Permeability Variations Associated
With Shearing and Isotropic
Unloading During the SAGD Process
P. LI*, R.J. CHALATURNYK
University of Alberta
*Now with Petro-Canada
Abstract
This paper discusses the variations of reservoir permeability
within unconsolidated sands due to isotropic and deviatoric
(shear) processes during the SAGD process. Isotropic unloading
results from the steam injection pressure, which is generally
higher than the initial reservoir pressure. The high steam injection
pressure results in the increase of pore pressure and reduces the
confining effective stress within the drained zone and part of the
partially drained zone. The shearing process is induced primarily
by changes in total stress but can also occur with increased pore
pressures. High steam temperatures associated with the SAGD
process result in significant volumetric expansion of the reser-
voir material within the steam chamber. Thus, total stress is in-
creased and the shearing process may occur beyond the steam
chamber surface. Both the isotropic unloading and shearing pro-
cesses can induce reservoir permeability variation because of the
change in pore space, pore shape, and pore throat. For isotropic
unloading, the configuration of the grains or their relative posi-
tion is, for the most part, unchanged and the grains simply move
apart without relative rearrangement. Whereas the shearing pro-
cess induces substantial relative motion of the grains and sig-
nificant changes in pore geometry. Based on lab test results, it is
summarized that the isotropic unloading process produces much
smaller changes in volume, absolute permeability, and water ef-
fective permeability compared to the shearing process. Conse-
quently, incorporating stress-induced permeability change within
coupled reservoir geomechanical simulations requires different
relationships or models for these two conditions. In addition,
some empirical permeability relationships, such as the Kozeny-
Carman model, Tortikes equation, and Chardabellass terms, are
also discussed.
Introduction
The SAGD process has become a commercially viable tech-
nology to develop the vast oil sands resources in Canada. Reser-
voir simulation is used as valuable design tool for development
projects and to predict SAGD production performance. However,
in some instances, conventional reservoir simulation may not be
suitable for predicting performance because it does not take into
account the coupled mechanisms between fluid flow and reservoir
deformation. In these cases, a coupled reservoir geomechanical
simulation may be required. A requirement for these coupled simu-
lations is suitable relationships between reservoir permeability and
geomechanical behaviour.
Oldakowski conducted a series of tests based on both reconsti-
tuted oil sands specimens and drilled oil sands cores to characterize
the relationships of permeability and geomechanical processes(1).
PEER REVIEWED PAPER (REVIEW AND PUBLICATION PROCESS CAN BE FOUND ON OUR WEB SITE)
54
Permeability variations of sandstone and other high strength mate-
rials experiencing the isotropic unloading and shearing processes
have also been studied(2). Chalaturnyk summarized test results on
oil sands and provided the relationship between oil sands com-
pressibility and confining effective stress(3). It is known that oil
sands samples can be easily disturbed during the sampling pro-
cess. In order to minimize the disturbance, Touhidi-Baghini(4) took
the test specimens from an exposed outcrop of bitumen-free Mc-
Murray Formation sandstone, northeast of Fort McMurray. Based
on these specimens, experimental studies were conducted on the
permeability variation during the shearing process.
This paper systematically discusses the oil sands permea-
bility variations due to the isotropic unloading and shearing pro-
cess based on Oldakowskis and Touhidi-Baghinis test results. In
addition, some empirical permeability relationships, such as the
Kozeny-Carman model, Tortikes equation, and Chardabellass
terms, are also discussed.
Geomechanical Behaviour During SAGD
During SAGD operations, steam is continuously injected into
the oil sands reservoir. The steam injection pressure is generally
greater than the initial reservoir pore pressure. The steam chamber
propagates under the influence of the injection pore pressures and
high temperatures. Because of different mechanisms of pore pres-
sure transmission and heat flow, the high pore pressure front is ap-
proximately four to five metres in front of the high temperature
front(5). During the SAGD process, two major geomechanical phe-
nomena may occur. One is the isotropic unloading process and the
other is the shearing process. Both of them can induce variations of
reservoir permeability. However, their mechanisms are much dif-
ferent, as discussed below in detail.
Isotropic Unloading Process
The two concepts, isotropic unloading and anisotropic un-
loading, need to be clarified first. Isotropic unloading means that
the confining effective stress is decreased by the same amount in all
directions. So, both pore pressure injection and confining pressure
reduction can induce the isotropic unloading process. Anisotropic
unloading means that the confining effective stress is decreased by
different amounts in different directions. Pore pressure injection
cannot result in the anisotropic unloading process unless the initial
stress state is anisotropic.
In the SAGD process, reservoir pore pressures are increased due
to steam injection and they basically are maintained at a constant
within the high pore pressure front. This region includes the whole
drained zone and part of the partially drained zone(5). Pore pressure
decreases gradually from the highest pore pressure value to the
Journal of Canadian Petroleum Technology
 1.8
1.6
Volume Change (%) 1.4
1.2
1 PI 1
0.8
0.6
0.4
0.2
0
0 1,000 2,000 3,000 4,000
Effective Confining Stress (kPa)
FIGURE 1: Volume change vs. confining effective stress(1).
initial value in the far field. It is clear that the whole region where
pore pressure is greater than the initial value undergoes isotropic
stress unloading. This process causes the grain packing to become
loose (e.g., porosity increases), but the relative position of grains
does not change. It should be noted that any porosity variations in-
duced by the isotropic unloading process are generally recovered
when the pore pressure is reduced and the confining stress returns
to its original value.
Shearing Process
High steam temperature in the SAGD process induces signifi-
cant thermal expansion of the reservoir material. During the SAGD
process, the thermal expansion of the reservoir region within the
high temperature front creates a large thermal stress normal to the
front surface. The volumetric expansion coefficient is defined
by:
1 d
= oil sands specimens as a function of isotropic confining effective
dT p
1 V
=
V T p
........................................................................................ (1)
The volumetric thermal expansion coefficient is related to the
linear thermal expansion coefficient by:
= 3 ..................................................................................................... (2)
Thermal stress due to the volumetric expansion can be calcu-
lated by:
T = E T ..................................................................................... (3)
In the SAGD operations at the Athabasca oil sands deposits, the
net temperature increase is normally as high as 250 C. Assuming
the coefficient of thermal expansion of oil sands is 2 10-5/ C,
and Youngs modulus is 1,200 MPa, the induced thermal stress can
be as high as 6 MPa. This thermal stress can significantly increase
the total stress beyond the high temperature front. However, the
pore pressure in this region is high and the effective stress is low,
which leads to the strength of the reservoir material in this region
becoming low. If the stress state reaches the failure envelope of the
material, shear failure occurs. The reservoir material located above
the steam chamber may behave differently compared to that beside
the steam chamber. This shearing process is also dependent upon
the in situ vertical and horizontal stress conditions.
The shearing process causes reservoir grains to roll, ride, or di-
late. Thus, it results in the rearrangement of grains and changes the
grain packing style. This mechanism causes significant variations
January 2006, Volume 45, No. 1
5
Sn 1-D Chalaturnyk (1995) UTF
Kosar (1989) Interbedded UTF
Sn 2-D Plewes (1987), Syncrude
Sn 3-D 4 Plewes (1987), Saline, Conventional Platens, ID<10%
Sn 5-D Plewes (1987), Saline, Frictionless Platens, ID<10%
Volumetric Strain (%)
Plewes (1987), Saline, Frictionless Platens, ID>10%
PI 2-D Agar (1984), Saline
PI 3-D Kosar (1989) Interbedded UTF unloading curve
3 Kosar (1989), Cold Lake Oil Sands
PI 2 Scott et al. (1994) Cold Lake Oil Sands
Wong (1993) Cold Lake Oil Sands
PI 3
PI 4 2
1
5,000 0
0 2,000 4,000 6,000 8,000 10,000
Isotropic Effective Confining Stress (kPa)
FIGURE 2: Volumetric strain vs. isotropic confining effective
stress(3).
of reservoir parameters, such as porosity and permeability. In this
paper, only permeability variations are discussed.
Absolute Permeability Variations in
Response to Isotropic Unloading
As discussed above, during the isotropic unloading process,
tight grain packing becomes loose, but the relative position among
grains does not change. This makes it possible to relate the perme-
ability variation to volumetric strain of the material.
Adhikary (6) and Oldakowski(1) conducted a series of tests based
on reconstituted oil sands specimens to measure the absolute per-
meability change due to isotropic stress unloading. Those tests,
with both volume change and permeability measurement, are taken
into account. The initial properties of the specimens can be found
from Oldakowski(1). Figure 1 shows the volume change of ten
stresses. It can be seen that, except for PI-3D, all specimens have
maximum confining effective stresses of about 4,000 kPa. With
decreasing effective confining stress, the volume of the specimens
(porosity) increases. Although Specimen PI-3D started at the con-
fining effective stress of 1,000 kPa, it has the same variation trend
of volume change as the other specimens. When the confining ef-
fective stresses are reduced to about 200 kPa, the volume is in-
creased by 1.2% to 1.5%.
Based on the relationship between oil sands compressibility
and confining effective stresses, the volumetric strain as a function
of isotropic effective stress can be found and is shown in Figure
2. The data on Cold Lake oil sands from Scott et al.(2) and Wong
et al.(7) are also included. All these data indicate that the volume
change of oil sands material during the isotropic unloading pro-
cess is less than 2%. The data of interbedded UTF oil sands from
Kosar(8) shows a volumetric strain of about 4.7% due to isotropic
unloading. This is not consistent with other testing results because
these samples were highly disturbed.
Permeability variation can be expressed as the ratio of the modi-
fied permeability during the isotropic unloading process to its ini-
tial value. Figure 3 is the permeability ratio as a function of volume
change for ten specimens from Oldakowski and Adhikary. It is
seen that the permeability ratio increases with increasing the volu-
metric strain. In addition, for the same volumetric strain, a lower
initial permeability results in larger permeability increases. Only
the specimen Sn-3 has untypical variation trend, which is due to
operational problems during the experiment. When the initial per-
meability is greater than 1 m2 (which is roughly consistent to the
real in situ oil sand conditions), the permeability ratio is less than
1.2 for the maximum volumetric strain. Cold Lake oil sands have
similar permeability variation characteristics(2).
55
 1.8
Initial Permeability (m2)
1.7 Sn-1 0.9007
Sn-3 2.3270
Sn-5 0.5367
1.6 PI3-D 0.4270
PI2-D 0.4119
1.5 PI-1 3.7830
PI-2 3.3057
k/k0
1.4 PI-3 3.4530
PI-4 1.9746
1.3
1.2
1.1
1.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Volumetric Strain (%)
FIGURE 3: Absolute permeability ratio vs. volumetric strain(1).
Absolute Permeability Variations in
Response to Shearing
An absolute permeability measurement on a bitumen saturated
oil sands specimen is difficult because bitumen cannot flow at room
temperature. The extraction of bitumen from oil sands material
will disturb the oil sands structure seriously. Therefore, Touhidi-
Baghini used block oil sands samples to measure the absolute per-
meability(4) during the shearing process. The block samples were
taken from the McMurray Formation along the banks of the High
Hill River, approximately 60 km east of Fort McMurray, Alberta.
They are bitumen free, so it is possible to avoid the disturbance
of gas evolution and bitumen extraction. The oil sands specimens
were obtained by coring the block sample both vertically and hor-
izontally. If the specimens were drilled perpendicular to the bed-
ding direction, they were called vertical core specimens, such as
Vertical T27, T28, T29, T36, T38, and T39. If the specimens were
drilled parallel to the bedding direction, they were called hori-
zontal core specimens, such as Horizontal T40, T41, T42, and T43.
Touhidi-Baghini measured absolute permeability of these oil sands
specimens during shearing based on four stress paths (Figure 4).
The test results show that the volumetric strain increases with
increasing the axial strain. For horizontal specimens, the max-
imum volumetric strain was about 11% when its axial strain was
10% (Horizontal T43 tested with stress path 4). Meanwhile, for
the same axial strain, volumetric strain decreases for specimens
tested along stress path 1. Horizontal T40, which was tested with
stress path 1, even contracted when the axial strain was less than
4%. Vertical specimens exhibit similar behaviour. The specimens
tested with stress paths 2, 3, and 4 (Vertical specimens T38, T36,
and T39) reach their maximum volumetric strain at an axial strain
of approximately 4%. However, Vertical specimens T27, T28, and
T29, which were tested along stress path 1, had a volumetric strain
of less than zero when the axial strain was less than about 5%.
Their maximum volumetric strain was also much smaller (3 4%)
compared with other vertical specimens.
Stress path 1 is a shearing loading process because it increases
the confining effective stress. Although the shearing process can
cause the specimen to dilate and increase its volume, the increasing
confining effective stress results in a decrease of the specimen
volume. The combination of these two mechanisms controls the
variation characteristics of the volumetric strain. In fact, the shear
stress causes contraction of the specimen initially, and the increase
in confining effective stress compresses the specimen. So, the volu-
metric strain is negative (contraction) at the beginning of shearing.
When the volumetric strain reaches the minimum value, shear di-
lation becomes dominant and the specimen volume increases as
shear stress and axial strain increase. Owing to the effect of in-
creasing confining effective stress, the final volumetric strain is
much smaller than that of the specimens tested with stress paths
2, 3, and 4. Stress path 2 has no change on the effective confining
stress and the volumetric strain of the specimen is only controlled
56
Sn-1
Sn-3
Sn-5
PI 3-D
PI 2-D
PI-1
PI-2
PI-3
PI-4
1.6 1.8
FIGURE 4: Stress paths for absolute permeability test during
shearing(4).
by the shearing process. However, stress paths 3 and 4 involve the
isotropic unloading process because the confining effective stress
decreases with increasing shear stress and axial strain. This un-
loading causes specimen volume to increase. More significantly,
this process affects the specimens tested with stress path 4 because,
for the same shear stress, the confining effective stress of stress
path 4 is smaller than that of stress path 3. Clearly, permeability
variations due to shear stresses are stress path dependent.
Although the initial permeability of these two types of spec-
imens is similar, the modified value by the shearing process is
much different. The permeability of vertical specimens can be in-
creased approximately from 2 m2 to the maximum value of 7
m2 (Vertical T36) for a volumetric strain of 2.5%. In contrast, the
permeability of horizontal specimens can be increased from ap-
proximately 1.5 m2 to the maximum value of 4 m2 (Horizontal
T43) for a volumetric strain of about 2.2%. The permeability of the
specimens tested with stress path 1 is not increased significantly
because of the volumetric contraction induced by the loading pro-
cess. It is noted that when the permeability is increased to the max-
imum value, it starts to decrease. The reason for this phenomenon
is that with increasing the axial strain and volumetric strain, fines
migration may occur and block some pore throats. Hence, the per-
meability decreases.
Permeability ratios as a function of volumetric strain are shown
in Figures 5 and 6 for horizontal and vertical specimens, respec-
tively. Except the specimens tested with stress path 1, the vertical
specimens permeability (Figure 6) can be increased by approx-
imately 400% (Vertical T36), 500% (Vertical T38), and 700%
(Vertical T39) for the volumetric strains of 2.5%, 6%, and 10%, re-
spectively. However, Figure 5 shows that the maximum horizontal
specimens permeability is increased by about 250% (Horizontal
T43) for the volumetric strain of 2.5%. So, for the same volumetric
3
2
k/k0
1
Horizontal T40, SP1, sig3'= 250 kPa
Horizontal T41, SP2, sig3'= 250 kPa
Horizontal T42, SP3, sig3'= 250 kPa
Horizontal T43, SP4, sig3'= 250 kPa
0
-4 -2 0 2 4 6 8 10 12
Volume Change (%)
FIGURE 5: Absolute permeability ratio vs. volume change for
horizontal specimens.
Journal of Canadian Petroleum Technology
 analyzed here because these samples have been tested with both
7
isotropic unloading and pure shearing process.
6
Figure 7 shows the relationship of volumetric strain as a func-
tion of confining effective stress. It is seen that the volume increase
5
is less than 2% when the confining effective stress is reduced from
approximately 6,000 kPa to 150 kPa.
4 Because of high bitumen saturation, pore space, and pore throat
geometry, the region available for water to flow is very small. The
k/k0

3 initial effective permeability to water of these five samples is ap-

proximately from 10-5 m2 (Sample S20) to 10-4 m2 (Sample
2 S19). The variation of effective permeability to water can be clearly
shown by the ratio of the improved value to its initial value (Figure
1 8). It is seen from Figure 8 that the isotropic unloading process can
increase the effective permeability to water more significantly if
0
-4 -2 0 2 4 6 8 10 12
the initial permeability is small (Sample S20)(1). The final effective
permeability to water of the five samples is about two to six times
Volume Change (%) the initial value for the maximum volumetric strain of 1.6 to 1.8%.
The increase of effective permeability to water is greater than that
FIGURE 6: Absolute permeability ratio vs. volume change for of the absolute permeability for the same volumetric strain due to
vertical specimens.
the isotropic unloading process (Figures 3 and 8). The reason is
that the initial effective permeability to water of Sample S19 to
strain, the vertical specimens permeability is increased more sig- S23 is much smaller than the initial absolute permeability of recon-
nificantly than the horizontal specimens permeability. stituted soil sands specimens Sn-1 to Sn-5 and PI-1 to PI-4 (Figure
3). This is consistent with the statement that the smaller the initial
permeability, the larger the increase of permeability due to the iso-
tropic unloading process.
Effective Permeability to Water Effective permeability to water varies differently during the
Oldakowski(1) also measured effective permeability to water and pure shearing process. In the test with the pure shearing stress path,
its variations of oil sands specimens during isotropic unloading and the applied initial confining effective stresses (3i) are 3,000 kPa,
shearing processes. These oil sands cores were obtained from wells 890 kPa, 1,950 kPa, 570 kPa, and 1,390 kPa for Samples S19, S20,
S21, S22, and S23, respectively. The volumetric strain for Sample
drilled in 1987 at the AOSTRA Underground Facility Phase A
S19 is the smallest because of the maximum initial confining effec-
site. The testing results of samples S19, S20, S21, S22, and S23 are tive stress. It is only 3% for the axial strain of 8%. Other samples
reach the maximum volumetric strain of about 6% for the axial
2.0
strain of 5% to 6%. The volumetric strain increases dramatically
S19
1.8
S20
when shear failure occurs. Examining the first derivative of volu-
Volume Change (%)

1.6 S21 metric strain to the axial strain can prove this. The axial strains of
1.4
S22 the five samples at failure are 3.5%, 2.2%, 1.6%, 1.7%, and 1.2%,
S23
1.2
respectively.
Figure 9 shows the effective permeability to water as a func-
1.0
tion of volumetric strain for Samples S19 to S23. Effective per-
0.8
meability increases significantly as volumetric strain increases.
0.6 Also, it increases dramatically when shear failure occurs. Again,
0.4 using the first derivative of effective permeability to axial strain
0.2 helps identify the change in permeability. For the same increment
0.0 of axial strain, effective permeability increases more significantly
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 after shear failure than before failure. The magnitude of the im-
provement of effective permeability to water can be clearly seen
Effective Confining Stress (kPa) in Figure 10. For all samples except Sample S19, which has the
highest initial confining effective stress, the effective permeability
to water is 100 to 500 times higher than the initial value, for a volu-
FIGURE 7: Volume change vs. effective confining stress(1).
metric strain of 4% to 6%.

7
Effective Permeability to Water (m2)

S19 1.0E-01
6 S20
S21
S22 1.0E-02
5 S23
Tortike's eq.
k/k0

1.0E-03
4
S19, sig3i = 3,000 kPa
S20, sig3i = 890 kPa
3 1.0E-04 S21, sig3i = 1,950 kPa
S22, sig3i = 570 kPa
S23, sig3i = 1,390 kPa
2
1.0E-05

1
0.0 0.5 1.0 1.5 2.0 1.0E-06
-1 0 1 2 3 4 5 6 7
Volume Change (%) Volume Change (%)

FIGURE 8: Effective permeability ratio vs. volume change FIGURE 9: Variation of effective permeability to water vs. volume
including the curve with Tortikes equation. change during pure shearing(1).

January 2006, Volume 45, No. 1 57


1,000
S19, sig3i'= 3,000 kPa
S20, sig3i'= 890 kPa
S21, sig3i'= 1,950 kPa
S22, sig3i'= 570 kPa
S23, sig3i'= 1,390 kPa
S19, isotropic
100 S20, isotropic
kw/kw0
S21, isotropic
S22, isotropic
S23, isotropic
10
1 equation cannot be used to determine the modified permeability
-1 0 1 2 3
Volume Change (%)
FIGURE 10: Comparison of water effective permeability ratio due
to isotropic unloading and shearing.
Discussions
Comparison of Permeability Changes Due
to Different Geomechanical Behaviour
The maximum absolute permeability change due to the isotropic
unloading process, based on the reconstituted oil sands specimens,
is not more than two times of its initial value (Figure 3). How-
ever, it can be five times higher than its initial value due to the
pure shearing process based on vertical specimens (Vertical T38)
(Figure 6).
The initial effective permeability to water of oil sands core spec-
imens is very low, and its maximum value due to the isotropic un-
loading process can be two to six times higher than its initial value.
In contrast, the pure shearing process can induce an even more sig-
nificant increase of effective permeability to water, which can be
100 to 500 times higher than its initial value for a volumetric strain
of 4% to 6% (Figure 10). Therefore, it can be concluded that what-
ever the initial permeability is for oil sands specimens, the shearing
process can induce a more significant improvement to oil sands
permeability. Of course, this shearing process involves shear dila-
tion. If the confining effective stress is too high, the shearing pro-
cess may cause oil sands permeability to decrease(2). In addition,
both the isotropic unloading process and the shearing process play
an important role when the initial permeability value is low.
The comparison of the effects of isotropic unloading and
shearing is shown in Figure 10. For Samples S19 to S23, the volu-
metric strain is less than 2% when shear failure occurs. The effec-
tive permeability to water can be greatly increased if shear failure
occurs. Before shear failure occurs, the pure shearing process plays
the same role as the isotropic unloading process (Figure 10).
Tortikes Equation and the Kozeny-Carman
Model
Tortike derived an equation based on the Kozeny-Carman model
to determine the modified permeability as a function of volumetric
strain, which is induced by geomechanical behaviour. This equa-
tion is expressed as follows:
( )
3
1 + v
o 1.4
k
=
k0 ( v)
1 +
..................................................................................... (4)
It can be applied to calculate the absolute permeability varia-
tions due to the isotropic unloading process. The results are shown
in Figure 11. This figure shows that Tortikes equation matches
the absolute permeability data measured during the isotropic un-
loading process when the initial value is greater than 1 m2. It has
large deviation if the initial permeability is small.
The initial absolute permeability of horizontal specimens from
the block oil sands sample is greater than 1 m2. The comparison
58
of the permeability ratio as a function of volumetric strain due to
the shearing process and the curve based on Tortikes equation
show that Tortikes equation matches the absolute permeability
data of horizontal specimens very well. However, if Tortikes equa-
tion is plotted on a similar figure for vertical specimens, it does
not match the testing results. When the volumetric strain is less
than 4%, Tortikes equation approximately matches the measured
absolute permeability data. However, if the axial strain is greater
than 4%, a large deviation exists between the measured data and
the calculated results from Tortikes equation. The reason is that
the shear failure occurs at the volumetric strain of about 2% to 4%
for vertical specimens. After shear failure, absolute permeability
of vertical specimens improve more significantly and Tortikes
4 5 6 7
value under this condition. Although the modified absolute per-
meability of horizontal specimens matches Tortikes equation, it is
likely coincidental because the improvement of horizontal perme-
ability due to the shearing process is smaller than that of vertical
permeability.
Consequently, it is concluded that if the initial absolute perme-
ability is greater than 1 m2, Tortikes equation is applicable to
calculate the modified absolute permeability due to the isotropic
unloading process. It is also possible to determine the absolute per-
meability value due to the shearing process before shear failure
occurs. After shear failure, Tortikes equation will induce signifi-
cant errors in the calculation of the modified absolute permeability
due to the shearing process. In addition, if the initial absolute per-
meability is much smaller than 1 m2, Tortikes equation cannot
be applied to calculate the modified absolute permeability by both
processes.
It should be noted that another expression of absolute perme-
ability as a function of porosity:
3
k 1 0
2
=
k0 0 1
.............................................................................. (5)
based on the Kozeny-Carman model(9) should be similar to Tor-
tikes equation because both of them do not take the variation of
tortuosity into account. Tortuosity can be defined as the ratio of the
length of a fluid particles path to the length of a straight line be-
tween the beginning and ending points of the path.
Based on the modified Kozeny-Carman model(10), the absolute
permeability ratio can be written as follows:
k C 3 0 S0
=
k0 C0 30 S
................................................................................... (6)
Although Tortikes equation considered the porosity change, it
assumes that the tortuosity is constant during the geomechanical
1.8
1.7
1.6
1.5
k/k0
1.3
1.2
1.1
1.0
0.0 0.5 1.0 1.5 2.0 2.5
Volumetric Strain (%)
FIGURE 11: Absolute permeability ratios from isotropic unloading
test and Tortikes equation.
Journal of Canadian Petroleum Technology
 2 2
Experimental data
Experimental data Chardabellas (Cn1 = 14.76)
1.5 Chardabellas (Cn1 = 5.9) 1.5 Linear (experimental data)
Linear (experimental data)
y = 0.0639x
R2 = 0.5345
Ln(k/k0)

Ln(k/k0)
1

0.5 0.5

y = 0.1409x
0 R2 = 0.7617
0

-0.5
-4 -2 0 2 4 6 8 10 12 -0.5
-4 -2 0 2 4 6 8 10 12
Volume Change (%) Volume Change (%)

FIGURE 12: Determination of absolute permeability due to FIGURE 13: Determination of absolute permeability due to
shearing for horizontal specimens. shearing for vertical specimens.

processes, such as isotropic unloading and shearing. The deviation k

may result from this assumption. Ln = Cn 1 v
k0
....................................................................................... (8)

Permeability of Horizontal and Vertical where

Specimens
Although the vertical and horizontal oil sands specimens were (1 0 ) a + b 0
Cn1 =
all obtained from the block oil sands sample, their permeability 0
varies differently during the shearing process. This difference may ...................................................................... (9)
result from the internal structures of both types of specimens. The
bedding direction in horizontal specimens is consistent with the Using the Chardabellas terms of a = b = B, he obtained:
core specimen axis. In a microscopic scale, the long axis of all the
grains is approximately parallel to the bedding direction because k B
this alignment of grains is more stable. During the shearing pro- Ln = v
k0 0
cess, axial stress applied to the specimen is increased gradually. ...................................................................................... (10)
If the shear stress is large enough to cause the specimen to yield,
shear dilation occurs and oil sands grains may start to rotate, roll, The average initial porosity is 0.339 for both horizontal and
and ride to each other. When the stress path touches the failure en- vertical specimens. However, he chose B = 2 and B = 5 to ob-
velope, peak state is reached and pore space increases significantly tain a good agreement with the experimental results for the hori-
around the shear bands. As shown by Wong(11), the density of shear zontal and vertical specimens, respectively. The parameter Cn1 is
bands can occupy 16% to 47% of the specimen body during the tri- 5.9 and 14.76 (Figure 12 and Figure 13), respectively. It is possible
axial compression tests with Athabasca oil sands specimens. Due to apply these equations to calculate the improved absolute perme-
to the mechanisms associated with shear dilation, the tortuosity ability due to the shearing process in the field if the confining ef-
of fluid flow is reduced and absolute permeability improved fective stress and the stress path are approximately consistent with
accordingly. that used in the testing.
Similar mechanisms can occur in the vertical specimens during It should be noted that although the shape factor C n1 in Equation
the shearing process. The long axis of most grains in these speci- (8) was assumed as a constant, the parameter Cn1 was chosen as
mens is also parallel to the bedding direction. However, it is per- 5.9 and 14.76 to match the testing results of horizontal and vertical
pendicular to the core specimen axis. Thus, when the specimens specimens, respectively. The reason is that, as discussed above,
yield and fail during the shearing process, grain rotation and rolling horizontal and vertical specimens change their pore shapes and tor-
may induce even more significant reductions of tortuosity than that tuosity differently. The different selection of B values or Cn1 values
of horizontal specimens. That is why absolute permeability of ver- has actually taken the different pore shape and tortuosity changes
tical specimens is much larger than that of horizontal specimens into consideration.
for the same volumetric strain.
7
Calculation of Absolute Permeability Due to 6
y = 0.907x
R2 = 0.8767
Shearing 5

Touhidi-Baghini(4) defined the following general equation for

Ln (kw/kw0)

4
absolute permeability: 3

2
a
k =C Ds2 1
(1 ) b ................................................................................. (7) 0

-1
where C is a function of the particle shape and pore shape, Ds is -1 0 1 2 3 4 5 6 7

the mean size of the solid particles, and a and b are constants to be Volumetric Strain (%)
determined from experimental data. Based on this equation and the
FIGURE 14: Determination of effective permeability to water due
assumption that C is constant, he derived the following equation to
to the pure shearing process.
calculate the oil sands permeability:
January 2006, Volume 45, No. 1 59
 The variation of effective permeability to water due to the iso-
tropic unloading process cannot be matched by Tortikes equation.
However, Equation (8) can be applied to fit the tested data with B
value of 6 (Cn1 = 81.69).
Figure 14 shows the effective permeability to water modified
by the shearing process as a function of volumetric strain. The cal-
culation results from Equation (8) are also shown in this figure.
The water porosity of the oil sands specimens S19 to S23 is from
6.1% to 9.3%, and the average value is 8.1%. If B value is 7.5 (Cn1
= 92.59), the calculated results match the tested results very well.
This is consistent with the discussion above that if initial perme-
ability and porosity are very small, shearing has a more significant
impact on absolute permeability.
Water Relative Permeability
The effect of isotropic unloading and shearing process on rela-
tive permeability to water can be discussed based on the testing re-
sults and the proposed calculation methods discussed above. The
modified porosity as a function of volumetric strain can be ex-
pressed as the following equation(4):
v + 0
= is applied, the pore pressure in the partially drained zone can still
1 + v
............................................................................................(11)
Because the tests were done at room temperature or the initial
reservoir temperature(1), the increased porosity due to the isotropic
unloading and the shearing process can be considered as the in-
crease of water porosity. The ratio between the improved water
porosity and the improved total porosity is the modified water sat-
uration. The absolute permeability change due to the isotropic un-
loading process is determined by Tortikes equation and the change
of effective permeability to water due to isotropic unloading is de-
termined by Equation (8) with the Cn1 value of 81.69. The ratio of
these two numbers provides the water relative permeability change
as a result of isotropic unloading. Correspondingly, the absolute
permeability change for horizontal specimens because of shearing
is calculated by Equation (8) with the Cn1 value of 5.9, and for ver-
tical specimens with the Cn1 value of 14.76. The modified value
of effective permeability due to shearing can be decided by Equa-
tion (8) with the Cn1 value of 92.59. Therefore, the water rela-
tive permeability change as a result of shearing is determined by
the ratios of modified water effective permeability and absolute
permeability.
Figure 15 shows the variations of water relative permeability
curves as a result of isotropic unloading and shearing. It can be
clearly seen that the improvement of water relative permeability
due to the isotropic unloading process is much smaller than that
due to the shearing process. Water relative permeability for both
horizontal and vertical specimens can be greatly increased by the
shearing process. If the following theoretical equation
0.00014
Isotropic unloading
Theoretical 1
0.00012
(isotropic unloading)
Shearing (horizontal)
Shearing (vertical)
0.00010 Theoretical 2
0.00008
krw
0.00006
0.00004
0.00002
0.00000
0.2 0.25 0.3
Water Saturation
FIGURE 15: Impact of isotropic unloading and shearing on water
relative permeability.
60
n
kw Sw Swi
krw = =
k 1 Swi
......................................................................... (12)
from the Petroleum Production Handbook(12) is used with the expo-
nent n = 3.70(1), the calculated water relative permeability curves
shown in Figure 15 can be much smaller than those modified by
geomechanical behaviours. In this calculation, the initial water sat-
urations are 20.97% and 23.96% at the beginning of the isotropic
unloading and shearing processes, respectively. It should be noted
that if the temperature increases, the relative permeability to oil
also increases as a function of the isotropic unloading and shearing
processes. Meanwhile, the relative permeability to water will de-
crease accordingly.
Field Applications
As discussed above, both the isotropic unloading process and the
shearing process can improve reservoir permeability. The shearing
process plays an even more important role in this aspect. In SAGD
operations, steam injection pressure is the key factor to induce both
the isotropic unloading process and the shearing process. Even if
the steam injection pressure similar to the initial reservoir pressure
be significantly increased due to the thermal expansion effect.
Clearly, to realize the benefits that may accrue from the geome-
chanical behaviour of an oil sands formation, higher steam injection
pressure is required. It helps reduce the confining effective stress
and the higher temperatures produce greater shear stress in the res-
ervoir material around the interface between the drained zone and
the partially drained zone. As a result, the shearing process will in-
duce better improvements on reservoir permeability (including ab-
solute permeability and relative permeability), particularly for low
permeability reservoirs. In addition, high steam injection pressure
helps the gas lifting process, increasing the oil production rate, and
reducing the entire operation period. However, it should be noted
that geomechanical behaviour is but one aspect in the selection
of steam injection pressure. The practical steam injection pressure
should be determined based on all the aspects associated with the
economic benefits of the SAGD process(13).
Conclusions
1. Isotropic stress and shear stress changes are two major geo-
mechanical processes during SAGD operations. The former
occurs within the high pore pressure front and the latter oc-
curs predominantly around the interface between the drained
zone and the partially drained zone.
2. The shearing process induces more significant improve-
ments on absolute permeability and effective permeability to
water compared to the isotropic unloading process. Partic-
ularly, after shear failure, oil sands permeability increases
0.040
dramatically.
3. Tortikes equation is applicable for the isotropic unloading
0.035
process if the initial absolute permeability is greater than
0.030 1 m2. Otherwise, it may yield significant errors. Tortikes
equation is not appropriate to calculate the modified absolute
0.025
permeability due to the shearing process.
(shearing)
0.020 4. The absolute permeability of vertical specimens increases
krw
0.015 more significantly than that of horizontal specimens for the
same volumetric strain. This may result from the significant
0.010 decrease of tortuosity for vertical specimens.
0.005 5. Geomechanics induced absolute permeability and effective
0.000
permeability to water can be determined based on Equation
0.35 0.4 (8) with different Cn1 values for geomechanical cases.
6. Water relative permeability increases due to the isotropic un-
loading and shearing processes. The shearing process pro-
duces more significant improvement, particularly after shear
failure.
Journal of Canadian Petroleum Technology
 6. ADHIKARY, D.P., Laboratory Investigation of Effects of Stress
Suggestions Changes on Hydraulic Conductivity of Reconstituted Oil Sands;
The variation of absolute permeability due to isotropic un- M.Eng. Report, Department of Civil Engineering, University of Al-
berta, p. 99, 1991.
loading and shearing at higher temperatures needs to be studied.
7. WONG, R.C.K., BARR, W.E., TO, N.M., and PAUL, R., Laboratory
Moreover, the impact of temperature on effective permeability to Measurement of Effective Permeability to Water and Compressibility
water is also an issue for further research. in Unconsolidated Athabasca Oil Sands Cores; proceedings of the
The variation of relative permeability to oil due to isotropic un- 44th Canadian Geotechnical Conference, Calgary, AB, Vol. 2, pp. 56-
loading and shearing also needs to be investigated as a function of 1 to 56-7, 1991.
temperature. 8. KOSAR, K.M., Geotechnical Properties of Oil Sands and Related
It is reasonable to incorporate the geomechanical effect on res- Strata; Ph.D. dissertation, Department of Civil Engineering, Univer-
ervoir permeability into a SAGD numerical study. In other words, sity of Alberta, p. 795, 1989.
the coupled reservoir geomechanical simulations are appropriate 9. DULLIEN, F.A.L., Porous MediaFluid Transport and Pore Struc-
ture: Academic Press, New York, London, 1979.
for the prediction of SAGD performance.
10. COLLINS, R.E., Flow of Fluids Through Porous Materials; The Pe-
troleum Publishing Company, Tulsa, OK, 1976.
11. WONG, R.C.K., Shear Deformation of Locked Sand in Triaxial Com-
Acknowledgements pression; Geotechnical Testing Journal, Vol. 23, No. 2, pp. 158-170,
June 2000.
This research was funded by a NSERC research grant awarded 12. Petroleum Production Handbook, Society of Petroleum Engineers of
to Dr. Chalaturnyk. The authors would also like to acknowledge AIME, Frick, Thomas C., Ed., Dallas, Texas, 1962.
Mr. Kazimierz Oldakowski for re-finding his original laboratory 13. LI, P. and CHALATURNYK, R.J., Discussion of SAGD and Geo-
testing results used in the paper. mechanics; Journal of Canadian Petroleum Technology, Vol. 42, No.
9, September 2003.
NOMENCLATURE
a, b = constants determined by experiment ProvenanceOriginal Petroleum Society manuscript, Permeability Vari-
B = constant in Equation (10) ations Associated With Shearing and Isotropic Unloading During the
C = shape factor in Kozeny-Carman equation SAGD Process (2004-240), first presented at the 5th Canadian International
Cn1 = constant in Equation (8) Petroleum Conference (the 55th Annual Technical Meeting of the Petroleum
Ds = mean size of solid particles Society), June 8 - 10, 2004, in Calgary, Alberta. Abstract submitted for re-
E = Youngs modulus view December 10, 2003; editorial comments sent to the author(s) May 2,
k = permeability 2005; revised manuscript received August 10, 2005; paper approved for
Ln = natural logarithm pre-press October 13, 2005; final approval December 1, 2005.
n = exponent in Equation (12)
S = specific area; saturation
T = temperature
V = volume Authors Biographies
= coefficient of volumetric thermal expansion
= coefficient of linear thermal expansion Pingke Li is a reservoir engineer at Petro-
v = volumetric strain Canada and is currently working with the
= porosity SAGD projects for the company. He has
= density been a Ph.D. student in the Department of
= stress Civil and Environmental Engineering at the
= effective stress University of Alberta since 1999. His Ph.D.
= tortuosity research is focused on coupled reservoir
geomechanical simulations for the SAGD
Subscripts process. Before he joined the University
of Alberta, he worked as a reservoir engi-
0 = initial value
neer at the China National Petroleum Cor-
i = initial
p = constant pressure in Equation (1) poration in Beijing from 1990 to 1998. He holds a B.Sc. degree
r = relative from the South Western Petroleum Institute, and an M.Sc. degree
T = temperature from the Research Institute of Petroleum Exploration and Devel-
v = volumetric opment (RIPED) in China. Pingkes professional interests include
w = water both reservoir engineering and reservoir geomechanics, specifi-
cally in thermal recovery processes including CSS, steam drive,
and SAGD. He has authored and co-authored 19 technical papers
REFERENCES regarding steam drive and the SAGD process. He is a member of
1. OLDAKOWSKI, K., Stress Induced Permeability Changes of Atha- SPE, CHOA, and the Petroleum Society.
basca Oil Sands; M.Sc. Thesis, Department of Civil Engineering,
University of Alberta, 1994. Rick Chalaturnyk is an associate pro-
2. SCOTT, J.D., ADHIKARY, D., and PROSKIN, S.A., Volume and fessor of geotechnical engineering at the
Permeability Changes Associated With Steam Stimulation in an Oil
Sands Reservoir; paper 91-63, proceedings of the Petroleum Sop-
University of Alberta, having obtained his
ciety/AOSTRA Technical Conference, 1991. Ph.D. in geotechnical engineering from
3. CHALATURNYK, R.J., Geomechanics of SAGD in Heavy Oil the University of Alberta. Before joining
Reservoirs; Ph.D dissertation, Department of Civil Engineering, Uni- the university, he was involved in petro-
versity of Alberta, 1996. leum geomechanics research in SAGD and
4. TOUHIDI-BAGHINI, A., Absolute Permeability of McMurray For- heavy oil cold production with the Centre
mation Oil Sands at Low Confining Stresses; Ph.D dissertation, for Engineering Research, and co-founded
Department of Civil Engineering, University of Alberta, p. 339, a downhole instrumentation and monitoring
1998. company. He has been involved in a wide
5. LI, P., CHALATURNYK, R.J., and POLIKAR, M., Issues With Res-
ervoir Geomechanical Simulations of the SAGD Process; paper
range of research activities ranging from mine tailings manage-
2002-130, presented at the Petroleum Societys Canadian Interna- ment technologies, geological storage of greenhouse gases, gas-
tional Petroleum Conference, Calgary, AB, June 11 13, 2002. Also, over-bitumen issues pertaining to SAGD development, and coalbed
Journal of Canadian Petroleum Technology, Vol. 43, No. 5, pp. 30-40, methane geomechanics. He is a registered professional engineer in
May 2004. Alberta, and is a member of SPE and the Petroleum Society.
January 2006, Volume 45, No. 1 61