Você está na página 1de 27

SPE-171089-MS

Can SAGD Be Exported? Potential Challenges


J.M. Alvarez, R. Moreno, and R.P. Sawatzky, Alberta Innovates Technology Futures

Copyright 2014, Society of Petroleum Engineers


This paper was prepared for presentation at the SPE Heavy and Extra Heavy Oil Conference - Latin America held in Medellin, Colombia, 24 26 September 2014.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
In spite of its relative youthfulness as a commercial IOR technology, SAGD has had a profound impact
on the development of the oil sands in northern Alberta and consequently on Albertas economy. The path
from a concept as articulated and developed by an individual research engineer, Roger Butler, to a
commercial recovery technology for exploiting Albertas oil sands has at times been a rocky one. More
than 30 years of applied research and development at both the laboratory scale and the field scale by a
community of researchers has been necessary to advance the technology to its current state of development. The SAGD technology has faced countless challenges since its origin. The public literature is
replete with examples of challenges it has faced in the past, and others that are predicted to be just over
the horizon. For example, more research and development will be required to expand SAGD to more
challenging areas of the Athabasca region. The challenges will include reservoirs that have shale barriers,
those that have bottom/top water zones, and those that are marginally too thin for conventional SAGD.
Variants of SAGD, such as fast-SAGD, X-SAGD, and ES-SAGD, were conceived in response to the need
to expand SAGD beyond the sweet spots in Albertas oil sands where the original technology was
demonstrated. Lessons learned from the application of SAGD technology in Albertas oil sands may be
useful to international producers from around the world that are interested in trying to adopt SAGD for
use in their reservoirs. This paper will discuss potential challenges that may be encountered in the
implementation of SAGD as the technology is migrated from its birthplace in the Athabasca oil sands to
heavy oil and bitumen reservoirs around the world.

Introduction
Arguably, SAGD was developed for the specific conditions of sweet spots in Athabasca reservoirs. In
these reservoirs, the high viscosity of the bitumen was detrimental to conventional steam-based recovery
processes such as steam flooding and cyclic steam stimulation (CSS). Injectivity to steam was low and
sweep efficiency was poor due to steam fingering/channeling into the formation. However, the sands were
unconsolidated, relatively clean and sufficiently thick to allow gravity forces to serve as an adequate
production mechanism. The issues arising from low injectivity and poor sweep efficiency were mitigated
in the SAGD process by developing suitable initialization schemes, involving the recirculation of steam
in the injector and producer wells, and by using gravity as the driving force. The advent of horizontal
drilling allowed SAGD to become a commercial recovery technology as the area that could be contacted

SPE-171089-MS

by steam was enlarged tremendously, enabling production rates to reach an order of magnitude required
for a successful commercial project.
The first field scale demonstration of the SAGD process using horizontal wells was carried out at the
AOSTRA Phase A UTF pilot project in 1987. Following the success of the AOSTRA field pilot project,
commercial projects have been developed by the oil industry over the past 10-15 years. Learnings from
the AOSTRA project were used in the implementation of the commercial SAGD projects. Many of these
projects were performed in areas where reservoir properties were quite similar to the ones at the UTF site.
Screening criteria for SAGD have been developed based on the experiences of the AOSTRA pilot project
and the commercial projects that have been implemented since then. It is commonly accepted in the
industry that for SAGD to be viable, pay thickness needs to be greater than 15 meters, oil content should
be greater than 10 wt.%, the ratio of vertical permeability to horizontal permeability (kv/kh) should be
greater than 0.2, reservoir pressure must be higher than 1,000 kPa, bottom or top water or a gas cap cannot
be present, and a competent caprock capable of acting as a seal must be present.
Further, relatively simplistic models have been developed to predict the performance of SAGD
projects. For example, Shin1 developed a very good predictor, based on field results, that required only
four reservoir parameters as input: vertical permeability, pay thickness, oil saturation and porosity. He
found that SAGD performance indicators such as cumulative steam oil ratio (CSOR), calendar day oil rate
(CDOR), and recovery factor (RF) are highly correlated with these four reservoir parameters. He
concluded that SAGD projects tended to be uneconomical when any one of the following conditions
occurred: porosity was lower than 0.3, pay thickness was less than 15 m, oil saturation was lower than
0.65, or vertical permeability was lower than 1 m2.
Jimenez2 performed an examination of most of the SAGD projects in Canada that existed at the time
of his review, with a view to determining what type of reservoirs are best-suited for SAGD. He found very
interesting and somewhat surprising trends. In about 30% of the cases that were reviewed, the CSOR was
less than 2.5 while in only about 10% of the cases was the CSOR above 4. The cases with higher CSOR
values were associated with poor quality reservoirs (low oil content and poor vertical connectivity).
Jimenez also found that usually about 30% of the recovery was obtained after 3.5 to 5 years of production,
although it might take up to 7.5 years to reach this value in the worst cases. The CSOR tended to reach
a plateau or even decrease with time after two years. Therefore the ultimate performance of a SAGD pad
might be estimated reasonably well 1.5 to 2 years after starting operations.
Along with many other experts, Jimenez2 observed that geology and operating conditions were the
most important parameters in a SAGD operation. Based on his study, it appeared that the following well
configurations were good choices for a successful SAGD project: horizontal well lengths ranging from
500 to 750 m, lateral spacing of well pairs of about 100 m, and 4 or more well pairs per pad. With regard
to geology, Jimenez2 identified certain SAGD projects such as MacKay River, where the geology appears
to be the dominant factor controlling SAGD performance. The comparatively poor performance obtained
in MacKay River pad A can be explained by a relatively poor geology dominated by breccia rather than
clean sand. The oil saturation and vertical connectivity is lower in this pad than in other pads in the same
area. Thief zones such as top/bottom water/gas zones can also have a big impact on SAGD performance.
In general, SAGD is considered by many experts as the ideal recovery process for a perfect reservoir,
as defined by the thick, clean sands in the sweet spots in Athabasca.
However, as the commercial application of SAGD in the Athabasca region expands, it is becoming
necessary for operators to learn how to apply the process in more challenging reservoirs. Predictive tools
have been useful in designing SAGD projects for clean sand reservoirs. However, the limitations of
these predictive tools increase substantially as the reservoir moves farther apart from the perfect
reservoir. This leads to the following question: can SAGD be extended to more challenging areas of the
Athabasca region? On a broader level, the question can be expanded: can SAGD be exported beyond the
borders of Alberta?

SPE-171089-MS

The purpose of this paper is to begin to try to answer this last question. Through the use of a relatively
simple numerical simulation approach, the effects of parameters that are representative of candidate
reservoirs in other parts of the world on SAGD are explored. These parameters include lower oil viscosity
(lower API), higher dip angle, higher GOR, greater heterogeneity, higher water saturation, higher
reservoir pressure and a wider range of pay thickness.

Potential Challenges in Exporting SAGD


SAGD has been tested in areas outside of the Athabasca region. The first SAGD pilot was conducted by
Imperial Oil at Cold Lake in 19783. The viscosity of the bitumen in this region (100,000-300,000 mPas)
tends to be about one order of magnitude lower than the viscosity of the bitumen in the Athabasca region
(1,000,000-3,000,000 mPas). A deviated production well near the base of the reservoir together with a
vertical injection well 45 m above the producer was used for the pilot. After 10 years of operation, the
cumulative bitumen production in the pilot was about 50,000 m3 with a CSOR of about 2.5. Although the
results appear encouraging from this vantage point, Imperial Oil chose not to continue evaluating SAGD
in the Cold Lake area after the pilot had been completed.
More recently, Blackrock evaluated SAGD at the Hilda Lake project in the Cold Lake region. Jimenez2
identified Hilda Lake as one of the worst-performing SAGD pads in his study. He suggests that one of
the possible reasons for this poor performance is that reservoirs in the Cold Lake area tend to be more
heterogeneous and less saturated with bitumen than reservoirs in the Athabasca region.
Several pilot tests of SAGD have been performed in Lloydminster-area heavy oil reservoirs. The
viscosity of the heavy oil in these reservoirs (10,000-30,000 mPas) tends to be about two orders of
magnitude lower than the viscosity of the bitumen in the Athabasca region. Moreover, the initial GOR for
this heavy oil (7-10 m3/m3) is about 5 times higher than the initial GOR for Athabasca bitumen (1.5-3.5
sm3/m3). Miller and Xiao4 discussed the results of a SAGD test in the Lloydminster region using a single
vertical well. This test performed much more poorly than had been expected, with an SOR of approximately 10. The authors suggested several potential reasons for the poor performance. These included the
presence of basal breccia that allowed edge water to migrate into the drainage area of the SAGD well and
the presence of pressure-depleted areas adjacent to the well resulting in steam migration away from the
well.
Tavallali et al.5 presented a brief summary of two SAGD pilot tests performed in the Lloydminster
region, the North Tangeflags project and the Senlac project. He observed that SAGD has not been applied
extensively in the region due to the widespread belief that SAGD will not be economical there. The main
reason for this belief is that Lloydminster-area reservoirs tend to be relatively thin compared with
Athabasca reservoirs, so the oil rate generated by SAGD in the Lloydminster region will tend to be lower
and the heat losses higher than in the Athabasca region. Another critical issue in many Lloydminster
SAGD pilots is the presence of bottom water6,7,8.
Laricina has tested SAGD in the Grosmont Formation, a bitumen carbonate reservoir located in
northern Alberta. The bitumen in this fractured carbonate reservoir is more viscous than the bitumen in
Athabasca oil sands reservoirs. More recently, however, Laricina announced that their preferred well
configuration and recovery technology for the Grosmont Formation is a single horizontal well using a
cyclic SAGD process rather than a conventional SAGD horizontal well pair.
SAGD has also been tested in other parts of the world. For example, Mendoza et al.9 presented the
results obtained in a SAGD pilot test performed in western Venezuela, in the Tia Juana field, in late 1997.
The viscosity of the heavy oil in this reservoir ranges from 10,000 to 20,000 mPas under initial reservoir
conditions, the initial oil saturation is 85%, and the initial GOR is about 11 m3/m3. The reservoir thickness
varied from 12 m at the heel of the horizontal well pair to 26 m at the toe. Mendoza et al.9 concluded that
the SAGD pilot had been implemented successfully. After 3 years of operation, the oil production rate
averaged about 110 m3/d, the CSOR was about 4.5, and the RF was expected to be about 60%. Based

SPE-171089-MS

partially on the results of this SAGD pilot, a screening criteria was developed by Llaguno et al.10 to select
candidate reservoirs in Venezuela for SAGD. A total of 35 reservoirs containing heavy oil and extra-heavy
oil were selected for further examination using the screening tool.
More recently, Guinand et al.11 reported the preliminary results of the first SAGD pilot performed in
the Orinoco Oil Belt. The pilot test was performed in the Bare field. The thickness of the reservoir in the
pilot area is about 24 m, the oil viscosity under initial reservoir conditions is about 700 mPas, and the
reservoir permeability is about 5 m2. Production started in early 2009. The production rate was about 130
m3/d and the expected recovery factor was about 40%. Dusseault et al.12 generated an estimate for the
recoverable reserves that could be achieved in the Orinoco-Belt (FPO) using SAGD. Their screening
criteria differed from conventional SAGD screening criteria in that they included other parameters such
as water saturation (should be lower than 0.25), Vshale (should be less that 0.35) and reservoir pressure
(should be lower than 12,400 kPa). Using their screening criteria they found several areas in the FPO
where SAGD could be performed successfully. Another interesting result from their paper is that their
estimate of the recovery factor associated with SAGD in the FPO was relatively low, below 30%.
Two SAGD pilot tests carried out at the Liaohe oil field in China have been reported by Yang13. The
pilot was performed by using bottom horizontal wells as producers and existing vertical wells as injectors.
In this reservoir there are five layers of sandstone alternated with conglomerates. The average pay
thickness is 90 m, the permeability is about 5.5 m2, and the oil viscosity is about 230,000 mPas under
reservoir conditions. This reservoir is classified as a heavy oil reservoir. It is a challenging reservoir in that
it contains bottom/top water zones along with inter-bedded zones. Yang13 observed that even though the
Liaohe field is relatively deep compared with Athabasca bitumen reservoirs (the depth at Liaohe is
530-640 m with a dip angle of 2-3), the results of the SAGD pilot tests have been encouraging. In these
tests, the expected ultimate RF is about 55% with an average oil production rate of over 100 m3/d. Yang
concludes that SAGD may be applicable in deeper reservoirs.
Although SAGD pilot tests have been performed in several regions outside the Athabasca area, the
reality is that SAGD has been applied commercially only in Athabasca. There are many reasons for this.
Some of the potential causes will be explored through numerical simulation, and discussed in the
following section. The results of the numerical simulations will be compared with anecdotes, observations
and results obtained during SAGD applications around the world.

Numerical Simulation Results


Numerical Model
A simplistic numerical model was created to serve as a platform for investigating and illustrating some
of the potential challenges that will be faced in trying to export SAGD to other areas outside of the
Athabasca region. The simulation model was a 3D Cartesian model, populated with generic homogeneous
data obtained from the public domain that is representative of a clean Athabasca oil sand reservoir. The
model contained one horizontal SAGD well pair. Each of the wells was 1,000 m long. It was assumed that
lateral well spacing was 100 m. Vertical well spacing was set at 5 m, with the producer situated almost
at the base of the oil sand. The oil sand was 30 m thick. The top of the oil sand was located at a depth
of 500 m.
The grid blocks in the simulation model were uniform in size. The horizontal dimension of each grid
block perpendicular to the horizontal well pair was set at 1 m, as was the vertical dimension of each grid
block. The horizontal dimension of each grid block parallel to the well pair was set at 50 m. Heat losses
to the overburden and underburden were included in the model.
The fluid phase behavior was determined using a black oil model approach. At the initial pressure and
temperature of the reservoir, 3,000 kPa and 12 C respectively, the initial GOR was set at approximately
3.5 sm3/m3 and the oil viscosity at 3.5 106 mPas. Relative permeability was handled with Stones

SPE-171089-MS

Table 1Model properties: base case


second three-phase/water-wet model. The properProperty
Value
ties used in the base case simulation model, including the porosity, horizontal and vertical permeabilThickness
30 m
ity, and initial water saturation, are summarized in
kh
3 m2
kv/kh
0.7
Table 1.
Porosity
0.3
A sensitivity analysis studying the effects of
Oil density @ reservoir conditions
1008.5 kg/m3
some of the critical parameters/conditions that afInitial GOR
3.47 sm3/m3
fect the performance of SAGD was carried out. The
Oil viscosity
3.52 106 mPas
parameters/conditions that were examined in the
Initial water saturation
0.13
Initial
oil
saturation
0.87
sensitivity study include: oil viscosity, initial GOR,
dip angle, reservoir heterogeneity, initial water saturation, reservoir/operating pressure, and pay thickTable 2Scenarios for oil viscosity
ness. All of the simulations were run for a little over
Scenario
Viscosity (mPas)
10 years, with the same steam injection rate (300
1
3.5 106
m3/d CWE) and maximum liquid production rate
3
2
3.2 106
(600 m /d). Except in the cases in which the effects
3
2.9 106
of reservoir/operating pressure were explored,
4
2.5 106
steam injection was constrained by a maximum
5
2.2 106
6
1.9 106
bottom hole injection pressure of 4,000 kPa and
7
1.5 106
liquid production by a minimum bottom hole pro8
1.2 106
duction pressure of 3,000 kPa.
9
0.8 106
Oil Viscosity
10
0.42 106
3
11
14,000
Butler speculated that if SAGD is operated in
12
3,200
heavy oil reservoirs, there could be more flexibility
in terms of the well configuration and distance
between the injector and the producer wells. Since
the oil mobility is much higher in a heavy oil reservoir than in a bitumen reservoir, the injector may be
placed near the top of the formation while the producer can be located at the bottom of the reservoir.
Therefore a displacement stabilized by gravity can be induced, as the steam and steam condensate
displaces the heavy oil downwards. Butler performed scaled experiments under conditions similar to the
ones that existed in the North Tangleflags field in the Lloydminster region, and found that high oil
production rates and a high recovery factor were obtained. A potential drawback to the placement of the
injector near the top of the formation is higher heat losses to the overburden. Another potential
complication of using the well configuration suggested by Butler is the possibility of steam channeling
between the injector and the producer, since they are relatively far apart.
Das14 investigated the effect of oil viscosity on the SAGD process via numerical simulation. He
evaluated three cases. In each case, the in situ oil viscosity changed by an order of magnitude: the viscosity
was set at 300,000, 30,000 and 3,000 mPas, respectively. As might have been anticipated, as the oil
viscosity decreased the oil recovery improved, the steam injectivity increased, and the SOR was
significantly lower. Singhal et al.15 recommended staggered SAGD well pairs when the oil viscosity is
less than 35,000 mPas. Conversely, he recommended that the injector be placed directly above the
producer if oil viscosity is higher than 65,000 mPas. Das14 also recommended that staggered well pairs
should be considered in the case of lower viscosity reservoirs. Staggered wells allow oil to be transported
to the producer by a pressure drive as well as by gravity drainage. Miller and Xiao8 suggested that in the
presence of mobile oil, the movement of injected steam and/or mobile water will not be contained as
readily. This may exacerbate the channeling/fingering of steam and water.

SPE-171089-MS

Figure 1Effect of oil viscosity on oil rate

Figure 2Effect of oil viscosity on CSOR

To explore the effects of oil viscosity on SAGD further, a set of twelve (12) simulations with different
oil viscosities was performed, in which the oil viscosity ranged from 3.5 106 mPas to 3,200 mPas. In
all other respects, including the GOR, the simulations were identical. The entire list of oil viscosities used
in this set of simulations is shown in Table 2. As noted in the previous sub-section summarizing the
general properties of the numerical model, all of the simulations were run for a little over 10 years. Figures
1 and 2 show the oil production rate and the cumulative SOR, respectively, for the simulations as a
function of time. The trend is generally as expected. As the oil viscosity is reduced, the oil production rate
becomes higher and the CSOR drops. However, some numerical instabilities were observed for the more
mobile oil cases. This was particularly acute in the case of the lowest oil viscosity, 3,200 mPas; the
simulation crashed after about 5 years.
Figures 3 and 4 show cumulative oil produced and recovery factor, respectively, for the simulations as
a function of time. As the oil viscosity is reduced, the cumulative oil produced and RF increase. This is
consistent with the observations of previous authors, as noted above. However, a more detailed examination of Figures 3 and 4 indicates that the results can be grouped into three clusters. At high oil
viscosities, in the range 3.5 106 to 1.9 106 mPas, the cumulative oil produced and RF are relatively
insensitive to changes in oil viscosity. The cumulative oil produced and RF also form a cluster at more
intermediate oil viscosities, in the range 1.5 106 to 0.4 106 mPas, although the spread of this cluster

SPE-171089-MS

Figure 3Effect of oil viscosity on cumulative oil

Figure 4 Effect of oil viscosity on RF

is a slightly greater than the cluster at high oil viscosities. Finally, at relatively low oil viscosities, below
14,000 mPas, a third cluster is formed. The cumulative oil produced and RF track very closely for the two
simulations in this cluster.
Figure 5 shows the steam chamber shape for three selected cases, one from each cluster described
above. A conventional cupcake-shaped steam chamber is observed for the cases of a high oil viscosity and
a more intermediate oil viscosity. However, in the case of a relatively low oil viscosity, the shape of the
steam chamber is much less oval. It becomes broader much more rapidly than in the other two cases of
more viscous oils. This could account for the behaviour of the simulated oil production rate observed in
Figure 1 for the case of one of the relatively low oil viscosities, 14,000 mPas. After about 7 years, the
oil production rate for this oil viscosity deviated from the general trend for the higher oil viscosities. In
this case, the general decline in production rate that occurred with higher oil viscosities was avoided
somewhat, remaining more or less at a plateau until it finally began to decline toward the end of the
simulation period. The onset of the relative plateau in oil production rate for the case of the 14,000 mPas
oil viscosity coincided with the approach of the steam chamber to the lateral boundary of the simulation
model, as observed in Figure 5. Subsequently, the steam chamber expanded downwards in this case, likely
causing the oil production rate to maintain its plateau for a period of time. The same behaviour might also

SPE-171089-MS

Figure 5Effect of oil viscosity on the steam chamber shape

Figure 6 Effect of oil viscosity on the vertical chamber velocity

have occurred in the case of the lowest oil viscosity, 3,200 mPas, if that simulation had run to completion
successfully.
The vertical growth of the steam chamber was also substantially different in the case of the relatively
low viscosity oil. For the two cases of the more viscous oils, the chamber takes more than 7 years to reach
the top of the oil sand formation. However, the rate of steam chamber growth is quite different in the case
of the relatively low viscosity oil. The steam chamber takes less than 2 years to reach the top of the oil
sand formation in this case. The vertical velocity of the steam chamber is plotted in Figure 6. During the
first two years of the simulation, the steam rise velocity for the two cases of the more viscous oils is very
similar. This velocity, in the range of 1.0 to 1.5 cm/d, is consistent with values reported in the literature
from field evidence and numerical simulations. On the other hand, the steam rise velocity in the case of

SPE-171089-MS

Figure 7Effect of increasing vertical distance on oil rate for the 3.5 million mPas case

Figure 8 Effect of increasing vertical distance on cumulative oil for the 3.5 million mPas case

the relatively low viscosity oil is up to three times faster than in the cased of the more viscous oils. Heat
losses to the overburden could become a concern after only two years for the case of relatively low
viscosity oil, noting that the initial GOR was kept fairly low in this case. This suggests that changes in
operating strategies such as reducing the steam injection rate or the steam injection pressure may be
helpful to manage SAGD more effectively in relatively low viscosity heavy oils.
The strategy of placing the injector and production wells farther apart vertically for more mobile heavy
oils was also examined in this set of simulations. Figures 7-9 present the effect of having the wells 3, 7
and 9 m apart vertically for the case in which the oil viscosity was 3.5 106 mPas. As expected, as the
vertical distance between the wells increases it takes longer to create communication between the wells
so it takes longer to start the process. For this case, as the vertical distance between wells increases the
performance of the SAGD process, as measured cumulative oil produced and CSOR, becomes much
worse. For comparison, the results of separating the wells by the same vertical distances (3, 7, 9 m) for
the case of a relatively low oil viscosity, 14,000 mPas, are presented in Figures 10-12. It can be observed
from these figures that communication between the wells is not an issue in this case. It took approximately
the same length of time to create communication between the wells for each vertical distance, so oil
production started at nearly the same time in these simulations.

10

SPE-171089-MS

Figure 9 Effect of increasing vertical distance on CSOR for the 3.5 million mPas case

Figure 10 Effect of increasing vertical distance on oil rate for the 14,000 mPas case

Initial GOR
Non-condensable gases can be generated in situ by two sources during SAGD. As a result of the
temperature increase caused by steam injection, non-condensable solution gas is released, mainly
methane. Non-condensable gas can also be generated via aquathermolysis reactions, mainly CO2 and H2S.
Yuan et al.16 indicated that there are many theories on the effects of non-condensable gas on SAGD
performance, and that the data is contradictory. Typically, results from numerical simulation show that
non-condensable gas tends to accumulate at the steam front. This accumulation of non-condensable gas
reduces the partial pressure of the steam, thereby reducing its temperature. As a result, a reduction in the
temperature gradient perpendicular to the steam front occurs, slowing down the heat transfer into the cold
bitumen. This reduces the rate of steam chamber expansion, and consequently the bitumen production
rate16-19.
Using both an analytical approach and numerical simulation, Bharatha et al.17 concluded that the
presence of dissolved gas is likely to reduce the bitumen production rate during the SAGD process.
However, this effect diminishes as the operating pressure increases. Other researchers16,18-20 suggest that
non-condensable gases can also have a positive impact on the SAGD process. They claimed that
non-condensable gas accumulation at the top of the steam chamber will provide a gas blanket for the
chamber. This will serve as an insulator, thereby reducing heat losses to the overburden.

SPE-171089-MS

11

Figure 11Effect of increasing vertical distance on cumulative oil for the 14,000 mPas case

Figure 12Effect of increasing vertical distance on CSOR for the 14,000 mPas case

However, Edmunds21 emphasized that if this gas blanket is not removed the steam will stop at the
bottom of the blanket. This will affect the SAGD process adversely, since the temperature will be lower
in the non-condensable gas blanket than in the steam chamber. Consequently, the viscosity of the bitumen
at the gas blanket-edge will not be reduced as much as the viscosity of the bitumen in the steam chamber,
reducing its mobility. This phenomenon may explain the why there has been evidence in several field
projects that the steam chamber can stall before reaching the top of the formation. Further, Edmunds
concluded that the rate of steam rise will be controlled by the accumulation of non-condensable gas and
the resulting thickness of the gas blanket above the steam chamber. The steam rise rate may be decreased
by as much as ten times where there is a moderate initial GOR, compared with the case of dead oil.
Butler3 discussed the effects of methane and CO2 generation on SAGD performance extensively. CO2
may be produced more readily than methane as its solubility in the produced water is relatively high
compared with methane. To mitigate the effects of methane accumulation in the steam chamber, Butler
suggested performing continuous or periodic blow-downs to accelerate methane production from the
SAGD chamber.
To explore the effects of initial GOR on SAGD further, a set of four (4) simulations with different
initial GORs was performed. The initial GOR in this set of simulations was 3.5, 4.0, 6.4 and 13.4 sm3/m3,
respectively. For the base case with an initial GOR of approximately 3.5 sm3/m3, gas-liquid K-values for

12

SPE-171089-MS

Table 3Scenarios for initial GOR


Initial GOR sm3/m3
3.5
4.0
6.4
13.4

Viscosity (mPas)
3.5 106
2.75 106
1.4 106
0.4 106

Figure 13Effect of increasing initial GOR on oil rate

Figure 14 Effect of increasing initial GOR on cumulative oil

the oil and methane components were calculated as a function of temperature and pressure with the
assumption that the bubble point was 3,200 kPa. For the remaining three scenarios with a higher GOR,
the K-values from the base case were modified in accordance with the increase in the initial GOR. The
oil viscosity under reservoir conditions was also adjusted for these scenarios to account for the increase
in initial GOR. In Table 3, the initial GOR and oil viscosity are tabulated for each scenario.
Figures 13-16 show the oil production rate, cumulative produced oil, recovery factor and CSOR,
respectively, for the simulations as a function of time. A change in the initial GOR from 3.5 to 4.0 sm3/m3
had a negligible effect on SAGD performance. However, in the other two cases with a higher initial GOR,
there was a significant reduction in the oil production rate, cumulative produced oil and recovery factor.

SPE-171089-MS

13

Figure 15Effect of increasing initial GOR on RF

Figure 16 Effect of increasing initial GOR on CSOR

Another observation from the previous set of simulations on oil viscosity is consistent with this behaviour.
The oil viscosity for the scenario with an initial GOR of 13.4 sm3/m3 in this set of simulations is
approximately 400,000 mPas, almost identical to the oil viscosity for one of the scenarios run in the
previous set of simulations with a much lower initial GOR, 3.5 sm3/m3. The recovery factor was reduced
considerably at the higher initial GOR, dropping to about 30% from about 50% in the previous scenario
with a higher GOR.
Figure 17 shows the steam chamber shape for two of the scenarios, one with the lowest initial GOR
(3.5 sm3/m3) and one with the highest initial GOR (13.4 sm3/m3). In both scenarios, a conventional
cupcake-shaped steam chamber was observed. However, in the scenario with the higher initial GOR, the
growth of the steam chamber was much slower and the halo representing the temperature outside of the
steam chamber was much thicker, implying a smaller temperature gradient perpendicular to the surface
of the steam chamber. These observations from the simulations can be explained by the presence of
non-condensable solution gas released during the heating process, and its impact on heat transfer and the
oil production rate. In support of this, Figure 18 shows the accumulation of methane at the top of the steam
chamber for the two scenarios at years 1 and 9. The accumulation of non-condensable gas is considerably
higher in the scenario with a higher initial GOR. These results indicate the effect of a relatively low
viscosity oil (relatively high mobility) in increasing the rate of steam chamber growth may be countered

14

SPE-171089-MS

Figure 17Effect of initial GOR on the steam chamber shape

Figure 18 Gas mole fraction and gas saturation for low and high initial GOR

by the effect of a higher initial GOR in this relatively low viscosity oil. It may be that the situation in
which a higher initial GOR exists will work in favour of controlling the vertical advance of the steam
chamber in more mobile heavy oil reservoirs and reducing the time required for the steam chamber to
reach the top of the reservoir.
Dip Angle
As noted above, Das14 suggested that in heavy oil reservoirs staggered injector and producer wells could
be considered for the SAGD process, taking advantage not only of the gravity drainage process, but also
of the steam drive process. However, Miller and Xiao21 emphasized that while laterally staggered injectors
and producers have been proposed by several authors, they have not been successfully tested in the field.
In the case of mildly or steeply dipping reservoirs, laterally staggered wells may have a real advantage.
The LAK Ranch in Eastern Wyoming is a sandstone steep-dip reservoir (dip angle 30-40)22
containing a moderately viscous heavy oil with a viscosity of 3,300 mPas under reservoir conditions.
Therefore it is expected that steam will migrate up dip and oil will be produced down dip. This reservoir
has an average porosity of 22%, an average permeability of about 0.8 m2, and an average oil saturation
of about 65%. A SAGD pilot was performed in the reservoir using a single well pair. Steam injection

SPE-171089-MS

15

Figure 19 Effect of dip angle on oil rate and cumulative oil

started in March 2001. The well pair was not drilled as a conventional SAGD well pair, with the injector
directly above the producer. Rather, the wells were drilled with a lateral stagger of 5.3 to 8 m and a vertical
separation of 4 to 5.5 m. The steam injection rate was fairly low, at about 110 m3/d. Consequently, the
oil production rate was quite limited, only 7 m3/d. The production of an oil emulsion was considered a
major problem during the execution of the pilot.
Pina et al.23 proposed dipping the horizontal wells for the case of a low-dip reservoir, one with an
average dip angle of 3.2. They proposed to take advantage of gravitational segregation in this low-dip
reservoir by placing the heel of the well up dip from the toe to favour a more uniform placement of steam.
An evaluation of this well configuration was undertaken by using a numerical model with properties
representing a heavy oil reservoir located in eastern Venezuela. The following reservoir and fluid
properties were used in the simulation study: reservoir thickness of 25 to 30 m, porosity ranging from 28
to 32%, permeability varying from 0.1 to 0.8 m2, average water saturation of 21%, a ratio of vertical to
horizontal permeability (kv/kh) of 0.2, and an oil viscosity of 540 mPas under reservoir conditions. They
found that by placing the heel of the well up dip from the toe, the recovery factor increased by about 9%
as compared with a flat horizontal well pair.
Mojarab et al.24 supported a similar approach. In a numerical study of Athabasca and Cold Lake oil
sands reservoirs, they found that SAGD performance could be improved by configuring the injector to dip
towards the toe while leaving the production well flat. They observed that the thermal efficiency of the
process was improved by 5% in the case of the dipping injector.
Yang and Butler25 published experimental results evaluating the effect of the dip angle on the SAGD
process. The dip angle in their experiments was 5. They found that higher production rates were achieved
for a reservoir dipping upwards from the production well than for a horizontal reservoir. They noted that
the increase in cumulative oil production in the case of the dipping reservoir was caused by the higher
drainage head.
Card et al.26 identified challenges in simulating SAGD in inclined reservoirs. They found that a
conventional steam trap constraint on production was effective in the situation where the reservoir did not
dip, but not so in the case in a dipping reservoir. In their simulation study the block containing the heel
was operated at a lower pressure (due to the difference in head). As a consequence, steam was produced
in the simulation. By controlling the steam injection rate they were able to solve this numerical problem.
However, their approach had an impact on the final results of the SAGD simulations. Both the steam
injection rate and the oil production rate were reduced in comparison with the non-dipping case.
To examine the effects of dip angle on SAGD further, a set of six (6) simulations with different dip
angles was performed. The dip angle in this set of simulations was 0, 5, 10, 15, 20 and 25, respectively.
All the cases were run with an oil viscosity of 1.9 106 mPas. Figure 19 shows the oil production rate
and cumulative produced oil, respectively, for the simulations as a function of time. The change in the dip

16

SPE-171089-MS

Figure 20 Effect of dip angle on the steam chamber shape after 9 years of injection

angle from 0 to 25 had a negligible effect on SAGD performance for this viscous oil, although there was
an effect on the shape of the steam chamber. As shown in Figure 20, it moved up dip preferentially. The
influence of the dip angle may be greater in reservoirs with a more mobile heavy oil.
Heterogeneity
The effect of heterogeneity on SAGD performance has been studied extensively by many authors3,25, 27-36.
In both Butler3 and Albahlani and Babadagli36 there are very comprehensive discussions on the effect of
reservoir heterogeneity on the SAGD process. For example, reservoir heterogeneities such as vertical
barriers may affect the head available for gravity drainage. Thief zones, on the other hand, such as lean
zones (areas with a relatively high water saturation), top/bottom water zones, and top gas zones may act
as sinks (areas with a relatively high water mobility) allowing steam to channel through them. Jimenez2
suggested that the MacKay River SAGD project provides an excellent example of the effects of geology
on SAGD performance.
Yang and Butler25 published experimental results evaluating the effects of reservoir heterogeneity
through an idealized model, one in which the reservoir consisted of two distinct layers. Two experiments
were performed. In case 1 a high permeability layer overlay a low permeability layer. In case 2 the
situation was reversed; a low permeability layer overlay a high permeability layer. They concluded that
for a reservoir with two horizontal layers the initial oil production rate is controlled by the top layer while
the effects of the bottom layer increase as the steam chamber grows. The influence of the lower layer
becomes dominant after most of the oil has drained from the top layer. However, SAGD performance was
better in case 1 than in case 2. Yang and Butler25 also evaluated the effect of horizontal barriers in their
experiments. They concluded that a short horizontal barrier does not affect SAGD performance significantly. On the other hand, the oil production rate decreases when a long horizontal barrier was present,
although not as much as had been expected.
Chen et al.28 studied the effects of heterogeneity on SAGD through numerical simulation by using a
stochastic model of the shale distribution. They evaluated the impact of heterogeneity on two flow
regions, the near well region (NWR) and the above well region (AWR). They concluded that the drainage
of hot fluids is affected drastically by the presence of shale in the NWR, as a consequence of poor vertical
communication. However, their results indicated that the presence of shale in the AWR affected SAGD
performance only when the shale was continuous or occupied a high fraction of this region shale. They
proposed that hydraulic fracturing could be used as a remediation technology for reservoirs with low
vertical continuity, noting that vertical fractures parallel to the well direction would be more effective in
improving SAGD performance in heterogeneous reservoirs than vertical fractures perpendicular to the
well direction.

SPE-171089-MS

17

Figure 21Effect of kv/kh on oil rate and cumulative oil

Dang et al.27 also investigated the effect of shale on SAGD performance via numerical simulation.
They studied the effect of shale barriers between the injector and the producer (BIP) and the effect of
barriers above the well pairs (AWP). They concluded that SAGD performance would be affected only if
the barriers above the well pairs were longer than 50 m. However, similar to Chen et al.28, they found that
SAGD performance was very sensitive to near well barriers. These findings are also in agreement with
commonly reported results from SAGD pilots such as the results from the UTF pilot. Considerable shale
material was present within the reservoir, as observed from cores, in the pilot area. Despite the restrictions
to flow that the shale material could have presented, the UTF pilot was considered to be a successful trial
of SAGD. In the sweet spots of the Athabasca reservoir it is not uncommon to finding shale layers that
are only a few centimeters thick and of limited horizontal length. However, this fortunate circumstance
is not likely to be the case in many reservoirs around the world.
Singhal et al.15 suggested that in reservoirs where steam override may be too severe, reservoir
impediments to vertical communication such as fining upwards sequences, stacked sands, a low ratio of
vertical to horizontal permeability, and the presence of discontinuous shale barriers or a high shale content
may help to confine the steam chamber.
Baker et al.33 concluded that top water thief zones influence the height of the steam chamber and heat
losses in the SAGD process, while bottom water thief zones may control the quantity of produced heat.
Law et al.29 evaluated the effects of top thief zones, either water or gas zones, on SAGD performance.
Their results showed that when the pressure gradient between the oil zone and the thief zone was
increased, more oil and steam would move into the thief zone causing a reduction in oil production and
an increase in the SOR. Doan et al.30 used numerical simulation to evaluate the effect of thief zones on
SAGD performance. They concluded that the SAGD process is hindered if a water zone is present, that
the presence of bottom water has less impact than an overlying water zone, and that the RF is correlated
negatively with both the thickness and the areal extent of an overlying water zone. Miller and Xiao8
suggested that inner wells on a SAGD pad can be shielded from the detrimental effects of bottom water
by operating the edge wells on the pad appropriately.
To illustrate the effects of heterogeneity on SAGD further, numerical simulations were carried out
under different geological settings: various ratios of vertical horizontal permeability (kv/kh), layers with
permeability either increasing or decreasing upwards, and a barrier. All the scenarios were run with an oil
viscosity of 3.5 106 mPas.
A set of three (3) simulations was performed with different values of kv/kh. Figure 21 shows the oil
production rate and cumulative produced oil, respectively, for the simulations as a function of time. As
kv/kh is reduced, the cumulative oil produced and oil rate decrease. This is consistent with the observations
of previous authors, as noted above. However, a more detailed examination of this figure indicates that
there is a relatively greater impact at the lowest value of kv/kh in the simulations, 0.3. Mainly, this is
caused by a longer delay in starting the process, since it takes longer to create communication between

18

SPE-171089-MS

Figure 22Effect of kv/kh on the steam chamber shape after 9 years of injection

Figure 23Effect of increasing permeability upwards or downwards on oil rate and cumulative oil

the wells. Figure 22 shows the shape of the steam chamber shape for each case after 9 years of steam
injection. The top of the chamber tends to flatten as kv/kh decreases. In additional, as expected, it takes
longer for the steam chamber to reach the top of the formation as kv/kh decreases.
A set of two (2) simulations was performed in which the reservoir was divided into three layers, each
with a different permeability. In one scenario the permeability was increased moving upwards through
successive layers, and in the other scenario the permeability was increased moving downwards through
successive layers. The permeabilities used in these scenarios were 5, 3 and 1 m2, respectively. In what
follows, the scenario with a homogeneous reservoir is also presented for comparison. Figure 23 shows the
oil production rate and cumulative produced oil, respectively, for the simulations as a function of time.
The scenario in which the permeability increased moving downwards through successive layers is very
similar to the scenario with a homogeneous reservoir. This is not consistent with the results of Yang and
Butler25, who concluded that for a reservoir with two horizontal layers the initial production rate is
controlled by the permeability of the top layer. As expected, the scenario in which the permeability
increased moving upwards performs significantly more poorly than the scenario with a homogeneous
reservoir. In this scenario it takes longer for the process to be initiated because it takes longer to create
communication between the wells when the lowest permeability occurs in the bottom layer. This is
consistent with the observations of previous authors27, 28, as noted above, where it was observed that
SAGD performance was affected drastically by the presence of reduced vertical communication in the
near well region.
A set of four (4) simulations was performed with a barrier located at different distances from the top
of the formation. The distance of the barrier from the top of the formation in this set of simulations was
5, 10, 15, and 20 m, respectively. In each of the scenarios, the horizontal length of the barrier was set at
60% of the distance across the entire model. Figure 24 shows the oil production rate and cumulative
produced oil, respectively, for the simulations as a function of time. As expected, and in agreement with

SPE-171089-MS

19

Figure 24 Effect of the barrier position on oil rate and cumulative oil

Figure 25Effect of the barrier position on the steam chamber shape

the observations of previous authors, as noted above, the scenario in which the barrier is closest to the
wells (20 m from the top of the formation) is the one in which the cumulative oil production is the lowest.
However, the sensitivity of the cumulative oil production to the location of the (partial) barriers is
relatively small. Figure 25 shows the shape of the steam chamber after 9 years of steam injection. As
expected, the vertical growth of the chamber shape is diminished by its proximity to the barrier.
Initial Water Saturation
Although some authors12 have included initial water saturation as an important reservoir parameter in their
screening criteria, there is not much information in the public literature on the effects of initial water
saturation on the performance of SAGD. A high initial water saturation can have a negative impact on the
SAGD process through at least four mechanisms: steam migration, lateral communication between steam
chambers, reduction in the relative permeability to oil, and reduction in heat efficiency37. Baker et al.33
indicated that the incremental water saturation above the irreducible water saturation plays a critical role
on overall SAGD performance. Water saturations in the range 20 to 25%, relatively close to the irreducible
water saturation, have a small impact on SAGD performance. However, in cases where the initial water
saturation is quite high relative to the irreducible water saturation, the SAGD process will be inefficient.
Oskouei et al.37 used an experimental approach to investigate the effects of initial water saturation on
SAGD performance. Two experiments were carried out with different water saturations, one with a water
saturation of about 15% (close to the irreducible water saturation) and one with a water saturation of 32%.
In these experiments, the recovery factor was reduced by about 7% in the case of the higher water
saturation compared with the case of the lower water saturation. The CSOR was two times higher in the
case of the higher water saturation. The authors also observed that the high water saturation induced an
increase in the vertical growth of the steam chamber in the short term, improving the oil production rate
during this period. However after about 1.5 years the oil production rate decelerated and remained
considerably lower than in the case of the experiment with the lower water saturation. This was attributed
mainly to the increase in heat losses. Further research is required on the effects of water saturation on the
SAGD process, and particularly on the amount of mobile water saturation that can be tolerated to retain
an acceptable level of SAGD performance.
To explore the impact of initial water saturation on SAGD further, a set of four (4) simulations with
different mobile water saturations was performed. The initial water saturation in this set of simulations

20

SPE-171089-MS

Figure 26 Effect of the initial water saturation on oil rate and cumulative oil

Figure 27Effect of the initial water saturation on the steam chamber shape

was 13, 15, 26 and 30%, respectively. The water relative permeability was kept the same in each of the
scenarios, with the irreducible water saturation set at 13%. Figure 26 shows the oil production rate and
cumulative produced oil, respectively, for the simulations as a function of time. As the initial water
saturation is increased, the cumulative oil produced and oil rate decrease. However, in the scenarios with
a higher mobile water saturation initially, the SAGD process starts earlier than in the scenarios where the
mobile water saturation is lower initially. These observations are consistent with the comments of
previous authors, as noted above. Figure 27 shows how the shape of the steam chamber evolves over time
for two selected scenarios, with initial water saturations of 15% and 30%. In the case of a 30% initial water
saturation, a very narrow steam chamber is formed initially. It reaches the top of the formation in less than
a year, making the SAGD process much less efficient after this due to heat losses to the overburden.
Reservoir/Operating Pressure
Recently, there has been considerable debate about the advantages and disadvantages of operating SAGD
at relatively high pressure versus relatively low pressure. Low pressure SAGD relies mainly on heat
conduction for heat transfer. On the one hand, the latent heat of steam will be higher so that the process

SPE-171089-MS

21

Figure 28 Effect of the initial reservoir pressure on oil rate and cumulative oil

will be more energy efficient. On the other hand the operating temperature will be lower so the reduction
in oil viscosity will not be as great. In high pressure SAGD conductive heat transfer will be assisted by
convection, and the reduction in oil viscosity will be greater since the operating temperature will be
higher. However, the energy efficiency of the process will be reduced since the latent heat of steam will
be lower at a higher operating pressure. Based on his analysis of many SAGD projects in Canada,
Jimenez2 suggested that a relatively high operating pressure should be used in the early stages of the
SAGD process, followed up by a relatively low operating pressure as the process matures. The analysis
of Jimenez is specific to the reservoir conditions of SAGD projects in the Athabasca region. These
conditions may differ considerably from those in deeper reservoirs around the world. Therefore, further
investigation on the applicability of SAGD in deeper reservoirs is required.
To explore the impact of initial reservoir pressure on SAGD further, a set of three (3) simulations with
different initial reservoir pressures was performed. The initial reservoir pressure in this set of simulations
was 2,000, 3,000 and 4,000 kPa, respectively. The constraint on injection pressure in each scenario was
increased along with the increase in the reservoir pressure. All the scenarios were run with an oil viscosity
of 3.5 106 mPas. Figure 28 shows the oil production rate and cumulative produced oil, respectively,
for the simulations as a function of time. There are relatively small differences in the cumulative oil
produced and the oil production rate in each scenario. The cumulative oil production after 10 years is very
similar in the low pressure scenario and the high pressure scenario. The largest difference in the simulation
results is that it takes longer for the SAGD process to start (i.e., for vapour injection to start) in the high
pressure scenario than in the lower pressure scenarios. This delay is observed in Figure 29, where the
steam chamber is smaller in the high pressure scenario at early times.
Pay Thickness
There is widespread agreement in the literature that an increase in reservoir thickness will have a positive
impact on SAGD performance. Results derived or obtained from theory, analytical models, laboratory
experiments, or numerical simulations tend to be consistent on this point, although there are exceptions.
Shins correlation shows that thickness is the most important parameter in its influence on the recovery
factor for SAGD1. However, Jimenez2 offered a contrary opinion. He found a poor correlation between
SAGD performance and reservoir thickness. He claimed that log resolution is not sufficiently good to
identify thin shale beds that could restrict the vertical continuity of the reservoir and thereby adversely
affect SAGD performance. Moreover, other geological and operating conditions could restrict the growth
of the steam chamber, masking the effect of pay thickness on overall SAGD performance. For example,
when the steam chamber is cupcake-shaped (growing both laterally and vertically), the influence of pay
thickness on SAGD performance will less important until later times in the process. However, when the
steam chamber takes on the shape of a chimney (growing first vertically and then laterally) the influence
of pay thickness on SAGD performance is expected to be much larger37. As a rule of thumb, the
recommended minimum pay thickness for the SAGD process is between 12 to 15 m. However, the

22

SPE-171089-MS

Figure 29 Effect of the initial reservoir pressure on the steam chamber shape

minimum pay thickness is expected to vary depending on reservoir conditions and operating considerations, such as the value of the oil being produced and the cost of steam generation. Therefore the
minimum pay thickness will vary from reservoir to reservoir and from country to country.
Numerical simulations examining the effect of pay thickness on the performance of SAGD would
reach the same conclusions as the ones reported above. However, it would be interesting and challenging
to try to determine the minimum pay thickness for which the SAGD process could be economic under
different reservoir and economic conditions, in different countries. Such a task would require a titanic
effort. This is beyond the scope of this paper. The task could be accomplished by defining dimensionless
numbers that include key reservoir properties and operating conditions for SAGD on the one hand and
CAPEX and OPEX parameters and measures on the other hand. An approach in which numerical
simulation was coupled with economic software may be useful for such a task. This could be a potential
topic for future papers as SAGD migrates from its birthplace.

Other Issues
There is limited knowledge about several issues that could affect the performance of SAGD. These issues
include the role of geomechanics, the non-condensable gases generated by aquathermolysis, and the flow
of emulsions created at the steam chamber interface. The impact of these issues may be magnified in
reservoirs where the reservoir and fluid properties differ significantly from the ones in the Athabasca area.

SPE-171089-MS

23

Carlson38 outlined the necessity of incorporating geomechanics in tools for predicting the performance
of SAGD, such as numerical simulators. He noted that this was important for the proper description,
analysis and design of the SAGD process. He emphasized that shearing and dilation can occur during the
SAGD process, and that this would enhance near wellbore permeability. Chalaturnyk and Li39 used
numerical simulation to evaluate the effect of injection pressure on permeability enhancement for shallow,
medium and deep Athabasca reservoirs. They found that under representative injection pressures the shear
stress changes sufficiently to cause the oil sand to approach failure. In this state the oil sand will dilate,
particularly near the wellbore. In the case of a deep Athabasca reservoir they reported a 26% increase in
the absolute permeability for an injection pressure of 15,000 kPa. Collins et al.40 presented field evidence
supporting the generation of shearing and dilation of oil sands due to the SAGD process. They indicated
that in the UTF area an upward vertical displacement of the formation was recorded.
The generation of CO2 and H2S due to aquathermolysis reactions is common in the SAGD process.
Kapadia et al.41 presented gas analysis data reported from different SAGD projects in Athabasca. The
trend observed from the field data shows that acid gas generation increases significantly when the
temperature increases above 240C. A method for predicting acid gas generation in Athabasca SAGD
projects was developed by Thimm42, 43. The acid gas generation rate depends on steam temperature and
the water, oil and rock composition. Therefore, the amount of acid gas generated will be reservoir-specific.
The effect of the generated acid gases on SAGD performance is not well understood. Studying this effect
via numerical simulation is challenging since commercial reservoir simulators do not include aquathermolysis reactions. Kapadia et al. 41 proposed aquathermolysis reaction systems and an upscaling
procedure to translate experimental kinetic parameters to the field scale for use in full field scale SAGD
simulations. Their results indicated that a high gas concentration and a low partial pressure to steam exist
at the steam chamber edge. This will hinder heat transfer and chamber growth. As well, the amount of H2S
at the top of the reservoir will increase as the SAGD process progresses.
The formation of water-in-oil emulsions has been reported in thermal oil operations44. Water-in-oil
emulsions have also been observed in a two-dimensional visual scaled model with the aid of a
high-resolution optical-fiber scope45. However, the physics of emulsion formation and emulsion flow in
porous media is poorly understood. There are no commercial reservoir simulators available that model
these mechanisms directly. However, pseudochemical reactions have been used in commercial reservoir
simulators to model emulsion formation indirectly, enabling the effect of emulsion formation and
transport on SAGD performance46, 47 to be explored. The increased drag force of emulsions at the
chamber walls and the high oil relative permeability (with oil as the continuous phase in water-in-oil
emulsions) are potential factors that may partially account for the high oil recovery obtained in SAGD.

Conclusions and Recommendations


1. The application of SAGD in other reservoirs with reservoir and fluid properties different from
those in perfect Athabasca reservoirs will bring new challenges to the SAGD technology. Some
of these challenges have been explored in this study. However, many others may arise that cannot
be addressed with current predictive tools such as analytical and numerical models and scaled
physical models. Therefore these predictive tools may need to be adapted/re-engineered if different
physics or recovery mechanisms need to be incorporated in them.
2. Current screening criteria are based on successful cases from the Athabasca area. Therefore, their
applicability is biased towards Canadian experiences. New or modified screening criteria will
likely need to be developed as SAGD expands beyond Canadian borders.
3. Numerical simulations were performed in this study to investigate the effects of some of the
critical parameters/conditions that affect the performance of SAGD, such as oil viscosity, initial

24

SPE-171089-MS

GOR, dip angle, reservoir heterogeneity, initial water saturation, reservoir/operating pressure, and
pay thickness. The results of these simulations are summarized under the following points.
4. The impact of oil viscosity on SAGD performance formed three different clusters. At high oil
viscosities, in the range 3.5 106 to 1.9 106 mPas, the cumulative oil produced and RF are
relatively insensitive to changes in oil viscosity. The cumulative oil produced and RF also form
a cluster at more intermediate oil viscosities, in the range 1.5 106 to 0.4 106 mPas, although
the spread of this cluster is a slightly greater than the cluster at high oil viscosities. Finally, at
relatively low oil viscosities, below 14,000 mPas, a third cluster is formed. The cumulative oil
produced and RF track very closely for the two simulations in this cluster.
5. The velocity of the steam chamber front is in the range of 1.0 to 1.5 cm/d for homogeneous viscous
oil reservoirs. On the other hand, the steam rise velocity in the case of a relatively low viscosity
oil is up to three times faster than in the case of more viscous oils.
6. At a relatively high initial GOR, the growth of the steam chamber is much slower and the halo
representing the temperature outside of the steam chamber is much thicker, implying a smaller
temperature gradient perpendicular to the surface of the steam chamber.
7. The effect of a relatively low viscosity oil (relatively high mobility) in increasing the rate of steam
chamber growth may be countered by the insulating effect of a higher initial GOR in this relatively
low viscosity oil.
8. SAGD performance in a layered reservoir in which the permeability increases moving downwards
through successive layers is very similar to SAGD performance in a homogeneous reservoir.
However, when the permeability in the layered reservoir increases moving upwards through
successive layers SAGD performs significantly more poorly than in the scenario with a homogeneous reservoir.
9. As reported by other authors, the ratio of vertical to horizontal permeability has an impact on
SAGD performance. This impact was considerably greater at the lowest value of kv/kh examined
in the simulations, 0.3.
10. When a relatively large barrier is present in the oil sand reservoir, with a horizontal length set at
60% of the distance across the entire model, it has a relatively small impact on SAGD performance.
11. In the scenarios with a higher mobile water saturation initially, the SAGD process starts earlier
than in the scenarios where the mobile water saturation is lower initially. However, as the initial
water saturation is increased, SAGD performance diminishes.
12. The impact of geomechanics, acid gas generation, and formation of water-in-oil emulsions on the
performance of SAGD needs further investigation. The role of these parameters may be more
important in deeper reservoirs, with a high clay content and oil containing natural surfactants.
13. Defining dimensionless numbers that include key reservoir properties and operating conditions for
SAGD on the one hand and CAPEX and OPEX parameters and measures on the other hand may
be helpful to generate screening criteria to try to determine the minimum pay thickness for which
the SAGD process could be economic under different reservoir and economic conditions, in
different countries.

References
1. Shin, H., A Reservoir Economic Indicator to Evaluate a SAGD Project, paper SPE 2008-072,
presented at the Canadian International Petroleum Conference/SPE Gas Technology Symposium,
Calgary, Alberta, Canada, 17-19 June, 2008.
2. Jimenez, J., The Field Performance of SAGD Projects in Canada, paper IPTC 12860, presented
at the International Petroleum Technology Conference, Kuala Lumpur, Malaysia, 3-5 December,
2008.

SPE-171089-MS

25

3. Butler, R. M., Horizontal Wells for the Recovery of Oil, Gas and Bitumen, Gulf Publishing
Company, 1996
4. Miller, K. A. and Xiao, Y., Lloydminster, Saskatchewan Vertical Well SAGD Field Test Results,
Journal of Canadian Petroleum Technology, November 2010, Volume 49, No. 11, pp 2229.
5. Tavellali, M., Maini, B., Harding, T. and Busahmin, B., Assessment of SAGD Well Configuration
Optimization in Lloydminster Heavy Oil Reserve, paper SPE 153128, presented at the SPE/EAGE
European Unconventional Resources Conference and Exhibition, Vienna, Austria, March 20-22,
2012.
6. Jespersen, P. J. and Fontaine, T. J., The Tangleflags North Pilot: A Horizontal Well Steamflood,
The Journal of Canadian Petroleum Technology, May 1993, Volume 32, No 5, pp 5257.
7. Wong, F. Y., Anderson, D. B., ORourke, J. C., Rea, H. Q. and Scheidt, K. A., Meeting the
Challenge to Extend Success at the Pikes Peak Steam Project to Areas with Bottom Water, paper
SPE 71630, presented at the SPE Annual Technical Conference and Exhibition, New Orleans,
Louisiana, September 30-October 3, 2001.
8. Miller, K. A., and Xiao, Y., SAGD Over Bottomwater-Optimum Utilization of Edge Pairs. Paper
2008-077, presented at the Canadian International Petroleum Conference/SPE Gas Technology
Symposium, Calgary, Alberta, Canada, June 17-19, 2008.
9. Mendoza, H. A., Finol, J. J. and Butler, R. M., paper SPE 53687 presented at the SPE Latin
American and Caribbean Petroleum Engineering Conference, Caracas, Venezuela, April 21-23,
1999.
10. Llaguno, P. E., Moreno, F., Garcia, R., Mendez, Z. and Escobar, E., A Reservoir Screening
Methodology for SAGD Applications, Paper 2002-124, presented at the Petroleum Societys
Canadian International Petroleum Conference, Calgary, Alberta, Canada, June 11-13, 2002.
11. Guinand, P., Ruiz, F., Mago, R., Hernandez, A., Ospina, R., Soto, R., Papaterra, E. and Mendez,
O., Drilling the First SAGD Wells in the Orinoco Oil-Belt Bare Field: A case History, paper
163059, presented at the South America Oil and Gas Congress, Maracaibo, Venezuela, October
18-21, 2011.
12. Dusseault, M. B., Zambrano, A., Barrios, J. R. and Guerra, C., Estimating Technically Recoverable Reserves in the Faja Petrolifera del Orinoco-FPO, paper 2008-437, presented at the World
Heavy Oil Congress, Edmonton, Alberta, Canada, March 10-12, 2008.
13. Yang, L., Field Test of SAGD as Follow-Up Process to CSS in Liaohe Oil Field of China, Journal
of Canadian Petroleum Technology, April 2007, Volume 46, No. 4, pp 1215.
14. Das, S., Application of Thermal Recovery Process in Heavy Oil Carbonate Reservoirs, paper SPE
105392, presented at the Middle East Oil and Gas Show and Conference, Kingdom of Bahrain,
March 11-14, 2011.
15. Singhal, A. K., Ito, Y. and Kasraie, M. Screening and Design Criteria for Steam Assisted Gravity
Drainage (SAGD) Projects, paper SPE 50410, presented at the SPE International Conference on
Horizontal Well Technology, Calgary, Alberta, November 1-4, 1998.
16. Yuan, J-Y., Chen, J., Pierce, G., Wiwchar, B., Golbeck, H., Wang, X. Beaulieu, G. and Cameron,
S., Non-Condensable Gas Distribution in SAGD Chamber, paper CSUG/SPE 137269, presented
at the Canadian Unconventional Resources & International Petroleum Conference, Calgary,
Alberta, Canada, October 19-21, 2010.
17. Bharatha, S., Yee, C. T. and Chan, M. Y. Dissolved Gas Effects in SAGD, paper 2005-176,
presented at the Canadian International Petroleum Conference, Calgary, Alberta, Canada, June
7-9, 2005.
18. Yuan, J-Y, Nasr, T. R. and Law, D. H., Impacts of Initial Gas-to-Oil Ratio (GOR) on GAGD
Operations, Journal of Canadian Petroleum Technology, January 2003, Volume 42, No. 1, pp
48 52.

26

SPE-171089-MS

19. Yang, L., Changfeng, X., Shangqi, L. and Chunze, L., Impact of Non-Condensable Gas on SAGD
Performance, paper SPE 150539, presented at the Heavy Oil Conference Canada, Calgary,
Alberta, Canada, June 12-14, 2012.
20. Butler, R., The Steam and Gas Push (SAGP), Journal of Canadian Petroleum Technology, March
1999, Volume 38, No. 3, pp 54 61.
21. Miller, K. A., and Xiao, Y., Improving the Performance of Classic SAGD with Offsetting Vertical
Producers, Journal of Canadian Petroleum Technology, February 2008, Volume 47, No. 2, pp
2227.
22. Grills, T. L., Vandal, B., Hallum, F. and Trost, P. Case History: Horizontal Well SAGD
Technology is Successfully Applied to Produce Oil at LAK Ranch in Newcastle Wyoming, paper
SPE/CIM/CHOA 78964, presented at the International Operations and Heavy Oil Symposium and
International Horizontal Well Technology Conference, Calgary, Alberta, Canada, November 4-7,
2002.
23. Pina, J. A., Bashbush, J. L. and Fernandez, E. A., Applicability and Optimization of SAGD in
Eastern Venezuela Reservoirs, paper SPE/PS/CHOA 117622, presented at the International
Thermal Operations and Heavy Oil Symposium, Calgary, Alberta, Canada, October 20-23, 2008.
24. Mojarab, M., Harding, T. and Maini, B., Improving the SAGD Performance by Introducing a New
Well Configuration, Journal of Canadian Petroleum Technology, April 2011, Volume 50, No. 4,
pp 9 18.
25. Yang, G. and Butler, R. M., Effects of Reservoir Heterogeneities on Heavy Oil Recovery by
Steam-Assisted Gravity Drainage, Journal of Canadian Petroleum Technology, October 1992,
Volume 31, No. 8, pp 3743.
26. Card, C., Oballa, V., Kisman, K., Fu, M. and Wang, X. Three Dimensional SAGD Simulator of
a Dipping Oil Sand Reservoir, paper 96-56, presented at the Annual Technical Meeting of The
Petroleum Society, Calgary, Alberta, Canada, June 10-12, 1996.
27. Dang, C., Nguyen. N., Bae, W., Nguyen, H. and Tu, T., Investigation of SAGD Recovery Process
in Complex Reservoir, paper SPE 133849, presented at the Asia Pacific Oil & Gas Conference
and Exhibition, Brisbane, Queensland, Australia, October 18-20, 2010.
28. Chen, Q., Gerritsen, M. G. and Kovscek, A. R., Effects of Reservoir Heterogeneities on the
Steam-Assisted Gravity Drainage Process, paper SPE 109873, presented at the Annual Technical
Conference and Exhibition, Anaheim, California, November 11-14, 2007.
29. Law, D. H., Nasr, T. N. and Good, W. K., Lab-Scale Numerical Simulation of SAGD Process in
the Presence of Top Thief Zones: A Mechanistic Study, paper 2000-21, presented at the Petroleum
Societys Canadian International Petroleum Conference, Calgary, Alberta, Canada, June 4-8,
2000.
30. Doan, L. T., Baird, H., Doan, Q. T. and Farouq Ali, S. M., An Investigation of the Steam-Assisted
Gravity-Drainage Process in the Presence of a Water Leg, paper SPE 56545, presented at the
Annual Technical Conference and Exhibition, Houston, Texas, October 3-6, 1999.
31. Alturki, A., Gates, I. D. and Maini, B., On SAGD in Oil Sands Reservoirs With No Cap Rock,
paper CSUG/SPE 137234, presented at the Canadian Unconventional Resources & International
Petroleum Conference, Calgary, Alberta, Canada, October 19-21, 2010.
32. Nasr, T. N., Golbeck, H., Korpany, G. and Pierce, G., SAGD Operating Strategies, paper SPE
50411, presented at the International Conference on Horizontal Well Technology, Calgary,
Alberta, Canada, November 1-4, 1998.
33. Baker, R., Fong, C., Li, T., Bowes, C, and Toews, M., Practical Considerations of Reservoir
Heterogeneities on SAGD Projects, paper SPE/PS/CHOA 117525, International Thermal Operations and Heavy Oil Symposium, Calgary, Alberta, Canada, October 20-23, 2008.

SPE-171089-MS

27

34. Jorshari, K. R., OHara, B. and Jones, R. W., SAGD Pair Performance Optimization: A Field Case
of Recovery Enhancement, paper CSUG/SPE 149230, presented at the Canadian Unconventional
Resources Conference, Calgary, Alberta, Canada, November 15-17, 2011.
35. Yuan, J-Y. Jiang, Q., Russel-Houston, J., Thornton, B. and Putnam, P., Evolving Recovery
Technology Directed Towards Commercial Development of the Grosmont Carbonate Reservoirs,
paper CSUG/SPE 137941 presented at the Canadian Unconventional Resources & International
Petroleum Conference, Calgary, Alberta, Canada, October 19-21, 2010.
36. Albahlani, A. M. and Babadagli, T., A Critical Review of the Status of SAGD: Where Are We
and What is Next?, paper SPE 113283, presented at the Western Regional and Pacific Section,
Bakersfield, California, March 31-April 2, 2008.
37. Oskouei, S. J., Maini, B., Moore, R. G. and Mehta, S. A., Effect of Initial Water Saturation on the
Thermal Efficiency of the Steam-Assisted Gravity-Drainage Process, Journal of Canadian
Petroleum Technology, September 2012, Volume 51, No. 5, pp 351361.
38. Carlson, M., SAGD Geomechanics, Journal of Canadian Petroleum Technology, June 2003,
Volume 42, No. 6, pp 19 25.
39. Chalaturnyk, R. J. and Li, P., When Is It Important to Consider Geomechanics in SAGD
Operations? Journal of Canadian Petroleum Technology, April 2004, Volume 43, No. 4, pp
5361.
40. Collins, P. M., Carlson, M. R., Walters, D. A. and Settari, A., Geomechanical and Thermal
Reservoir Simulation Demonstrates SAGD Enhancement Due to Shear Dilation, paper SPE/ISRM
78237, presented at the Rock Mechanics Conference, Irving, Texas, October 20-23, 2002.
41. Kapadia, P. R., Wang, J., Kallos, M. S. and Gates, I. D., Reactive Thermal Reservoir Simulation:
Hydrogen Sulphide Production in SAGD, paper CSUG/SPE 149448, presented at the Canadian
Unconventional Resources Conference, Calgary, Alberta, Canada, November 15-17, 2011.
42. Thimm, H. F., Prediction of Hydrocarbon Sulphide Production in SAGD Projects, paper 2007094, presented at the Canadian International Petroleum Conference, Calgary Alberta, Canada,
June 12-14, 2007.
43. Thimm, H. F., Prediction of Carbon Dioxide Production in SAGD Operations, paper 2008-007,
presented at the Canadian International Petroleum Conference, Calgary Alberta, Canada, June
17-19, 2008.
44. Butler, R. M., Thermal Recovery of Oil and Bitumen, Published by Prentice-Hall, Inc., 1997
45. Sasaki, K., Akibayashi, S., Yazawa, N. and Kaneko, F., Microscopic Visualization with High
Resolution Optical-Fiber Scope at Steam Chamber Interface on Initial Stage of SAGD Process,
paper SPE 75241, presented at the Improved Oil Recovery Symposium, Tulsa, Oklahoma, April
13-17, 2002.
46. Azom, P. and Srinivasan, S., Mechanistic Modelling of Emulsion Formation and Heat Transfer
During the Steam-Assisted Gravity Drainage (SAGD) Process, paper SPE 124930, presented at
the Annual Technical Conference and Exhibition, New Orleans, Louisiana, October 4-7, 2009.
47. Ezeuko, C. C., Wang, J. and Gates, I. D., Investigation of Emulsion Flow in Steam-Assisted
Gravity Drainage, SPE Journal, June 2013, Volume 18, No. 3, pp 440 447.

Você também pode gostar