SPE 109873

Effects of Reservoir Heterogeneities on the Steam-Assisted Gravity Drainage Process

Q. Chen, M.G. Gerritsen, and A.R. Kovscek, SPE, Stanford University

Copyright 2007, Society of Petroleum Engineers

pressure, thick, permeable, and homogeneous sands where
This paper was prepared for presentation at the 2007 SPE Annual Technical Conference and benign surface conditions exist. Recent advances in horizontal
Exhibition held in Anaheim, California, U.S.A., 11–14 November 2007.
well technology have emerged that expand the applicabability
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, as
of thermal recovery2. Of the thermal methods, steam-assisted
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to gravity drainage (SAGD) appears to be quite promising,
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at especially for bitumen.
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
A typical SAGD process is illustrated in Fig. 1. Two
for commercial purposes without the written consent of the Society of Petroleum Engineers is horizontal wells are placed close to the bottom of a formation.
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 The injector well is placed directly above the producer a short
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, Texas 75083-3836 U.S.A., fax 01-972-952-9435.
vertical distance (4–10 m). Steam is injected continuously into
the upper well, and rises in the formation, forming a steam
Abstract chamber. Cold oil surrounding the steam chamber is heated,
The success of steam-assisted gravity drainage (SAGD) has becomes mobile as its temperature increases, and flows
been demonstrated by both field and laboratory studies mostly together with condensate along the chamber boundaries
based on homogeneous reservoirs and reservoir models. A toward the lower well that functions as a producer3. The
comprehensive understanding of the effects of reservoir SAGD technique enjoys many advantages over other thermal
heterogeneities on SAGD performance is required, however, methods, especially the conventional steam flooding methods.
for wider and more successful implementation. This work SAGD overcomes the shortcomings of steam override by
presents a numerical investigation of the effects of reservoir employing only gravity as the driving mechanism, which leads
heterogeneities on SAGD using a stochastic model of shale to a stable displacement and a potentially high oil recovery.
distribution. Two flow regions, the near well region (NWR) Moreover, in the SAGD process, the heated oil remains hot
and the above well region (AWR), are identified to decouple and movable as it flows toward the production well; whereas,
the complex effects of reservoir heterogeneities on the SAGD in conventional steam flooding, the oil displaced from the
process. Numerical simulations are conducted with various steam chamber cools and consequently oil-phase viscosity
realizations of shale distribution to compare SAGD increases as oil flows to the production well.
performance in terms of the effects of NWR and AWR. To design an effective SAGD field project, a good
Hydraulic fracturing is proposed to enhance steam chamber understanding of the complicated physics is needed for
development for reservoirs with poor vertical communication reliable prediction of performance. A vast literature on the
and the feasibility of hydraulic fracturing is discussed in terms SAGD concept has been developed since it was first
of in-situ stresses and well orientations. Fracturing the introduced by Butler and his colleagues in the late 1970s4,5.
formation around both the injection and production wells is Butler developed a gravity drainage theory based on
found to improve steam distribution, oil production rate, and numerous simplifying assumptions, such as steam chamber
the oil-steam ratio. pressure remains at the original reservoir pressure and the
steam chamber must remain connected to the producer. He
Introduction derived a semi-analytical numerical solution to predict the oil
Vast quantities of heavy and extra-heavy oil (bitumen) drainage rate. Additinoally, he and his coworkers also reported
resources have been found worldwide. For example, an experimental data obtained with a scaled visual laboratory
estimated original heavy oil in place of more than 1.8 trillion model. Reiss6 proposed modifications to Butler's gravity
barrels is present in Venezuela, 1.7 trillion barrels in Alberta, drainage model by using an empirical dimensionless
Canada, and 20–25 billion barrels on the North Slope of temperature coefficient and the maximum velocity, and Akin7
Alaska, USA1. Due to the significant viscosity of the crude at also modified the model by incorporating asphaltene content
reservoir temperature, however, the technical and economic dependent viscosity to better match the experimental data in
recovery of these resources presents a significant challenge. the literature. Nasr et al.8 studied steam-liquid counter-current
Thermal recovery, and steam injection in particular, is and co-current flows for different permeabilities and initial gas
tremendously successful for such resources. To date, more saturations with a non-steady state, laboratory steam-front
than 4 billion bbl of oil have been recovered as a result of dynamic tracking technique.
steam injection. Nevertheless, conventional steam injection Numerical simulation has been widely used by many
candidates are limited to on-shore, relatively shallow, low researchers to investigate the physical process and practical
2 SPE 109873

operation of SAGD as well. For example, Edmunds9 analyzed develop a chamber above the wells. An idealized steam
SAGD steam trap control with 2D and 3D simulation models. chamber in a homogeneous reservoir is shown schematically
He found that establishing a liquid saturated leg above the in Fig. 1. The development of this upside down triangular
producer was feasible by controlling the temperature of the steam chamber involves complicated steam condensate flows
produced fluid. The producer is shut in when the temperature and thermal processes. Injected steam rises within the chamber
of the produced fluid approaches the temperature of the under buoyancy forces and flows continuously to the
injected fluid. perimeter of the chamber, where it condenses and releases a
These analytical, numerical, and experimental studies, large amount of latent heat. The heat is transferred, by both
however, were generally performed for homogenous, isotropic conduction and convection, first to the condensate that flows
porous media. In reality, no reservoir is homogeneous because inside the steam chamber, and then the adjoining oil16,17. The
of natural geological features, such as shale, faults, and mobilized oil and the condensate drain by gravity along the
fractures. One example is the oil sand deposit in Peace River, steam chamber toward the production well. As the oil is
Alberta, Canada that contains a good deal of marine shale and removed and more steam is injected, the boundary of steam
mudstone, that forms continuous and discontinuous shale chamber expands upwards and sideways.
barriers throughout the formation10. The heterogeneity Butler's3 gravity drainage theory yields an analytical
introduced by the shale barriers and other geological features equation to predict the oil rate drained along the sidewalls of
may play an important role in the propagation of steam11. the steam chamber, with the assumptions that only steam
Therefore, without understanding the effects of reservoir flows in the steam chamber, the steam pressure is constant in
heterogeneities, the knowledge of SAGD gained with the steam chamber, oil saturation is residual, and heat transfer
homogeneous systems cannot be applied directly to provide ahead of the steam chamber to cold oil is only by conduction:
accurate, reliable predictions for field type systems.
Over the past decades, several researchers have 1.5φΔSo kgαh
investigated the role of reservoir heterogeneities on steam qo = L (1)
mν s
chamber development for a SAGD process. Joshi and
Threlkeld12 studied reservoirs with shale barriers and
compared the effects of various well configuration schemes as where, L is the length of the horizontal well, φ is porosity, ΔSo
well as vertical fractures experimentally. Yang and Butler11 is the difference between initial oil saturation and residual oil
conducted SAGD experiments with reservoirs of two different saturation to steam, k is the effective permeability for the flow
types: reservoirs with thin shale layers and reservoirs with of oil, g is the acceleration due to gravity, α is the thermal
horizontal layers of different permeabilities. These studies diffusivity, h is the steam chamber height, m is a constant
were subject to experimental limitations. Reservoir between 3 and 4 depending on oil viscosity versus temperature
heterogeneities were simulated by including a limited number curve, and νs is the kinematic viscosity of oil at steam
of impermeable barriers at designated locations. Given the temperature. This equation indicates that steam chamber
complex geological nature of shale, it is better to use a growth is necessary for oil production and that the rate of
stochastic model of shale distribution generated by drainage is a function of the vertical height and permeability
geostatistical methods12. of the steam chamber.
The objective of this study is to investigate the effects of Considering a heterogeneous reservoir that contains
reservoir heterogeneity on SAGD using such a stochastic randomly-distributed shale barriers, the development of a
model of shale distribution and a thermal reservoir simulator. steam chamber is affected to some extent and the ideal steam
To decouple the complex effects of reservoir heterogeneity on chamber shown in Fig. 1 does not form. However, because of
the SAGD process, two flow regions of the steam chamber are the unique well arrangement and gravity driving mechanism,
identified according to the characteristic. Numerical the steam chamber is expected to attain the inverse triangular
simulations are conducted with equal-probability realizations shape during its development, although with some distortions.
to compare SAGD performance. For reservoirs with poor The drainage path of hot oil and condensate is still along the
vertical communication, hydraulic fracturing is proposed to slopes of the steam chamber. One analogy to such a drainage
enhance steam chamber development and the feasibility of path is the fluid flow along the surface of a funnel.
hydraulic fracturing is discussed in terms of in-situ stress and The triangular shape and orientation of the steam chamber
well orientations. results in different characteristic lengths. In the upper part of
the chamber, the heated volume is large, and the steam flow
Features of the SAGD process inside and hot oil drainage around the chamber have wider
Figure 1 shows the common implementation of SAGD with paths to propagate. The flows in this region are of relatively
two horizontal wells in a vertical cross section view. In long characteristic length, e.g., half of the formation height.
practice, the SAGD process is normally initiated with a On the other hand, in the bottom of the steam chamber, all the
preheating period to overcome the difficulty of steam injection flows are limited to the small region around the well.
due to extremely unfavorable mobility ratio15. During the Accordingly, the characteristic length is relatively short.
preheating period, steam is circulated in the tubing and out of Recall that gravity drainage theory indicates that the success
the annulus for both horizontal wells, thus heating the of a SAGD process depends on the balance between a rising
surrounding oil by conduction. Once thermal communication steam chamber and draining hot fluids. The effects of
is established between wells and the oil in the inter-well reservoir heterogeneity on steam rising and oil drainage are
region becomes mobile, steam is injected into the reservoir to not the same throughout the formation because of the
SPE 109873 3

difference in the characteristics of the flows in different has no effect on the horizontal permeability. Therefore, a
regions. This observation leads to the definition of two reduction factor of 10−5 is applied to the vertical permeability
regions, the near well regions (NWR) and the above well of the shaly sand blocks in this study. As exact geological
region (AWR). Each region is indicated in Fig. 1 by dashed- information of sand and shale sequences is not available to us,
line boxes. As demonstrated later, identification of the NWR we model the distribution of shaly sands with a stochastic
and AWR makes it possible to decouple the complex effects representation based on a geostatistical method, sequential
of reservoir heterogeneities on the SAGD process. indicator simulation (SISIM)18. In the geostatistical model, the
probability of the shaly-sand occurrence (Ps) and correlation
Reservoir Simulation Model length of shale (Ls) are the two key parameters that determine
This section lays out a description of the numerical model as the fraction of shaly sands and the continuity of shale in the
well as describes how various shale distributions are created distribution, respectively. These two characteristics of shale
and how hydraulic fractures are represented in the geological distribution, as demonstrated later, play important roles in the
model. SAGD process. For each pair of Ps and Ls, SISIM generates a
number of realizations, all honoring the predetermined data
Description of the synthetic reservoir
(e.g., hard data) and, thus all realizations are equally probable.
A synthetic reservoir representing a generic heavy-oil
Figure 3a shows one of the realizations obtained with
formation is selected. The conditions are broadly
Ps = 30 % and Ls = 1 m.
representative of the Alberta (Canada) oil. The reservoir is
300 m deep with a pay zone thickness of 20 m. Initial oil and Representation of fracture
water saturations in the pay zone are 0.8 and 0.2, respectively, Another common case of reservoir heterogeneity is the
and reservoir temperature is 50 oF. The reservoir formation presence of fractures, either naturally-existing or
consists of clean sands and shaly sands that contain laterally- hydraulically-induced. Fractures have substantial permeability
orientated thin shale. The absolute permeability of the clean and very small pore-volume. In this study, we only consider
sands is 3,000 mD in the horizontal direction and 1,800 mD in hydraulic fractures. The effect of an hydraulic fracture on
the vertical direction. The representation of shale in the model petrophysical properties of the sand block where it is located
is detailed in the next subsection. Both clean sands and shaly is approximated as follows. A fracture permeability of 106 mD
sands have a porosity of 32 %. Water-oil and gas-oil relative and aperture of 0.01 m are assumed. The absolute permeability
permeability curves are depicted together with the oil in the direction perpendicular to the fracture plane remains
temperature versus viscosity curve in Fig. 2. Detailed reservoir identical to the case without a fracture. For the directions
properties are listed in Table 1. parallel to the fracture plane, the permeability of the grid block
For the SAGD process, horizontal production wells with a is changed to the equivalent permeability computed by
length of 1,000 m are placed 1.5 m above the bottom of the arithmetic averaging according to fracture and block
pay zone. Horizontal injection wells with the same length are geometries.
drilled parallel to the producers with a vertical well spacing of
Simulation runs
4 m. The horizontal spacing between well pairs is 100 m. A
The thermal, compositional simulator model, STARS,
confined formation unit with one well pair in the center is
developed by Computer Modeling Group (CMG) was used for
considered, assuming symmetry between well pairs.
all simulation runs. These simulation runs are classified into
Figure 3 shows two grid systems adopted for the reservoir
three groups: (1) varying NWR, (2) varying AWR, and (3)
simulation. The two-dimensional system represents the
adding fractures. For the baseline simulation runs, electrical
vertical cross section (perpendicular to well) of the reservoir in
preheating is first carried out at both well locations for 90 days
which no variation in reservoir properties along the well
to mobilize the oil around wells and to establish hydraulic
direction (y-direction) is assumed. This grid contains 67 grid
communication between the two wells. Then, 95 % quality
blocks each 1.5 m wide in the horizontal (x) direction except
steam at 435 psi (i.e., 15 psi greater than the initial pay zone
the center one that is 1.0 m, and 20 grid blocks each 1.0 m
pressure) is injected continuously at the upper well. The lower
thick in the vertical (z) direction. The grid size in the vertical
production well is operated using steam trap conditions to
cross section is fine to resolve complex flows occurring in the
avoid excessive steam production. This steam trap control is
vertical plane. The three-dimensional model is specially
achieved in the simulation by setting production temperature
designed to compare the effect of fractures. To achieve
18 oF below steam temperature to establish a definite liquid
sufficient spatial resolution in all the three dimensions while
leg above the producer9,19. The simulation runs are terminated
limiting the total number of grid blocks for reasonable
after 10 years of injection.
machine running time, non-uniform grids of 37 by 37 by 20
are adopted to simulate the synthetic reservoir with 1/10 of its
Results and Discussion – Shale Barriers
original length.
Results regarding identification of flow regions around the
Shale distribution wells and above the well are reported first. Results and
Reservoir heterogeneity is introduced by including randomly- discussion as to possible improvements in recovery effiiciency
distributed thin shale. The shale is characterized by extremely obtained by hydraulically fracturing the formation around the
low vertical permeability, typically in the range of 10−6 to injection and production wells are found in the next section.
10−3 mD. For laterally-oriented thin shale, it is acceptable to
assume that the occurrence of shale in sand reduces
dramatically the vertical permeability of the sand block, but
4 SPE 109873

Near Well Region—NWR NWR are illustrated in Fig. 5b–c. Note that the three
The study of the role of the NWR on the SAGD process relies realizations each share the exact same configuration in the
on the appropriate definition of the NWR size corresponding NWR, but have different AWR with the same fraction of shaly
to the minimum volume allowing the NWR to be decoupled sand and shale continuity.
from the AWR. In other words, if the NWR is chosen As expected, the curves in the same color that represent the
appropriately, consistent SAGD performance is obtained for a cases with the same NWR collapse together with acceptable
number of equal-probability realizations, conditioned to the variance. This confirms that the determined NWR size is
same NWR data that contain the same fraction of shaly sand appropriate. Secondly, two sets of curves, in red and blue, as
(Ps) with similar shale continuity (Ls). illustrated in Fig. 5b–c, reveal dramatic differences in SAGD
Figure 4a illustrates three choices of NWR sizes, labeled performance between realizations that have different shaly-
as small, medium and large, with respect to the vertical well sand configurations in the NWR. For the case of NWR1, all
spacing. For each case, three equal-probability realizations three realizations yield an average oil production rate of about
were generated. First, a random distribution of shaly sands is 300 bbl/d. This rate is more than 2 times less in comparison to
created and used as synthetic hard data. The fraction of shaly the oil production rate of 800 bbl/d in the case of NWR2.
sands is 30 % and shale correlation length is 1 m. The volume Similar results are observed in the comparisons of oil recovery
defined by a NWR size in this synthetic realization is then and cumulative oil steam ratio.
used as hard conditioning data to generate three realizations of This considerable discrepancy is mainly attributed to the
shaly sand distribution. As a result, the three realizations in manner in which the permeability distribution in the NWR
each case share the exact same configuration in the NWR, but affects steam chamber development. Effectively removing
have different AWR with the same fraction of shaly sand and heated oil and condensate from the reservoir is necessary for
shale continuity. continuous steam injection and thus successful steam chamber
Reservoir simulation runs were conducted with the three expansion. Hot fluids must pass through the NWR before
equal-probability realizations for each NWR size. Figure 4b–d being produced. Therefore, a NWR with substantial vertical
compare SAGD performance in terms of oil production rate, and horizontal connectivity facilitates fluid drainage and thus
oil recovery versus cumulative steam injection, and aids steam chamber development. If the NWR contains shale
cumulative oil-steam ratio for the three sets of realizations. In barriers that impede vertical flow, the drainage path of hot oil
all the figures, three types of curves, solid, dashed, and dotted, may be blocked in the NWR. Moreover, because of the
in the same color represent the three realizations for the same relatively short characteristic length of flows in the NWR, the
NWR size. As seen in the figures, the curves in blue drainage flow is sensitive to the distribution of shale barriers
corresponding to the case of small NWR size exhibit in the NWR, especially when shale continuity is increased.
considerable variations between realizations. For instance, for The above analysis is easily verified by visual comparison of
the case of small NWR size, the startup time of oil production NWR1 and NWR2 in Fig. 5a. The comparison of SAGD
in Fig. 4b varies from 300 days to 1,900 days, and the oil performance between the cases of NWR1 and NWR2 suggests
recovery after 10 years of operation differs from 40 % to that in practice, horizontal well pairs should be placed in the
70 %. In contrast, the curves in green and red, for the medium high quality region (less shale) of the formation to optimize
and large NWR sizes, respectively, show consistent results SAGD performance.
between realizations. This comparison suggests that the
Above Well Region—AWR
medium size of the NWR (2Dv by 1.5D where Dv is the
Two sets of simulation runs were conducted to investigate the
vertical well spacing) is likely to be the appropriate definition
effect of AWR in terms of shale percentage and shale
of NWR size.
continuity. In the first set, the fraction of shaly sands is fixed
This result is explained by examining the physical flows
to 30 %, and the shale correlation length is varied from 1 m, to
occurring in the near well region. As described above, a steam
4 m, 8 m, and 16 m. In the second set, the shale correlation
chamber in a heterogeneous formation still forms with an
length is fixed at 4 m, and the fraction of shaly sands is
inverse triangular shape anchored at the production well
changed from 10 % to 30 % and 50 %. All the realizations are
(Fig. 1). The NWR affects the SAGD process mainly by
conditioned to the same pre-determined NWR data (NWR2
influencing the drainage flow of hot fluids (water and oil)
shown in Fig. 5a).
along the chamber boundary in the bottom portion of the
Figure 6 compares the effect of shale continuity in the
chamber. Because the angle of the chamber wall with respect
AWR on SAGD performance. For each case, reservoir
to the horizontal plane changes gradually as the steam
simulation runs were conducted with three realizations, and
chamber expands, it is reasonable to assume an average value
one of them is plotted in the figure as their responses are
of 45o for the qualitative analysis. A rectangular region is
consistent. The figure shows that oil production is strongly
determined with the angle to be 2Dv by 1Dv that covers fully
correlated to the shale continuity. The oil production rate
the bottom part of steam chamber. This leads to a NWR size
curve for the more continuous shale, for example, is below
corresponding to the case of the medium NWR size in Fig. 4a.
that for the shale with shorter correlation length.
After determining the correct NWR size, we investigate
As the shale becomes more continuous, from 1 m to 16 m,
the effect of the NWR on SAGD performance. Figure 5a
the oil recovery factor decreases from 70 % to 23 % and the
shows two configurations of NWR with the medium size. Two
cumulative oil steam ratio reduces from 0.3 to 0.15. Notice
random realizations with 30 % shale occurrence and
that this decreasing trend is not uniform. For changes in shale
correlation length of 1 m were used to select the NWR. SAGD
continuity from 1 m to 4 m, the resulting difference in SAGD
performance with three realizations conditioned to each fixed
SPE 109873 5

performance is not obvious; but when the shale correlation stress corresponds to the overburden stress. Once a fracture is
length is significant, they cause dramatic reduction in oil induced hydraulically, the dominant orientation of the fracture
production. This is because the steam chamber expansion plane is horizontal (Fig. 9a). For deep SAGD projects, the
mainly occurs in the AWR. The flows associated with the reservoirs have a minimum horizontal stress corresponding to
steam chamber expansion are of relatively long characteristic the least principal stress. The dominant orientation of the
flow length depending on the steam chamber height. As a induced fracture plane is vertical. In addition, depending on
result, the horizontal barrier formed by shale can only affect the drilling direction of the horizontal well pairs in a SAGD
the steam chamber development when it is long enough; process, a vertical fracture can be parallel or perpendicular to
otherwise, steam easily bypasses the discontinuous shale and the wells (Fig. 9b and c). In the following subsection, we
extends the chamber further into the un-touched zones. The examine the effects of the fractures with three different
comparison in Fig. 6 indicates that shaly sand with correlation orientations on the SAGD process by reservoir simulation.
length longer than 8 m impairs SAGD performance
Effect of hydraulic fractures
substantially. This critical value is just about half of the
Figure 10 compares the SAGD performances with three
formation thickness. Such results confirm our analysis of the
shale/fracture configurations. The three cases share the same
flow characteristic length in the AWR.
shale distribution that has 50 % shale and shale correlation
When the fraction of shaly sand increases, we observe a
length of 1 m. In the base case (red curve), the oil production
similar reduction trend in SAGD performance, as presented in
rate starts with a low value and then experiences a jump at
Fig. 7. The case with 10 % shale gives the highest oil
about 700 days that leads to the main production period. The
production rate as well as the best oil recovery factor. Note
oil production rate, after reaching a peak, drops back to
that there is a sharp fluctuation in the oil production rate at
300 bbl/d and remains at a plateau rate with a slow decline for
3,000 days for the 10 % shale case. This is likely attributed to
the rest of production time. The final oil recovery is only 21 %
the steam trap control that triggers an increase in the producer
and the cumulative oil steam ratio is 0.2. For the case with
BHP at 3,000 days to avoid steam breakthrough. There is a
horizontal fractures, the oil production rate curve (in blue)
small reduction in oil production when the shale percentage is
shows a similar shape as the base case except that the main
increased by 20 %, whereas, another 20 % increase in shale
production period occurs 300 days earlier. For the case with a
percentage results in substantial change in oil production, i.e.,
vertical fracture, the main oil production period starts shortly
approximately 60 % reduction in both the average oil
after steam injection and exhibits a much greater average oil
production rate and oil recovery factor. This observation is
rate, more than twice the oil rates of the other two cases. As
expected because when shale percentage increases, although
seen in Fig. 10b, the base case yields an oil recovery of only
with correlation length of only 4 m, the shaly sand blocks are
21 % after 4,500 bbl CWE steam injection. Adding horizontal
likely to become inter-connected and form a more continuous
fractures increases the oil recovery to 24 %, while the
barrier to vertical flow, which, in turn, limits the development
presence of the vertical fracture improves the oil recovery
of steam chamber.
dramatically, up to 56 %.
The observed different effects of horizontal and vertical
Results and Discussion – Hydraulic Fracture
fractures are explained by examining how the steam chamber
Next, we consider the ability of hydraulic fractures to mitigate
profile is affected by the presence of fractures. Figure 11
the poor vertical communication accompanying the cases with
shows the temperature profiles in the vertical cross section of
a high percentage of shale.
the formation after 3 years of steam injection. Because of
Fracture orientation shaly sands (causing very low vertical permeability), the steam
Hydraulically-induced fractures always propagate chamber in the base case develops very little in the formation
perpendicular to the least principal stress20. Therefore, to after 3 years of steam injection. Compared with the base case,
determine the orientation of hydraulic fractures, the tectonic the horizontal fractures (Fig. 11b) help the steam chamber
stresses in a reservoir should be analyzed to obtain knowledge extend laterally better, but show very little improvement in the
of the least principal stress. Figure 8a shows the orientation of vertical direction. The steam chamber in the case with a
the maximum horizontal stress (SHmax) in Alberta21 that is one vertical fracture, however, is developed fully through the
of the major heavy oil resources attracting interest. The whole thickness of the formation. This is because the presence
orientation of SHmax is determined from the borehole breakout of a vertical fracture provides a highly permeable vertical path
analysis of vertical wells. The highly consistent NE-SW trend for steam, that substantially improves the vertical development
of the maximum horizontal stress is observed throughout the of the steam chamber. Recall that steam trap control helps to
Alberta Province. Figure 8b shows the magnitudes of the three limit steam breakthrough.
principal stresses as a function of depth. As seen in the figure, According to Butler's analytical equation (Eq. (1)), the oil
for the formation shallower than 150 m, the vertical stress drainage rate is proportional to the square root of the chamber
corresponds to the least principal stress. Beyond a depth of height. Hence, the improvement in the vertical development of
200 m, the least principal stress changes to the minimum steam chamber accelerates oil drainage, and consequently the
horizontal stress. This observation indicates that the performance of SAGD is enhanced dramatically. As seen in
orientation of hydraulic fractures strongly correlates to depth. Fig. 11, the volume of a steam chamber with a vertical fracture
Consequently, we consider two categories of SAGD is much larger than the other two cases.
projects according to the formation depth: shallow SAGD and The vertical fracture considered above is along the well
deep SAGD. In a shallow SAGD project, the least principal direction. According to the argument in the previous section,
6 SPE 109873

the other orientation of a vertical fracture is perpendicular to the well pair along the direction of the minimum horizontal
the well if the well is drilled along the direction of the stress (Shmin) and creating a series of vertical fractures
minimum horizontal stress. The effects of these two situations perpendicular to the well direction may provide an alternative
are investigated with the three-dimensional grid system to enhance the performance of the SAGD process in a shaly
(Fig. 3b). Similar to the 2-D model, we consider three cases: a reservoir.
base case with no fracture and two cases with a vertical A final discussion item is the penetration of the injector
fracture oriented parallel to the well and with a vertical and producer by the same hydraulic fracture. This puts great
fracture perpendicular to the well, respectively. reliance on steam trap control to limit short circuiting of steam
Figure 12 compares the predicted oil production for the from the injector to the producer as well as to limit steam
three cases. As shown in Fig. 12a, the case with the vertical intrusion into the producer. Further simulations (not shown)
fracture along the wells gives an oil production rate more than indicate that shifting the injector slightly (order of the vertical
twice that of the base case. It is interesting that with a vertical well separation) so that it does not lie in the same vertical
fracture perpendicular to the well, the oil production starts plane as the producer, hydraulically fracturing the injector,
with a lower rate than the base case and then catches up and and then performing SAGD is also effective. This completion
exceeds the latter in the late stage. This happens because of technique provides further assurance that steam is injected
the steam trap control we put at the wells to avoid direct deep into the reservoir.
stream production. The vertical fracture perpendicular to the
well creates a very permeable channel between the two wells Conclusions
at the perpendicular intersection plane. This actually increases This work presents a numerical investigation of the role of
the heterogeneity contrast along the wells. To avoid steam reservoir heterogeneity on SAGD process efficiency using a
breakthrough, the steam trap control has to set a low steam stochastic model of shale distribution and a thermal numerical
injection rate at the beginning because of high injectivity reservoir simulator. Potential improvement of SAGD
contrast along the well. As a result, the oil production rate is performance by hydraulic fracturing has also been studied.
low in the early stage. When the steam chamber develops a The following conclusions are derived:
little bit around the wells, the injectivity contrast along the 1. The complex effect of reservoir heterogeneity on the
well is smoothed out and then steam is injected at the SAGD process is decoupled by identifying two flow
reservoir's full capability in the later stages. Because the regions: the near well region (NWR) and the above well
vertical fracture, as demonstrated in the 2-D model, aids the region (AWR).
vertical development of steam chamber, the oil production 2. The drainage and flow of hot fluid within the NWR is of
shows an increase in the late time. Figure 12b shows that the short characteristic length and is found to be very sensitive
oil recovery factors are 23 %, 29 %, and 45 % for three cases, to the presence of shale that impairs vertical permeability.
respectively. 3 The AWR affects the (vertical and horizontal) expansion of
Figure 13 presents the steam chamber profiles after 6 years the steam chamber. It is of characteristic flow length on the
of steam injection in the three cases. As expected, for the case order of half of formation height. SAGD performance is
with a vertical fracture along the well (Fig. 13b), the steam affected adversely only when the AWR contains long,
chamber is well developed and a large volume of the oil sand continuous shale or a high fraction of shale.
is swept by steam. For the case with a vertical fracture 4. Hydraulic fracturing coupled with steam trap control of the
perpendicular to the well (Fig. 13c), steam extends along the producer improves well injectivity dramatically to achieve
fracture plane to the top of the formation and forms a nice an effective oil production rate in reservoirs with poor
chamber. As shown, the well-developed steam chamber is vertical communication.
limited to the region near the fracture plane. In the regions 5. In some cases, introduction of a vertical hydraulic fracture
around the two ends of well, the development of steam increases the oil steam ratio by a factor or 2. The
chamber is relatively poor. This means that the improvement orientation of hydraulic fractures depends on the depth of
in production brought by the vertical fracture perpendicular to the formation of interest. Fractures are usually horizontal
the well direction is moderate. Note that this result is obtained for shallow SAGD projects and vertical for deep SAGD
with the assumption of one vertical fracture perpendicular to projects. Vertical hydraulic fractures are predicted to
the well within the interval of interest. It is possible to induce enhance SAGD performance more dramatically in
hydraulically multiple such fractures which likely results in comparison to hydraulic fractures. Thus, hydraulic
more successful steam chamber development along the whole fracturing may be desired for deep SAGD projects.
length of well. 6. It is also found that a vertical hydraulic fracture along the
The above simulation results suggest that vertical fractures well direction is superior to one perpendicular to the well
enhance the SAGD process. The real challenge of such an idea direction. The field practice that a horizontal well is drilled
is the feasibility of generating desired vertical fractures in the along the direction of the maximum horizontal stress to
field. To achieve vertical fractures propagating along the well, ensure well stability coincides with the requirement of
it is required that the horizontal wells be drilled along the vertical hydraulic fractures parallel to the well direction.
direction of the maximum horizontal stress (SHmax). This
requirement, fortunately, coincides with the general field Nomenclature
practice that horizontal wells are normally drilled exactly in AWR = above well region
such a way to ensure well stability. If the well stability is not COSR = cumulative oil-steam ratio
an issue (e.g., strong rock) for a particular reservoir, drilling Dv = vertical well spacing
SPE 109873 7

g = gravity acceleration 2004 SPE International Thermal Operations and Heavy

h = steam chamber height Oil Symposium, Bakersfield, CA, U.S.A.,Mar.16–18.
k = effective permeability 8. Nasr, T.N., Law, D.H.S., Golbeck, H. and Korpany G.:
kr = relative permeability “Counter-Current Aspect of the SAGD Process,” Journal
L = length of horizontal well of Canadian Petroleum Technology (Jan. 2000) 39(1) 41.
Ls = shale correlation length 9. Edmunds, N.R.: “Investigation of SAGD Steam Trap
m = a constant between 3 and 4 from oil viscosity Control in Two and Three Dimensions,” paper SPE 50413
versus temperature curve presented at the 1998 SPE International Conference on
NWR = near well region Horizontal Well Technology, Calgary, Alberta, Canada,
P = pressure, psi Nov. 4–7.
Ps = probability of occurrence of shale; shale percentage 10. Webb, A.C., Schroder-Adams, C.J. and Pedersen, P.K.:
qo = oil drainage rate “Regional Subsurface Correlations of Albian Sequences
S3 = least principle stress North of the Peace River in NE British Columbia:
SAGD = steam-assisted gravity drainage Northward Extent of Sandstones of the Falher and
Sg = gas saturation Notikewin Members along the Eastern Flank of the
SHmax = maximum horizontal stress Foredeep,” Bulletin of Canadian Petroleum Geology
Shmin = minimum horizontal stress (Jun. 2005) 53(2) 165.
SISIM = sequential indicator simulation 11. Richardson, J.G., Harris, D.G., Rossen, R.H. and VanHee,
So = oil saturation G.: “The Effect of Small, Discontinuous Shales on Oil
Sw = water saturation Recovery,” Journal of Petroleum Technology (Nov. 1978)
Sv = overburden stress 1531.
T = temperature 12. Joshi, S.D. and Threlkeld, C.B.: “Laboratory Studies of
x,y,z = three principle directions in a Cartesian system Thermally Aided Gravity Drainage Using Horizontal
α = thermal diffusivity Wells,” AOSTRA J. Res. (1985) 2(1) 11.
φ = porosity 13. Yang, G. and Butler, R.M.: “Effects of Reservoir
νs = kinematic viscosity Heterogeneities on Heavy Oil Recovery by Steam Assisted
Gravity Drainage,” Journal of Canadian Petroleum
Acknowledgements Technology (Oct. 1992) 31(8) 37.
This work was prepared with the support of the U.S. 14. Pooladi-Darvish, M. and Mattar, L.: “SAGD Operations
Department of Energy, under Award No. DE-FC26- in the Presence of Overlying Gas Cap and Water Layer—
04NT15526. The support of the Stanford University Effect of Shale Layers,” Journal of Canadian Petroleum
Petroleum Research Institute (SUPRI-A) Industrial Affiliates Technology (Jun. 2002) 41(6) 1.
is gratefully acknowledged. 15. Saltuklaroglu, M., Wright, G.N., Conrad, P.R., Conrad,
J.R. and Manchester, G.J.: “Mobil’s SAGD Experience at
References Celtic, Saskatchewan,” Journal of Canadian Petroleum
1. Burton, R.C., Chin, L.Y., Davis, E.R., Enderlin, M., Fuh, Technology (Apr. 2000) 39(4) 45.
G., Hodge, R., Ramos, G.G., VanDeVerg, P., Werner, M., 16. Farouq-Ali, S.M.: “Is There Life After SAGD?” Journal of
Mathews, W.L. and Petersen, S.: “North Slope Heavy-Oil Canadian Petroleum Technology (Jun. 1997) 36(6) 20.
Sand-Control Strategy: Detailed Case Study of Sand 17. Ito, Y. and Suzuki, S.: “Numerical Simulation of the
Production Predictions and Field Measurements for SAGD Process in the Hangingstone Oil Sands Reservoir,”
Flaskan Heavy Oil Multilateral Field,” paper SPE 97279 paper in Proc. 47th ATM of the Petroleum Society of
presented at the 2005 SPE Annual Technical Conference CIM (Jun. 1996), Calgary, Alberta, Canada.
and Exhibition, Dallas, Texas, U.S.A., Oct. 9–12. 18. Goovaerts, P.: “Geostatistics for Natural Resources
2. Butler, R.M.: “Application of SAGD, Related Process Evaluation,” Oxford University Press, Inc., New York,
Growing in Canada,” Oil & Gas Journal (May 2001) 74. Oxford (1997).
3. Butler, R.M.: “Thermal Recovery of Oil and Bitumen,” 19. Egermann, P., Genard, G. and Delamaide, E.: “SAGD
Prentice-Hall, Inc. Calgary, Alberta (1998). Performance Optimization through Numerical
4. Butler, R.M. and Stephens, D.J.: “The Gravity Drainage Simulations: Methodology and Field Case Example,”
of Steam Heated Heavy Oil to Parallel Horizontal Wells,” paper SPE 69690 presented at the 2001 SPE International
Journal of Canadian Petroleum Technology (Jun. 1981) Thermal Operations and Heavy Oil Symposium,
20(2) 90. Porlamar, Margarita Island, Venezuela, Mar. 12–14.
5. Butler, R.M., Mcnab, G.S. and Lo, H.Y.: “Theoretical 20. Hubbert, M.K. and Willis, D.G.: “Mechanics of Hydraulic
Studies on the Gravity Drainage of Heavy Oil during In Fracturing,” Pet Trans AIME (1957) 210 153.
Situ Steam Heating,” Canadian Journal Chemical 21. Collins, P.M.: “Geomechanical Effects on the SAGD
Engineering (May 1981) 59(4) 455. Process,” paper SPE/PS-CIM/CHOA 97905 presented at
6. Reiss, J.C.: “A Steam-Assisted Gravity Drainage Model the 2005 SPE International Thermal operations and
for Tar Sands: Linear Geometry,” Journal of Canadian Heavy Oil Symposium, Calgary, Alberta, Canada, Nov.
Petroleum Technology (1992) 31(10) 14. 1–3.
7. Akin, S.: “Mathematical Modeling of Steam Assisted
Gravity Drainage,” paper SPE 86963 presented at the
8 SPE 109873

Table 1. Reservoir properties.

Reservoir depth 300 m
Reservoir thickness 20 m
Porosity 0.32
Horizontal permeability 3,000 mD
Vertical permeability 1,800 mD
Oil viscosity (RC) 1,000,000 cp
Reference depth 315 m
Initial pressure 420 psi
Initial temperature 50 oF
Initial oil saturation 0.80
Initial water saturation 0.20
Oil density (RC) 8.8 oAPI

Fig. 3. Numerical grids for SAGD simulation (a) 2D model and

(b) 3D model. A representative 50% shale distribution is
illustrated.

Fig.1. Schematic steam chamber growth, SAGD process

Fig. 4. Comparison of different NWR sizes. (a) definition of

three sizes, (b) oil production rate, (c) oil recovery versus
cumulative steam injection, and (d) cumulative oil-steam ratio.

Fig. 2. Rock and fluid properties. (a) water-oil, (b) gas-oil

relative permeability, and (c) oil viscosity versus temperature.
SPE 109873 9

Fig. 8. In-situ stresses in Alberta oil sand. (a) maximum

horizontal stress orientation and (b) stress magnitude versus
depth18.

Fig. 5. Comparison of SAGD performance between two shaly-

sand distributions in the NWR. (a) NWR1 and NWR2, (b) oil
production rate, (c) oil recovery versus cumulative steam
injection, and (d) cumulative oil-steam ratio.
Fig. 9. Schematic of possible orientations of hydraulic fractures
(a) horizontal fracture, (b) vertical fracture parallel to, and (c)
vertical fracture perpendicular to well.

Fig. 6. Effect of correlation length of shaly-sand in AWR on

SAGD performance. (a) oil production rate and (b) oil recovery
versus cumulative steam injection.

Fig. 7. Effect of shaly-sand percentage in AWR on SAGD Fig. 10. Comparison of no fracture, horizontal, and vertical
performance. (a) oil production rate and (b) oil recovery versus fractures. (a) oil production rate, (b) oil recovery versus
cumulative steam injection. cumulative steam injection, and (c) cumulative oil-steam ratio.
10 SPE 109873

Fig. 11. Temperature profiles after 3 years of steam injection.

(a) no fracture, (b) horizontal fracture, and (c) vertical fracture.
Fig. 13. Temperature profiles after 6 years of steam injection.
(a) no fracture, (b) horizontal fracture, and (c) vertical fracture.

Fig. 12. Comparison of formation with/without vertical fractures.

(a) oil production rate, (b) oil recovery versus cumulative steam
injection, and (c) cumulative oil-steam ratio.