first break volume 30, March 2012 technical article

Estimation of stress and geomechanical

properties using 3D seismic data
David Gray,1* Paul Anderson,2 John Logel,3 Franck Delbecq,4 Darren Schmidt4 and Ron Schmid5

Abstract
Principal stresses, that is vertical, maximum horizontal and minimum horizontal stresses, and elastic moduli related to rock
brittleness, like Young’s modulus and Poisson’s ratio, can be estimated from wide-angle, wide-azimuth seismic data. This
is established using a small 3D seismic survey over the Colorado shale gas play of Alberta, Canada. It is shown that this
information can be used to optimize the placement and direction of horizontal wells and hydraulic fracture stimulations.

Introduction Closure pressure or closure stress is defined in the

In 2008, the lead author was asked to derive a method that
would increase the relevance of seismic data in shale plays.
After a little research, it became apparent that geomechanical
parameters derived from seismic data could be very important
in shale plays. In most shale plays, and indeed in many other
plays, stimulation of the reservoir by hydraulic fracturing is the
primary mechanism used to increase rates of oil and gas returns
to the surface. In ‘tight’ plays, such as shale gas, shale oil, and
coal-bed methane, stimulation is required to make the wells eco-
nomically viable. Rickman et al. (2008) described four geome-
chanical considerations necessary for the design of a stimulation
treatment. These are: brittleness, closure pressure, proppant size
and volume, and the location of the initiation of the fracture.
Seismic data allow for estimation of all of these values between
existing wells except the proppant size and volume.
The term ‘brittle’ is defined by the Oxford Dictionary as
‘hard but liable to break easily’ and by Webster’s New World
Dictionary as ‘easily broken or shattered’. Both of these defini-
tions are appropriate for use in fracture stimulation, but are
subjective. Although the term brittleness is used frequently in
tight plays, an objective definition was missing until Rickman
et al. (2008) defined a relationship expressing brittleness as a
percentage and related it to Young’s modulus and Poisson’s
ratio, which can both be derived from seismic data (e.g.,
Mallick, 1995 for Poisson’s ratio; Gray, 2005a for Young’s
modulus). In this paper, a generalized form of the Rickman
et al. (2008) percentage brittleness is estimated from Young’s
modulus and Poisson’s ratio, which are both derived from
inversion of seismic data.
Schlumberger Oilfield Glossary as ‘an analysis parameter used
in hydraulic fracture design to indicate the pressure at which
the fracture effectively closes without proppant in place’. In
general it is empirically related to the minimum horizontal
stress (e.g., Iverson, 1995). Expanding on Gray (2010b), a
method for the estimation from seismic data of all three prin-
cipal stresses, the vertical stress, ıv, the maximum horizontal
stress, ıH, and the minimum horizontal stress, ıh, is introduced
here.
Warpinski and Smith (1989) state that in-situ stresses
are clearly the most important factor controlling hydraulic
fracturing. Iverson (1995) shows that knowledge of shear
anisotropy allows for a reformulation of the closure stress
equation to account for non-equal horizontal stresses,
which is generally the case in the earth’s crust, e.g., Bell et
al. (1994). A simplification of Hooke’s law using linear slip
theory (Schoenberg and Sayers, 1995) allows the estimation
of principal stresses from wide-angle, wide-azimuth seismic
data (Gray, 2011). The rock properties required in its imple-
mentation are derived from wide-angle seismic data, e.g.,
Gray (2005a). The method is demonstrated by estimating the
principal stresses and rock properties for the Second White
Speckled Shale using seismic data acquired in central Alberta,
Canada. The results show that only about one quarter of the
Second White Speckled Shale in the survey area will fracture
as a network, while most of the rest will fracture linearly and
some areas will not fracture at all, which implies that this
information can assist in the location of both wells and the
optimal location to initiate fractures, which is the last of the

1
Nexen, 801 - 7th Avenue S.W., Calgary, Alberta, Canada, T2P 3P7.
2
Apache Energy, Level 9, 100 St Georges Terrace, Perth W. Australia, 6000.
3
Talisman Energy Norge, Verven 4, Sentrum, 4003 Stavanger, Norway.
4
Hampson-Russell, Suite 510, 715 - 5th Avenue SW, Calgary, Alberta, Canada, T2P 2X6.
5
CGGVeritas Canada Land Library, 715 - 5th Avenue S.W., #2200, Calgary, Alberta, Canada, T2P 5A2.
*
Corresponding author, E-mail: FDavid_Gray@nexeninc.com

© 2012 EAGE www.firstbreak.org 59

 technical article
Rickman et al. (2008) geomechanical parameters derivable
from seismic data for this shale gas reservoir.
In order to calibrate the closure stress to actual measure-
ments taken in the field, an additional term, called the tectonic
stress term, is often required to be added to the closure stress
equation. Blanton and Olson (1999) state: ‘When independent
measures of horizontal stress magnitudes are available from
micro-fracs or extended leak-off tests, there is often a discrep-
ancy between the log-derived and measured values, leading to
the conclusion that the uniaxial strain assumption inherent to
[their] Eq. 1 is inadequate. In order to improve the estimated
stress values, an adjustment (calibration) is made by adding an
additional stress term to [their] Eq. 1, thereby shifting the profile
to match the measured values.’ In this paper, the strain-corrected
method of Blanton and Olson (1999) is used to calibrate the clo-
sure stress to known values. This method is adopted because it is
expected that strain will change smoothly across bedding planes,
whereas horizontal stress can change rapidly from formation to
formation due to changes in elastic rock properties.
Method
Seismic waves travelling through the earth create small, short-
term, strains in the rock that lie in the elastic regime of the
rocks. Therefore, using the assumptions outlined below, we are
able to calculate useful elastic moduli such as Young’s modulus
and Poisson’s ratio (Rickman et al., 2008) or the LMR (Lamé’s
modulus, shear modulus, and density) parameters of Goodway
et al. (1997) (e.g., Gray, 2002). These moduli are described
as dynamic because they are measured by relatively high-
frequency measurements of velocities of elastic waves (Fjaer,
2009). These dynamic moduli measured by seismic and well
logs occur in the high frequency range and with low strain
amplitudes (Olsen et al., 2004).
When the rock is subject to long-term strain in geotechnical
tests, such as in compressive failure tests used to judge the
strength of rocks, static moduli are estimated from the slope
of the stress-strain relationship (Fjaer, 2009). Therefore, the
moduli related to hydraulic fracturing are most likely the
static moduli, because hydraulic fracturing takes some time to
build up the pressure required to fracture the rock. Olsen et
al. (2004) suggested that the static modulus is a combination
of the dynamic modulus and a non-elastic modulus related
to permanent deformation. The correction of the dynamic
moduli to equivalent static moduli is often done via such a
method or by scaling (e.g., Fjaer, 2009). In addition, there can
be a frequency-dependent effect related to the pore size, the
squirt-flow mechanism described by Dvorkin and Nur (1993).
Seismic-derived dynamic moduli should indicate lateral
changes in elastic moduli and brittleness, which is related to
them (Rickman et al., 2008) between wells. There are many
relationships between static and elastic moduli in the petro-
physical literature, suggesting that the best way to calibrate
them is by comparing local measurements of both. If such rela-
tionships can be determined, then they can be used to constrain
the extrapolation of static moduli away from wells within,
60
first break volume 30, March 2012
say, the area of a seismic survey. This is because the seismic
response is calibrated to the well moduli through the process of
amplitude versus offset (AVO) inversion. AVO inversion using
wide angles can also be used to estimate density from which
we can, in turn, calculate the vertical stress, ıv, through integra-
tion. From ıv and the theory presented below, the minimum
and maximum horizontal stresses, ıh and ıH, respectively, can
be estimated from wide-angle, 3D seismic data.
If we assume that the rocks in-situ in the subsurface are
constrained horizontally, i.e., the horizontal strain is zero in
their natural state, and that the rocks are undergoing elastic
deformation, then we can estimate the in-situ stress state of
these rocks from elastic information (Iverson, 1995). In fact,
estimates of the three principal stresses can be obtained from
anisotropic elastic parameters (Iverson, 1995). Since elastic
information is derivable from seismic data via AVO inversion,
then the in-situ stress state can be estimated from anywhere
where seismic waves have intersected the rocks at a wide
(>40°) angle. Furthermore, the anisotropic stress equations of
Iversen (1995) can be recast using parameters from linear slip
theory (Schoenberg and Sayers, 1995), which can be derived
from wide-angle, wide-azimuth, 3D seismic data (e.g., Lefeuvre
et al., 1992; Varela et al., 2009), as shown here.
Using linear slip theory, Schoenberg and Sayers (1995)
showed that Hooke’s Law can be simplified to:
(1)
where, İi are the strains in the rock, ıj are the stresses in the
rock, and S represents the compliance of the rock. Linear
slip theory simplifies the compliance terms by separating
it into two independent parts: Sb – the compliance of the
background rock, and Sf – the compliance of the fractures
and micro-fractures in the rock (Schoenberg and Sayers,
1995). The incorporation of the concept of micro-fractures is
important because it allows for the relationship of fractures
to stress (e.g., Crampin, 1994), which is required for this
formulation relating linear slip theory to the in-situ principal
stresses. It is assumed that the maximum horizontal stress is
parallel to a single set of parallel vertical fractures or micro-
cracks and the minimum horizontal stress is perpendicular to
these fractures or micro-cracks (e.g., Schoenberg and Sayers,
1995). Expanding Equation (1) with linear slip theory leads
to the following system of equations:
(2)
www.firstbreak.org © 2012 EAGE
first break volume 30, March 2012
where E is Young’s modulus, v is Poisson’s ratio, and μ is
the shear modulus. The parameters ZN and ZT represent the
normal and tangential compliances, respectively.
Compliance is defined by the Merriam-Webster Dictionary
as ‘the ability of an object to yield elastically when a force is
applied’. In the case of linear slip theory, this definition should
be applied to both the background material and the faces of
fractures, i.e., it is the ability of the background material and/
or a fracture in it to yield elastically when a force is applied.
The normal strain component perpendicular to a fracture or
a plane of weakness in the material is related to the material’s
normal compliance, ZN. Shear strain components tangential to
a fracture or a plane of weakness in the material are related to
the material’s tangential compliance, ZT.
Making the assumption that one principal stress is vertical,
so the other two are horizontal, Equation (2) may be used
to write the horizontal strains as functions of these stresses.
Given the assumption that the rocks, in-situ, are constrained
horizontally, these horizontal strains are equal to zero:
(3)
and
(4)
Solving Equations (3) and (4) for the principal horizontal
stresses yields the following equations:
(5)
and
(6)
All of the parameters in Equations (5) and (6) can be derived
from seismic data (e.g., Mallick, 1995 for v; Gray, 2005a for
E). Downton and Roure (2010) use modern wide-angle, 3D,
seismic data to estimate ZN. The vertical stress is generally
estimated by integrating logged density over depth and mul-
tiplying by the acceleration due to gravity. As density can be
estimated from seismic data (e.g., Van Koughnet et al., 2003),
the vertical stress can also be estimated integrating the density
estimated from seismic data. This means that the anisotropic
in-situ stress state of the rock and its elastic moduli and brittle-
ness can be estimated between wells through the use of modern
3D seismic data.
Since we know how to derive all these parameters from
seismic data, we can also estimate all three principal stresses
from seismic data. These stresses need to be calibrated to
static stress measurements obtained from wells, and when this
is done it is found that the horizontal stresses are frequently
underestimated. Fortunately, the concept of tectonic stress has
already been introduced in the engineering world to deal with
this discrepancy. The formula for solving for stress is obtained
© 2012 EAGE www.firstbreak.org
technical article
by assuming that the vertical stress is set equal to the overbur-
den stress and horizontal strains are set to zero. Thus, the only
deformation allowed is uniaxial strain in the vertical direction.
When in-situ minimum horizontal stress measurements
such as micro-fracture(1) tests or extended leak-off tests are
available, there is often a discrepancy between the log-derived
values (and therefore, seismic-derived values, since they are
based on inversions to match log properties) and these meas-
ured values, leading to the conclusion that the uniaxial strain
assumption inherent is inadequate. In order to improve the
estimated stress values, the stress is shifted by adding an
additional ‘tectonic stress’ term to the equation to match the
measured values (Blanton and Olson, 1999). Rather than
adding a tectonic stress term, we use the tectonic strain method
described by Blanton and Olsen (1999) to estimate a vertically
variable shift in stress. The primary rationale for using tectonic
strain rather than tectonic stress is that the horizontal strains
are expected to be more continuous than the horizontal stress-
es. This can be conceptualized through the realization, from
Equations (5) and (6), that the horizontal stresses are functions
of the continuous vertical stress multiplied by a function of the
discontinuous parameters, E, v, and ZN. Then again, the fact
that this calibration needs to be done implies that the rocks in
question are undergoing horizontal strain, thereby revising our
earlier assumption that the rocks are horizontally constrained.
Nevertheless, this strain can be captured in the above equations
by incorporating the tectonic strain term of Blanton and Olsen
(1999). Thereafter, at the well, the horizontal stresses are
calibrated to the well measurements and can then be used as
estimates of inter-well stresses and, therefore, for well planning.
It turns out that a differential horizontal stress ratio
(DHSR), i.e., the differential ratio of the maximum and
minimum horizontal stresses as shown in Equation (7), can
be calculated from seismic parameters derived from these
equations alone, without any knowledge of the stress state of
the reservoir, because ıv in Equations (1) and (2) cancels when
calculating it:
(7)
DHSR is a very important parameter in determining how a res-
ervoir is likely to fracture. In general, in hydraulic fracturing, it
is preferable if it is small. When the DHSR is large, hydraulic
fractures will tend to occur as non-intersecting planes parallel
to the maximum horizontal stress, since fractures tend to be
created parallel to it. In contrast, when the DHSR is small,
fractures induced by hydraulic fracturing will tend to grow
in a variety of directions and therefore intersect. This multi-
directional fracture network tends to provide much better
access to the hydrocarbons in the reservoir.
The knowledge of elastic parameters, such as Young’s
modulus or the shear modulus, indicates which rocks are brit-
tle and therefore likely to fail. Rickman et al. (2008) derived a
means of estimating brittleness, B, from logs. Equation (8) is a
generalized form of these equations. Brittleness is expressed
61
 technical article
as a percentage. These same equations can be applied to the
inversion results because they are tied to these same logs:
(8a)
(8b)
and
(8c)
where BE is brittleness estimated from Young’s modulus, Bn is
brittleness estimated from Poisson’s ratio and subscripts min
and max denote minimum and maximum values.
In order to determine how these rocks will fracture under
the stresses induced by hydraulic fracturing, knowledge of all
these parameters is important. To fracture the rock, the pres-
sure in the well must first overcome the hoop stress created in
the rocks around the borehole. Then it must be greater than
the far-field stresses away from the borehole in order to con-
tinue its growth. The hoop stress and far-field stresses may be
similar, but the possibility exists that they can be completely
different from each other in both maximum stress direction
and magnitude. In the example below, the hoop stress is about
twice the far-field stress, so knowledge of both is critical. Such
differences can have a significant impact on wellbore design.
The hoop stress is the additional tangential stress in the
rock close to the borehole wall that is induced by the presence
of the borehole. It can be estimated once the principal stresses
are known. Since it has been shown above that the principal
stresses can be estimated from seismic data, the hoop stress
can also be estimated. The hoop stress, ıșș, at the borehole
wall for three cases is considered here. For a vertical borehole,
the minimum hoop stress is oriented parallel to ıh and is given
by
(9)
For the top of a horizontal borehole oriented perpendicular to
ıH, with ıv > ıH,
(10)
For the side of a horizontal borehole oriented perpendicular to
ıH, with ıH > ıv,
(11)
The general case was given by Kirsch (1898). Following Lund
(2000) and in the example below, it is assumed that the differ-
ence, ΔP, between the fluid pressure in the borehole and the
formation pore pressure is zero. It is necessary to overcome
the hoop stress in order to initiate a fracture in the formation;
therefore, these stresses can be considered as the formation
breakdown pressure, PFB.
62
first break volume 30, March 2012
Knowledge of the stress state prior to drilling can also
be used to predict areas at risk for wellbore failure, thereby
allowing for bypassing of these areas or preparation for the
drilling problems likely to be encountered therein. For example,
note that any of the formation breakdown pressures could be
negative, then the formation will likely fracture solely due to the
presence of the borehole. It is important to know this possibility
before drilling because there is a risk of lost circulation in that
borehole. This scenario does occur and is shown in the example
below. The hoop stress can also be significantly higher than
the closure stress. Therefore, it is important to be aware of this
prior to attempting to initiate hydraulic fracture. Wide-angle,
wide-azimuth, 3D conventional P-wave, or multi-component
seismic data allow the estimation of these parameters before
wells are drilled. For conventional seismic data, azimuthal
inversion is used. For multi-component seismic data, birefrin-
gence, registration, and inversion can be used. These parameters
must be calibrated to known points of stress and elastic moduli
within the reservoir. Then, using this additional knowledge, bet-
ter well planning should be achievable in order to optimize well
location, direction and location of hydraulic fractures. How this
might be done is demonstrated in the following example.
Example
A small, 9 km2, conventional P-wave, wide-azimuth, 3D
seismic survey shot southeast of Red Deer, Alberta, Canada
in the Western Canadian Sedimentary Basin (WCSB) is used
to demonstrate the method. The target area is the Cretaceous
Colorado Shale Group, highlighted by a dashed box in the
figures, which sits above weak shales of the Joli Fou Formation
and sandier Lower Cretaceous Mannville Formation. The
Mannville Formation sits above an unconformity, below which
the strata consist predominantly of carbonate rocks. It will be
shown that the Second White Speckled Shale unit is the most
brittle part of the Colorado Shale Group and will be the focus
of this example. Figure 1 shows the DHSR as plates displayed
over estimates of Young’s modulus for the entire 3D volume.
The direction of the plates indicates the estimated direction of
ıH, and their size, the magnitude of DHSR (Figure 2). These
show significant variations over this very small area, largely
driven by ıh (Figures 3 and 4).
Estimates of the principal stresses are derived as described
above and calibrated to stress estimates for the whole WCSB
available from the WCSB Atlas (Bell et al., 1994) using the
tectonic strain concept of Blanton and Olsen (1999). Local
calibration is preferred, but no wells with stress estimates in the
Colorado Shale Group are available in this area. For example,
subsequent stress estimation projects performed by the authors
have used information derived from local well-based closure
stress estimates to calibrate the minimum horizontal stresses.
Figures 3 and 4 show significant vertical and horizontal
variations in ıh. Very low values are associated with Mannville
Formation coals and the weak shales of the Joli Fou Formation.
High values are associated with carbonates in the bottom of the
figures below the unconformity.
www.firstbreak.org © 2012 EAGE
first break volume 30, March 2012 technical article

Figure 1 Dynamic Young’s modulus from the 3D seismic volume shown in colour, with the scale numbered in units of GPa. Plates, indicated by the arrow, show
the differential horizontal stress ratio (DHSR). The size of the plate is proportional to the magnitude of the DHSR and the direction of the plate shows the
direction of the local maximum horizontal stress. The long axis of the survey is E-W and survey area is 9 km2.

Figure 2 Maximum horizontal stress, ıH , showing the target Colorado Shale and the Second White Speckled Shale (2WS) formation within it. The coloured stripes
at the well locations are the vertical stress, ıv , for reference.

These variations are probably due to different litholo-
gies being stressed differently as they drape over underlying
pinnacle reefs of the Leduc Formation (Gray, 2005b), dif-
ferential compaction stresses, stresses induced by underlying
salt collapse, or stresses due to the presence of open fractures
in this formation. The horizontal slice in Figure 1 shows 100%
variations in Young’s modulus ranging from 12–24 MPa.
Variations in Young’s modulus and Poisson’s ratio should be
expected due to variations in lithology, porosity, fluid content,
fractures and cement in any sedimentary rock. Brittleness is
estimated using the generalization of the approach of Rickman
et al. (2008), described above, combining Young’s modulus and
Poisson’s ratio as shown for these data in Figure 5. Furthermore,
there are significant variations in the DHSR ranging from
zero to over 100% (Figure 6). Both of these parameters have
significant impact on whether and how the rock will fracture:
higher values of brittleness indicate rocks that are more likely
to fracture and lower values of the DHSR indicate areas where
rocks will have a greater tendency to fracture into a network.
Clearly, we want to find areas where both these properties

© 2012 EAGE www.firstbreak.org 63

 technical article first break volume 30, March 2012

Figure 3 Minimum horizontal stress, ıh , showing the target Colorado Shale and the Second White Speckled Shale (2WS) formation within it. The coloured stripes
at the well locations are the vertical stress, ıv , for reference.

Figure 4 Map of minimum horizontal stress, ıh, over the Second White Speckled Shale. Note its variability over a short distance from about 20 MPa around
well 14-13 to 26 MPa around well 6-13. North is to the top and east to the right. The red line towards the top of the figure shows the location of the sections
shown in other figures. After Gray 2010a.

occur in order to find the best areas for hydraulic fracturing.
Young’s modulus is used as a proxy for brittleness throughout
the remainder of this study.
This requirement for high brittleness and low DHSR
suggests that a crossplot of these values should indicate areas
that are suitable for the creation of fracture networks. Figure 6
is a demonstration of such a crossplot using the seismic data
for the Colorado Shale Group, which consist primarily of
Late Cretaceous shales. The result of applying this crossplot to
the Second White Speckled Shale Formation of the Colorado
Group is shown in Figure 7. The zones and cut-offs in the
crossplot should be optimized through well control as the field
is developed. Areas where the DHSR is small and Young’s
modulus is large (green in Figure 7) are confined to small areas
of this already small survey. These areas, such as the green
zone in the middle of Figure 7, should be drilled first, which
should move the best production forward, thereby increasing
net present value.
It is significant that there are such large differences in the
properties that are important for hydraulic fracturing over very
short distances. Examination of the variations to the southwest
of Well 14-13 in the north-central part of Figures 7 and 4 show
a change from a very good area for hydraulic fracturing to a
bad area over a distance of less than 200 m, much shorter than
the length of most horizontal wells drilled in these plays. The
results indicate that most of the Second White Speckled Shale
will fracture with parallel fractures, implying that it will be very
important to direct the horizontal wells intended to be fractured
as nearly parallel to the average ıh direction as possible.
Therefore, knowledge of the rapid variation in the direction of

64 www.firstbreak.org © 2012 EAGE

first break volume 30, March 2012 technical article

Figure 5 Brittleness in percent, estimated using our generalization of the methods of Rickman et al. (2008). The green stripes at the well locations are there
because there is no equivalent brittleness log in the project.

Figure 6 Crossplot of DHSR versus Young’s

modulus. Preferred areas for hydraulic fractur-
ing are indicated in green, less desirable areas
in yellow, and poor areas in red.

Figure 7 Map of Second White Speckled Shale showing zones highlighted in the crossplot in Figure 6. Green suggests where fracture swarms will form, red sug-
gests the rocks are more ductile and thus less likely to fracture, and yellow suggests where aligned fractures are likely to occur. There are considerable variations
in the north-central part of the survey, where the data suggest areas good and bad for hydraulic fracturing within about 100 m of each other.

© 2012 EAGE www.firstbreak.org 65

 technical article first break volume 30, March 2012

ıh, or ıH, as shown in Figure 1, is important information for

optimal hydraulic fracturing of this reservoir. This rapid varia-
tion in the stress direction is consistent with the interpretation
of the causes of fracture orientation in underlying coalbeds and
sandstones described by Gray (2005b). He ascribed these frac-
tures to local stress induced by drape over underlying pinnacle
reefs and differences in brittleness due to lithology. They could
also be caused by differential compaction or collapse associated
with underlying salt dissolution.
Areas indicated in red in Figure 7 have a low likelihood
of fracturing and therefore should be assessed carefully before
drilling. There will likely be difficulty fracturing these rocks.
However, one possible reason for the low level of brittleness is
high total organic carbon content, which is potentially beneficial
to production. This reason is just one example of why all levels
of brittleness should be properly assessed before proceeding
with drilling, which can only be done with seismic data. Assume
that one well per square kilometre is required to develop this
shale gas play with drilling and completion costs of US$8 mil-
lion per well, which are similar to numbers that are commonly
reported (e.g., Crum, 2008), and assume that the areas in red
in Figure 7 will not fracture. This study shows that only three-
quarters of the prospect will fracture, meaning that two of the
nine wells should not be drilled. The cost saving, by not drilling
the two wells, would be $16 million, reducing the total cost to
develop this 9 km2 area from $72 million to $56 million. These
savings easily pay for the acquisition, processing, and analysis
of the seismic data used to derive this information, and are in
addition to the uplift in net present value derived by drilling and
completing the best areas first and additional value of deriving
field scale development plans (e.g., orientation of horizontal
wells on a large scale for optimum pad placement and planning).
Figure 8 The hoop stress concept: ‘If a circular hole is made in a homogeneous
body experiencing a homogeneous stress field, stress will concentrate around
the hole since no force can be carried through the interior void. [This] Figure
… shows the stress concentration around the hole in a body under uniaxial
compression, of magnitude ı1R in the far-field, in the x-direction’ (Lund, 2000).
Kirsch’s (1898) concept of hoop stress (Figure 8) can be
used to estimate PFB given the principal stresses that have been
estimated already. Hoop stress is generated by the removal of a
cylinder of material caused by the drilling of a well. By estimat-
ing hoop stresses from the principal stresses derived earlier, we
can get an idea of what will happen while drilling before a well
is drilled. Here we assume that the pressure in the borehole is
close to the formation pressure in this estimation (although if
wellbore pressures, e.g. mudweights, are known then they can
be used). Since from the above analysis we now have estimates
of principal stresses everywhere, we can see what will happen
everywhere before a well is drilled and thereby avoid areas
where there are potential wellbore stability issues.
In Figure 9, we can see that there are potential vertical
wellbore stability issues (green zones indicating very low PFB)
in vertical wells in the Mannville coalbeds (below the Colorado

Figure 9 Fracture breakdown pressure (PFB ) at the side of a vertical borehole. Areas shown in green indicate areas of very low or negative PFB indicating a sig-
nificant risk of borehole failure at these locations. Many of these are in the Mannville coals at about 1000 ms, which is not surprising, but there are also some
in the primarily carbonate section under the unconformity marked by the blue line at about 1075 ms. This information has potential importance for drilling.
The coloured stripes at the well locations are the vertical stress, ıv , for reference. After Gray 2010a.

66 www.firstbreak.org © 2012 EAGE

first break volume 30, March 2012 technical article

Figure 10 PFB at the side of a horizontal borehole. Lower values indicate where it is easier to initiate fractures from the side of the borehole. Values ranging
from about 20 MPa to over 100 MPa show that it is important this information be considered in well planning. The target Second White Speckled Shale (2WS)
Formation has moderate PFB of 40-50 MPa, which is significantly greater than its closure stress, indicating that a significant amount of force will be required to
initiate the fracture at the borehole, but that once the fracture is initiated, then it should continue to propagate until the pressure drops below the closure
pressure of 20-50 MPa. The coloured stripes at the well locations are the vertical stress, ıv, for reference. After Gray (2010a).

Figure 11 Difference of fracture breakdown pressure for a horizontal well, PFB(side) – PFB(top). Positive values mean that the fracture will initiate out of the sides
of a horizontal borehole. Negative values mean that fractures will initiate from the top of the borehole.

Shale), which are expected, and in the carbonate section below
the unconformity at the base of the Mannville section, which
is unexpected. We can also drill wells in areas where the hoop
stress is in a range such that the pressure generated in the well-
bore during drilling is lower and the pressure generated during
hydraulic fracturing is sufficient to overcome it and break the
rock without damaging the casing. Since horizontal wellbores
will be drilled here, we should examine the fracture breakdown
pressure for a horizontal wellbore (Figure 10) before drilling.
We can also estimate whether the fracture will initiate in the
top of the borehole or on its side (Figure 11). The closure stress
and rock properties, derived earlier, predict how the fracture
will behave away from the borehole.
Discussion
It has been demonstrated that seismic data can be used to
estimate rock brittleness and stresses away from boreholes.
The combination of these estimates allows for assessment of
the potential for hydraulic fractures and geomechanical issues
in boreholes before drilling any wells. An example from the
Colorado Shale of Alberta is used to demonstrate the method.
This method is based on physics and so is applicable in any shale
gas play as well as any other play where hydraulic fracturing is
being considered, such as tight gas sands, coalbed methane, and
carbonates, and where stress or geomechanics are at issue, such
as in heavy oil plays and new exploration. The seismic data
should be used to extrapolate information derived at wellbores

© 2012 EAGE www.firstbreak.org 67

 technical article
to the inter-well regions. However, this analysis can be done
without well information using measured seismic velocities and
assumed petrophysical relationships. Once principal stresses are
estimated, other useful stress estimates, such as hoop stress, can
be used to predict how the well will behave while drilling and
under stimulation.
Conclusions
There is a tremendous amount of information on stresses and
rock properties that can be estimated from wide-angle, wide-
azimuth, 3D seismic data. The horizontal stresses derived from
this method need to be calibrated to stress values estimated
from well data. Once this is done, these stresses can be used to
estimate various stresses likely to be encountered between wells.
Furthermore, the elastic properties derived during this process,
such as shear modulus, Young’s modulus, and Poisson’s ratio,
which also require static moduli for calibration, can be used to
estimate the brittleness of the rock between wells. This informa-
tion can then be used to plan well paths and optimize hydraulic
fracture programmes.
Acknowledgements
We thank CGGVeritas, EnCana, Apache, Talisman Energy,
Magnitude, Fairborne Energy, Nexen, Bill Goodway, Doug
Anderson, Ron Larson, Jonathan Miller, Louis Chabot, Eric
Andersen, Jared Atkinson, John Varsek, Tobin Marchand,
Dominique Holy, Colin Wright, Wayne Nowry, Jessie Arthur,
Philip Janse, Christophe Maison, Dragana Todorovic-Marinic,
Christian Abaco, Lee Hunt, Scott Reynolds and Jon Downton.
Note
(1)
A short duration, small volume fracturing operations
where a small amount (<100 BBLS) of KCL water is
pumped until fracture initiation. At that point the wing
or frac valve is closed allowing the well’s pressure to fall-
off naturally over the course of 24 to 48 hrs. (or longer).
References
Bell, J.S., Price, P.R. and McLellan, P.J. [1994] In-situ stress in the
Western Canada Sedimentary Basin. In: Mossop, G.D. and Shetsen,
I. (Compilers) Geological Atlas of the Western Canada Sedimentary
Basin, Canadian Society of Petroleum Geologists and Alberta Research
Council, Special Report 4. http://www.ags.gov.ab.ca/publications/
wcsb_atlas/atlas.html
Blanton, T.L. and Olsen, J.E. [1999] Stress magnitudes from logs: effects of
tectonic strains and temperature. SPE 54653.
Crampin, S. [1994] The fracture criticality of crustal rocks. Geophysical
Journal International, 118, 428–438.
Crum, J. [2008] 2008 Analyst Review, Apache Corporation, Canada.
http://www.apachecorp.com/resources/upload/prevarticlefiles/files/
apache_2008_analyst_review_08_canada.pdf
Dvorkin, J., Hoeksema, R.N. and Nur, A. [1994] The squirt-flow mecha-
nism: macroscopic description. Geophysics, 59, 428–438.
Fjær, E. [2009] Static and dynamic moduli of a weak sandstone. Geophysics,
74, WA103–WA112.
68
first break volume 30, March 2012
Goodway, W., Chen, T. and Downton, J. [1997] Improved AVO fluid
detection and lithology discrimination using Lamé petrophysical
parameters; ‘Ȝȡ’,’μȡ’, & ‘Ȝ/μ fluid stack’, from P and S inversions. 67th
SEG Annual Meeting, Expanded Abstracts, 16, 183–186.
Gray, D. [2002] Elastic inversion for Lamé parameters, 72nd SEG Annual
Meeting, Expanded Abstracts, 21, 213–216.
Gray, D. [2005a] Estimating compressibility from seismic data. 67th EAGE
Conference & Exhibition, Extended Abstracts, P025.
Gray, D. [2005b] Seismic anisotropy in coal beds. 75th SEG Annual
Meeting, Expanded Abstracts, 24, 142–45.
Gray, D. [2010a] Targeting horizontal stresses and optimal hydraulic
fracture locations through seismic data. 6th Annual Canadian Institute
Shale Gas Conference. http://www.cggveritas.com/technicalDocu-
ments/cggv_0000006453.pdf
Gray, D., Anderson, P., Logel, J., Delbecq, F. and Schmidt, D. [2010b]
Estimating in-situ, anisotropic, principal stresses from 3D seismic. 72nd
EAGE Conference & Exhibition, Extended Abstracts, J046.
Gray, F.D. [2011] Methods and systems for estimating stress using seismic data.
United States Patent Application, 20110182144A1. http://www.google.
com/patents/US20110182144.pdf?source=gbs_overview_r&cad=0.
Iverson, W.P. [1995] Closure stress calculations in anisotropic formations.
SPE 29598.
Kirsch, G. [1898] Die Theorie der Elastizität und die Bedürfnisse der
Festigkeitslehre. Zeitschrift des Vereines deutscher Ingenieure, 42,
797–807.
Lefeuvre, F., Nicoletis, L., Ansel, V. and Cliet, C. [1992] Detection and
measure of the shear-wave birefringence from vertical seismic data:
theory and applications. Geophysics, 57, 1463–1481.
Lund, B. [2000] Stress around a borehole. http://www.geofys.uu.se/bl/Avh/
node8.html
Mallick, S. [1995] Model-based inversion of amplitude-variations-with-
offset data using a genetic algorithm. Geophysics, 60, 939–954.
Olsen, C., Fabricius, I.L., Krogsbøll, A. and Prasad, M. [2004] Static
and dynamic Young’s Modulus for Lower Cretaceous chalk. A low
frequency scenario. AAPG International Meeting, Extended Abstracts,
89511.
Rickman, R., Mullen, M., Petre, E., Grieser, B. and Kundert, D. [2008] A
practical use of shale petrophysics for stimulation design optimization:
all shale plays are not clones of the Barnett Shale. SPE 115258.
Schoenberg, M. and Sayers, C.M. [1995] Seismic anisotropy of fractured
rock. Geophysics, 60, 204-211.
Van Koughnet, R.W., Skidmore, C.M., Kelly, M.C. and Lindsay, R.O.
[2003] Prospecting with the density cube. The Leading Edge, 22,
1038–1045.
Varela, I., Maultzsch, S., Chapman, M. and Li, X. [2009] Fracture density
inversion from a physical geological model using azimuthal AVO
with optimal basis functions. 79th SEG Annual Meeting, Expanded
Abstracts, 28, 2075–2079.
Warpinski, N.R. and Smith, M.B. [1989] Rock mechanics and fracture
geometry in recent advances in hydraulic fracturing. In: Gidley, J.L.
(Ed.) Recent Advances in Hydraulic Fracturing. SPE, Monograph 12,
57–80.
Received 24 January 2011; accepted 10 January 2012.
doi: 10.3997/1365-2397.2011042
www.firstbreak.org © 2012 EAGE