You are on page 1of 13

SPE 157855

A Geomechanical Methodology for Determining Maximum Operating


Pressure in SAGD Reservoirs
Dale Walters, Taurus Reservoir Solutions Ltd., Jin Wang, Suncor Energy, Inc., A. Settari, University of Calgary

Copyright 2012, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Heavy Oil Conference Canada held in Calgary, Alberta, Canada, 12–14 June 2012.

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

Abstract
The thermal recovery of bitumen reservoirs by steam assisted gravity drainage (SAGD) is often designed to maximize the
operating pressure while maintaining a safe and economic operation. In general, higher operating pressure can reduce
thermal efficiency due to heat losses to over/underburden formation, but the other benefits usually compensate. To name a
few, higher steam temperatures can maximize the reduction of oil viscosity, enhance permeability associated with lower
effective stress and shear dilation, and give a larger pressure window to allow flexible control of the producer. This is
especially important for shallow reservoirs where the pressure window for injection and production is smaller. The limitation
of the maximum operating pressure is then based on maintaining caprock integrity. Thus, shear and tensile failure
mechanisms should be quantified and managed.

This paper presents a methodology to perform a geomechanical analysis of caprock integrity for SAGD operation and
illustrates the available approaches. Both analytical and numerical approaches are compared demonstrating their usefulness.
Main factors in the analysis are the knowledge of the initial stress state and proper representation of the complexity of the
geomaterials. A typical initial stress state for a northern Alberta SAGD property, Suncor’s MacKay River project, is
presented showing the potential for low initial minimum total stress and elevated initial shear stress levels. The stress-strain
behavior for the MacKay River sand and caprock materials is discussed focusing on the potential for shear dilation in the
sand and shear strength behavior in the caprock. An elasto-plastic constitutive model is used to represent the sand and
caprock materials. The increase in pressure and temperature alter the stress state and disturb the soil matrix. This
disturbance results in shear dilation of the sand matrix creating regions of enhanced permeability and porosity. Also, the
transfer of stress and strain to the caprock causes dynamic stress changes and, therefore, dynamic behavior of shear and
tensile failure conditions. Calculations are presented showing the stress paths associated with SAGD operations, suggesting
better design of lab testing programs and the implications for shear dilation in the sand and shear failure in the caprock.
Finally, the results are used to demonstrate locations that are most likely at risk for potential tensile and shear failure. Stress
ratios are used to summarize the analysis and quantify and monitor the failure mechanisms.

The above methodology has been developed and applied in several studies of other SAGD projects and aided the operators in
the optimization and permitting the operating conditions.

INTRODUCTION
In situ bitumen recovery in Alberta’s Athabasca oil sands is currently dominated by the SAGD (Steam
Assisted Gravity Drainage) recovery method. SAGD technology allows bitumen to be recovered from oil
sands deposits that are impractical for surface mining and extraction methods, but not deep enough for high-
pressure steam techniques to work. Only five per cent of oil sands can be recovered through surface mining,
and other in situ extraction techniques recover around another 15 per cent. The development of SAGD
potentially doubles the economic viability of Canada’s oil sands reserves (Alberta Innovates, 2010). The
SAGD method was first studied by R.M. Butler (Butler, 1980). The first field test of the SAGD process was
the Underground Test Facility (UTF) project initiated by the Alberta Oil Sands Technology and Research
Authority (AOSTRA – since restructured and rebranded Alberta Innovates - Energy and Environment
2 SPE 157855

Solutions) (O’Rourke et al., 1997). This pilot-test project, which was carried out between 1986 and 1998,
proved the process and stimulated the development of commercial SAGD projects in Alberta.

The SAGD process relies on the gravity drainage of the bitumen with decreased viscosity while heating. High
pressure SAGD maximizes the steam temperature, latent heat and high reduction in oil viscosity. This aids the
SAGD start-up procedures, early steam chamber development and reduced time to peak oil rate. There are
drawbacks to higher temperature and pressure SAGD due to inefficiencies with heat loss to the overburden and
potential energy loss to thief zones. However, in optimum oil sand reservoirs without thief zones, maximizing
operating pressure in the first 2 – 3 years is common practice.

The sand-dominated McMurray Formation found in northeastern Alberta is the focus of most of the current
commercial SAGD projects. The McMurray Sand is a dense granular material with the best reservoirs
containing clean, channel sand deposits. In Alberta, the oil sands exhibit a dense packing of grains due to the
glacial history. This dense packing and unconsolidated nature of the sands increase the potential for
geomechanical effects with increased pressure and temperature during SAGD. Increased pressure with SAGD
lowers the effective stress and decreases shear strength, causing a nonlinear increase in the pore volume and
permeability of the sand due to softening and shear dilation. These geomechanical effects may therefore cause
an increase in the efficiency of the gravity drainage, which is another aspect operators try to capitalize on by
maximizing the SAGD operating pressure. In general, the caprock associated with the McMurray Sand is
typically the Clearwater Shale, the thickest shale formation, laterally continuous over the development area
and impermeable. However, the interbedded mudstones in the Wabiskaw units (situated above just below the
Clearwater Shale), although thinner and of higher silt content and higher permeability, have been shown to
arrest the steam chamber growth over the life of a SAGD well pair and, therefore, may also be used as an
effective caprock (Collins et al. 2011).

The maximum operating pressure (MOP) for SAGD is regulated to ensure steam containment within the
reservoir zone and protect the safety of operators, the public and the environment. Ensuring steam containment
depends on maintaining caprock integrity. Caprock integrity has more commonly been associated with gas
storage projects where an increase in maximum operating pressure results in increased deliverability and
working gas capacity (Bruno et al. 1998, Bruno et al., 2000, Khan et al. 2010, Hawkes et al. 2005, Dusseault et
al. 2001). More specifically, issues of caprock failure and associated well failures due to steaming operations
in oil sands have been studied (Smith, 1997; Collins, 2007). Similar concepts control the caprock integrity for
the SAGD process. The general geomechancial processes at play during the SAGD process have been
discussed in detail by Collins (2000). Finally, there has been some general discussion on the specific problem
of caprock integrity for SAGD (Chalaturnyk, 2011; McClellan and Gillen, 2000). The focus of these studies
has been the characterization of fracture networks and constitutive models of the caprock and mudstone
materials and coupled modeling to assess the dynamic nature of the stresses.

This paper reviews the main factors controlling the caprock integrity of SAGD processes: material behavior
and stress state. The failure mechanisms will be defined and evaluated in analytical and numerical solutions to
the problem. Finally, coupled reservoir and geomechanical models of Suncor’s MacKay River lease will be
used to illustrate the locations for potential risk of failure.

GEOLOGY
Figure 1 shows a typical well from the Suncor’s MacKay River project development area (MacKay River
commercial approval 8668O). The McMurray Formation is the lowermost unit within the Lower Cretaceous
Mannville Group in northeast Alberta. The lower part of McMurray is a heavily channelized, fluvial-
dominated succession that overlies Devonian carbonates, and the upper part of McMurray is a heavily
channelized, mixed fluvial-tidal succession that overlies the lower McMurray fluvial deposits. These channel
sands typically grade upward into inclined heterolithic stratified (IHS) units of rippled sand/heavily burrowed
silty mudstones. The upper part of the McMurray is overlain by the Wabiskaw Member of the Clearwater
Formation with an upper erosional contact which often contains mudstones. The Wabiskaw deposits differ
considerably due to a transition to a full marine depositional environment. The Clearwater Formation overlies
the McMurray Formation. The Wabiskaw Member at the base of the Clearwater contains two important, well-
SPE 157855 3

defined units at its base: the Wabiskaw C (marine sand) underlain by the Wabiskaw D Mudstone
(marine/estuarine muddy sand to sandy mudstone). This continuous and predominantly Wabiskaw D
Mudstone is used as a base of caprock for SAGD operations in MacKay River. The Wabiskaw C Sand is
important as it is typically a relatively high permeability formation allowing pressure monitoring above the
SAGD operations. Above the Wabiskaw units the Clearwater Formation is dominated by regionally extensive
marine mudstones and siltstones.

Figure 1. Stratigraphy of MacKay River PDA area

The Wabiskaw unit (Wab D Mudstone, Wab C Sand, and Wab A Shale) and the Clearwater Shale make up the
potential caprock layers. However, the reservoir material, the McMurray Sand, plays in integral role in the
deformations and stresses transferred to the caprock. Therefore, both materials require detailed representation
in any engineering calculations or modeling exercise. The McMurray Sand, the best quality reservoir sand, is
unconsolidated sand, with grains typically angular in shape and in a dense packing arrangement. The historical
ice ages resulted in ice loading of over 1 km thickness resulting in severe overconsolidation. This dense
packing arrangement results in a “locked” sand (Dusseault and Morgenstern, 1979) that is relatively stiff under
small strains, but softens with larger strains due to shearing and rearrangement of the packing.
4 SPE 157855

GEOMECHANICAL DATA DESCRIPTION

A geomechanical analysis is very much dependent on the data describing the problem. In general this data can
be grouped into the material description and the stress initialization, one often dependent on the other. The
data important within these two groupings will be discussed next.

Geomechanical material description

Although many geological variations exist in the materials describing the base rock, reservoir sand and
caprock, geomechanically these materials can be grouped into hard rock, fine grained and coarse grained soils.
The Devonian or base rock is usually assumed to behave as a hard rock represented by linear elastic properties.
The reservoir sand and caprock can be characterized as soils. The geotechnical community has an established
soil classification system that can be used to categorize soil types for their geotechnical uses (Dunn et al.,
1980). The same classification governs the geomechanical behavior of these materials with respect to SAGD
operations and caprock integrity. Many variations exist within these groups such as grain size distribution, clay
content, clay type, grain shape, etc.; however, geotechnically they can be separated based on their mechanical
behavior.

The McMurray oil sand behavior is typically consistent with dense sand particles. It shows distinct peak and
residual shear strength with a strong tendency to dilate through shear failure. An example of MacKay River
McMurray sand stress-strain behavior is shown in Figure 2. The stress-strain behavior is fairly linear
(depending on the disturbance of the core, which can be measured using disturbance index (Wong, 2004)) until
the peak shear strength is reached. During shear failure rearrangement of the sand grains causes significant
shear dilation and a reduction of the shear strength to a residual level representing the critical state. An elasto-
plastic (EP) constitutive model was used to describe the sand, which includes both the void ratio and average
effective stress in the strength calculation thereby incorporating micromechanical effects in a macroscopic
continuum treatment of the material (Wan and Guo, 1998). The constitutive model is general enough to
capture the behavior of both loose (normally consolidated) or dense (over consolidated) sands. The yield
surface of the EP model in I1 – sqrt(J2) plane is shown in Figure 3, where I1 is the first invariant of the stress
tensor, and J2 is the second invariant of the deviatoric stress tensor. In brief the EP constitutive model is
characterized by:
 Shear yield surface (fs) dependent on void ratio
 Compaction yield surface (fc) dependent on preconsolidation stress and void ratio
 Tension cut-off yield surface (ft)
 Modified Rowe’s stress dilatancy behavior dependent on current void ratio, critical void ratio and
material parameters
Figure 2a shows that the constitutive model is able to adequately capture the material stress-strain behavior of
the Mackay River McMurray sand. The peak shear strength is governed by fs, thus as the effective confining
stress increases the peak shear strength should also increase. The lab data showed some variation, potentially
due to sample disturbance or sample variation. As expected, the volumetric expansion with shear dilation
tends to increase with a decrease in effective confining stress. In fact, at high effective confining stresses shear
dilation becomes very difficult. This is one of the main reasons for operating at the maximum operating
pressure (low effective stress) – to maximize the potential for shear dilation and increase in permeability.
SPE 157855 5

(a) (b)
Figure 2. Lab data (symbols) and EP match (solid lines): (a) McMurray Sand; (b) Clearwater Shale

J2

fs

gM
fc Qs 1
b
compressio n N tension
1 ft
I1
T
I1apex

Figure 3. Yield surface used by Elasto-Plastic Constitutive model

The caprock material is characterized by mudstones from the Clearwater Shale, Wabiskaw A Shale and
Wabiskaw D Mudstone. Data for the Clearwater Shale of the MacKay River lease is presented here and shown
in Figure 2b. The same EP constitutive model was used to match the stress strain behavior of the Clearwater
Shale. Although designed for sand, the EP model is general enough to characterize any material that is
normally or overly consolidated. The Clearwater Shale is expected to be highly over consolidated and shows
this behavior with peak shear strengths and some expansion with shear failure. To match the initial linear
portion of the stress-strain data, the Young’s modulus was allowed to increase with minimum effective
confining stress. Some dilation of the shale materials was observed with shear, but very little net volumetric
strain increase.

The two main materials affecting the SAGD operation (sand and caprock) have been characterized by an EP
constitutive model. These two material descriptions were used in the following calculations of stress changes
and caprock integrity.

Stress data

In the Western Canadian sedimentary basin, the regional stress state is dominated tectonically by movement of
the North American Plate (Bell et al., 2010). This results in a dominant minimum horizontal stress ( hmin)
orientation of NW-SE or in general parallel to the Rocky Mountains. For the caprock analysis completed
herein a stress regime calibrated to the MacKay River lease area was used. In most cases for this shallow
depths and this area of Alberta, the minimum horizontal stress measured is somewhat greater than that
generated from gravity loading. This suggests tectonics or residual stress from glaciation altered the stress state
developed with deposition. Two minifrac test programs were performed at the MacKay River lease. The two
programs (minifracs #1 and minifracs #2 shown in Figure 4) were completed on different wells, however, they
were compared and used to calibrate a 1D stress profile for MacKay River by using the fracture closure
6 SPE 157855

gradient (h divided by depth) for each zone tested in one consistent log profile. A log with compressional and
shear sonic data close to the minifrac locations was used for the calibration process. The following procedure
was taken:
1. Integrate density log to calculate vertical total stress
2. Establish pressure profile based on field observations and regional description
3. Interpret elastic rock properties from shear and compressional wave velocity data
4. Calculate horizontal total stress profile based on uniaxial strain boundary conditions and gravity
equilibrium
5. Adjust the horizontal stress state using tectonic strain and/or glacial loading hysteresis
6. Calculate average stress gradients through each stratigraphic zone to populate simulation model

Figure 4. Stress profiles interpreted for MacKay River Lease

It may be observed that the stress profile in general shows the minimum horizontal stress is equal to or greater
than the vertical stress. In fact the minifrac data showed stress gradients of about 23 kPa/m for the caprock
zones and 22 kPa/m in the McMurray sand. In order to adjust the horizontal stress from uniaxial strain gravity
loaded conditions a tectonic strain approach was used. The governing equations of poro- and thermoelasticity
(presented in equation (2)) may be used to calculate the adjusted stress profile by making some assumptions
and simplifying assumptions:
1. x = H; x = H = strain in maximum stress direction
2. y = h; y = h = strain in minimum stress direction
3. Vertical boundary is free to deform
Using these assumptions, the horizontal stresses may be calculated as:
 E E
H  ( z  p)  p    h
1  1  2 H
1  2
 E E
h  ( z  p)  p    H (1)
1  1  2 h
1  2

The calibrated stress profile shown in Figure 4 used only a strain in the maximum horizontal stress direction
(  H   H ;  h  0 ). The resulting difference between maximum and minimum horizontal stress becomes a
function of the elastic rock properties (E, ). Thus, the mudstones layers showed a maximum horizontal stress
(H) about 20% larger than the minimum horizontal stress (h), while the McMurray sand showed a larger
contrast in horizontal stresses (about 30%). The magnitude of the maximum horizontal stress is difficult to
quantify from field testing. Minifrac breakdown analysis, break-out analysis and borehole ellipticity can all be
used to make some inferences on the degree of anisotropy of the horizontal stresses. The values used in this
analysis, especially in the sand are at the high end of expected values. The implications of this stress state are
SPE 157855 7

twofold. First, the z is the lowest total stress indicating fracturing pressure (tensile failure) at the overburden
stress. Thus horizontal fractures would be expected. Second, the initial deviatoric stress (max–min) is fairly
large. The deviatoric stress is important when considering shear failure. This will be discussed next.

MOHR-CIRCLE ANALYSIS

This next section will introduce a method to analytically investigate the stress changes due to SAGD
operations. First, the equations are presented to express the relations between the deformations and change in
stress as the pressures and temperatures change during a SAGD operation. Special forms of these equations are
then used to estimate potential failure mechanisms and the change in pressure and/or temperature required to
cause those failures.

Poro- and thermoelasticity

The theory of poro-thermoelasticity is developed by extending the theory of linear elasticity to include the
effect of pressure and temperature (Coussy, 1994). These equations may be written in incremental form (and
geomechanics sign convention – compression positive) as:
 x  y  z p
 x     a L T  
E E E 3K
 y  x  z p
 y     a L T  
E E E 3K
 z  y  x p
 z     a L T  
E E E 3K (2)
where  is strain,  is total stress,  is Poisson’s ratio, E is Young’s modulus, K is the bulk modulus, T is
temperature, p is pressure, aL is the drained linear thermal expansion coefficient and  is Biot’s poroelastic
constant. Biot’s constant may be calculated as:
K
  1
Ks (3)
where Ks is the grain modulus (bulk modulus of the solid grains making up the soil/rock skeleton).
These governing equations may be simplified to represent the typical reservoir geometry by making uniaxial
strain assumptions (shown as a cross-section in Figure 5). The left side of Figure 5 shows the actual 3-D semi-
infinite medium. The right side shows a portion of the medium extracted for the calculation of the stresses. The
simplified model assumes that in-situ conditions can be approximated by a 2D block of material, which is
constrained on all surfaces except the top surface, which has full freedom of movement. The sides of the
model are constrained in the lateral direction. These assumptions give a good approximation for an isotropic,
homogeneous reservoir with a width that is much greater than its height and much greater than the depth to
surface (since it neglects arching effects).
Free Surface – Ground Level

Reservoir

Material Properties:
Reservoir E, aL

Loads:
p, T

Figure 5. Uniaxial strain boundary conditions

The general formulation of equation (2) may be simplified, based upon the idealization shown on the right side
of Figure 5. This requires applying the following assumptions:
 x   y   h  0
8 SPE 157855

 x   y   h
 z  0 (4)
Using the above assumptions and rearranging the equations allows the stress solution to be written as a
horizontal stress change divided into two components: the thermoelastic effect and the poroelastic effect:
Total stress change:
 h   ht   hp
(5)
Thermo-elastic stress change:
E
 ht  aL T (6)
1 
Poro-elastic stress change:
1  2
 hp   p (7)
1 

Mohr-Circle analysis for shear failure

A convenient way to analytically evaluate the shear strength of the materials under SAGD conditions is to use
a Mohr-Circle (MC) analysis (Dunn et al., 1980). Knowing the initial stress state and using Equations (6)
through (6) one may calculate the changes in total stress and evaluate shear failure. The main assumptions in
this analysis are as follows:
1. Uniaxial strain boundary conditions (i.e., no lateral strain, free vertical strain)
2. No arching
3. Homogeneous pressure and temperature distribution
These assumptions oversimplify the stress changes going on during SAGD operations, but may be used to
provide a first estimate of safe operating pressure. Three calculations with different levels of complexity can
be used for the MC analysis. These are explained next.

The initial stress state may be plotted as three MC’s using the principal stresses (1, 2 and 3). These are
shown as three MC’s in Figure 6 (navy blue, green and orange). The shear failure analysis will be performed
with the largest of the three MC’s in each case. The first analysis is to plot the MC accounting for the current
effective stress state and only the change in effective stress associated with the pressure changes during
SAGD. In this case the MC would remain the same size as initially (navy blue line in Figure 6), but move to
the left until it touches the shear failure envelope (yellow circle intercepting the red line in Figure 6). For the
Mackay River stress state the maximum principal stress is the maximum horizontal stress and the minimum
principal stress in the vertical stress.

(a) (b)
Figure 6. Mohr-Circle analysis of Shear Failure (a) Clearwater Shale and (b) McMurray Sand

The second analysis uses Equation (7) to calculate the change in horizontal total stress, increasing the pressure
until it touches the shear failure surface. The stress regime plays a key role in how the MC develops. The
horizontal stresses will increase with an increase in pressure. Thus, if the initial horizontal total stresses are
greater or equal to the vertical stress the MC will increase in size as it moves to the left and touches fs. This
SPE 157855 9

causes the MC to become larger as it moves to the left to touch the shear failure envelope (light blue line in
Figure 6a and 6b).

The third analysis uses Equation (6) and (7) to calculate the change in horizontal total stress, due to pressure and
temperature increase until it touches the shear failure surface (assumes saturated steam conditions). As noted with
the pressure changes, the assumed reservoir boundary conditions cause the total stress in the lateral direction to
increase with an increase in temperature, while the total vertical stress remains unchanged. This effect when
combined with the poroelastic stress changes causes the MC to become even larger as it moves to the left (due to
increased temperature and pressure decreasing the effective stress) to touch the shear failure envelope (purple line
in Figure 6). The stress paths of this third analysis for both stress regimes are plotted in Figure 9 (labeled MC).
They are plotted together with simulation results, which will be discussed next. It is the authors’ experience that
the results are conservative especially for the Clearewater Shale as they assume that the pressure and temperature
changes fully penetrate the caprock. In reality, vertical pressure and temperature gradients exist such that these
conditions may not be realized.

COUPLED MODELING APPROACH - CRITICAL ZONES FOR FAILURE

A coupled reservoir and geomechanical simulator (Settari and Walters, 2001) was used to investigate the stress
changes during typical SAGD operations. The vertical profile of stratigraphic zones, pressure gradients and stress
gradients shown in Figure 4 was used for the simulator description. The simulation model was created for a
reservoir description slightly shallower than the well used for the log analysis. To populate the model the
structure tops were used to select appropriate pressure and stress gradients to populate the model. The 2D, single
well pair, element of symmetry model included all the material from ground level (GL) to about 500 m below GL.
The model assumed the McMurray (McM) Sand (of 25 m thickness) was entirely reservoir quality sand such that
the steam chamber when fully developed in height would stop at the base of the Wabiskaw D Mudstone zone.
The MacKay River lease uses three caprock zones to evaluate the maximum operating pressure (MOP):
Wabiskaw D (Wab D) Mudstone, the Wabiskaw A (Wab A) Shale, and the Clearwater (CW) Shale. The Wab D
Mudstone is the lower part of the Wabiskaw member. Important depths for the evaluation of caprock stability are
the base of the CW (76 m), the base of the Wab (88 m) and the base of the McM (113 m). The geometry of the
model, in a cross-sectional plane perpendicular to the wells, is illustrated in Figure 7 showing the pressure and
temperature distribution after 273 days of operation. The injector and producer wells were positioned on the left
side of the model at depths of 106.5 m and 111.5 m, respectively. The geomechanical analysis of the results of 10
years of SAGD operation at a constant operating pressure of 1663 kPaa are shown in the next section. The models
were calibrated to current field measurements (pressure, temperature, surface heave, and extensometer data) and
to extensive laboratory data for the Wabiskaw D Mudstone, Clearwater Shale and McMurray Sand. The iterative
loop of calibrating the geomechanical models with additional data has been successfully employed historically at
Suncor MacKay River to increase the reliability of the modeling results.

1 6
2 7
3 8

4 9

5 10

a) (b)
Figure 7. (a) Pressure and (b) Temperature distributions after 912 days in 2D simulation model

Failure modes and stress ratios

When evaluating caprock integrity two modes of failure require evaluation: tensile and shear. In order to easily
evaluate the simulation results for these modes of failure two stress ratios are used:
10 SPE 157855

P/Smin = ratio of pressure (P) to total minimum principal stress (Smin) - tensile failure (P>Smin)
StrLev  1
   3 
 1   3  f = ratio of maximum shear stress to shear strength - shear failure (StrLev>1)
Where  1   3  f   dev  f  2c cos   2 3 sin 

(8)
1  sin  
Here c is the material cohesion,  is the material friction angle and 3 is the minimum effective stress. An
example of the simulation results for 600 days and 273 days after the beginning of SAGD operations is shown
in Figure 8a and 8b. Figure 8a shows the tensile stress ratio (P/Smin) at 600 days. The results show that a zone of
higher values (about 0.85 max) occur to the right of the steam chamber. This is due to the vertical stress
unloading adjacent to the steam chamber (so-called thermal jacking as shown in Figure 7b) and pressure leak-
off ahead of the steam chamber. No values greater than 0.9 were observed during the history of the
simulation. Some increase in P/Smin the Wab D Mudstone was observed, but nothing greater than 0.6. This
increase in P/Smin in the Wab D is due to an increase in pressure (due to vertical leak-off) and a horizontal
stretching affect (Figure 7b) similar to the thermal jacking of the vertical stress discussed earlier. The
horizontal stretching is due to the expansion of the pressurized and heated zone in the McM. The caprock
must deform to accommodate the expansion occurring below and as the caprock bends some horizontal
stretching occurs as well. This effect will be discussed later. Figure 8b shows the shear stress level after 273
days. The sand adjacent to the steam chamber very quickly goes into shear failure (the StrLev is calculated
based on the residual shear strength, when it reaches the residual shear yield surface fs) due to pressure
increase quickly moving the material into shear failure (initial deviatoric stress in the sand is relatively large).
The Wab D Mudstone shows a small amount of shear failure at its very base later in the history, but in general
the Wab D Mudstone and CW remain free of shear failure. The results indicate that based on the stress ratio
analysis the operating pressure used in the simulation is reasonably safe. However, the stress level at base of
the Wab D has a relatively large value for a region adjacent to the shoulder of the steam chamber. This is the
typical location for shear failure due to the additional shear loading and bending around the heave zone. The
results illustrate that a safe MOP analysis cannot be performed using the initial stress state alone, but must
account for the dynamic changes in stress during SAGD operations. Typically the horizontal total stresses (h,
H) decrease above the steam chamber and increase adjacent to it. Also, the total vertical stress (z) can
decrease adjacent to the steam chamber as the steam zone expands, mainly in the vertical direction, and
stretches the materials around it. The changes in horizontal stress above the steam chamber were investigated
further. A decrease in the minimum horizontal total stress effectively decreases the fracture gradient in those
locations and may increase the risk of tensile failure. To investigate this, changes in the minimum horizontal
total were plotted at the base of the Wab D Mudstone directly above the well pairs. This is shown in Figure
99. The history of the minimum horizontal total stress is plotted along with the pressure at the location and the
pressure in the sand one meter below. The results show that although the stress does decrease significantly (by
about 160 kPa or about 7%), the pressure in the block and the gridblock below are still far from reaching
fracturing pressures. Also, this effect happens fairly early in the life of the well pair, in this case within the
first year. Figure 99b shows a cross-sectional plot of the change in minimum total horizontal stress after 1
year. It shows that the maximum decrease in horizontal total stress above the well pairs actually occurs in the
sandy or silty zones (up to 30% decrease). This is because those zones were assigned McM sand properties (as
shown in Figure 2) and are stiffer than the mudstone layers. Therefore, for a given amount of stretching
(horizontal strain) a stiffer material will give a larger stress change. Figure 99a also shows that once the
pressure and temperature penetrate the Wab D Mudstone the horizontal stresses increase and the risk is gone.
This further illustrates that the magnitude of stress decrease in the caprock is dependent on a specific
distribution of pressure and temperature below. If the distribution changes the resulting stress change in the
caprock will also be different.
SPE 157855 11

(a) (b)
Figure 8. Stress Ratio plots (a) P/Smin and (b) StrLev

(a) (b)
Figure 9. Stress Changes occurring above steam chamber

A closer look at the stress changes occurring during SAGD may be taken using stress path plots. 2D stress path
plots show the history of deviatoric stress or q, q=(1-3)/2 (top of the MC), versus mean effective stress or
p, p=(1+3)/2 (center of the MC). Figure 7 shows the locations at which stress paths were plotted. Figure 9
shows the stress path results. The stress paths illustrate in greater detail the shear stress levels observed above.
They are shown separately for caprock blocks and sand blocks as the two materials were characterized by
different shear failure envelopes and parameters describing the EP constitutive models. Two gridblocks for at
the base of the Wab D are shown in Figure 9a. The point directly above the well pairs shows the horizontal
stress decrease discussed above as the q drops initially before moving up and to the left. The stress path
results show that the very base of the Wab D exceeds the residual shear strength, but does not reach the peak
shear strength. The upper part of the Wab D and CW did not show any shear failure. Most of the results for
the sand show an increase in q and decrease in p as the stress state moves toward the shear yield surface, fs.
This is due to the increase in total horizontal stress with an increase in pressure and temperature. A closer
inspection clearly shows the pressure effect first (predominant movement to the left and slightly upward)
followed by the temperature effect (movement upward and to the right) as the pressure leak-off reaches the
sand before the steam temperatures. After this there is a continual shearing of the sand (with shear dilation).
This is desired as shear dilation can cause permeability enhancements. The stress path plots for the MC
analysis previously discussed are also shown with the simulation results. They show that if the nonlinear
behavior of the material is captured with an incremental MC analysis, the average stress path can be captured.
Finally, by evaluating the stress paths from modeling results, the stress paths tested in the lab can be refined.
Each of the plots in Figure 9 also shows the lab test results (plotted as symbols) from the respective material. It
may be observed that the stress paths during SAGD operation differ considerably from the typical triaxial
stress path.
12 SPE 157855

(a) (b)
Figure 9. Stress Path Plots (a) Caprock blocks; (b) Sand blocks

CONCLUSIONS

1. The shale, mudstone and sand materials have complex stress-strain behaviour due to the overconsolidated
nature of the materials. These materials may be characterized quite well using a void ratio dependent
elasto-plastic model with three yield surfaces.
2. A calibrated initial stress state regime, as developed for Suncor’s Mackay River lease, is critical to
evaluating the potential risk of each mode of caprock failure: tensile and shear.
3. Mohr circle analysis yields valuable insight into the failure mechanisms of the SAGD process, but this
analysis cannot capture the bending effect at the steam chamber shoulder. Moreover, MC analysis can
only give a good prediction of shear failure for a permeable sand (McM Sand), not for low permeable
materials (such as the Wab D Mudstone), and may give a very conservative prediction for MOP values.
4. Tensile and shear stress ratios can be used to evaluate failure modes in simulation results, and act as an
indicator of the locations most at risk for tensile and shear failure.
5. Total horizontal stress decreases occur above the well pair in the caprock early in the life of the SAGD
operation. These stress changes are strongly dependent on the distribution of pressure and temperature
below and the stiffness of the caprock.
6. These stress changes should be considered when selecting a reasonable operating pressure.
7. Stress paths provide additional information on the stress history and can be used to guide lab testing and
identify areas at risk for shear failure.
8. It was crucial to have an iterative approach in place to calibrate the geomechanical models with field
measurements (pressure, temperature, surface heave, and extensometer data) and extensive laboratory
data for the Wabiskaw D Mudstone, Clearwater Shale and McMurray Sandstone. This approach increases
the level of confidence and reliability of the modeling results.

ACKNOWLEDGEMENT
We would like to appreciate Suncor Energy Inc. for the permission of the publication for this paper.

REFERENCES

ALBERTA INNOVATES, 2010. “Steam Assisted Gravity Drainage (SAGD)”. http://www.albertainnovates.ca /energy/investing/success-
stories/sagd.
BELL, J.S., PRICE, P.R. and MCLELLAN P.J. 2010. “In-situ Stress in the Western Canada Sedimentary Basin, Chapter 29 Geological
Atlas of the Western Canada Sedimentary Basin. 2010.
DUSSEAULT, M.B. and MORGENSTERN, N.R. 1979. “Locked sands”, Quarterly Journal of Engineering Geology, vol.12, pp.117-131.
BRUNO M.S., DUSSEAULT M.B., BALAA T.T., and BERRERA J.A., 1998. “Geomechanical Analysis of Pressure Limits for Gas
Storage Reservoirs”. Paper Number USA-328-5, presented at the North American Rock Mechanics Symposium, NARMS ’98, June 3-
5, 1998.
BRUNO, M.S. DEWOLF, G. and FOH, S., 2000. “Geomechanical Analysis and Decision Analysis for Delta Pressure Operations in Gas
Storage Reservoirs” Gas Research Institute. Paper presented at the American Gas Association Operations Conference, Denver, CO,
May 7-9, 2000.
SPE 157855 13

BUTLER, R.M., and STEPHENS D.J., 1981. “The Gravity Drainage of Steam Heated Heavy Oil to Parallel Horizontal Wells”, Journal of
Canadian Petroleum Technology, 20(2): 90-96.
CHALATURNYK, R. 2011. “Observations on SAGD Caprock Integrity”. Recovery – 2011 CSPG CSEG CWLS Convention.
COLLINS, P.M., WALTERS, D., PERKINS, T., KUHACH, J. and VEITH, E. 2011. “Effective Caprock Determination for SAGD
Projects”. SPE 149226-MS. Presented at Canadian Unconventional Resources Conference, Calgary, Alberta, Canada, November 15-
17.
COLLINS, P.M., 2007. “Geomechanical Effects on the SAGD Process”. SPE 97905. August 2007 SPE Reservoir Evaluation and
Engineering.
COUSSY, O. 1994. Mechanics of Porous Continua. John Wiley & Sons. 455pp.
DUNN, I.S., L.R. ANDERSON and F.W. KIEFER. 1980. Geotechnical Analysis. John Wiley & Sons. 414pp.
DUSSEAULT, M., BRUNO, M. and BARRERA, J., 2001. “Casing Shear: Causes, Cases, Cures”. SPE 72060. June 2001, SPE Drilling
and Completion.
HAWKES, C.D., MCLELLAN, P.J. and BACHU S., 2005. “Geomechanical Factors Affecting Geological Storage of CO2 in Depleted Oil
and Gas Reservoirs”. October 2005, Volume 44, No. 10. Journal of Canadian Petroleum Technology.
KHAN S., HAN H., ANSARI S., KHOSRAVI N., 2010. “An Integrated Geomechanics Workflow for Caprock-Integrity Analysis of a
Potential Carbon Storage”. SPE 139477. presented at the SPE International Conference on CO2 Capture, Storage, and Utilization held
in New Orleans, Louisiana, USA, 10-12 November 2010.
MCLELLAN, P. and GILLEN, K. 2000. “Assessing Caprock Integrity for Steam-Assisted Gravity-Drainage Projects in Heavy-Oil
Reservoirs” (GEM Presentation) prepared for presentation at the 2000 SPE/Petroleum Society of CIM International Conference on
Horizontal Well Technology held in Calgary, Alberta, Canada, 6-8 November, 2000. SPE 65521.
O’ROURKE, J.C., BEGLEY A.G., BOYLE H.A., YEE C.T., CHAMBERS J.I., LUHNING R.W., 1997. “UTF project status update May
1997”. Paper 98-08, first presented at the 48th Annual Technical Meeting of the Petroleum Society in Calgary, Alberta, Canada, June
8-11, 1997.
SETTARI, A. and WALTERS, D.A. 2001. “Advances in Coupled Geomechanical and Reservoir Modeling With Applications to Reservoir
Compaction”, SPE Journal, Vol. 6, No. 3, Sept. 2001, pp. 334-342.
SMITH, R.J., 1997. Geomechanical Effects of Cyclic steam Stimuation on Casing Integrity. University of Calgary, Master’s Thesis, 1997.
WAN, R.G. and GUO P.J., 1998. “A Simple Constitutive Model for Granular Soils: Modified Stress-Dilatancy Approach”. Computers and
Geotechnics, Vol. 22, No. 2, pp. 109-133.
WONG, R.C.K. 2004 Effect of Sample Disturbance Induced by Gas Exsolution on Geotechnical and Hydraulic Properties Measurements
in Oil Sands PAPER 2004-071, Petroleum Society of CIM, CIPC, Calgary, Alberta, Canada, June 8-10, 13pp.

Nomenclature
I1 : the first invariant of the stress tensor,
J2 : the second invariant of the deviatoric stress tensor
H, h : Maximum and minimum stress
H, h : strain in Maximum and minimum stress
E: Young’s modulus
 Poisson’s ratio
K:bulk modulus,
T:temperature,
p: pressure,
aL: the drained linear thermal expansion coefficient
 : Biot’s poroelastic constant
Ks : grain modulus
StrLev  1
   3 
 1   3  f = ratio of maximum shear stress to shear strength - shear failure (StrLev>1)
C: the material cohesion,
 the material friction angle
3 : the minimum effective stress
q : the deviatoric stress
p: the mean effective stress
fs : Shear yield surface
fc : Compaction yield surface
ft : Tension cut-off yield surface

You might also like