Petroleum Engineers of AIME (February 1978): 27-32.
Teufel, L. W., Clark, J. A. 1984. Hydraulic fracture propagation in layered rock: Exerimental studies of fracture
containment. Society of Petroleum Engineers of AIME (February 1984): 19-32.
Thiercelin, M., Jeffrey, R. G., Ben Naceur, K. 1989. Influence of fracture toughness on the geometry of hydraulic fractures.
SPE Production Engineering (November 1989): 435-442.
van Eekelen, H. A. M. 1982. Hydraulic fracture geometry: Fracture containment in layered formations. Society of
Petroleum Engineers of AIME (June 1982): 341-349.
Wu, H., Chudnovsky, A., Dudley, J. W., Wong, G. K. 2004. A map of fracture behavior in the vicinity of an interface.
ARMA/NARMS 04-620. The 6th North America Rock Mechanics Symposium (NARMS): Rock Mechanics Across
Borders and Disciplines, Houston, Texas, June 5-9.
Zhang, X., Jeffrey, R. G. 2008. Reinitiation or termination of fluid-driven fractures at frictional bedding interfaces. Journal
of Geophysical Research (Vol. 113, B08416, 2008).