Professional Documents
Culture Documents
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- almost a necessary component. Without flowback, a test may take
iour before and after the fracture closure is magnified by the flow- days to complete properly when multiple cycles are used. With
back, which makes the closure defined more clearly. Fig. 4 shows flowback, a test with multiple cycles can be completed in 2 to 3
one example plot of pressure vs. square root of the shut-in time. hours. Fig. 5a shows one example test performed in the Clear-
The inflecting pattern is clear on the plot, and the intersection of water capping shale. It took approximately 3 hours to finish nine
two different slopes points to the fracture closure. The detailed cycles. A consistent, and thus reliable, fracture-closure pressure
description of fracture flowback tests is outside the scope of this was obtained as shown in Fig. 5b.
paper but can be found in some published papers (Raaen and It is our observation that although minifrac tests are used rou-
Brudy 2001; Raaen et al. 2001). tinely in the industry, significant attention should be paid to the
Natural leakoff from the fracture can be slow if the water mo- unique conditions in the oil-sands development and the corre-
bility is small, which is common when testing caprock shales. It is sponding demands for high-quality data and interpretation accu-
also likely to be small in the oil-sands pay zone if it has only mini- racy. Some measures that can enhance the quality include
mum initial water mobility. In these cases, the flowback becomes multiple injection/shut-in cycles to check for consistency in the
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 low permeability requires slow loading rates; therefore, tests on
flowback technique can help define the fracture closure better and shales take a long time. Osmosis and hydration may be involved
also increase test efficiency. Third, a proper minifrac test should if fluids of different activity contact the shales. Shale properties
not measure an in-situ minimum stress larger than the density- will change if shale is hydrated or dehydrated. Therefore, it is nec-
derived vertical stress. If this does occur, a definitive explanation essary to determine the chemistry and activity of the pore fluid in
should be provided. 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
Geomechanical Laboratory Tests the Clearwater shales in the current shallow well. Compared with
Mechanical properties of the caprock, including the compressive the conventional hard-rock shales, the Clearwater shales above
strengths, are essential for determining the maximum thermal the target reservoir have a greater quartz content. Correspond-
operating pressure for the reservoir. Geomechanical laboratory ingly, the clay content is lower. Among the clay minerals, the
tests on the site-specific cores are the most-comprehensive and - illite/smectite (I/S) mixed layers are dominant. A higher smectite
effective means of deriving the necessary mechanical properties. content foretells a high hydration tendency. These characteristics
Mechanical tests on rocks in general are complex, and tests on are observed in the field. Without proper attention to the coring
shales are even more complicated. Attention to the complexity is mud, Clearwater shale cores can readily become hydrated when
important because shale is clay rich with low permeability. The brought to the surface. Therefore, careful attention is required
P–√Δt
1.2
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.
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
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. tion and that for C24T1S1 from the lower part. C14T3S1 was
Hydration inhibitors added to the mud become necessary. tested under an effective confining pressure, CP¼0.2 MPa. But its
Both room-temperature and high-temperature triaxial tests peak stress is higher than C24T1S1, which was tested under a
were conducted on more than 10 samples for this project. They higher CP¼1.5 MPa. One possible cause is the difference in the
were used to derive the elastoplastic deformation properties and mineralogy between the two samples. As shown in Fig. 6, the
strength parameters to be used for the geomechanical simulations. upper part of the Clearwater caprock, where C14T3S1 came from,
A detailed description of the test procedures and results (e.g., how is more silty, while the lower part where C24T1S1 was situated
to derive the Mohr-Coulomb failure envelope from the tests) are has a higher clay content and I/S mixed layers. A similar phenom-
beyond the scope of this paper. Interested readers can refer to a enon has been reported in literature on the mechanical behaviour
paper by Xu et al. (2011) for the triaxial-test procedures and refer of soils. For example, some researchers (Bishop 1966; Kenney
to relevant publications by the ERCB for the test results. In the 1967; Mitchell and Soga 2005) documented that soil with a higher
following, the effect of high I/S contents on the mechanical clay content, especially when the clay minerals are mostly mont-
behaviour of the Clearwater shales is elaborated upon. Little morillonite, exhibits a small friction angle. Contribution from a
attention has been directed to this factor so far in the context of higher normal stress or confining pressure to the peak stress is
caprock-integrity studies, and thus it warrants a dedicated discus- muted for a smaller friction angle.
sion herein. Sedimentary rocks (e.g., the Clearwater shales in our current
Fig. 7 plots the stress/strain curves for two representative sam- discussion) rely on the friction for their mechanical strength. The
ples: C14T3S1 from the upper part of the Clearwater caprock sec- normal stress condition or confining pressure in the triaxial tests
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- late the induced stresses and deformation in the caprock caused by
tions suggest that for shallow depths, the mineralogy, in con- the evolving pore-pressure ( p) and temperature (T) conditions in
junction with the stress condition, also affects the strength the reservoir. The simulation models start with the original in-situ
significantly. For greater depths, the confining pressure is much stress condition measured in the minifrac tests. They use the me-
higher. There, the stress condition may overshadow the mineral- chanical properties extracted from the laboratory tests to describe
ogical influence. Therefore, it is important to characterize the the constitutive deformation behaviour and allow safety factors to
mineralogical variation in the Clearwater shale. Different depth be evaluated for a given thermal operating pressure. Finally, the
intervals may exhibit different clay/quartz contents. Therefore, optimal maximum operating pressure (MOP) is derived by a series
these intervals should be grouped and tested separately. Different of sensitivity-analysis runs for a given geology in a project site.
mechanical properties may be measured for each of such groups. 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
Geomechanical Simulation because of its strong nonlinear solution capability and unrivaled
Geomechanical simulations are used to investigate the impact of strength in simulating the coupled thermal/hydraulic/mechanical
thermal operating conditions on the caprock integrity. They calcu- processes. For the caprock strata, our simulations are coupled
2.0
1.8
1.6
1.2
1.0
0.8
0.6
C14T3S1: CP = 0.2 MPa
0.4
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 ditions calculated independently by a thermal simulator or from
deformation. The hydraulic/mechanical coupling is two-way— analytical solutions. Their induced reservoir deformation is thus
namely, the pressure diffusion and rock deformation influence simulated by ABAQUS. Reservoir dilation is automatically
each other fully. But the thermal/mechanical coupling is one-way accounted for by the chosen elastoplastic model and input dilation
(i.e., the temperature diffusion affects the rock deformation only, angle. But the thus-calculated reservoir deformation is not allowed
while the mechanical deformation does not affect the thermal to feed back into the thermal simulation, affecting the p and T con-
properties and temperature distribution). Temperature is allowed ditions in the reservoir. It is our opinion that such one-way cou-
to diffuse into the caprock because of the SAGD temperature con- pling from the p and T conditions to the mechanical response in
ditions in the reservoir. But no thermal convection is considered. the reservoir is an adequate approximation and the most cost-
This is valid given the low permeability and thus the slow fluid- effective if the major focus is on the caprock deformation only.
flow velocities in the caprock. The following discussion will compare the induced stresses in
Within the reservoir layer, the coupling is only one-way: Our the caprock for two different reservoir depths: 60 and 200 m. The
specially-designed ABAQUS driver enters the SAGD p and T con- 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-
Stress-free ground surface 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
Clearwater current discussions. An oil-sands pay zone 27 m thick is sand-
caprock, ho 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-
z izontal SAGD wells. The top of the model is stress free,
Oil-Sands
representing the real ground surface, while the bottom of the
pay zone, hr model is constrained from moving in the vertical direction,
x
reflecting the support from underneath. It is an artificial boundary
condition, but it does not affect the computations in the pay zone
Underburden 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
Fig. 8—A schematic model used for the geomechanical simula- Drucker-Prager (D-P)-type plasticity model was used to account
tions and analytical derivations concerning induced stresses for the shear-induced dilation in the reservoir. In addition, a po-
and deformation in the caprock caused by SAGD steaming in rous nonlinear elastic model is used in the reservoir to account for
the reservoir. the pressure-dependent elastic moduli (SIMULIA 2012). A
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.
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 Lower shear stresses and a more compressive nature in the
as plotted increases in the caprock during the SAGD operation in induced stresses all favour the integrity of the caprock. Unfortu-
both shallow and deep reservoir depths. The increase is more in nately, because of its shallow depth, the initial stress state is under
the deeper reservoir (> 2 MPa) than in the shallow one (approxi- a smaller confinement in the shallow reservoir than in the deep
mately 1.5 MPa). In the case of a deeper reservoir, the stress state one. This is reflected by the fact that the original state of stresses
becomes less compressive (i.e., the stress path moves to the left for the shallow reservoir depth is located to the left of (i.e., less
from the original condition). This is caused by the higher steam compressive than) that for the deep reservoir depth. The rock
pressure and thus a higher pore pressure diffusing into the cap- strength is mostly derived from its friction. As a result, the cap-
rock. On the contrary, for the shallow reservoir, the stress state rock touches the failure envelope more readily in the shallow
moves toward more compressive or toward the right on the plot. depth than in the deep one. As shown in Fig. 11, the stress path
This reflects the low operating pressure and the dominating nature for the shallow reservoir has passed the 30 failure envelope at
of the compressive thermally induced stresses. 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)
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.
1.8 45°
1.0
0.8
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Mean Effective Stress (MPa)
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 stresses become significantly smaller because of the smaller
30 failure envelope at Elem 1, but stays just below, although Young’s modulus, so they are comparable to or even overshad-
close to, the 45 failure. Location of Elem 2 is safe for the two owed by the decreasing effective stresses caused by the pore-pres-
strength envelopes. sure increase. The latter makes the stress state approach the less-
Limited Sensitivity Analyses. Caprock integrity is a complex compressive state (i.e., to the left in Fig. 12). In summary, the
issue and numerous factors contribute (Yuan 2008). Ideally, a se- softer or more-compliant nature of the shallow caprock shale sug-
ries of sensitivity analyses should be run in the simulations. How- gested by the laboratory tests helps maintain the caprock integrity
ever, the major purpose of the simulation works presented in this where only the induced stresses are concerned.
paper is to compare the induced stresses in the caprock during the Thermal expansion is affected dramatically by the drained or
SAGD operation between a shallow (60 m) and deep (200 m) res- undrained heating. The pore fluid has a thermal-expansion
ervoir, respectively. Therefore, to limit the scope of this presenta- capacity generally at an order of magnitude larger than that of the
tion, the sensitivity analysis is limited to the following two rock matrix. For low-mobility porous media (e.g., the bitumen-
material parameters: (1) Young’s modulus of the caprock shale saturated oil sands), the excessively expanded fluid phase cannot
and (2) thermal-expansion coefficient of the reservoir. The analy- dissipate in time and, consequently, more volumetric expansion
sis is further limited to the shallow reservoir depth and illustrated results in the undrained heating (Agar et al. 1986) or thermally
by the changes in the surface heave (Fig. 10) and induced stresses induced pore pressure (Butler 1986). With the intention of con-
in the caprock (Fig. 12). Details are summarized next. servative design, the earlier simulations used a larger thermal-
The earlier presentation demonstrated that the mineralogical expansion coefficient for the reservoir. Fig. 10 shows that a
content of the caprock shale varies across the depth and the smaller thermal-expansion coefficient for the reservoir,
impact on the mechanical strength can be significant for the shal- aT¼410–5/ C, causes a much smaller surface heave. But Fig. 12
low reservoirs. Fig. 7 suggests that the Young’s modulus E for demonstrates that the ultimate level of the induced stresses in the
Sample C24T1S1 is approximately 20 MPa if measured by a se- caprock at the end of the SAGD operation is similar between the
cant line across the major portion of the stress/strain curve data two thermal-expansion coefficients. The stress path is simpler in
before the peak stress. The surface heave is not significantly the case of small thermal-expansion coefficients aT¼410–5/ C.
affected by the Young’s modulus of the caprock (Fig. 10). This It is a nearly straight line throughout the SAGD operation, extend-
can be explained by the analytical derivations in Appendix A. The ing right and upward. On the contrary, the larger thermal-expan-
vertical displacement caused by the thermal expansion (Eq. A-5) sion coefficient, aT¼210–4/ C, causes a nonmonotonic evolution
has no variable E involved. E does affect the poroelastically history in the induced stresses. The difference reflects the two
induced vertical displacement (Eq. A-6), which is responsible for major mechanisms causing the caprock to deform. Reservoir de-
the difference observed in the numerical calculation, as shown in formation and thermal stress take place when the temperature
Fig. 10. front diffuses into the caprock. These two mechanisms differ in
But much smaller induced stresses will result if the caprock is time and magnitude when they contribute to the caprock deforma-
softer or more compliant—having a smaller Young’s modulus tion. The similar end state of the induced stresses shown in Fig.
(Fig. 12). When E¼20 MPa, the state of stresses all stay below 12 suggests that, ultimately, the thermal stress induced by the
the strength lines (i.e., the caprock at the shallow depth is safe temperature changes in the caprock dominates the stress state
against the two strength envelopes: 30 or 45 for the D-P friction therein.
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 Conclusions
Young’s modulus E and the thermal-expansion coefficient. A The Alberta oil-sands/heavy-oil industry is exemplary and proac-
smaller E generates a smaller induced thermal stress. Also note tive in safeguarding caprock integrity. The intent is to design a
that when E¼20 MPa, the stress path moves toward the left (i.e., safe operating pressure and steam temperature condition so that
less-compressive state, as shown in Fig. 12). The induced thermal the caprock maintains the sealing capacity during thermal