You are on page 1of 13

Geomechanical-Data Acquisition,

Monitoring, and Applications in SAGD


F. Gu, M. Chan, and R. Fryk*, Statoil Canada

Summary Several types of field-monitoring methods have been applied in


Steam-assisted gravity drainage (SAGD) has proven to be a com- SAGD projects to understand geomechanical responses and steam-
mercially viable method to extract bitumen from oil-sands res- chamber dynamics. Piezometers and thermocouples are commonly
ervoirs in western Canada. To understand the influence of steam used to monitor pressures and temperatures at observation wells
injection on reservoir and surrounding rocks and potential impacts and SAGD injectors and producers (Chalaturnyk 1996). Downhole
of surface deformation on the environment, various types of in- extensometers and inclinometers (tiltmeters) were also installed
strumentation and 4D-seismic surveys have been applied in SAGD at the underground test facility to measure the downhole and sur-
projects. The effect of geomechanics on SAGD has been well doc- face deformations (Chalaturnyk 1996). Tiltmeters were applied to
umented. Collecting essential geomechanical data, properly in- monitor surface deformations, and triaxial geophones (that have
terpreting them, and incorporating them into numerical models vibration-sensing elements in three directions) were used to de-
are necessary to ensure meaningful history matching and under- tect microseismic events during SAGD operations (Maxwell et al.
standing of reservoir performances. 2007). Interferometric synthetic aperture radar (InSAR) has been
This paper outlines geomechanical-data acquisition and field- applied to measure surface deformations in Suncors Firebag pro-
monitoring methods from a reservoir-engineering perspective, ject (Suncor Energy 2008), Shells Hilda Lake project (Shell Canada
and the applications of geomechanics in SAGD analyses. Min- Energy 2008), and Totals Joslyn project (Total E&P Canada 2007).
imal-data-acquisition programs are suggested to collect the neces- 4D-seismic surveys have been used to understand steam conform-
sary geomechanical data for different analysis purposes in SAGD ance during SAGD in Suncors Firebag project (Suncor Energy
projects. Primary instrumentation is briefly overviewed, and rec- 2008), ConocoPhillips Surmont project (Forgues et al. 2006), and
ommendations for instrumentation selection are provided. Using EnCanas Christina Lake project (Zhang et al. 2007).
generic Canadian-oil-sands reservoir and rock properties, the sub- Because geomechanics plays an important role in the SAGD
surface and surface changes and deformations are simulated, in- process, collecting appropriate geomechanical data is crucial for
cluding permeability changes, reservoir movements, and strains and incorporating geomechanical effect into various studies (reser-
surface uplifts. Simulations were conducted using a commercially voir modelling, understanding IHS response to steam, improving
available simulator. The limitations of the simulator are also noted. the accuracy of geostatistic models, ensuring meaningful inter-
The method to couple the results of geostatistics modelling, reser- pretation of field observations, understanding the deformations
voir simulation, and geomechanics in SAGD simulation and to link of oil-sands reservoirs and surrounding rocks, and optimizing the
them with a 4D-seismic survey in history matching is provided. SAGD process). This paper discusses the essential geomechanical-
data-acquisition program necessary for these studies, through an
Introduction example demonstrates coupled reservoir and geomechanical simu-
SAGD has proven to be a commercially viable method to extract lation, and proposes a method and work flow to match all meas-
bitumen from oil-sands reservoirs in western Canada (Butler ured data, including 4D-seismic surveys, with coupled reservoir
1998; Donnelly 2000; Butler 2001). In the SAGD process, high- and geomechanical simulation.
temperature steam is injected into reservoirs at pressures close
to or higher than initial reservoir pressures. Steam pressures and Geomechanical-Data-Acquisition Program
temperatures can exceed 5 MPa and 264C, respectively. This On the basis of the purpose of applications, geomechanics analyses
can cause significant geomechanical effect on the reservoir and can be grouped into four broad categories.
surrounding rocks, and may also affect surface facilities and the 1.Ensuring safe SAGD operationanalyzing caprock integrity,
surrounding environment. casing integrity, fault reactivation, surface uplift, wellbore stability,
The influence of geomechanics on bitumen recovery from oil and sand production.
sands with SAGD has been investigated by a number of researchers 2.Optimizing bitumen recovery in oil-sands reservoirsopti-
(Chalaturnyk 1996; Collins 2005; Li 2006), and some of their find- mizing startup, ramp-up, and operating conditions.
ings will be referred to in this paper. In addition to ensuring safe 3.Making integrated history matching and model validation
SAGD operations, understanding geomechanics will be necessary matching injection/production data (fluid rates, cumulative fluid
to investigate the potential breaking of interbedded mudstone and volumes, temperatures, pressures, and steam qualities), observa-
inclined heterolithic stratification (IHS), a commonly found geo- tion-well and surface-monitored data (temperatures, pressures, de-
logical facies that is often a baffle to the upward growth of SAGD formation, and microseismic events), and 4D-seismic-survey data.
steam chambers. As demonstrated by Li (2006), if IHS can be 4.Improving SAGD efficiency in the reservoirs with vertical-flow
broken in the reservoir where IHS is prevalent, bitumen recovery barriers and thief zonesusing geomechanics to investigate if and
would increase significantly. how injection-induced stress alterations can be used to break flow bar-
riers and thus increase oil recovery, and to minimize the influence of
top water, top gas, and basal water by operational optimization.
Copyright 2011 Society of Petroleum Engineers. This paper (2009-177) was accepted
for presentation at the 10th Canadian International Petroleum Conference (the 60th An-
There is no standardized fit between data-acquisition programs
nual Technical Meeting of the Petroleum Society), Calgary, 1618 June 2009, and re- and geomechanical analyses in the oil-sand industry. Table 1 shows
vised for publication. Original manuscript received for review 2 February 2009. Revised
paper received for review 20 December 2010. Paper peer approved 4 January 2011 as *Currently with Velvet Energy Ltd.
SPE paper 145402.

June 2011 9
TABLE 1DATA FOR GEOMECHANICAL ANALYSES

Recommended Data Acquisition Program


Purpose of Analysis Essential Rock Mechanical Data Data From Laboratory or Field Testing
Ensuring Caprock integrity MP, PS and RS (CR ) MP, MP, IS, PP, PS and RS (CR )
safe SAGD IS,
operation Surface uplift IS, CT (CR adjacent to OS, OS, UB)
PP
PP,
Wellbore instability and
PS
Sand production CT Strength degradation (OS) and Strength degradation (OS)
(OB,
Fault reactivation PS, RS and k/k 0 (fault) RS PS, RS and k/k 0 (fault)
OS,
(OS)
Casing integrity UB); PS (CM)
Optimization of bitumen recovery k/k 0 (OS) MP, PS, RS, IS and PP (CR );
PS k/k 0 (OS)
and
Integrated history matching and RS k/k 0 (OS); k/k 0 (OS);
model validation MP (different rock facies in OS) MP (different rock facies in OS)
(OS)
Improving Breaking IHS PS, RS and k/k 0 (IHS) PS, RS and k/k 0 (IHS)
SAGD
efficiency Top water/gas issue MP, PS, PP and k/k 0 (rock in top MP, PS, PP and k/k 0 (rock in top
water/gas zone) water/gas zone)

Basal water issue MP, PS, PP and k/k 0 (rock in MP, PS, PP and k/k 0 (rock in basal
basal water zone) water zone)

Abbreviation of Data Type Abbreviation of Material Type in Brackets


CT = coefficient of thermal expansion CM = cement
MP = mechanical properties (Youngs modulus and Poissons ratio) CR = caprock
IS = in-situ vertical, horizontal minimum and horizontal maximum stresses IHS = inclined heterolithic stratification
k/k 0 = permeability ratio of deformed rock to its initial state OB = overburden
PS = peak strength (cohesion, friction angle and dilation angle) OS = oil sand
RS = residual strength (cohesion, friction angle and dilation angle) UB = underburden
PP = pore pressure

the primary data that can be collected from laboratory or field mechanical properties measured only represent the rock properties
testing for each analysis in the previously described four groups. under the specified stress states and temperature conditions.
Mechanical properties; in-situ stresses; pore pressures; the coeffi- Dynamic Moduli. To compute dynamic moduli, bulk density
cients of thermal expansion of overburdens, oil sands, and under- and compressional- and shear-wave velocities of rock are required.
burdens; and peak and residual strengths of oil sands are commonly Bulk density can be estimated from the density log or core mea-
required for all analyses. For a specific analysis, additional data surement. Dynamic moduli are calculated with the following equa-
may also be needed. For example, to analyze the risk of caprock tions (Fjr et al. 1992):
failure caused by SAGD operations, the peak strength and residual
strength of caprock are also needed. To analyze IHS breaking, the G = u ( vs )2 ..........................................................................(1)
peak strength, residual strength, and permeability change with in-
situ conditions of IHS are also necessary. It is desirable to acquire K u = u ( vs )2 [( vps )2 43 ] .........................................................(2)
v

mechanical properties, peak strength, and residual strength for oil


sands through laboratory testing, and collecting in-situ stress and v
3( vps )2 4
pressure data by field testing. Eu = u ( vs ) 2 .........................................................(3)
v
( vps )2 1
Mechanical Properties and Strengths. Mechanical properties v
can be obtained from log analyses and laboratory tests on cores. ( vps )2 2 .......................................................................(4)
= v
The dynamic moduli can be derived from logs, and static moduli 2[( vps ) 2 1]
from laboratory tests. The former are normally larger than the lat-
ter. Dynamic moduli may be continuous if log data are available, It should be noted that the velocity of shear wave may be aniso-
while static moduli are discontinuous because it is impossible to tropic because of the anisotropy of in-situ rocks and stresses
test the entire profile. Dynamic moduli can also be derived from (Chardac et al. 2005).
laboratory acoustic measurements with cores. However, it is static Static Moduli and Strengths. Static moduli are obtained from
moduli that are used in geomechanical analyses. Dynamic moduli triaxial tests, and the testing process has been more or less stan-
can be converted to static moduli if the correlation between them dardized for rock mechanics (Fjr et al. 1992). In triaxial tests,
is established. specimens with a diameter of 1 to 1.5 in. (1 in.=25.4 mm) and a
Note that mechanical properties of rocks are not constant and length of twice the diameter are used. Confining stresses and in-
may change with the state of in-situ stresses and temperatures. With ternal pore pressures are applied to cylindrical specimens, then the
triaxial tests, Wong et al. (1993) showed that the mechanical prop- axial stresses are increased until the failure or yielding of spec-
erties of oil sands change significantly because of the changes of imens occurs. To quantify a failure envelope (e.g., a Mohr-Cou-
stresses and temperatures. By measuring the velocities of shear lomb failure envelope), at least three triaxial tests under different
waves and compressional waves and calculating dynamic bulk and confining-stress levels are necessary. From triaxial tests, Youngs
shear moduli, Zou (2005) demonstrated that the dynamic moduli of moduli, Poissons ratios, and peak and residual strengths (cohesion
oil sands also vary with stresses and temperatures. This means that and friction angle) can be obtained.

10 Journal of Canadian Petroleum Technology


2r
= 12 (a c )
aa
a c (deviatoric stress)
Mohr-Coulomb failure envelope
cc cc f =c + f tan( )
L
peak strength
slip
planes
planes f
aa
residual (ultimate) strength
Mohr circle

r c
E =

L
1
elastic part of curve
a c f a
Fig. 1Typical stress/strain curve from the triaxial test. Fig. 2Determination of failure envelope.

Fig. 1 illustrates a typical stress/strain curve from a triaxial test in-situ stresses. Caliper data may be used to infer the direction of
and the determination of Youngs modulus and Poissons ratio from stresses (Chardac et al. 2005).
the stress/strain curve.
Fig. 2 shows the determination of a failure envelope from mul- Field-Monitoring Methods
tiple triaxial tests. A detailed review of field-monitoring methods is outside the scope
It is worth repeating that the moduli of oil sands and shale vary of this paper and can be found in the work of Chalaturnyk (1996),
with in-situ stresses and temperatures. With sufficient triaxial-test Dusseault (2007), Lowe and Donovan (1989), and Collins (1994).
data, the relation of static moduli with stresses and temperatures Monitoring temperatures and pressures and using 4D-seismic sur-
can also be established. veys (3D+time) are not new in the oil-sands industry, thus will not
Samples used for geomechanical laboratory testing require be discussed here. The following subsections summarize the main
a special core handling and preservation process. Collins (2005) categories of field-monitoring methdologies together with some
detailed the potential disturbance during coring, geomechanical general guidelines for selecting the fit for purpose monitoring and
coring, preparation, and other concerns for oil-sands cores. Gen- fiel data acquisition.
erally, oil-sands cores need to be frozen from well sites to before
tests, while mudstone, shale, and high-IHS-content cores need to be Surface Heave Monitoring Methods and Principles. The major
preserved in an unfrozen state (never lower than 5C). The former monitoring methods used in SAGD projects are geophones, tiltme-
is to decrease core disturbance caused by gas release, while the ters, InSAR, and the global positioning system (GPS). Geophones
latter is to decrease core disturbance caused by icing expansion. can only detect seismic events and frequency occurring in the SAGD
Correlation Between Dynamic and Static Moduli. The correla- process because of the shear failure of rocks and casings. The other
tion between dynamic moduli and static moduli can be established three methods can measure or use their data to infer displacements.
by laboratory tests through measuring acoustic velocities during Tiltmeters continuously measure tilts along both horizontal
triaxial tests. Lacy (1997) gave some examples of such correla- planes (x-axis and y-axis), using an array of tiltmeter instruments.
tions. It is worth noting that there is no universal correlation be- Surface deformations are integrated from tilt data.
tween static moduli and dynamic moduli, and correlations also vary InSAR is a remote sensing technique that uses satellite radar
with rock types. images to detect surface deformations. Satellites project micro-
wave beams toward the Earths surface and record them when they
Initial In-Situ Stresses and Pore Pressures. Initial in-situ stress- bounce back. The signal intensity of the bounced waves can be used
es and pore pressures are necessary to build a geomechanical to identify material (e.g., oil vs. water) on the Earths surface, and
model. Pore pressures can usually be measured with formation the phase difference between two images of the same area captured
testers that are widely applied in the petroleum industry. However, at two different times is used to estimate surface/object displace-
measuring pressures in oil sands and caprock formations with for- ments. Generally, InSAR does not require surface installations. But
mation testers would be a challenge because the former have little for poor ground conditions (wet snow and dense vegetation), re-
fluid mobility, while the latter have low permeability. Long-term flectors are needed. If installing reflectors is impossible because
measurements with piezometers can provide more-reliable pres- of poor ground conditions (ponds and muskegs), the data quality
sure data. would be degraded. In addition, several surveys need to be car-
Minimum in-situ stress can be measured in the field with ried out before production commences to provide the references for
microfrac tests (Proskin et al. 1990) or minifracture tests (Kular et comparing the displacements during production.
al. 1988). If the density log is available from the formation of in- GPS directly measures surface deformations using a network
terest to the surface, the integral of such density log can provide an of surface receivers and a constellation of satellites. The receivers
estimate of the vertical stress. If the vertical stress is the minimum are fixed high-precision GPS stations that record and interpret the
stress, the integral of the density log provides approximate min- radio signals broadcast by the satellites. Precision of subcentimetre
imum in-situ stress. scale in horizontal positioning, and approximately 20 mm in z,
It is challenging to directly measure in-situ stresses other than is feasible, depending on the spacing and density of the differen-
minimum stress. Sayers et al. (2007) proposed a method to estimate tial GPS sites.
the magnitude and orientation of in-situ stresses. In this method, To our knowledge, two combinations of surface-deformation
the variation of elastic wave velocities was expected to be caused monitoring have been applied in SAGD projects (i.e., InSAR com-
by the damage or failure of rock and was characterized with sonic bined with GPS and tiltmeters combined with GPS). They were
logging. In-situ stresses were estimated by matching the velocities used in Suncors Firebag project (Suncor Energy 2008, 2006) and
with the calculated velocities. Stress databases of past tests (Bell Totals Joslyn project (Total E&P Canada 2007). The results from
et al. 1994; Collins 2002) can provide a first-order estimation of InSAR and GPS are in agreement (Suncor Energy 2008).

June 2011 11
TABLE 2COMPARISON OF FIELD-MONITORING METHODS
Monitoring Monitoring Time Place of Resolution
Parameter Method Coverage Interval Installation Constraint (Dusseault 2007) Cost
8
Deformation Tiltmeter Points Continuous Surface or No 10 radians Expensive
downhole startup
InSAR Extensive area Days Surface Partial 25 mm Least
months expensive
GPS Large area Continuous Surface Partial 110 mm Expensive
Microseismic Geophone Local
event

Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1


9,268
Fully Solve fluid flow and 8,341
Any Process
Coupled deformation as a system 7,414
6,488
implicitly explicitly 5,561
4,634
Time i Time i Time i+1
Sequential 3,707
Coupled S
Soolv
Solvee fluid flow
ve
Any Process 2,780
Sol
Sol ve
Solve d
ve deformation
dee 1,854
927
S olve
Solve
lve fluid flow
flow 0
One Way Any Process
Coupled Solve
Solv
So lve deformation
e de
d efo
forrma
ormat
atio
tio Fig. 4Initial horizontal permeability in Plane i-k.

Fig. 3Methods of coupled reservoir and geomechanical simula- (Settari and Walters 2001; Rutqvist et al. 2002; Tran et al. 2002).
tion. An illustration of these methods is shown in Fig. 3.
In fully coupled simulation, flow variables (e.g., pressure, tem-
Selection of Monitoring Methods. The selection of monitoring perature) and geomechanical variables (e.g., stress, deformation in-
methods depends on various factors, including monitoring purpose, formation) are simultaneously calculated by solving a system of
area coverage, cost, resolution, weather constraint, and field acces- equations governed by and established on three basic principles:
sibility. Table 2 is a comparison of these characteristics and relative mass, momentum, and energy conservations.
costs. Of these factors, the dominant one is probably monitoring In sequentially coupled simulation, flow variables and
purpose. As listed in Table 1, the applications of geomechanics are geomechanical variables are sequentially solved from a reservoir
categorized into four types. More generally, they can be grouped simulator and a geomechanical simulator and the coupling param-
into two [i.e., preventing hazards (ensuring safe operations) and eters are iterated between two simulators. If coupling parameters
understanding and using geomechanical benefits]. The monitoring are iterated within each timestep, the simulation is referred to as an
purposes may vary for different reservoirs and operation conditions. implicitly sequential coupled simulation (Rutqvist et al. 2002) or
For shallow oil-sands reservoirs (e.g., depths less than 200 m), iterative coupled simulation (Tran et al. 2002). If coupling param-
preventing hazards should be the first priority; thus, continuous and eters are assumed to be constant during each timestep and evalu-
accurate monitoring is essential. For this purpose, a tiltmeter com- ation and iteration are executed at the end of each timestep, the
bined with GPS stations is recommended. simulation is called an explicitly sequential coupled simulation.
For deeper oil-sands reservoirs, if the expected surface uplifts Decoupled simulation is also called uncoupled simulation or
are on the order of centimetres per year, InSAR may be appropriate one-way coupled simulation. The changes of fluid pressures and
because it can provide a reasonable resolution and an overall lower temperatures influence stresses and strains, but the changes of
cost. If there is a need for continuous monitoring in specific areas, stresses and strains are assumed not to affect fluid pressures and
combining InSAR with some GPS stations may be a good solution. temperatures. Thus, fluid-flow equations can be solved indepen-
If the expected uplifts on the surface are on the order of millimetres dently and resulting pore pressures inserted into geomechanical
per year, a tiltmeter combined with GPS stations may be an appro- simulation as a time-independent parameter.
priate choice. According to simulator availability and computer capacities, se-
quential coupled simulation is probably the best solution for the
Coupled Reservoir and Geomechanical Simulation SAGD recovery process at present. This method is applied in the
An example of coupled reservoir and geomechanical simulation for simulation of this paper.
a SAGD pad with six well pairs is illustrated with a widely applied
simulator (CMG 2008). The simulators features for coupled simu- Model Setup for Coupled Simulation. Reservoir, Well, and
lation are summarized in Appendix A. On the basis of the learning Operation Conditions. The reservoir properties of the SAGD pad
from these simulations, the suggestions to improve the simulators were obtained from geostatistic modelling. For demonstration pur-
geomechanical capabilities for oil-sands applications are provided poses, a 2D slice model cut from the 3D geomodel of the pad was
in Appendix B. used. The grid counts are 460 (i)1 (j)57 (k, vertical) and the cell
sizes are 2 m (i)800 m (j)1 m (k, vertical). The horizontal wells
Method of Coupled Simulation. Coupled reservoir and geome- stretch in j-direction and perpendicular to Plane i-k. The top depth
chanical simulation is a numerical method often deployed to better of McMurray oil sands is approximately 390 m. The initial reser-
understand the influence of geomechanics on oil sands, surround- voir temperature and pressure are 12C and 2000 kPa, respectively.
ing rocks, surface facilities, and the environment during the SAGD The initial reservoir properties and saturations are shown in Figs. 4
recovery process. Coupled simulation can be divided into fully through 7. The initial vertical permeability is assumed to be 50%
coupled, sequentially coupled, and decoupled simulation methods of the initial horizontal permeability.

12 Journal of Canadian Petroleum Technology


Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1 0.40 Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1 1.00
0.37 0.90
0.32 0.80
0.28 0.70
0.24 0.60
0.20 0.50
0.16
0.40
0.12
0.30
0.08
0.20
0.04
0.10
0.00
0.00
Fig. 5Initial porosity in Plane i-k.
Fig. 6Initial oil saturation in Plane i-k.
Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1 1.00
By distributing geomechanical properties and stresses to the
0.90
cells on the basis of their geological layers, the geomechanical
0.80 model is obtained.
0.70 The geomechanical properties for the example case are shown
0.60 in Table 3. The gradients of the effective stresses are simply esti-
0.50
mated by assuming Biots coefficient to be 1.0. The initial effective
normal stresses are indicated in Figs. 8 through 10, and the shear
0.40
stresses are assumed to be zero. In the McMurray reservoir, the
0.30 maximum stress is along the horizontal wellbores, while the min-
0.20 imum stress is normal to the wellbores.
0.10 As shown in Fig. 11, the boundary conditions for the geomechan-
0.00 ical model are rollers at the bottom and on the right, left, front, and
back. These boundary conditions represent the symmetric situation
Fig. 7Initial water saturation in Plane i-k. in which several pads in a big pool are in SAGD operation and the
study pad is located in the middle of the pool. The Mohr-Coulomb
In Figs. 4 through 7, from right to left, six SAGD well pairs are failure criterion is applied. The empirical model by Li (2006) is
sequentially labelled as Inj1/Prod1 through Inj6/Prod6 (injector/ used for permeability coupling with Cn1=10.33 and *GCOUP-
producer). The well length is 800 m, the well-pair spacing is 100 m, LING 2 (CMG 2008) for porosity coupling.
and the injector/producer distance is 5 m.
The maximum injection bottomhole pressure is 3500 kPa (Tsat Results of Coupled Simulation. With the previously described
=242.6C), the maximum steam-injection rate is 800 m3/d (water reservoir and geomechanical models, and initial conditions, three
equivalent), and the steam quality is 95%. For the producers, the scenarios of coupled simulation were simulated. The first scenario
minimum production bottomhole pressure is 1000 kPa, the max- assumes all well pairs are in operation. The second scenario as-
imum live steam-production rate is 5 m3/d (water equivalent), and sumes only four middle well pairs (Inj2/Prod2 through Inj5/Prod5)
the maximum liquid-production rate is 1000 m3/d. are in operation and those remaining are shut in. The last scenario
Building Coupled Simulation Models. The following are key assumes only two middle well pairs (Inj3/Prod3 and Inj4/Prod4) are
steps for building the simulation models for this study. in operation, with those remaining shut in. The analysis will focus
Collect geomechanical properties, in-situ stresses, and pore on Well Pairs Inj3/Prod3 and Inj4/Prod4 because they have produc-
pressures for overburdens, oil-sands reservoir, and underburdens. tion data from all scenarios. The results from conventional simula-
Note that the simulator accepts only effective stresses; thus, they tion without geomechanical effect are also plotted for comparison
need to be calculated from stress and pressure gradients. purposes. In the legends of Figs. 12 and 13, coupling means the
Build a coupled simulation grid system by adding overburden results with geomechanical effect, while no coupling represents
and underburden layers and new grids to the reservoir cut from the the results without geomechanical effect.
geomodel because it does not contain sufficient overburden and Production Profiles. The influence of geomechanics and the
underburden layers needed for coupled simulation. To decrease the number of operative well pairs on production is shown in Figs. 12
cell number of added layers and increase simulation accuracy, vari- and 13. Fig. 12 indicates that the influence of geomechanics on
able grid sizes are used (i.e., using smaller grid sizes in the layers oil rates and cumulative oil production is evident, but the influ-
near the reservoir, while using larger grid sizes in the layers distant ence of the number of operative well pairs is small. Fig. 13 shows
from the reservoir). In the i- and j-directions, the cell sizes are the that because of geomechanical effect, the oil recovery factor (RF)
same as those generated from geostatistic modelling. increases by 2%-5%, compared with that without geomechan-
By transferring the properties of the reservoir from the geo- ical effect, but influence of the number of operative well pairs is
model to the coupled simulation grid system, the reservoir model small. The influence of geomechanics and the number of operative
is established. Because the new cells for the added over- and under- well pairs on the cumulative steam/oil ratio (CSOR) is negligible.
burden layers have no property data or saturations, those data need Overall, the fewer the operative well pairs, the more pronounced
to be added. It is crucial to ensure that the number of property the geomechanical effect.
values matches the number of cells for each layer. Set the porosity The influence of geomechanics is not significant in this study.
of the new over- and underburdens to zero to allow heat to flow The following four aspects may be the main reasons. First, the in-
through, but no fluid to flow. Permeability and saturations in the situ effective stresses in the reservoir are relatively high, approxi-
new over- and underburdens can be set to zero if desired. mately 6-8 MPa, while the injection pressure is not high at 3.5
Calculate initial effective normal stresses and shear stresses MPa. Shear failure was observed, but was not prevalent. Thus, the
for each layer in k-direction (the vertical direction) on the basis of geomechanical effect is less significant. Second, well pairs Inj3/
the coupled simulation grid system. If the cells in each layer have Prod3 and Inj4/Prod4 are located in relatively high-permeability
various depths or initial stresses, initial stresses should be compiled clean sand. This means there are no fluid-flow barriers, thus the
and distributed to each cell rather than to each layer. geomechanical effect is less important. Third, the permeability

June 2011 13
TABLE 3GEOMECHANICAL PROPERTIES FOR COUPLED SIMULATION OF SAGD

Grand Beaverhill
*
Parameters Quaternary Viking Rapids Clearwater Wabiskaw McMurray Lake Group

Domain Lithology Glacial Till Shale Shale Shale Mudstone Sand Limestone
/Sandstone /Sandstone
Fluid Type Water Water Water Water Water Oil + Water Water
Formation/Group Top Depth 0 100 205 292 379 390 445
(m, Ground Level)
Thickness (m) 100 104.5 87.3 86.9 11.3 55.3 100
Static Youngs Modulus, E 1.00E+05 1.60E+05 4.50E+05 1.35E+06 1.19E+06 7.20E+05 1.65E+07
(kPa)
Static Poissons Ratio, 0.28 0.35 0.35 0.32 0.32 0.30 0.30
Coefficient of Linear 2.0E06 1.6E06 1.6E06 1.6 06
E
5.0E06 2.0E05 3.0E06
1
Thermal Expansion, (C )
Peak Friction Angle, p () 30 35 38
Peak Cohesion, cp (kPa) 300 300 100

Peak Dilation Angle, p () 15 15 25


In Situ Stresses and Formation Pressures
Pore Pressure Gradient, Pr /z 9.6 8.5 7.8 7.5 4.7 4.7 7.3
(kPa/m)
Horizontal Stress Gradient, 21.0 21.0 21.0 20.5 20.0 19.0 21.0
h/z (kPa/m)
Horizontal Stress Gradient, 23.0 23.0 23.0 22.5 22.5 23.0 23.0
H/z (kPa/m)
Vertical Stress Gradient, 20.0 20.0 20.0 20.4 21.0 21.0 22.0
v/z
(kPa/m)
Horizontal Effective Stress 9.6 14.5 14.2 13.0 15.3 14.3 13.7
Gradient, h'/z (kPa/m)
Horizontal Effective Stress 9.6 14.5 14.2 15.0 17.8 18.3 15.7
Gradient, H'/z (kPa/m)
Vertical Effective Stress 9.4 10.7 12.2 12.9 16.3 16.3 14.7
Gradient, v'/z (kPa/m)
* The McMurray is an oil-sands formation in Athabasca area

Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1 Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1
7,034 8,153
6,581 7,361
6,128
5,675 6,570
5,222 5,779
4,769
4,316 4,988
3,863 4,196
3,410
2,958 3,405
2,505 2,614
2,052
1,823
1,599
1,146 1,031
693 240
240
Fig. 9Initial effective stresses in j-direction.
Fig. 8Initial effective normal stresses in i-direction.
model applied in the simulation calculated and used only one per- Overall, with the increase of operative well pairs, the vertical dis-
meability multiplier for all three directions. The anisotropic dilation placements increase.
of oil sands is not considered. Thus, the geomechanical influence is It is noted that in this study the influence of the number of oper-
underestimated. In addition, because of the limitation of the simu- ative well pairs on production profiles is less significant than its
lator, shear failure is not properly simulated. This is detailed in influence on the vertical displacement. This is because the three
Appendix B. scenarios have a similar increase of volumetric strain in the
Deformation and Shear Failure. The maximum vertical dis- drainage areas of the middle two well pairs, resulting in a similar
placements for the entire simulation period are shown in Fig. 14. permeability enhancement, thus similar production profiles. How-
The maximum surface uplift when all well pairs are operating is ever, the horizontal expansion is constrained for the scenario with
0.379 m; when only two middle well pairs are in operation, it is six operative well pairs because other well pairs are in operation
0.235 m. The former is approximately 50% more than the latter. and have similar volumetric expansion. Thus, the most volume ex-

14 Journal of Canadian Petroleum Technology


Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1
7,547 Pad Area
Area
J
6,816 I
6,085 K
5,354 Overburden
Overburden
Pad Area
Area
4,622
Oil Sands
Oil Sands
3,891
Underburden
Underburden
3,160 J
(a) Geomechanical Body I
2,429
1,697
(b) Plan View of
966
I Geomechanical Body
235 K under Constraints
Overburden
Overburden
Fig. 10Initial effective stresses in k-direction.
(c) Cross Section of
pands vertically. In comparison, the volume expands not only ver- Oil Sands
Oil Sands Geomechanical Body
tically but also horizontally for the scenario with only two middle Underburden
Underburden under Constraints
well pairs operative, because other well pairs are shut in. Therefore,
the overall vertical movement of the former is much larger than that Fig. 11Geomechanical model of coupled simulation.
of the latter.
The simulation results indicate that the shear failure of oil sands Work Flow. The work flow for proposed integrated history matching
was initiated after approximately 6 years of operation and con- is illustrated in Fig. 16, and major steps are listed as the following.
tinued until the end of simulation in various areas. Fig. 15 shows Build a reservoir model that is based on reservoir-property
two snapshots of failure state at 7.2 and 7.4 years. In the plots, yield output from geomodelling, and build a geomechanical model with
state=1.0 means shear failure state. geomechanical data and the locations of facies from geomodelling.
Implement coupled reservoir and geomechanical simulation
Integrated History Matching using measured pressures (and temperatures) at injectors and pro-
Concept. The integrated history matching of the SAGD process ducers as input data. From reservoir simulation, the temperatures
proposed here is to match all measured historical data with cou- and pressures at observation wells and the rates and volumes of
pled reservoir and geomechanical simulation. The historical data oil, gas, and water at injectors and producers are obtained and used
include injection/production data (fluid rates, cumulative fluid to match the measured values. From geomechanical simulation,
volumes, temperatures, pressures, and steam qualities), observa- strains, displacements or uplifts, and microseismic events are ob-
tion-well and surface-monitored data (temperatures, pressures, de- tained and applied to match the measured historical data. Mean-
formation, and microseismic events), and 4D-seismic-survey data. while, synthetic seismic data are generated and used to match
Coupled simulation models are built by integrating geostatistic, 4D-seismic data.
reservoir, and geomechanical data. The geomechanical model, es- If the matches are good enough, simulation can be stopped.
pecially, should include the mechanical properties of different fa- If the matches are not satisfactory, a new realization from geo-
cies in oil sands identified in geostatistic analysis and modelling. modelling may need to be generated or the geomodel used needs
To do so, the properties of each facies need to be acquired through to be modified. Geomechanical data may also need to be modi-
log analyses and/or laboratory tests. fied. Thus, the reservoir and/or geomechanical simulation models

400 800,000
Inj3/Prod3: oil rate (coupling) Inj3/Prod3: cum. oil (coupling)
Inj3/Prod3: oil rate (no coupling) Inj3/Prod3: cum. oil (no coupling)
Inj4/Prod4: oil rate (coupling) Inj4/Prod4: cum. oil (coupling)
350 Inj4/Prod4: oil rate (no coupling) Inj4/Prod4: cum. oil (no coupling) 700,000

300 600,000
Cumulative Oil SC, m3

250 500,000
Oil Rate SC, m3/d

200 400,000

150 300,000

100 200,000

50 100,000

0 0
0 365 730 1,095 1,460 1,825 2,190 2,555 2,920 3,285 3,650
(a) Time, day

Fig. 12(a) Comparison of oil rate and cumulative oil production of Well Pairs Inj3/Prod3 and Inj4/Prod4. All well pairs are operative.

June 2011 15
400 800,000
Inj3/Prod3: oil rate (coupling) Inj3/Prod3: cum. oil (coupling)
Inj3/Prod3: oil rate (no coupling) Inj3/Prod3: cum. oil (no coupling)
Inj4/Prod4: oil rate (coupling) Inj4/Prod4: cum. oil (coupling)
350 Inj4/Prod4: oil rate (no coupling) Inj4/Prod4: cum. oil (no coupling) 700,000

300 600,000

Cumulative Oil SC, m3


250 500,000
Oil Rate SC, m3/d

200 400,000

150 300,000

100 200,000

50 100,000

0 0
0 365 730 1,095 1,460 1,825 2,190 2,555 2,920 3,285 3,650
(b) Time, day
Fig. 12(b) Comparison of oil rate and cumulative oil production of Well Pairs Inj3/Prod3 and Inj4/Prod4. Four middle well pairs
are operative.

400 800,000
Inj3/Prod3: oil rate (coupling) Inj3/Prod3: cum. oil (coupling)
Inj3/Prod3: oil rate (no coupling) Inj3/Prod3: cum. oil (no coupling)
Inj4/Prod4: oil rate (coupling) Inj4/Prod4: cum. oil (coupling)
350 Inj4/Prod4: oil rate (no coupling) Inj4/Prod4: cum. oil (no coupling) 700,000

300 600,000

Cumulative Oil SC, m3


250 500,000
Oil Rate SC, m3/d

200 400,000

150 300,000

100 200,000

50 100,000

0 0
0 365 730 1,095 1,460 1,825 2,190 2,555 2,920 3,285 3,650
(c) Time, day

Fig. 12(c) Comparison of oil rate and cumulative oil production of Well Pairs Inj3/Prod3 and Inj4/Prod4. Two middle well pairs
are operative.

should be modified. Coupled simulation and history matching


should be reperformed until satisfactory matches are obtained. vs = G
u
...................................................................................(5)

Incorporating Seismic Data. Unlike matching field historical data


(pressures, temperatures, and displacements), matching seismic- vp =
Ku + 43 G ............................................................................(6)
u
survey data with the results of a coupled simulation is not straight-
forward and additional calculations and data processing to generate
synthetic seismic data are required before matching. The method In Eqs. 5 and 6, bulk density of undrained (saturated) rock is es-
and procedure are described in the following subsections. timated by
Fundamentals. The fundamental links between seismology and
geomechanics are the velocities of P-wave (compressional wave) u = s (1 ) + f ,...............................................................(7)
and S-wave (shear wave). From Eqs. 1 and 2, we have

16 Journal of Canadian Petroleum Technology


90 5.4
85 Inj3/Prod3: RF (coupling) 5.1
Inj3/Prod3: RF (no coupling)
80 Inj4/Prod4: RF (coupling) 4.8
Inj4/Prod4: RF (no coupling)
75 4.5
Inj3/Prod3: CSOR (coupling)
70 Inj3/Prod3: CSOR (no coupling) 4.2
Inj4/Prod4: CSOR (coupling)
65 Inj4/Prod4: CSOR (no coupling)
3.9
60 3.6
55 3.3

CSOR, m3/m3
50 3.0
RF, %

45 2.7
40 2.4
35 2.1
30 1.8
25 1.5
20 1.2
15 0.9
10 0.6
5 0.3
0 0.0
0 365 730 1,095 1,460 1,825 2,190 2,555 2,920 3,285 3,650
(a) Time, day

Fig. 13(a) Comparison of RF and CSOR of Well Pairs Inj3/Prod3 and Inj4/Prod4. All well pairs are operative.

90 5.4
Inj3/Prod3: RF (coupling)
85 5.1
Inj3/Prod3: RF (no coupling)
80 Inj4/Prod4: RF (coupling) 4.8
Inj4/Prod4: RF (no coupling)
75 Inj3/Prod3: CSOR (coupling) 4.5
70 Inj3/Prod3: CSOR (no coupling) 4.2
Inj4/Prod4: CSOR (coupling)
65 Inj4/Prod4: CSOR (no coupling) 3.9
60 3.6
55 3.3

CSOR, m3/m3
50 3.0
RF, %

45 2.7
40 2.4
35 2.1
30 1.8
25 1.5
20 1.2
15 0.9
10 0.6
5 0.3
0 0.0
0 365 730 1,095 1,460 1,825 2,190 2,555 2,920 3,285 3,650
(b) Time, day

Fig. 13(b) Comparison of RF and CSOR of Well Pairs Inj3/Prod3 and Inj4/Prod4. Four middle well pairs are operative.

where Bulk modulus of fluid (i.e., the reverse of compressibility) in


Eq. 9 can be calculated with the model proposed by Mavko and
f = So o + Sg g + Sw w ........................................................(8) Mukerji (1998):

Bulk modulus of undrained (saturated) rock is calculated with 1


=
So
+
Sg
+
Sw ...................................................................(11)
Kf Ko Kg Kw
the well-known Gassmanns equation (Gassmann 1951),
(1 KKds )2
K u = Kd + ,..................................................... (9) Correlations Established by Laboratory Tests. Before calculat-

Kf + 1Ks ( KKd)2 ing the velocities of seismic waves during SAGD, the following
s
correlations need to be established from laboratory tests.
and shear modulus is caluclated with The changes of density (rs) and moduli (Ks) of rock grain
solid with the changes of temperatures and pressures during SAGD
Gu = Gd = G .......................................................................(10) operationThe chemical composition of sand grains may be al-

June 2011 17
90 5.4
Inj3/Prod3: RF (coupling)
85 5.1
Inj3/Prod3: RF (no coupling)
80 Inj4/Prod4: RF (coupling) 4.8
Inj4/Prod4: RF (no coupling)
75 Inj3/Prod3: CSOR (coupling) 4.5
70 Inj3/Prod3: CSOR (no coupling) 4.2
Inj4/Prod4: CSOR (coupling)
65 Inj4/Prod4: CSOR (no coupling) 3.9
60 3.6
55 3.3

CSOR, m3/m3
50 3.0
RF, %

45 2.7
40 2.4
35 2.1
30 1.8
25 1.5
20 1.2
15 0.9
10 0.6
5 0.3
0 0.0
0 365 730 1,095 1,460 1,825 2,190 2,555 2,920 3,285 3,650
(c) Time, day

Fig. 13(c) Comparison of RF and CSOR of Well Pairs Inj3/Prod3 and Inj4/Prod4. Two middle well pairs are operative.

tered because of high-temperature steam and condensate. This Conclusions


change may influence the density and moduli of sand grains. The work of this paper is summarized as follows.
The correlations for calculating the moduli of oil, gas, and Geomechanical-data-acquisition programs are proposed ac-
water with the changes of temperatures and pressuresBatzle and cording to the purposes of intended applications.
Wang (1992) provided one example. The methods to acquire each type of geomechanical data are de-
The correlations for calculating the changes of dry bulk mod- scribed.
ulus and shear modulus of rock with the changes of temperatures Field monitoring methods are reviewed briefly, and suggestions
and pressuresZou (2005) established one example for the oil for selecting monitoring methods are provided.
sands of Pikes Peak field: Coupled reservoir and geomechanical simulations are demon-
strated using a commercial simulator, and suggestions for simu-
K d = 141.18 c 2 .57 T .............................................(12) lator improvement are discussed.
The concept and work flow of integrated history matching are
G = 7.08 c 2 .96 T ..................................................(13) proposed.
The method and procedure to incorporate seismic data into the
Note that Eqs. 12 and 13 were based on dynamic bulk and shear coupled reservoir and geomechanical simulation are established.
moduli that were derived from measured shear- and compressional-
wave velocities. Nomenclature
Procedure To Synthesize Seismic Velocities. For the simulator c = effective cohesion
used in this study, the following procedure can be used to compute Cn1 = a parameter for permeability coupling in STARS
seismic velocities of each cell in one timestep. Eu = Youngs modulus of undrained (saturated) rock
Calculate the density and moduli of rock grain solid at given G = shear modulus of rock
temperature and pressure with the correlations established by labo- Gd = shear modulus of dry rock
ratory tests. If no such correlations are available, assume they are Gu = shear modulus of undrained (saturated) rock
constant as a first-order estimate. k = permeability
Generate So, Sg, Sw, ro, rg, rw, and f from reservoir simula- kH = horizontal permeability
tion. Read the data and calculate fluid density with Eq. 8. Then, kV = vertical permeability
calculate the bulk density of undrained (saturated) rock with Eq. 7. Kd = bulk modulus of dry rock
Estimate the moduli of oil, gas, and water with the correlations Kf = bulk modulus of fluid
established by laboratory tests, then calculate the bulk modulus of Ks = bulk modulus of rock grain solid
fluids with Eq. 11. However, when calculating the modulus of gas, Ku = bulk modulus of undrained (saturated) rock
note that the gas phase in the SAGD process contains water vapour P = pressure
and other gases (CH4, CO2, and H2S). Qg = gas rate
Estimate the changes of dry bulk modulus and shear modulus Qo = oil rate
of rock with the correlations established by laboratory tests. Then, Qw = water rate
compute the bulk modulus of undrained (saturated) rock with Sg = gas saturation
Eq. 9. So = oil saturation
Synthesize seismic velocities using Eqs. 5 and 6. Sw = water saturation
The dynamic undrained bulk modulus and shear modulus cal- T = temperature
culated previously should be converted to static moduli with cor- vs = velocity of shear wave
relations between dynamic and static moduli and also should be Vg = gas volume
applied in geomechanical simulation. Vo = oil volume

18 Journal of Canadian Petroleum Technology


Maximum Surface Uplift=0.379 m Unit: m
Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1
0.43
0.40 Yield
0.37 State
0.34 1.00
0.31
0.29
0.26
0.23
0.20 0.50
0.17
0.14
0.11
0.08
0.06
0.00
0.03
0.00 (a)
(a)
Maximum Surface Uplift=0.320 m Unit: m
Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1
0.42
0.39 Yield
0.36 State
0.34 1.00
0.31
0.28
0.25
0.22
0.20
0.50
0.17
0.14
0.11
0.08
0.05
0.03 0.00
0.00
(b) (b)
Maximum Surface Uplift=0.235 m Unit: m
Inj6/Prod6 Inj5/Prod5 Inj4/Prod4 Inj3/Prod3 Inj2/Prod2 Inj1/Prod1
0.38
Fig. 15Yield state inside reservoir during SAGD of scenario
0.35
0.33 with six operative well pairs. (a) Yield state at 2,640 days (b)
0.30 Yield state at 2,700 days.
0.28
0.25
0.23
0.20
0.18 Acknowledgements
0.15 We thank Statoil for permission to publish this paper.JCPT
0.12
0.10
0.07
0.05 References
0.02 Batzle, M. and Wang, Z. 1992. Seismic properties of pore fluids. Geo-
0.00 physics 57 (11): 13961408. doi:10.1190/1.1443207.
(c) Bell, J.S., Price, P.R., and Mclellan, P.J. 1994. In-situ Stress in the Western
Canada Sedimentary Basin. In Geological Atlas of the Western
Fig. 14Comparison of vertical displacements. (a) All well pairs Canada Sedimentary Basin (WCSB), ed. G.D. Mossop and I. Shetsen,
are operative. (b) Four middle well pairs are operative. (c) Two Chap. 29. Available online from Alberta Geological Survey, http://
middle well pairs are operative.
www.ags.gov.ab.ca/publications/wcsb_atlas/atlas.html.
Butler, R. 1998. SAGD Comes of Age! J Can Pet Technol 37 (7): 912.
Vw = water volume JCPT Paper No. 98-07-DA. doi: 10.2118/98-07-DA.
= strain Butler, R.M. 2001. Some Recent Developments in SAGD. J Can Pet Technol
a = axial strain 40 (1): 822. JCPT Paper No. 01-01-DAS. doi: 10.2118/01-01-DAS.
n = Poissons ratio Chalaturnyk, R.J. 1996. Geomechanics of SAGD in Heavy Oil Reservoirs.
vp = velocity of compressional wave PhD dissertation, Department of Civil and Environmental Engin-
rf = density of fluid eering, University of Alberta, Edmonton, Alberta.
rg = density of gas Chardac, O., Murray, D., Carnegie, A., and Marsden, J.R. 2005. A Proposed
ro = density of oil Data Acquisition Program for Successful Geomechanics Projects.
rs = density of rock grain solid Paper SPE 93182 presented at the SPE Middle Oil and Gas Show and
ru = bulk density of undrained (saturated) rock Conference, Bahrain, 1215 March. doi: 10.2118/93182-MS.
rw = density of water Collins, P.M. 1994. Design of the Monitoring Program for AOSTRAs
s = stress Underground Test Facility, Phase B Pilot. J Can Pet Technol 33 (3):
sa = axial stress 4653.
sa = effective axial stress (axial stress minus pore Collins, P.M. 2002. Injection Pressures for Geomechanical Enhancement
pressure) of Recovery Processes in the Athabasca Oil Sands. Paper SPE 79028
sc = confining stress presented at the SPE International Thermal Operations and Heavy Oil
sc = effective confining stress (confining stress minus Symposium and International Horizontal Well Technology Confer-
pore pressure) ence, Calgary, 47 November. doi: 10.2118/79028-MS.
f = porosity Collins, P.M. 2005. Geomechanical Effects on the SAGD Process. SPE Res
j = effective friction angle Eval & Eng 10 (4): 367375. SPE-97905-PA. doi: 10.2118/97905-PA.

June 2011 19
Donnelly, J.K. 2000. The Best Process for Cold Lake: CSS vs. SAGD. J Can Maxwell, S.C., Du, J., Shemeta, J., Zimmer, U., Boroumand, N., and Griffin,
Pet Technol 39 (8): 57. JCPT Paper No. 00-08-TN. doi: 10.2118/00- L.G. 2007. Monitoring SAGD Steam Injection Using Microseismicity
08-TN. and Tiltmeters. Paper SPE 110634 presented at the SPE Annual Tech-
Dusseault, M.B. 2007. Monitoring and modeling in coupled geomechanics nical Conference and Exhibition, Anaheim, California, USA, 1114
processes. Paper 2007-028 presented at 8th Canadian International November. doi: 10.2118/110634-MS.
Petroleum Conference, Calgary, 1214 June. Proskin, S.A., Scott, J.D., and Chhina, H.S. 1990. Current Practice in the
Fjr, E., Holt, R.M., Raaen, A.M., Risnes, R., and Horsrud, P. 1992. Petroleum Interpretation of Microfrac Tests in Oil Sands. Paper SPE 20040 pre-
Related Rock Mechanics. Amsterdam, The Netherlands: Elsevier. sented at the SPE California Regional Meeting, Ventura, California,
Forgues, E., Meunier, J., Grsillon, F.X., Hubans, C., and Druesne, D. USA, 46 April. doi: 10.2118/20040-MS.
2006. Continuous high-resolution seismic monitoring of SAGD. Rutqvist, J., Wu, Y.-S., Tsang, C.-F., and Bodvarsson, G. 2002. A Mod-
SEG Technical Program Expanded Abstracts 25 (1): 3165-3169. eling Approach for Analysis of Coupled Multiphase Fluid Flow, Heat
10.1190/1.2370186. Transfer, and Deformation in Fractured Porous Rock. Int. J. Rock
Gassmann, F. 1951. Elastic waves through a packing of spheres. Geophysics Mech. Min. Sci. 39: 429442.
16 (4): 673685. doi:10.1190/1.1437718. Sayers, C.M., Kisra, S., Tagbor, K., Taleghani, A.D., and Adachi, J. 2007.
Kular, G.S., Chhina, H.S., Best, D.A., and MacKensie, W.T. 1988. Multiple Calibrating the Mechanical Properties and In-Situ Stresses Using
Hydraulic Fracture Propagation in Oil Sands. Paper SPE 17534 pre- Acoustic Radial Profiles. Paper SPE 11089 presented at the SPE
sented at the SPE Rocky Mountain Regional Meeting, Casper, Wyo- Annual Technical Conference and Exhibition, Anaheim, California,
ming, USA, 1113 May. doi: 10.2118/17534-MS. USA, 1114 November. doi: 10.2118/11089-MS.
Lacy, L.L. 1997. Dynamic Rock Mechanics Testing for Optimized Frac- Settari, A. and Walters, D.A. 2001. Advances in Coupled Geomechanical
ture Designs. Paper SPE 38716 presented at the SPE Annual Technical and Reservoir Modeling With Applications to Reservoir Compaction.
Conference and Exhibition, San Antonio, Texas, USA, 58 October. SPE J 6 (3): 334342. SPE-51927-PA. doi: 10.2118/51927-MS.
doi: 10.2118/38716-MS. Shell Canada Energy. 2008. In Situ Oilsands Progress Presentation: Hilda
Li, P. 2006. Numerical Simulation of the SAGD Process Coupled with Lake Pilot 8093, Orion 10103. Progress Presentation, Energy Re-
Geomechanical Behavior. PhD thesis, The University of Alberta, Ed- sources Conservation Board, Calgary, Alberta (8 April 2008).
monton, Alberta, Canada. STARS User Manual. 2008. Calgary, Alberta: Computer Modelling Group
Lowe, K.W. and Donovan, T.E. 1989. Thermal Zone Monitoring in an Atha- (CMG).
basca Oil Sands Pilot. Paper SPE 19854 presented at the SPE Annual Suncor Energy. 2006. Firebag Project: Commercial In-Situ Oil Sands Pro-
Technical Conference and Exhibition, San Antonio, Texas, USA, 811 ject. Annual Review Presentation, Approval No. 8870, Alberta Energy
October. doi: 10.2118/19854-MS. and Utilities Board, Calgary, Alberta (19 April 2006).
Mavko, G. and Mukerji, T. 1998. Bounds on low-frequency seismic vel- Suncor Energy. 2008 ERCB Annual Review: Suncor Firebag. Annual Re-
ocities in partially saturated rocks. Geophysics 63 (3): 918914. port, Energy Resources Conservation Board, Calgary, Alberta (30
doi:10.1190/1.1444402. April 2008).

Qo , Qg , Qw ,
T, P Vo , Vg , Vw
from OW from I/P

Match Match
T, P
I /P
f rom T,P Q,V

Reservoir T, P, So , Sg , Sw , Qo , Reservoir
k ,
, k H, VS Model Qw, Qg , Vo , Vg , Vw Simulation
, g
Geostatistic So, Sw
Modelling P, T
So Good
Modify Models Sg No Yes
k, Match?
Sw
Fa
cie

Geomechanical Geomechanical
s

, Simulation
S Model Simulation End
,R
, PS P
ES I S, P , DS SS

Match Match
Geomechanical
Data
SU, , DS, ME from 4D Seismic
OW and Surface
OW, Survey

Abbreviation of measured data


DS = displacement Q = fluid rate
I/P = injector and producer SS = synthetic seismic data
ME = microseismic events SU = surface uplift
OW = observation well T = temperature
P = pressure V = fluid volume

Fig. 16Work flow of integrated history matching.

20 Journal of Canadian Petroleum Technology


Total E&P Canada. 2007. Total E&P Canada 2007 Joslyn Creek Perform- When initial stresses are transported into a geomechanical
ance Presentation. Annual Performance Presentation, Approval No. model, equilibrium may be lost. A feature allowing re-equilibration
9272C, Alberta Energy and Utilities Board, Calgary, Alberta (2 Oc- would be an improvement.
tober 2007).
Tran, D., Settari, A., Nghiem, L. 2002. New Iterative Coupling Between a
Reservoir Simulator and a Geomechanics Module. Paper SPE 78192 Authors
presented at the SPE/ISRM Rock Mechanics Conference, Irving,
Texas, USA, 2023 October. doi: 10.2118/78192-MS. Fagang Gu is a senior reservoir engineer at
Wong, R.C.K., Barr, W.E., and Kry, P.R. 1993. Stress-strain response of Statoil Canada. He has worked on various
Cold Lake oil sands. Can. Geotech. J 30 (2): 220235. doi:10.1139/ reservoir-engineering studies of the Corner
t93-019. SAGD project since joining Statoil in 2008.
Zhang, W., Youn, S., and Doan, Q. 2007. Understanding Reservoir Archi- Previously, Gu had worked for Golder Asso-
tectures and Steam-Chamber Growth at Christina Lake, Alberta, by ciates, Weatherford International/Advanced
Using 4D Seismic and Crosswell Seismic Imaging. SPE Res Eval & Technology, and Southwest Petroleum Insti-
Eng 10 (5): 446452. SPE-97808-PA. doi: 10.2118/97808-PA. tute (SWPI)/China National Petroleum Cor-
Zou, Y. 2005. Integration of Seismic Methods with Reservoir Simulation, poration for 15 years. His technical expertise
Pikes Peak Heavy Oil Field, Saskatchewan. PhD thesis, The Univer- includes reservoir simulation of SAGD and
sity of Calgary, Calgary, Alberta. other thermal recovery, coalbed-methane (CBM) and enhanced
CBM (ECBM) recovery, and low-permeability and tight-reservoir
Appendix A: Features of Simulator oil- and gas-recovery processes; applications of geomechanics in
The simulator (CMG 2008) applied in this study has options for reservoir/geomechanical coupled simulation, caprock integrity, and
both implicitly and explicitly sequential coupled simulations of sand production in SAGD, CBM/ECBM, heavy-oil cold produc-
SAGD. Reservoir fluid flow is solved by a finite-difference solver, tion, and sour-gas enhanced-oil-recovery (EOR); optimization of
while geomechanical deformation is solved by a finite-element low-permeability and tight-reservoir development through stimula-
solver. The built-in failure criteria for rock and soil deformation tion (hydraulic fracturing and acidizing); and production optimiza-
are the Mohr-Coulomb and Drucker-Prager criteria. The constitu- tion. Gu holds a PhD degree in geotechnical engineering from the
tive models include elastic, nonlinear elastic, elastic-perfect plastic, University of Alberta and MSc and BSc degrees in petroleum engi-
and generalized plasticity models. Geomechanical simulation can neering from SWPI (now Southwest Petroleum University) in
be run in 2D plane-strain and 3D modes. The coupling of porosity China. He has published more than 30 technical papers and is a
has three modes. member of Association of Professional Engineers, Geologists, and
*GCOUPLING 0Porosity does not depend upon deforma- Geophysicists of Alberta (APEGGA), SPE, and the Canadian
tion. Heavy Oil Association (CHOA).
*GCOUPLING 1Porosity is a function of pressure, temper-
ature, and volumetric strain. Mark Y.S. Chan is a reservoir engineering
*GCOUPLING 2Porosity is a function of pressure, temper- lead for the Corner SAGD Project and up-
ature, and total mean stress. stream and in-situ oil sands at Statoil
The coupling of permeability has four modes. One is to use the Canada. Before joining Statoil, he worked
empirical model proposed by Li (2006), and the other three are to for Unocal, Petro-Canada, Chevron, and
use tables to specify the permeability multiplier as a function of Value Creation Inc. In the past 33 years,
mean effective-stress change, total stress change, or volumetric Chan has worked in various aspects of reser-
strain change from their initial states, respectively. In coupled sim- voir management for both conventional and
ulation, the related stresses and strains are calculated by the simu- EOR onshore oil and gas projects in the
lator. The permeability and porosity are calculated on the basis of western Canadian sedimentary basin and
the chosen permeability and porosity modes and input data, then offshore projects in the Grand Banks and the Scotian Shelf on the
are applied in reservoir simulation. All steps are implemented auto- east coast of Canada. He has also managed reservoirs in the North
matically without user interference. Sea and has experience with in-situ thermal recovery schemes for
oil-sands and heavy-oil assets in Alberta and Saskatchewan. Chans
Appendix B: Limitations of Simulator primary areas of expertise include numerical reservoir-flow model-
The refinements in the following areas will improve the simulators ling, the application of SAGD and cyclic steam stimulation thermal
geomechanical capabilities for oil-sands applications. recovery processes to heavy-oil and bitumen deposits, the applica-
It is well understood (Li 2006; Wong et al. 1993) that the me- tion of reservoir geostatistics, and reservoir optimization. He holds
chanical moduli, friction angle, and dilation angle of oil sands are a BSc degree in mechanical engineering and a Masters degree in
functions of stresses and strains. However, the present constitutive fluid mechanics from the University of Calgary. Chan is a member
models in the simulator cannot capture such characteristics fully. of APEGGA, CHOA, and SPE.
The deformation of oil sands, especially shear failure, is irre-
versible. However, the results of coupled simulation showed that Robert Fryk is a professional engineer with
some previously shear-failed cells became nonfailed in the fol- more than 26 years of combined engi-
lowing timesteps (e.g., in Fig. 15a and 15b). This could not happen neering, operations, and management expe-
physically. In addition, the deformation of oil sands in the post- rience, with expertise in the areas of
failure stage would be different from the deformation in the pre- reservoir engineering, exploitation, drilling,
failure stage. Capturing these deformation behaviours can enhance completions, operations, marketing, and
the simulators geomechanical-simulation capacity. asset and corporate acquisitions. He holds a
To avoid simulating fluid flow in over- and underburdens, the BSc degree in chemical engineering from
porosity has to be set to zero. However, this ignores the influence of the University of Calgary and is a member
pore-pressure development on effective stresses, deformation, and of APEGGA.
potential caprock failure during the SAGD process.
The initial stresses of each cell or layer must be entered through
effective stresses rather than stress gradients. This makes data entry
and geomechanical-model building slow and inefficient.

June 2011 21

You might also like