Design of Caprock Integrity in Thermal
Stimulation of Shallow Oil-Sands Reservoirs
Yanguang Yuan, SPE, and Bin Xu, SPE, BitCan G&E Incorporated; and
Claes Palmgren, SPE, Value Creation Incorporated
Summary
Stakeholders in in-situ oil-sands development take caprock-integ-
rity issues seriously. The industry is faced with the challenge of
determining an optimal operating pressure in the reservoir where,
in general, the pressure should stay significantly low to ensure the
caprock integrity while being significantly high for enhanced oil
production and economics. This paper presents a comprehensive
work program on the subject for a shallow oil-sands play.
Caprock integrity considers the induced stress and deformation
in a caprock during the thermal stimulation of an oil-sands reser-
voir. A minifrac-test program is undertaken to define the original
in-situ stress state. Laboratory tests are carried out to measure the
deformation and strength properties. Simulations are run to calcu-
late the induced stresses and evaluate them against the mechanical
strength. This paper describes some important quality-control
issues for these activities. For the minifrac tests, multiple cycles
and use of flowback are promoted for enhanced efficiency and ac-
curacy. Laboratory tests are recommended on whole cores in a
drained condition at a slow strain rate. Numerical simulations
should use site-specific and laboratory-measured material proper-
ties. On the basis of the limited sensitivity analyses, the thermal-
expansion coefficient of the reservoir and Young’s modulus of the
caprock are found to significantly affect the caprock deformation
and/or induced stresses.
Introduction
Collectively, caprock refers to a certain interval in the overburden
rock formations above a petroleum reservoir containing the reser-
voir fluids within the reservoir. It is normally shaly, with high
clay content and low permeability. Sometimes, it immediately
overlies the pay zone. In other cases, there is a buffer zone be-
tween the caprock and pay zone. During petroleum exploitation,
the caprock plays an important role in safeguarding against the
hydrocarbon fluid, stimulating materials, and/or their mixture
invading zones above the caprock. Often, these zones contain
groundwater aquifers.
Ultimately, caprock integrity considers hydraulic integrity—
no reservoir fluids should escape through the caprock into the
groundwater aquifers or to the surface. In general, the hydraulic
integrity is already maintained naturally, as in the geological his-
tory of the caprock preventing further upward hydrocarbon migra-
tion. It is the process-induced mechanical deformation and
potential failure of the caprock during thermal operations that
may introduce new hydraulic conduits and thus compromise the
hydraulic integrity. Therefore, hydraulic integrity becomes a me-
chanical integrity issue.
Caprock integrity also considers caprock mechanical integrity
(i.e., deformation and failure in the caprock strata). For example,
surface heave, which is rock deformation reflected on the ground,
can alter the environment by changing the landscape or the surface
or shallow subsurface hydrogeological conditions. Such surface
heave could damage surface installations and infrastructures, and
have other unintended impacts. Furthermore, rock deformation
Copyright V
C 2013 Society of Petroleum Engineers
This paper (SPE 149371) was accepted for presentation at the Canadian Unconventional
Resources Conference, Calgary, 15–17 November 2011, and revised for publication.
Original manuscript received for review 22 November 2011. Revised manuscript received for
review 18 December 2012. Paper peer approved 17 May 2013.
266
and failure (e.g., reactivation of pre-existing weak planes) can
damage the well casing, breaking its hydraulic-sealing capacity.
Whether hydraulic or mechanical, caprock integrity becomes a
geomechanical issue related to caprock deformation and potential
failure. Caprock-integrity analysis compares the prevailing stress
conditions against the material strength. The ongoing stress condi-
tion is the induced stresses superimposed on the virgin in-situ
stresses. Therefore, there are three major components to any geo-
mechanical work program: determination of the original in-situ
stresses, evaluation of the induced stresses, and measurement of
the mechanical properties. Minifrac tests are the most reliable
method to measure the in-situ minimum stress; the induced
stresses are normally inferred from geomechanical simulations;
and the mechanical properties should be measured ideally from
cores obtained from the project sites. This general geomechanical
work program is followed in the oil-sands industry now. How-
ever, special considerations are warranted for oil-sands develop-
ment where the formations are relatively shallow. This paper will
present some relevant examples of how to ensure quality control
in these special cases.
Both the industry and government regulatory agencies take the
caprock-integrity issue seriously. For the in-situ development of
the Alberta oil-sands reservoirs, systematic efforts in this regard
were first initiated for cyclic-steam-stimulation (CSS) operations
(Smith et al. 2004). More recently, steam-assisted gravity drain-
age (SAGD) has become another mainstream commercial in-situ
oil-sands recovery process. SAGD is normally thought to be gen-
tler on the caprock because it operates at a lower pressure than
CSS. However, as the operational experience with SAGD grows,
evidence indicates that proactive precautions are still necessary to
safeguard the caprock integrity. Yuan (2008) presented theoretical
arguments about why attention also should be paid to integrity
issues in SAGD. The current paper will apply these theoretical
principles in the context of shallow oil-sands reservoirs.
In the following sections, case histories are given to illustrate
important quality-control issues in various caprock-integrity stud-
ies. First, complexities encountered in the oil-sands and the corre-
sponding best practices during minifrac tests are described. Next,
geomechanical laboratory tests are explored. The subsequent sec-
tion is dedicated to a study of the geomechanical simulations
which combine both the field-obtained data and laboratory-meas-
ured material properties to derive the safe SAGD operating pres-
sure. The effect of reservoir depth on caprock integrity is also
specifically discussed.
The target reservoir discussed in this paper belongs to the
McMurray formation sandstones of the Lower Mannville in the
Fort McMurray area. Fig. 1 offers a schematic about the general
stratigraphic column. The reservoir is generally less than 100 m
deep, but its thickness is approximately 40 m, making the asset
attractive. The Clearwater formation constitutes the regionally
continuous caprock, which is mostly shale with interbedded silty-
to-sandy mudstone. The caprock thickness ranges from 46 to 61
m. A Wabiskaw member is sandwiched between the Clearwater
caprock and McMurray pay zone. It is a marine shore face system
conformably overlying the McMurray formation. It may also con-
tain sand facies that can be bitumen-saturated. Therefore, the
Wabiskaw member constitutes a transitional buffer zone from the
reservoir pay zone upward to the caprock. A solvent-assisted low-
pressure SAGD is proposed to develop the reservoirs (Palmgren
et al. 2011). Significant efforts have been commissioned by the
July 2013 Journal of Canadian Petroleum Technology

Quaternary Till
Upper Clearwater Shales
Lower Clearwater Shales
Wabiskaw Shales
McMurray Oil Sands
Fig. 1—A schematic about the general well stratigraphic col-
umns encountered in the shallow reservoir.
operator, some of which are still ongoing, to design an operating
pressure that maintains the caprock integrity.
Minifrac Tests
Minihydraulic-fracturing stress tests, commonly called minifrac
tests, have been routinely used by the industry to measure the in-
situ stresses. They are regarded widely as the most reliable
method to define the in-situ minimum stress, Smin. Using con-
trolled high-pressure water injection, the minifrac tests create a
fracture and propagate it for a sufficient distance from the injec-
tion well and into the formation. This ensures the fracturing
behaviour being dominated by the far-field stress condition. The
pressure data are analyzed to estimate the fracture-closure pres-
sure Pc. The latter can then be equated to the in-situ minimum
stress acting perpendicular to the fracture. For more background
information on minifrac tests in the industry, one may refer to the
review paper by Hudson et al. (1993). Bell et al. (1994) summar-
ized the in-situ stress measurements made mostly at deep depths
in the western Canadian sedimentary basin. Some tests were per-
formed in the context of oil sands [e.g., those reported in Settari
and Raisbeck (1979), Chhina and Agar (1985), Chhina et al.
(1987), Kry (1989), Proskin et al. (1990), Kry et al. (1992)]. Our
test procedures and analysis methods incorporate the learnings
from these earlier works and also implement new developments
and/or adjustments dedicated to the shallow depths and the uncon-
solidated nature of the oil sands and overlying caprock intervals.
Sv as a Quality-Check Index. At the shallow depths associ-
ated with the target oil-sands reservoir, the original in-situ stress
condition can be defined by three principal normal stress compo-
nents: the vertical stress Sv and two horizontal stresses, which are
commonly denoted as the maximum and minimum horizontal
stresses, SHmax and SHmin, respectively. Sv can be estimated reli-
ably from the overburden lithological column by integrating the
density log, which typically ranges from 20 to 22 kPa/m. As a
rule of thumb, Sv¼21 kPa/m.
In theory, a properly executed and interpreted minifrac test
should never measure an Smin larger than the density-derived Sv.
Hydraulically driven fracture propagation is always perpendicular
to the direction of Smin. This direction represents the least resist-
ance (i.e., requiring the smallest pressure to extend the fracture).
Away from the mechanical influence domain of the borehole, the
fracture should be perpendicular to the least of the three stress
components: Sv, SHmax, and SHmin. Therefore, if Sv is the mini-
July 2013 Journal of Canadian Petroleum Technology
mum stress, it should be detected by the minifrac tests and the frac-
ture being measured is horizontal. If the minifrac tests detect an
Smin smaller than the density-derived Sv, it means the fracture
being measured is vertical. The measured stress represents SHmin.
Thus, the fracture-closure pressure measured in a minifrac test is,
at most, equal to Sv and in any case should never be larger than Sv.
Three test programs were conducted for the target shallow res-
ervoir in the particular study. They were performed by three dif-
ferent service providers on four different wells. The first test
program was completed in 2009 by a wireline unit. It measured
fracture-closure-pressure gradients at 24 to 36 kPa/m [Appendix 2
and Appendix 3 of Clearwater Pilot Application submitted to
Energy and Resources Conservation Board (ERCB) of Alberta,
January 2010]. Doubts should be raised about these values
because they are larger than the density-derived Sv at approxi-
mately 21 kPa/m. Indeed, subsequent analysis led to a conclusion
that most measurements provided data of low confidence and
should be deemed inconclusive, although one successful measure-
ment did provide a closure-pressure gradient of 20.57 kPa/m.
A borehole is a stress concentrator. Stresses around a borehole
are higher than the far-field in-situ stress condition. Therefore, a
proper minifrac test should inject a sufficient volume of liquid for
the fractures to propagate outside of the stress-concentration area
and for the majority of test length analyses to focus on far-field
stresses. In the previously described tests completed in 2009, the
total injected volume was only up to 10 L per test. This small injec-
tion volume may be one of the reasons for the unusually high frac-
ture-closure pressure as described previously. The fracture may
have stayed within the influence domain of the borehole. There-
fore, the interpreted fracture-closure pressure reflects the stress
concentration near the borehole, not the far-field stress condition.
Multiple Test Cycles for Consistency Check. The in-situ
stress magnitudes are small at shallow depths. Thus, accurate
interpretations of the fracture closure become an important issue.
For example, a 50-kPa inaccuracy in the interpreted closure pres-
sure, Pc, is equivalent to 0.5 kPa/m at a 100-m depth, but only
0.05 kPa/m if the test depth is 1000 m. Two measures can help to
enhance the interpretation’s certainty. One is to use multiple
injection and shut-in cycles, and the other is to use flowback, as
described later. When using multiple injection and shut-in cycles,
the interpreted closure pressure from each cycle should remain
similar as compared with their average from the multiple cycles,
thus better representing the in-situ minimum stress. Fig. 2a shows
one example minifrac test performed in the McMurray sands with
nine cycles. The pressure declined quickly during the shut-in,
making it difficult to derive a reliable Pc. As a result of on-site
real-time analysis, the difficulty was spotted immediately and the
corresponding corrective measures were taken. More cycles were
used, which yielded consistent closure behaviour eventually in the
last four cycles (Fig. 2b). Therefore, real-time analysis and multi-
ple shorter cycles certainly assisted in enhancing the data quality
in this difficult situation.
High formation-breakdown pressures and fracture-propagation
pressures were observed during the test previously described (Fig.
2a). This was caused by the significant near-wellbore friction dur-
ing the injection. It can be observed during the step-rate test in
Cycle 6. Every time the injection rate increased or decreased
abruptly, the injection pressure rose or dropped correspondingly
by a relatively large amount (Fig. 3), which is evidence of flow-
induced friction in the injection system. Near-wellbore complex-
ities are responsible for the high friction. One example is perfora-
tion damage, where the explosive action of the perforating
operation and the associated high temperatures likely cause signif-
icant compaction to the rock.
Flowback. A further enhancement to the minifrac test is a
flowback procedure. For the flowback, a certain volume of water
is withdrawn manually from the injection system (wellbore plus
the fracture) during the shut-in period. Instead of waiting for the
fracturing pressure to decline in response to the natural leakoff
from the fracture, the fracture now closes because of the manually
induced pressure drop. As a result, the fracture closure is
267
 (a)
BHP@Wellhead Sv = 2.23 140
Hydrostatic, 1.03 Smin = 2.23
Rate 6 9 120
10
Cycle 3 8
2 4 5 100
8

Injection rate, L/min

80

BHP, MPa 6 60

40
4
20
2
0

0 –20
2:09 PM 2:24 PM 2:38 PM 2:52 PM 3:07 PM 3:21 PM 3:36 PM

(b) Summary
8

5
Pressure (MPa)

2 Preopen
ISIP
1 Fracture closure
Pc,sqrt
Pc, compliance pressure = 2.23 MPa

0
0 2 4 6 8 10
Cycle

Fig. 2—(a) Recorded pressure and injection/flowback rate history for a minifrac test in the McMurray reservoir at 105-m true vertical
depth (TVD). The bottomhole pressures (BHP) were calculated from a surface pressure sensor at the pump plus the hydraulic
head (“Hydrostatic”) from the water column. The overburden weight (Sv) was calculated from the density log. Smin was the in-situ
minimum stress or fracture-closure pressure interpreted from the pressure data. The negative rate means flowback. Similar con-
ventions are used in this paper unless otherwise specified. (b) Various characteristic pressures interpreted from the minifrac test.
Preopen denotes the fracture reopening pressure where the fracture starts to reopen during the subsequent injection. In the first
cycle, Preopen corresponds to the breakdown pressure when the fracture is formed. ISIP is the instantaneous shut-in pressure;
Pc, sqrt refers to the fracture-closure pressure extracted by the sqrt(dt) plot, and Pc, compliance is the fracture-closure pressure
extracted by the compliance plot from the flowback tests.

relatively quick. Moreover, the difference in the pressure behav-
iour before and after the fracture closure is magnified by the flow-
back, which makes the closure defined more clearly. Fig. 4 shows
one example plot of pressure vs. square root of the shut-in time.
The inflecting pattern is clear on the plot, and the intersection of
two different slopes points to the fracture closure. The detailed
description of fracture flowback tests is outside the scope of this
paper but can be found in some published papers (Raaen and
Brudy 2001; Raaen et al. 2001).
Natural leakoff from the fracture can be slow if the water mo-
bility is small, which is common when testing caprock shales. It is
also likely to be small in the oil-sands pay zone if it has only mini-
mum initial water mobility. In these cases, the flowback becomes
almost a necessary component. Without flowback, a test may take
days to complete properly when multiple cycles are used. With
flowback, a test with multiple cycles can be completed in 2 to 3
hours. Fig. 5a shows one example test performed in the Clear-
water capping shale. It took approximately 3 hours to finish nine
cycles. A consistent, and thus reliable, fracture-closure pressure
was obtained as shown in Fig. 5b.
It is our observation that although minifrac tests are used rou-
tinely in the industry, significant attention should be paid to the
unique conditions in the oil-sands development and the corre-
sponding demands for high-quality data and interpretation accu-
racy. Some measures that can enhance the quality include
multiple injection/shut-in cycles to check for consistency in the

268 July 2013 Journal of Canadian Petroleum Technology

 12 BHP@Wellhead 140
Hydrostatic, 1.03
Smin = 2.23 120
10 Rate
100

Injection rate, L/min

8
80

BHP, MPa
6 60

40
4
20
2
0

0 –20
14:55:41 14:57:07 14:58:34 15:00:00 15:01:26 15:02:53 15:04:19

Fig. 3—A history plot of pressure and injection rate in the step-rate test Cycle 6 for a minifrac test in the McMurray reservoir at 105-
m TVD. Other details are the same as in Fig. 2a.

data and on-site preliminary interpretations. Additionally, the
flowback technique can help define the fracture closure better and
also increase test efficiency. Third, a proper minifrac test should
not measure an in-situ minimum stress larger than the density-
derived vertical stress. If this does occur, a definitive explanation
should be provided.
Geomechanical Laboratory Tests
Mechanical properties of the caprock, including the compressive
strengths, are essential for determining the maximum thermal
operating pressure for the reservoir. Geomechanical laboratory
tests on the site-specific cores are the most-comprehensive and -
effective means of deriving the necessary mechanical properties.
Mechanical tests on rocks in general are complex, and tests on
shales are even more complicated. Attention to the complexity is
important because shale is clay rich with low permeability. The
low permeability requires slow loading rates; therefore, tests on
shales take a long time. Osmosis and hydration may be involved
if fluids of different activity contact the shales. Shale properties
will change if shale is hydrated or dehydrated. Therefore, it is nec-
essary to determine the chemistry and activity of the pore fluid in
the shales so that the pore fluid used in the laboratory tests is
chemically compatible.
Fig. 6 shows the X-ray diffraction (XRD) analysis results on
the Clearwater shales in the current shallow well. Compared with
the conventional hard-rock shales, the Clearwater shales above
the target reservoir have a greater quartz content. Correspond-
ingly, the clay content is lower. Among the clay minerals, the
illite/smectite (I/S) mixed layers are dominant. A higher smectite
content foretells a high hydration tendency. These characteristics
are observed in the field. Without proper attention to the coring
mud, Clearwater shale cores can readily become hydrated when
brought to the surface. Therefore, careful attention is required

P–√Δt
1.2

Closure Pressure: 0.859 MPa

1.1

1.0

0.9
BHP (MPa)

0.8

0.7

0.6

0.5

0.4
0 5 10 15 20 25 30 35 40
Square Root of Shut-In Time

Fig. 4—The fracture closure is better defined by a clear inflecting curve at the intersection of two different slopes on a p-sqrt(time)
plot. Cycle 9 with flowback in the test is shown in Fig. 5a.

July 2013 Journal of Canadian Petroleum Technology 269

 (a)
BHP@pump sensor Sv = 0.92 Hydrostatic, 0.42
2.0
Cycle 1 Smin = 0.88 Rate
1.8 140
3
5
1.6
120
7 9 10
1.4
100

Injection, L/min
1.2
80
1.0
60
0.8
40
0.6
20
0.4

0.2 0

0.0 –20
12:00 PM 12:28 PM 12:57 PM 1:26 PM 1:55 PM 2:24 PM 2:52 PM 3:21 PM

(b)
Summary
2.0

1.5
Pressure (MPa)

1.0

0.5 Preopen Fracture closure

ISIP pressure = 0.88 MPa
Pc, sqrt
Pc, compliance
0.0
0 2 4 6 8 10
Cycle

Fig. 5—(a) Pressure/rate history from a minifrac test in the Clearwater caprock shale at 43-m TVD. The negative injection rates
represent flowback. (b) Various characteristic pressures interpreted from the minifrac test shown in Fig. 5a. Other details follow
Fig. 2b.

when deciding on the coring mud and preserving the cores.
Hydration inhibitors added to the mud become necessary.
Both room-temperature and high-temperature triaxial tests
were conducted on more than 10 samples for this project. They
were used to derive the elastoplastic deformation properties and
strength parameters to be used for the geomechanical simulations.
A detailed description of the test procedures and results (e.g., how
to derive the Mohr-Coulomb failure envelope from the tests) are
beyond the scope of this paper. Interested readers can refer to a
paper by Xu et al. (2011) for the triaxial-test procedures and refer
to relevant publications by the ERCB for the test results. In the
following, the effect of high I/S contents on the mechanical
behaviour of the Clearwater shales is elaborated upon. Little
attention has been directed to this factor so far in the context of
caprock-integrity studies, and thus it warrants a dedicated discus-
sion herein.
Fig. 7 plots the stress/strain curves for two representative sam-
ples: C14T3S1 from the upper part of the Clearwater caprock sec-
tion and that for C24T1S1 from the lower part. C14T3S1 was
tested under an effective confining pressure, CP¼0.2 MPa. But its
peak stress is higher than C24T1S1, which was tested under a
higher CP¼1.5 MPa. One possible cause is the difference in the
mineralogy between the two samples. As shown in Fig. 6, the
upper part of the Clearwater caprock, where C14T3S1 came from,
is more silty, while the lower part where C24T1S1 was situated
has a higher clay content and I/S mixed layers. A similar phenom-
enon has been reported in literature on the mechanical behaviour
of soils. For example, some researchers (Bishop 1966; Kenney
1967; Mitchell and Soga 2005) documented that soil with a higher
clay content, especially when the clay minerals are mostly mont-
morillonite, exhibits a small friction angle. Contribution from a
higher normal stress or confining pressure to the peak stress is
muted for a smaller friction angle.
Sedimentary rocks (e.g., the Clearwater shales in our current
discussion) rely on the friction for their mechanical strength. The
normal stress condition or confining pressure in the triaxial tests

270 July 2013 Journal of Canadian Petroleum Technology

 80

70

Quartz Clay
60

Bulk content, %
50

40

30

20

10

0
m m m m m m m m m m m m m m ale ea n e e
47 .66 .87 .59 .06 .16 .18 .40 .70 .91 .94 .15 .91 .14 sh
. S gto ierr cen
2 2 2 2 22 3 2 3 3 3 3 3 8 3 8 4 4 4 4 4 9 5 0 5 6 5 7
d o r t h l l i n P o
2, 3, 1, 1, 3, 4, 3, 4, 1, 2, 2, 3, 1, 2, ra No We E
T 1S T1S T2S T3S T3S T3S T2S T2S T3S T3S T3S T3S T1S T1S olo
C
11 11 11 14 14 14 16 16 18 18 21 21 24 24 R,
C C C C C C C C C C C C C C O
I

100
Kaolinite
90 Chlorite
Illite
80
I/S mixed
Clay type/content, %

70

60

50

40

30

20

10

0
m m m m m m m m m m m m m m ale ea n e e
47 .66 .87 .59 .06 .16 .18 .40 .70 .91 .94 .15 .91 .14 sh
. S gto ierr cen
, ,
2 2
22
2 ,2 ,3 ,3 2 3
,
3 8 8 4
3 ,3 ,3 ,4 ,4 ,4 ,5 ,5 ,5 4 9 0 6 7
d o r t h l l i n P Eo
a o e
S2 S3 S1 S1 S3 S4 S3 S4 S1 S2 S2 S3 S1 S2 lor N W
1 T1 1T1 1T2 4T3 4T3 4T3 6T2 6T2 8T3 8T3 1T3 1T3 4T1 4T1 Co
1 1 1 1 1 1 1 1 1 1 2 2 2 2 ,
C C C C C C C C C C C C C C OR Conventional hard rock
I

Fig. 6—Bulk minerals and clay type/contents measured from XRD analysis on a shallow oil-sands well. “C11T1S2, 22.47 m”
denotes Sample identification and depth, respectively. Data for “IOR, Colorado shale” are from the CSS operation in Cold Lake,
Alberta (Smith et al. 2004). Data for “Conventional hard rock” are from the public literature whose sources are not listed here.

controls the strength. However, the previously described observa-
tions suggest that for shallow depths, the mineralogy, in con-
junction with the stress condition, also affects the strength
significantly. For greater depths, the confining pressure is much
higher. There, the stress condition may overshadow the mineral-
ogical influence. Therefore, it is important to characterize the
mineralogical variation in the Clearwater shale. Different depth
intervals may exhibit different clay/quartz contents. Therefore,
these intervals should be grouped and tested separately. Different
mechanical properties may be measured for each of such groups.
Geomechanical Simulation
Geomechanical simulations are used to investigate the impact of
thermal operating conditions on the caprock integrity. They calcu-
late the induced stresses and deformation in the caprock caused by
the evolving pore-pressure ( p) and temperature (T) conditions in
the reservoir. The simulation models start with the original in-situ
stress condition measured in the minifrac tests. They use the me-
chanical properties extracted from the laboratory tests to describe
the constitutive deformation behaviour and allow safety factors to
be evaluated for a given thermal operating pressure. Finally, the
optimal maximum operating pressure (MOP) is derived by a series
of sensitivity-analysis runs for a given geology in a project site.
ABAQUS (SIMULIA 2012), a large-scale commercial finite-
element stress-analysis software, was used for the simulations. It
is well-established in the industry and fulfills our requirements
because of its strong nonlinear solution capability and unrivaled
strength in simulating the coupled thermal/hydraulic/mechanical
processes. For the caprock strata, our simulations are coupled

July 2013 Journal of Canadian Petroleum Technology 271

 2.2

2.0

1.8

1.6

Deviatoric Stress, q (MPa)

1.4

1.2

1.0

0.8

0.6
C14T3S1: CP = 0.2 MPa
0.4

0.2 C24T1S1: CP = 1.5 MPa

0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
Aial Strain, Ea (%)

Fig. 7—Two example stress/strain curves measured from triaxial tests on the shallow Clearwater shales. The sudden stress drops
are likely caused by localized microcracks being activated in the samples. They do not affect the ultimate sample strength.

between the fluid flow, temperature diffusion, and elastoplastic
deformation. The hydraulic/mechanical coupling is two-way—
namely, the pressure diffusion and rock deformation influence
each other fully. But the thermal/mechanical coupling is one-way
(i.e., the temperature diffusion affects the rock deformation only,
while the mechanical deformation does not affect the thermal
properties and temperature distribution). Temperature is allowed
to diffuse into the caprock because of the SAGD temperature con-
ditions in the reservoir. But no thermal convection is considered.
This is valid given the low permeability and thus the slow fluid-
flow velocities in the caprock.
Within the reservoir layer, the coupling is only one-way: Our
specially-designed ABAQUS driver enters the SAGD p and T con-
Stress-free ground surface
Clearwater
caprock, ho
z izontal SAGD wells. The top of the model is stress free,
Oil-Sands
pay zone, hr
x
Underburden
Fig. 8—A schematic model used for the geomechanical simula-
tions and analytical derivations concerning induced stresses
and deformation in the caprock caused by SAGD steaming in
the reservoir.
ditions calculated independently by a thermal simulator or from
analytical solutions. Their induced reservoir deformation is thus
simulated by ABAQUS. Reservoir dilation is automatically
accounted for by the chosen elastoplastic model and input dilation
angle. But the thus-calculated reservoir deformation is not allowed
to feed back into the thermal simulation, affecting the p and T con-
ditions in the reservoir. It is our opinion that such one-way cou-
pling from the p and T conditions to the mechanical response in
the reservoir is an adequate approximation and the most cost-
effective if the major focus is on the caprock deformation only.
The following discussion will compare the induced stresses in
the caprock for two different reservoir depths: 60 and 200 m. The
Clearwater shale is assumed to be in the horizontal-fracture stress
regime where the vertical stress Sv is the minimum component.
As a reference point, an MOP at 80% of Sv is at the given reser-
voir depth. Thus, at 60 m, MOP¼80%21 kPa/m60 m 1
MPa. The corresponding steam temperature Ts is assumed at
1818C following the phase behaviour of steam at the 1-MPa pres-
sure. Similarly, at 200-m depth, MOP¼3.36 MPa and Ts ¼2458C.
Fig. 8 shows a conceptual geomechanical model used for the
current discussions. An oil-sands pay zone 27 m thick is sand-
wiched between the overburden and underburden. For the sake of
simplicity, the whole overburden is assumed to be the Clearwater
formation and the underburden is taken to be a limestone-like ma-
terial. The model is a 2D plane strain cut perpendicular to the hor-
representing the real ground surface, while the bottom of the
model is constrained from moving in the vertical direction,
reflecting the support from underneath. It is an artificial boundary
condition, but it does not affect the computations in the pay zone
and overburden because the bottom of the model is sufficiently far
from the area of interest. The left and right sides of the model are
a symmetrical boundary type, which reflects the physical reality
because the left side of the model cuts vertically through the mid-
dle of the SAGD wells and the right side is at the midway between
two neighbouring SAGD well pairs.
Material properties used in the previously described simula-
tions are listed in Table 1. For simplicity of presentation, a linear
elastic model is used for the over- and underburden. But a
Drucker-Prager (D-P)-type plasticity model was used to account
for the shear-induced dilation in the reservoir. In addition, a po-
rous nonlinear elastic model is used in the reservoir to account for
the pressure-dependent elastic moduli (SIMULIA 2012). A

272 July 2013 Journal of Canadian Petroleum Technology


TABLE 1—MATERIAL PROPERTIES USED IN
GEOMECHANICAL SIMULATIONS
Overburden
Young’s modulus (MPa) 200
Poisson’s ratio 0.25
Volumetric thermal-expansion coefficient (1/ C) 8.4010–5
Pay zone
Porous bulk modulus* 0.007
Poisson’s ratio 0.4
Volumetric thermal-expansion coefficient (1 /C) 2.0010–4
Drucker-Prager friction angle (degrees) 45
Drucker Prager cohesion (kPa) 62
Drucker-Prager dilation angle (degrees) 20
Underburden
Young’s modulus (MPa) 1000
Poisson’s ratio 0.25
* Logarithmic bulk modulus for porous elasticity in ABAQUS (SIMULIA 2012).
parameter called logarithmic bulk modulus is introduced for the
oil sands (Table 1). For readers who are more familiar with the
conventional linear elasticity theories, a logarithmic bulk modulus
of 0.007 corresponds to a Young’s modulus of 142 MPa at a con-
fining pressure of 1 MPa if other relevant material properties fol-
low those given in Table 1. It should be noted that the coefficient
of thermal expansion for reservoir sands came from the upper
bound of the high-temperature triaxial tests conducted by Agar
et al. (1986). Their laboratory results indicated that, at 100 C, the
coefficient of volumetric thermal expansion for the oil sands is
2.410–4/ C in the undrained tests and 410–5/ C in the drained
tests. Also, this simplified numerical simulation did not consider
the glaciation-induced slickenside structure inside the Clearwater
shale formation. Thus, cohesion is used to simulate the rock-ma-
trix deformation and failure. More-sophisticated numerical mod-
els are needed to simulate these weak slickenside structures,
which is beyond the scope of this paper.
In order to explain some of the numerical-simulation results,
analytical derivations are given in Appendix A for the induced
stresses and vertical displacement in an interval caused by a uni-
form temperature or pore-pressure increase. Despite the simplicity
of the model, the analytical expressions help greatly to explain
how various material and/or geometric parameters affect the rele-
vant caprock-integrity issues.
In order to calculate the evolving p and T distributions in the
reservoir, an analytical solution given by Butler and Stephens
(1981) and Butler (1991) is used. Fig. 9 plots a steam-chamber
shape calculated from the solution and imported into the geome-
chanical simulations. It captures the general steam-chamber shape
during the SAGD operation. However, differences do exist in the
evolving p and T conditions if they are calculated by a numerical
thermal simulator. It is expected that the differences are not im-
portant for the current objective of evaluating the impact of differ-
ent reservoir depths on caprock integrity.
Induced Deformation in the Caprock. Reservoir expansion
caused by the temperature increase is shown in Fig. 9 by magnify-
ing the deformed meshes. It shows the caprock is being bent by the
expanding reservoir underneath. The surface heave, which may be
a more familiar term in the industry, is given in Fig. 10. It shows
the evolving heave directly above the SAGD well pair during the
SAGD operation in both shallow and deep reservoirs. For these
simplified calculations, as much as 90 cm for the shallow reservoir
and 130 cm for the deep reservoir are registered at the surface after
approximately 10 years of SAGD operations. These large calcu-
lated surface heaves are a direct consequence of the large thermal-
expansion coefficient aT¼210–4/ C being used in the simula-
tions. As shown in Fig. 10, a smaller thermal-expansion coefficient,
aT¼410–5/ C reduces the surface heave to 30 cm for the 60-m-
deep reservoir. A higher steam temperature is responsible for the
larger deformation observed for the deeper reservoir. As analyti-
cally derived in Appendix A, the thermally induced displacement
is proportional to the thermal expansion. A large thermal-expan-
sion coefficient and/or a large temperature increase means more
thermal expansion and, correspondingly, greater surface heave.
Induced Stresses in the Caprock. Reservoir expansion indu-
ces additional stresses inside the caprock. In fact, it is these
induced stresses that are responsible for a caprock failure if it is to
occur. Stress paths are normally used to depict the evolving state
of stresses at a given location. Two elements in the caprock above
the SAGD chamber are used for the illustration as shown in
Fig. 11: Elem 1 is closer to the top of the reservoir than Elem 2.
Two failure envelopes of D-P type are also plotted for reference,
representing two different friction angles: 30 and 45 , respec-
tively. All of them use a D-P cohesion¼0.3 MPa, notwithstanding
that 45 is stronger than 30 . Failure occurs when the stress path
touches the failure envelope. Because elastic deformation is
assumed in the caprock, it is possible that the stress path goes
beyond the failure envelope. If the plastic deformation model is
used, the stress path can only follow the failure line.

NT11 S, S11
+1.810e+02 (Avg: 75%)
+1.666e+02 +2.329e+01
+1.522e+02 +0.000e+00
+1.378e+02 –5.942e+02
+1.233e+02 –1.188e+03
+1.089e+02 Caprock –1.783e+03
+9.450e+01 –2.377e+03
+8.008e+01 –2.971e+03
+6.567e+01 –3.565e+03
+5.125e+01 –4.159e+03
+3.683e+01 –4.754e+03
+2.242e+01 –5.348e+03
+8.000e+00 –5.942e+03
–6.536e+03
–7.131e+03

Reservoir

Y Y

X Step: Step–24, SAGD process t = 1225.35days X Step: Step–24, SAGD process t = 1225.35days
Increment 17: Step Time = 53.28
Increment 17: Step Time = 53.28
Primary Var: NT11 Primary Var: S, S11
Deformed Var: U Deformation Scale Factor: +5.000e+01 Deformed Var: U Deformation Scale Factor: +5.000e+01

Fig. 9—Left: temperature distribution (NT11 in the unit of 8C) in the reservoir and the caprock after 1,225 days of SAGD operation.
Right: the corresponding effective horizontal stress (S, S11 in the unit of kPa). Negative stresses are compressive in a 60-m reser-
voir depth. The mesh is magnified by 50 times to show the deformed shape.

July 2013 Journal of Canadian Petroleum Technology 273

 60 m, E = 200 MPa, Expansion = 2e–4/C
140
200 m, E = 200 MPa, Expansion = 2e–4/C
120 60 m, E = 20 MPa, Expansion = 2e–4/C

60 m, E = 200 MPa, Expansion = 4e–5/C

100

Surface Heave (cm) 80

60

40

20

0
0 500 1,000 1,500 2,000 2,500 3,000 3,500
SAGD Operating Time (Days)

Fig. 10—Temporal variation of the vertical displacement on the ground surface (also called heave) directly above the SAGD well
pair for the various sensitivity cases simulated. “60m, E5200 MPa, expansion52e–4/C” denotes, for example, a 60-m reservoir
depth, the Young’s modulus of the caprock shale of 200 MPa, and volumetric thermal-expansion coefficient of the reservoir of
2310–4/º C. The same holds for the legends in Figs. 11 and 12.

Fig. 11 demonstrates that the shear stress or deviatoric stress
as plotted increases in the caprock during the SAGD operation in
both shallow and deep reservoir depths. The increase is more in
the deeper reservoir (> 2 MPa) than in the shallow one (approxi-
mately 1.5 MPa). In the case of a deeper reservoir, the stress state
becomes less compressive (i.e., the stress path moves to the left
from the original condition). This is caused by the higher steam
pressure and thus a higher pore pressure diffusing into the cap-
rock. On the contrary, for the shallow reservoir, the stress state
moves toward more compressive or toward the right on the plot.
This reflects the low operating pressure and the dominating nature
of the compressive thermally induced stresses.
Lower shear stresses and a more compressive nature in the
induced stresses all favour the integrity of the caprock. Unfortu-
nately, because of its shallow depth, the initial stress state is under
a smaller confinement in the shallow reservoir than in the deep
one. This is reflected by the fact that the original state of stresses
for the shallow reservoir depth is located to the left of (i.e., less
compressive than) that for the deep reservoir depth. The rock
strength is mostly derived from its friction. As a result, the cap-
rock touches the failure envelope more readily in the shallow
depth than in the deep one. As shown in Fig. 11, the stress path
for the shallow reservoir has passed the 30 failure envelope at
both Elem 1 and Elem 2 and has touched the 45 envelope

6.0
30°
45°
60 m, E = 200 MPa, Expansion = 2e–4/C, Elem 1
5.0
60 m, E = 200 MPa, Expansion = 2e–4/C, Elem 2
200 m, E = 200 MPa, Expansion = 2e–4/C, Elem 1
Deviatoric Stress (MPa)

4.0 200 m, E = 200 MPa, Expansion = 2e–4/C, Elem 2

3.0

2.0

1.0

0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Mean Effective Stress (MPa)

Fig. 11—Stress paths at two elements in the caprock above the SAGD steam chamber for the two different reservoir depths. 30º
denotes the D-P friction angle of 30º . D-P cohesion is all at 0.3 MPa. Other details follow Fig. 10. Elem 1 or 2 denotes the location
shown in the insert where the stress path is taken.

274 July 2013 Journal of Canadian Petroleum Technology

 2.0
30°
1.8 45°
60 m, E = 200 MPa, Expansion = 2e–4/C, Elem 1
1.6
60 m, E = 200 MPa, Expansion = 4e–5/C, Elem 1
60 m, E = 20 MPa, Expansion = 2e–4/C, Elem 1
1.4
Deviatoric Stress (MPa)
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.0 0.2
Fig. 12—Different stress paths corresponding to the sensitivity-analysis parameters denoted in the legend. The other details follow
Fig. 10 or Fig. 11.
already. For the deeper reservoir, the stress path goes beyond the
30 failure envelope at Elem 1, but stays just below, although
close to, the 45 failure. Location of Elem 2 is safe for the two
strength envelopes.
Limited Sensitivity Analyses. Caprock integrity is a complex
issue and numerous factors contribute (Yuan 2008). Ideally, a se-
ries of sensitivity analyses should be run in the simulations. How-
ever, the major purpose of the simulation works presented in this
paper is to compare the induced stresses in the caprock during the
SAGD operation between a shallow (60 m) and deep (200 m) res-
ervoir, respectively. Therefore, to limit the scope of this presenta-
tion, the sensitivity analysis is limited to the following two
material parameters: (1) Young’s modulus of the caprock shale
and (2) thermal-expansion coefficient of the reservoir. The analy-
sis is further limited to the shallow reservoir depth and illustrated
by the changes in the surface heave (Fig. 10) and induced stresses
in the caprock (Fig. 12). Details are summarized next.
The earlier presentation demonstrated that the mineralogical
content of the caprock shale varies across the depth and the
impact on the mechanical strength can be significant for the shal-
low reservoirs. Fig. 7 suggests that the Young’s modulus E for
Sample C24T1S1 is approximately 20 MPa if measured by a se-
cant line across the major portion of the stress/strain curve data
before the peak stress. The surface heave is not significantly
affected by the Young’s modulus of the caprock (Fig. 10). This
can be explained by the analytical derivations in Appendix A. The
vertical displacement caused by the thermal expansion (Eq. A-5)
has no variable E involved. E does affect the poroelastically
induced vertical displacement (Eq. A-6), which is responsible for
the difference observed in the numerical calculation, as shown in
Fig. 10.
But much smaller induced stresses will result if the caprock is
softer or more compliant—having a smaller Young’s modulus
(Fig. 12). When E¼20 MPa, the state of stresses all stay below
the strength lines (i.e., the caprock at the shallow depth is safe
against the two strength envelopes: 30 or 45 for the D-P friction
angle in Fig. 12). Again, the analytical derivations in Appendix A
can help explain the previously described sensitivity-analysis
results. The thermally induced stresses are proportional to the
Young’s modulus E and the thermal-expansion coefficient. A
smaller E generates a smaller induced thermal stress. Also note
that when E¼20 MPa, the stress path moves toward the left (i.e.,
less-compressive state, as shown in Fig. 12). The induced thermal
July 2013 Journal of Canadian Petroleum Technology
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Mean Effective Stress (MPa)
stresses become significantly smaller because of the smaller
Young’s modulus, so they are comparable to or even overshad-
owed by the decreasing effective stresses caused by the pore-pres-
sure increase. The latter makes the stress state approach the less-
compressive state (i.e., to the left in Fig. 12). In summary, the
softer or more-compliant nature of the shallow caprock shale sug-
gested by the laboratory tests helps maintain the caprock integrity
where only the induced stresses are concerned.
Thermal expansion is affected dramatically by the drained or
undrained heating. The pore fluid has a thermal-expansion
capacity generally at an order of magnitude larger than that of the
rock matrix. For low-mobility porous media (e.g., the bitumen-
saturated oil sands), the excessively expanded fluid phase cannot
dissipate in time and, consequently, more volumetric expansion
results in the undrained heating (Agar et al. 1986) or thermally
induced pore pressure (Butler 1986). With the intention of con-
servative design, the earlier simulations used a larger thermal-
expansion coefficient for the reservoir. Fig. 10 shows that a
smaller thermal-expansion coefficient for the reservoir,
aT¼410–5/ C, causes a much smaller surface heave. But Fig. 12
demonstrates that the ultimate level of the induced stresses in the
caprock at the end of the SAGD operation is similar between the
two thermal-expansion coefficients. The stress path is simpler in
the case of small thermal-expansion coefficients aT¼410–5/ C.
It is a nearly straight line throughout the SAGD operation, extend-
ing right and upward. On the contrary, the larger thermal-expan-
sion coefficient, aT¼210–4/ C, causes a nonmonotonic evolution
history in the induced stresses. The difference reflects the two
major mechanisms causing the caprock to deform. Reservoir de-
formation and thermal stress take place when the temperature
front diffuses into the caprock. These two mechanisms differ in
time and magnitude when they contribute to the caprock deforma-
tion. The similar end state of the induced stresses shown in Fig.
12 suggests that, ultimately, the thermal stress induced by the
temperature changes in the caprock dominates the stress state
therein.
Conclusions
The Alberta oil-sands/heavy-oil industry is exemplary and proac-
tive in safeguarding caprock integrity. The intent is to design a
safe operating pressure and steam temperature condition so that
the caprock maintains the sealing capacity during thermal
275
stimulation that it has inherently displayed throughout geological
history. The design and analysis of the caprock integrity serves to
weigh the balance between the induced stresses and the material
strength. Major activities include (1) minifrac tests to define the
initial stress state before the manual disturbance introduced during
the SAGD operation; (2) geomechanical laboratory tests to mea-
sure the mechanical properties including compressive strengths of
the caprock; and (3) geomechanical simulations to calculate the
induced stresses. These components of a caprock-integrity pro-
gram are common to both deep and shallow reservoir depths.
However, when applied to an oil-sands development at shallow
depths, these components merit some special considerations. This
paper has discussed the various quality-control issues, as follows:
• Minifrac tests for oil-sands development require increased ac-
curacy in executing the tests, acquiring the field data, and inter-
preting the data. Interpretation of the density log can reliably
give the vertical stress Sv. A minifrac test, if properly executed
and interpreted, should not yield a minimum stress larger than
Sv.
• Multiple cycles supplemented with flowback can enhance the
quality of the minifrac tests. A consistent fracture-closure pres-
sure should be observed between multiple test cycles. The flow-
back can shorten the duration of each test without
compromising the test quality, and help define the fracture-clo-
sure better.
• Clearwater shale above the McMurray oil-sands reservoirs typi-
cally has a higher quartz content compared with the conven-
tional hard rock, but an I/S mixed layer is the dominant clay
type. In the case of shallow caprock depths, which are the focus
of the current paper, different clay contents affect the mechani-
cal properties and strength significantly. Tests on samples with
higher clay and I/S content showed a lower compressive
strength. Thus, it is important in the shallow reservoirs to group
the rock facies properly when planning the laboratory tests and
geomechanical simulations.
Geomechanical simulations carried out in this paper serve to
compare the induced stresses in the caprock by the underlying
SAGD operation in the shallow (60 m) or deep (200 m) reservoirs.
A limited sensitivity analysis is also run to investigate the impact
of Young’s modulus of the caprock and thermal-expansion coeffi-
cient of the reservoir. Some observations can be drawn as follows:
(1) It is appropriate to limit the SAGD operating pressure as a
function of the reservoir depth. In this paper, the 80% propor-
tionality according to the fracture-closure pressure in the cap-
rock is used as an example. A shallow SAGD operation
should be conducted at a lower pressure and therefore a lower
steam temperature. As a result of lower pressure and tempera-
ture conditions, the induced deformation and stresses in the
caprock become smaller for the shallow reservoirs than for
the deep ones. This is beneficial for the caprock integrity.
(2) However, also inherent with the shallow depths is the original
lower confining stress conditions—a disadvantage for main-
taining the caprock integrity. Consequently, despite the
smaller induced stresses by the lower steaming temperature
and pressure in the reservoir, it is still possible in the case of
shallow reservoir depths that the caprock may enter into plas-
tic failure state. This highlights the challenges in developing
the shallow reservoirs and the necessity for a rigorous cap-
rock-integrity study before production and for monitoring pro-
grams during production.
(3) In addition to the lower operating pressure and temperature,
the integrity of the shallow caprock can be enhanced by a
softer and more-compliant mechanical nature. The sensitivity
analysis shows that if the caprock shale has a smaller Young’s
modulus, the induced stresses are further diminished and the
caprock becomes more resilient against the plastic yielding
under the same SAGD operating pressure and temperature.
(4) The mechanical properties in a given location and depth must
be adequately characterized and considered in effective mod-
elling. Moreover, geomechanical laboratory tests show that
some intervals can indeed have a lower Young’s modulus and
276
lower strength. Mineralogical contents can vary across the
depth. The impact on the mechanical properties is more pro-
nounced for the shallow depths. All of these speak for the
need of a detailed geomechanical laboratory program to mea-
sure the mechanical properties of the caprock shale and their
potential variations across the depth.
(5) Young’s modulus of the caprock shale does not greatly affect
the surface heave. A smaller thermal-expansion coefficient of
the reservoir reduces the surface heave and influences the evo-
lution of the induced stresses in the caprock. But for the simu-
lated cases, it appears that the ultimate stress state in the
caprock is not altered significantly by the thermal expansion in
the underlying reservoir. It reflects the dominating role played
by the thermal stresses in the caprock induced by the diffusing
temperature front from the underlying steam chamber.
This paper is not intended to support or argue against develop-
ing a particular type of reservoir depth. In fact, it is our expecta-
tion that no reservoirs should be categorically branded as safe or
unsafe for undergoing processing. Such generalizations can result
in certain valuable bitumen resources remaining unextracted, or
conversely some resources being extracted without proper regard
to protecting the groundwater aquifers and the environment. The
key is a rigorous engineering program, including the geomechani-
cal engineering, to design the operation and then an effective
monitoring program to detect signals of negative impacts on the
caprock integrity. Frequent updating and inversion of the moni-
tored data will offer a continuous health check on the caprock
integrity.
It cannot be overstated that the caprock integrity is a complex
issue. Numerous factors contribute which may compete against
each other. Dominance of some factors over the others may
change with time and in space. They all depend on geology, mate-
rial properties, and operating conditions. Numerous mechanisms
or factors have not been considered in the current paper. For
example, the current paper does not consider the thermally
induced pore pressure. It can be important in both thye reservoir
and the caprock shale. Second, the present paper does not con-
sider impact of weak planes. Examples of weak planes include
slickensides, joints, fractures, or even thin bentonite layers. In all,
readers are warned not to apply the results in this paper in design-
ing their operation without a proper site-specific geomechanical
study program. But the methodologies and quality-control meas-
ures illustrated and promoted in this paper should benefit the
industry in carrying out geomechanical works.
Nomenclature
Cp ¼ effective confining pressure in triaxial tests, kPa
E ¼ Young’s modulus
G ¼ shear modulus
MOP ¼ maximum operating pressure in, kPa
Pc ¼ fracture-closure pressure, kPa
SHmin ¼ minimum horizontal stress, kPa
SHmax ¼ maximum horizontal stress, kPa
Sv ¼ vertical stress, kPa
Ts ¼ steam temperature in SAGD,  C
a ¼ volumetric thermal-expansion coefficient
eij ¼ strain tensor
g ¼ Biot’s stress coefficient, g¼a(1þ)/[2(1þ)] with a
here being the Biot’s coupling coefficient
 ¼ Poisson’s ratio
ui ¼ displacement vector
rij ¼ stress tensor with subscripts i ( j) running from x, y, to z
Acknowledgements
We thank the management of Alberta Oilsands Incorporated and
BitCan G&E Incorporated for allowing the publication of this pa-
per. Thanks are also due to Jeffrey Tailleur, Natalie Murray, and
Donald Nicholson for editing the manuscript. Xu is partly sup-
ported by Alberta Innovates through an industrial research and de-
velopment associate fund program.
July 2013 Journal of Canadian Petroleum Technology
References Technol 43 (2): 39–46. PETSOC-04-02-03. http://dx.doi.org/10.2118/
Agar, J.G., Morgenstern, N.R., and Scott, J.D. 1986. Thermal expansion 04-02-03.
and pore pressure generation in oil sands. Canadian Geotechnical Xu, B., Yuan, Y.G., and Wang, Z.C. 2011. Thermal Impact On Shale De-
Journal 23 (3): 327–333. http://dx.doi.org/10.1139/t86-046. formation/Failure Behaviors—Laboratory Studies. Presented at the
Bell, J.S., Price, P.R., and Mclellan, P.J. 1994. In-situ Stress in the West- 45th US Rock Mechanics/Geomechanics Symposium 2011, San Fran-
ern Canada Sedimentary Basin. In Geological Atlas of the Western cisco, California, USA, 26–29 June. ARMA-11-303.
Canada Sedimentary Basin (WCSB), G.D. Mossop and I. Shetsen, Yuan, Y. 2008. Overburden/Casing Integrity in SAGD without High Oper-
Chap. 29. Available online from Alberta Geological Survey, http:// ating Pressures. Presented at the 2008 Canadian International Petro-
www.ags.gov.ab.ca/publications/wcsb_atlas/atlas.html. leum Conference/Petroleum Society s 59th Annual Technical Meeting,
Bishop, A.W. 1966. The Strength of Soils as Engineering Materials. Geo- Calgary, 17–19 June. CIPC-2008-206.
technique 16 (2): 89–130.
Butler, R.M. 1986. The Expansion Of Tar Sands During Thermal Recov-
Appendix A: Thermoelastic and Poroelastic
ery. J Can Pet Technol 25 (5). PETSOC-86-05-05. http://dx.doi.org/
10.2118/86-05-05. Solutions to the Stress and Deformation in a
Butler, R.M. 1991. Thermal Recovery of Oil and Bitumen. Englewood Uniformly Heated or Pressured Formation
Cliffs, New Jersey: Prentice Hall. In order to derive analytical solutions, the reservoir is assumed to
Butler, R.M. and Stephens, D.J. 1981. The Gravity Drainage of Steam- be flat and laterally infinite with a constant thickness, as shown in
Heated Heavy Oil to Parallel Horizontal Wells. J Can Pet Technol 20 Fig. 8. The temperature increase, T, is uniform throughout the
(2): 90–96. JCPT Paper No. 81-02-07. http://dx.doi.org/10.2118/81- reservoir.
02-07. Relevant equilibrium equations according to the theory of ther-
Chhina, H.S., Luhning, R.W., Bilak, R.A. et al. 1987. A Horizontal Frac- moelasticity can be written as
ture Test In The Athabasca Oil Sands. Presented at the 38th Annual   
Technical Meeting of the Petroleum Society of CIM/SPE Annual
@rij aT E @T
 ¼ 0; . . . . . . . . . . . . . . . . ðA-1Þ
Technical Meeting, Calgary, 7–10 June. CIM 87-38-56. http:// @xj 3ð1  2Þ @xi
dx.doi.org/10.2118/87-38-56.
Chhina, H.S. and Agar, J.G. 1985. Potential Use of Fracture Technology
with the constitutive relation as
for Recovery of Bitumen from Oil Sands. Presented at the 3rd Interna-  
2G aT E
tional Conference on Heavy Crude and Tar Sands, Long Beach, Cali- rij ¼ 2Geij þ ekk dij  Tdij ; . . . . . . ðA-2Þ
1  2 3ð1  2Þ
fornia, USA, 22-31 June.
Hudson, J.A., Brown, E.T., Fairhurst, C. et al. 1993. Rock Testing and and boundary conditions of
Site Characterization. In Comprehensive Rock Engineering: Princi-
ples, Practice & Projects, Vol. 3, Chap. 17–21. Oxford, UK: Perga- ux ¼ uy ¼ 0 x ¼ 61
mon Press. rzz ¼ rxz ¼ ryz ¼ 0 z ¼ ho þ hr ; x 2 ð1; 1Þ :
Kenney, T.C. 1967. The Influence of Mineral Composition on the Residual
Shear Strength of Natural Soils. Proc., Geotechnical Conference on uz ¼ 0 z ¼ 0; x 2 ð1; 1Þ
Shear Strength Properties of Natural Soils and Rocks, Oslo, Norway,                    ðA-3Þ
Vol. 1, 123–129, TRB 00237620.
Kry, P.R. 1989. Field Observations of Steam Distribution during Injection The coordinate system is shown in Fig. 8. rij is the stress
to the Cold Lake Reservoir. In Rock at Great Depth: Rock Mechanics tensor with subscripts i ( j) running from x, y, to z. eij is the strain
and Rock Physics at Great Depth—Proceedings of an International tensor, ui is the displacement vector, aT is the volumetric thermal-
Symposium, Pau, 28–31 August 1989, V. Maury and D. Fourmain- expansion coefficient,  is the Poisson’s ratio, G is the shear mod-
traux, Vol. 3. London: Taylor & Francis. ulus, and E is the Young’s modulus.
Kry, P.R., Boone, T.J., Gronseth, J.M. et al. 1992. Fracture Orientation Using the boundary conditions at the ground surface (Eq. A-3),
Observations from an Athabasca Oil Sands Cyclic Steam Stimulation it is easy to show that
Project. Presented at the 43rd Annual Technical Meeting of the Petro-
rxz ¼ 0
leum Society of CIM, Calgary, 7–10 June. CIM 92-37.
Mitchell, J.K. and Soga, K. 2005. Fundamentals of Soil Behavior. New
ryz ¼ 0 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ðA-4Þ
York: John Wiley & Sons. rzz ¼ 0
Palmgren, C., Walker, I., Carlson, M. et al. 2011. Reservoir design of a
The remaining solution variables are
shallow LP-SAGD project for in-situ extraction of Athabasca bitumen.
 
Presented at the 2011 World Heavy Oil Conference (WHOC11), Ed- aT E
monton, Alberta, Canada, 14-17 March. WHOC11-520. rxx ¼ ryy ¼  T
3ð1  Þ
Proskin, S.A., Scott, J.D., and Chhina, H.S. 1990. Current Practice in the   ; . . . . . . . . . . . . . . . . . ðA-5Þ
Interpretation of Microfrac Tests in Oil Sands. Presented at the SPE aT ð1 þ Þ
Duz ¼ Thr
California Regional Meeting, Ventura, California, USA, 4–6 April. 1
SPE-20040-MS. http://dx.doi.org/10.2118/20040-MS.
Raaen, A.M. and Brudy, M. 2001. Pump-in/Flowback Tests Reduce the Esti- in which Duz is the vertical displacement in the system which is
mate of Horizontal in-Situ Stress Significantly. Presented at the SPE An- reflected on the ground as the surface heave. Upward movement
nual Technical Conference and Exhibition, New Orleans, 30 September–3 or tensile stresses are positive.
October. SPE-71367-MS. http://dx.doi.org/10.2118/71367-MS. A similar result can be derived for the poroelastic solution
Raaen, A.M., Skomedal, E., Kjørholt, H. et al. 2001. Stress determination (i.e., stress changes and displacements in the reservoir caused by
from hydraulic fracturing tests: the system stiffness approach. Int. J. a uniform pore-pressure change, p):
Rock Mech. Min. Sci. 38 (4): 529–541. http://dx.doi.org/10.1016/
rxx ¼ ryy ¼ 2gp
S1365-1609(01)00020-X.  
Settari, A. and Raisbeck, J.M. 1979. Fracture Mechanics Analysis in In- g : ..................... ðA-6Þ
Duz;i ¼ hr p
Situ Oil Sands Recovery. J Can Pet Technol 18 (2): 85–94. PETSOC- G
79-02-07. http://dx.doi.org/10.2118/79-02-07.
SIMULIA 2012. ABAQUS Version 6.12 documentation, http:// Yanguang Yuan is both a professional engineer and a geo-
www.3ds.com/products/simulia/support/documentation/. scientist in Alberta. In his more than 25 years of experience, he
Smith, R.J., Bacon, R.M., Boone, T.J. et al. 2004. Cyclic Steam Stimula- has consulted with petroleum companies in North America,
tion Below a Known Hydraulically Induced Shale Fracture. J Can Pet the Middle East, and Far East, and has authored or

July 2013 Journal of Canadian Petroleum Technology 277

coauthored more than 150 technical reports and papers, one
monograph, and a 600-page course note on petroleum geo-
mechanics. In 2000, Yuan founded BitCan Geosciences & En-
gineering Incorporated, which has now grown into being the
only independent integrated petroleum geomechanics firm in
the world. He is a member of SPE.
Bin Xu is a senior geomechanics specialist in research and de-
velopment with BitCan Geosciences & Engineering, where he
is involved in geomechanical numerical simulations, labora-
tory testing, and DFIT/minifrac testing. Xu holds a PhD degree
from the University of Calgary and MSc and BSc degrees from
Tongji University in China.
278
Claes Palmgren is Vice President – Resource Development,
Technical at Alberta Oilsands Incorporated. He has more than
20 years of experience with design, development, and imple-
mentation of bitumen and extraheavy-oil in-situ production
technologies at the laboratory scale, the field pilot scale, and
for larger commercial projects. Palmgren also has manage-
ment and executive experience with pioneering oil-sands
companies and development of companies after working as
reservoir-engineering lead during Suncor’s Mackay River
startup and as overall lead for Petro-Canada vapour-assisted
petroleum-extraction pilot. He holds MSc and PhD degrees
from Delft University of Technology’s faculty of Mining and Pe-
troleum Engineering in the Netherlands.
July 2013 Journal of Canadian Petroleum Technology