Você está na página 1de 11

IPTC 12860

The Field Performance of SAGD Projects in Canada


Jaime Jimenez, Shell International Exploration & Production

Copyright 2008, International Petroleum Technology Conference


purpose was to prove the SAGD concept. The first SAGD
This paper was prepared for presentation at the International Petroleum Technology pilot was run from 1987 to 1990 under the name of Phase A.
Conference held in Kuala Lumpur, Malaysia, 3–5 December 2008.
This ‘proof of concept’ pilot consisted of three short well pairs
This paper was selected for presentation by an IPTC Programme Committee following review
of information contained in an abstract submitted by the author(s). Contents of the paper, as
closely spaced (50 m in horizontal length and 25 m apart). The
presented, have not been reviewed by the International Petroleum Technology Conference success of this pilot led to a joint venture between AOSTRA
and are subject to correction by the author(s). The material, as presented, does not necessarily
reflect any position of the International Petroleum Technology Conference, its officers, or and industry. This resulted in the first commercial pilot built
members. Papers presented at IPTC are subject to publication review by Sponsor Society
Committees of IPTC. Electronic reproduction, distribution, or storage of any part of this paper
from 1990 to 1992 in the same facility called Phase B. This
for commercial purposes without the written consent of the International Petroleum Technology pilot had three wells pairs as well, which were 70 m apart and
Conference 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 had lengths of 500 m. This pilot was operated until June 2004,
acknowledgment of where and by whom the paper was presented. Write Librarian, IPTC, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax +1-972-952-9435.
with an ultimate recovery in excess of 65% and an Oil-Steam-
Ratio (OSR) of 0.42.
Abstract Since then, more than 10 commercial SAGD projects have
The rise of the SAGD (steam-assisted gravity drainage) been operating in Canada, mainly in the Athabasca area, while
technology over the last ten years as the leading technology to CSS is used only by three operators (Imperial Oil, Shell and
develop oil sands in-situ is unquestionable. This despite the CNRL) in the two other Alberta oil sands areas. Therefore,
youth and questions that still surround this technology. over the last decade SAGD has become the preferred in-situ
Therefore, a review of most of the existing operations in technology for developing oil sand leases, mainly in the
Canada has been undertaken (32 pads in 8 different Athabasca area. Presently, the biggest in-situ operation in
operations), which includes an analysis of their current Canada, Imperial Oil at Cold Lake, uses CSS, with a
performance and particularities, trying to understand what production of ~140 kbpd. Similarly, CSS is currently the
makes a SAGD project successful, and what determines its technology used at Shell’s Peace River as SAGD has not
performance. Moreover, SAGD’s performance has been shown its promise there.
compared with the performance of CSS (cyclic steam Consequently, the rise of SAGD, a fairly young technology
stimulation), the other leading in-situ technology for oil sands. compared to CSS, has raised a lot of questions such as what is
The main finding of this work is that geology and reservoir its efficiency, its ultimate recovery, how sound is it, what type
properties are by far the most dominant features for a of reservoirs are better suited for SAGD? Therefore, an
successful SAGD operation. SAGD targets must be reservoir analysis of the performance of the current operations must
areas with average thickness above 15 m, good vertical address whether SAGD is the preferred in-situ technology to
communication and no thief zones. Moreover, if the geological develop oil sands, and under which geological conditions. The
conditions are known, the SAGD process has to be operated present document reviews a large database of SAGD
properly as lack of operational excellence can be detrimental operations in Canada (8 operations and 32 pads) and tries to
to the performance of any SAGD project. SAGD operations address some of the questions mentioned hitherto.
are badly compromised by lack of steam mainly, but also by
long boiler shutdowns, and by losing confined injectors early Methodology and Analysis Of Operations
on in the process, which lead to splitting big pads into smaller Methodology
ones. A total of 32 pads that are part of 8 different SAGD
Finally, as long as the steam chamber can grow, the operations in Alberta, Canada were included in the present
ultimate recovery of a SAGD operation can be expected to be analysis. The analysis of the performance from different
in the order of 60 to 70%. The cOSR (cumulative Oil-Steam- operators uses the following methodology:
Ratio) can fluctuate between 0.30 and 0.50, with the higher • All the analyses are made on a pad basis instead of by
end values associated with high quality reservoirs (mainly oil well or by field. This is because the coalescence of steam
content), excellent operations, and large pads; while chambers increases the recovery and energy efficiency,
operations at the lower end values have usually a combination and allows for optimization. Moreover, recoveries are
of operational issues, smaller oil content and shale baffles better defined over pads rather than single well pairs.
(poor vertical connectivity). • The technology is young, with only two pads over the
spectrum of the analysis that have terminated operation.
Introduction Therefore, it is necessary to remove somehow the time
From 1985 to 1987 the government of Alberta, represented by component in order to have a fair comparison of the
AOSTRA, built the Underground Test Facility (UTF), whose different operations. In so doing, the number of
2 IPTC 12860

hydrocarbon pore volumes injected versus hydrocarbon JACOS’ Hangingstone


recovery will be used as the main assessment criterion. Hangingstone is a unique operation in the sense that it has
• Most of the information used is public as reported by the not been developed as pads but more as couples of well pairs.
operator to the Alberta Energy Utilities Board (AEUB)1, Nonetheless the existing operation is one of the oldest, it has
in both the annual presentation and the monthly rates of 15 well pairs and a production in the order of 8000 bbl/d by
commercial operations. The rate data available to the the end of 2006. Moreover, its oldest pad (2 wells) is finished,
author is up to December 2005. and with the UTF Phase B are the only two pads where
On the other hand this analysis has some weakness and production has been completed.
short-comings such as: EnCana’s Christina Lake
• OOIP is based on the values reported to the AEUB, but Christina Lake is EnCana’s demonstration pilot and
different operators have different criteria, and they are not ‘playground’. There are 6 well pairs, which have been drilled
always known to the author. The OOIP cut off is based on two at a time. EnCana is trying the use of solvent in some of
the bitumen weight as percentage, which is defined as: its wells, and it has used 3D seismic as monitoring tool. The
So *φ operation is not commercial so EnCana is not obligated to
Wt.Bitumen(%) = * 100 report monthly volumes. Therefore cumulative volumes are
(2.65 * (1 − φ )) + (1.0 * φ ) known but rates are unknown.
where 2.65 (g/cm3) is the density of the reservoir rock and ConocoPhillips’ Surmont
1.0 (g/cm3) the density of the oil and water. Surmont is a three well pair pilot, with two relatively short
• The detailed geology of every site is not well known to pairs (350 m) and one long pair (1000 m). It is a pilot and a
outsiders. significant amount of information has been made public such
• Operational issues are mostly unknown to outsiders, as pressures and temperatures. Unfortunately the availability
therefore some of the conclusions in this regard are of steam is limited, which has limited the operation of the pilot
mainly conjectural. in general, and especially of the long well. Therefore, its
• Detailed reservoir properties such as permeability, performance is not really representative of a SAGD operation.
saturation and so on, and pad configuration details such as Nonetheless some of the data are used for analysis purposes.
well spacing, lengths and so on are average, so some Suncor’s Firebag
details that may or may not be meaningful could be Probably the biggest SAGD operation being undertaken in
missed. Canada with 40 well pairs (4 pads) in place by the middle of
Analysis of Operations 2006, 20 more pairs being drilled or prepared (2 pads), and at
EnCana’s Foster Creek least 20 more approved (2 pads). By the middle of 2006 the
Foster Creek is currently the largest SAGD operation in production was more than 30000 bbl/d out of 22 wells (steam
Canada with 54 well pairs distributed in 8 pads, and 1 pad that is limited presently but more capacity is being built).
is a series of infill well pairs (between pads). Out of the 9 Unfortunately the operation is too young to be included in-
pads, one is fairly mature, 5 are mature, and 3 are young. depth in the analysis, so the results shown must be taken
Production by the first quarter of 2006 was in the order of cautiously.
40000 bbl/d. The cut-off criterion used by EnCana to estimate Blackrock’s Hilda Lake
its OOIP is unknown to the author, but due to the high Hilda Lake is a two well pair pilot operated in the Cold
porosity and saturation in this area of Athabasca, a cut-off of Lake area by Blackrock, and recently bought by Shell. Its
10% is expected. performance cannot be compared on a one-to-one basis with
PetroCanada’s MacKay River the other operations as the steam chambers of the wells have
MacKay River is currently the SAGD operation with the not coalesced and there have been issues with steam
best OSR in Canada. There are 4 pads with 25 well pairs in availability. The data are included for completeness, and
operation and 3 more pads are going to be steamed soon. because it is the only SAGD operation outside Athabasca that
Production by the second quarter of 2006 was in the order of the author has had access data wise. A point worth noting here
25000 bbl/d. PetroCanada uses a 10% bitumen weight as cut- is that on the basis of typical saturations and porosities from
off to calculate the OOIP. Cold Lake, the cut-off to calculate OOIP must be in the order
Devon’s UTF of 8%, which is what Imperial Oil uses in its Cold Lake lease,
The UTF has been the laboratory for SAGD, it is instead in the 10% of Athabasca areas.
composed of 4 phases, A which was a ‘proof of concept’ pilot, Sagd Performance
B the first commercial demonstration, D the first SAGD wells The ultimate recovery of SAGD and its energy efficiency are
drilled from surface, E the first demonstration of a SAGD still very controversial issues by reasons such as the current
operation besides a depleted pad, B. Moreover there are two operation of most of the existing pads, the lack of
more well pairs (phases F&G), and the VapEx pilot called predictability of simulators, the variability of results in
DoVap. contiguous pads, and so on. Table A-1 shows the performance
details and size of the pads analyzed in this report, and Figure
1 shows the current performance of the 32 SAGD pads. In an
attempt to remove the time component, the data are plotted as
hydrocarbon pore volume produced versus hydrocarbon pore
1 The AEUB has been re-named to the Energy Resources Conservation volumes injected. Moreover this plot has an advantage that the
Board (ERCB) in 2007 slope of any line in this space is the inverse of the Oil-Steam
IPTC 12860 3

Ratio (OSR). This OSR is based on dry steam and is So*Phi2 in oil sands is between 0.20 and 0.27, with Cold
calculated using 96% or more steam quality. Out of the 32 Lake and Peace River towards the low end of the
pads analyzed only 2 have terminated operations, the A&B spectrum, and Athabasca towards the high end.
wells at Hangingstone (2 well pad) and the Phase B at the UTF Table 1 shows the overall performance of MacKay River,
(3 well pad). Moreover, the Hilda Lake information is for the Foster Creek and Hangingstone, three of the largest SAGD
two existing wells, whose steam chambers have not yet operations in Canada. Both MacKay River and Foster Creek
coalesced, so they perform presently as two separate single- have excellent cOSR, and the recovery is already of the order
well pads. of 30% despite the fact that most of their pads only have a few
years of operation. This emphasizes the statement of ultimate
2.7 recoveries in the 60 to 70% even over entire field
.2 developments. On the other hand Hangingstone, as explained
MacKay River =0 Finished
SR

2.4 .3
O

Foster Creek 0
R=
UTF (Dover) OS before, does not have the same cOSR but it is still fairly good
Hangingstone
2.1 Christina Lake at 0.31, and its recovery has already reached 37%, with only 2
Steam Injected (CEW)/OOIP

Surmont 0 .4
Firebag OS
R= well pairs out of the existing 15 finished. Notice also that these
1.8 Hilda Lake
recoveries, which already match Imperial Oil Cold Lake, are
1.5
Finished
achieved faster than the 15 to 20 years that a typical CSS pad
=0.5
OSR has in their operation [1]. Thus, SAGD does not only look
1.2
promising over the selected pads but also over large field
0.9 developments. Note that all these field developments have
been started in pay zones that are both thick and with a good
0.6 vertical permeabilit, and future expansions try to follow these
0.3
“nice” areas.

0.0 Table 1. Overall performance of three of the largest commercial


0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 SAGD operations in Canada
HCPV Produced Operation MacKay Foster Hangingstone
Figure 1. Current performance of the 32 SAGD pads used in this River Creek
analysis. OSR are calculated for dry steam
Well pairs in Operation 18 50 15
Months of Operation Pads 46/46 as of 7/108 as of 16/90 as of
There are quite a few interesting aspects that can be (Youngest/Oldest) Sept-06 May-06 Dec-06
inferred from this figure: Recovery Factor (%) 30 27 37
• Considering that only two pads are finished, and cOSR 0.46 0.39 0.31
observing the trend of the still on-going mature pads, it is
reasonable to expect ultimate recoveries around 60 to In terms of cOSR there is a widespread range of values for
70% for most of the pads. different pads as can be seen in Figure 2. Nonetheless about
• The OSR can change significantly not only between 55% of the pads have a cOSR above 0.35, and about 30% of
projects but also between pads within the same lease (see them above 0.40, Figure 2. Moreover, among the worst
MacKay River or Foster Creek), but is consistently performing pads are those of operations such as Surmont,
around 0.3 and above, with the best pads operating around Hilda Lake and Firebag: all of them have suffered a serious
an OSR of 0.5, which is outstanding. lack of steam (and they account for 6 of the 14 pads with
• Among the three most mature operations MacKay River, cOSR less than 0.35). However, that issue is now being
Foster Creek and Hangingstone, the latest is the poorest addressed at Firebag and an important improvement in
while the other two are alike, although the best pads in performance was observed from 2006 to 2007 when more
MacKay River are superior to the best pads in Foster steam became available. The rest of the pads with cOSR
Creek. The reason for the difference in performance below 0.35 are towards the low end of oil content and/or are
between Hangingstone and the other two operations small pads (2 or 3 well pairs only).
seems to be associated with lower permeability in the
reservoir, smaller pads and a slightly smaller oil content.
The average horizontal permeability in MacKay River and
Foster creek is of the order of 6000 mD while in
Hangingstone it is 2400 mD. Both MacKay and Foster
Creek use large pads (6-7 well pairs), while Hangingstone
uses smaller pads with 2-3 wells.
• The apparently poor performance of Hilda Lake is mainly
due to two reasons; each dot is a single well pad so the
heat losses are much more significant that in multi-well
pads; and there has been a lack of steam to operate the
wells smoothly, which hurts both the short- and long-term
performance. Moreover it is worth noting that the oil 2 Oil saturation times porosity is commonly used as an oil content
content in Cold Lake is not as high as in Athabasca; i.e.
indicator in thermal projects
4 IPTC 12860

cOSR
0.8
0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55
10 100% Finished
MacKay River Finished
9 90% Foster Creek
UTF (Dover)
Number of Pads (Frequency)

8 80%

Cumulative Probability (%)


Hangingstone
0.6
Christina Lake
7 70% Surmont
Firebag

HCPV Produced
6 60%
Hilda Lake
5 50%
4 40% 0.4

3 30%
2 20%

1 10%
0.2
0 0% Hilda Lake and
0.20 0.25 0.30 0.35 0.40 0.45 0.50 Surmont have
cOSR Firebag lack suffered of a lack
steam early on of steam
0.0
Figure 2. Histogram of COSR for the 32 pads analyzed, and the 0 20 40 60 80 100 120 140
cumulative probability of a pad having a given COSR
Months in Production
Figure 4. Rate of production of the different SAGD operations
Another interesting point is the short time over which the analyzed
performance of a pad stabilizes, which in practical terms
means that, if there are no serious operational issues or a lack In summary, it seems that as long as the steam chamber
of steam availability, the performance of a pad can be inferred can growth and there are no steam constrains, the SAGD
within 15 to 25 months after the start of an operation, as can performance tends to be within a certain range in terms of
be seen in Figure 3. Moreover, the cOSR becomes fairly cOSR, ultimate recovery and rate of recovery. Nonetheless
constant with time to some degree although it gets worse with there is significant variability in performance between leases,
time in some of the pads (pad B in Foster Creek), but keeps and between pads within a lease. Consequently, it is worth to
improving for others (pad C in MacKay River). look in detail at some of them to understand what makes a
project a great performer or an average performer. Among the
0.6 issues to consider are the geology, operations, pad size, well
spacing, length, and pressure among others.
0.5 Geology and Thickness
SAGD is considered by many the ‘perfect process for the
perfect reservoir’; but what is the perfect reservoir, and what is
0.4
the impact of heterogeneities in its performance?
PetroCanada’s MacKay River is a great example that geology
COSR

0.3 is the dominant control in SAGD’s performance. Figure 5


Foster Creek - A shows the performance of the four operating pads in MacKay
Foster Creek - B
0.2 Foster Creek - C River. All the pads have been operated for the same amount of
Foster Creek - D
MacKay - A time, and have gone through the same steam shortages. They
MacKay - B

0.1 MacKay - C are contiguous, from East to West, pads A to D. Both A and B
MacKay - D
Hangingstone - AB have 7 well pairs while C and D have 6 and 5 well pairs,
Hangingstone - CDE
respectively (see layout in [2]).
0.0
0 20 40 60 80 100 Figure 5 shows the performance of these four pads, from
Time (months)
their start in August 2002 to December 2005. As seen in the
Figure 3. Evolution of the cOSR with time figure, pads B and C have similar performance both reaching
an OSR of around 0.5 over that time period. The performance
Finally, the rate of recovery of the various projects tend to of the other two pads was not as good, with pad D reaching an
be fairly uniform, especially over their entire operation. Thus OSR of about 0.37 and pad A of about 0.3.
trends for most of the pads of either Foster Creek, Dover or
Hangingstone are easily established as shown in Figure 4.
Moreover, there is a general trend in all of the operations
except the ones previously mentioned that suffered from a lack
of steam. Thus, a 30% recovery has been achieved after 40 to
60 months of production for most of the operations.
IPTC 12860 5

80 high. Athabasca is in general a cleaner reservoir with a


clean sand unlike Cold Lake or Peace River that are more
heterogeneous.
Steam Injected / OOIP (%)

60

2.7

.2
=0
40 ≤ 20m

SR
2.4 ≤ 25m 0.3

O
R=
≤ 35m OS
≥ 35m
2.1

Steam Injected (CEW)/OOIP


20 0 .4
R=
OS
1.8

0 1.5
0 5 10 15 20 25 30 =0.5
OSR
Recovery Factor (% )
1.2
MacKay - A MacKay - B MacKay - C MacKay - D OSR=0.2
OSR=0.3 OSR=0.4 OSR=0.5
0.9
Figure 5. Pad performance in MacKay River from a start date of
August 2002 to December 2005 0.6

0.3
Looking at the MacKay River lease geology and the
typical geology of each pad [2], it is clear that different pads 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
have different geologies and thicknesses. Although pad A is
the thinnest, more importantly the dominant facies is a breccia HCPV Produced
Figure 6. Average thickness of the different SAGD pads analyzed
rather than a clean sand. Therefore, its oil content and vertical and their performance
permeability would be lower than in the other pads, which is
reflected mainly in the OSR, which is of the order of 0.3 rather Operational Excellence and Optimization
than 0.5 for the best pads. On the other hand, pad C is the While favorable geological conditions are paramount to a
‘perfect reservoir’ mentioned previously, very thick, rich and successful SAGD project, the operation in the field, especially
with no apparent shale baffles. The reason as to why pad D at the pad scale, can either help or hurt the performance. As
does not behave quite as well as B or C seems to be related to examples of both scenarios are pads A and B in Foster Creek
thinning of the reservoir at the heel, less wells in the pad, and [3], whose performance with time is shown in Figure 7. Pad B
different length of wells, ranging from 700 to 500 m from initially looked like a good performer but with time its
north to south. In summary, a contrast in geology is the main performance got worse (OSR around 0.3). On the other hand,
controlling factor of performance in contiguous pads that have the more mature pad A seemed to have a constant OSR of
been operated virtually identically, which highlights why 0.35, and through optimization, has improved to around 0.42.
geology is the key parameter in SAGD. Both pads will be reviewed briefly.
This brings forward three key points from a reservoir point 140
of view for a successful SAGD: 75 months of

• A continuous thick reservoir is ideal. However, the 120 operation


Steam Injected / OOIP (%)

assessment of the ‘SAGD thickness’ is one of the biggest 100

challenges in reservoir characterization as thin but 80


continuous shales can easily bring the vertical growth of
the steam chamber to a halt. Moreover, it is common that 60

logs miss these thin shales, and with vertical wells, is 40


quite difficult to establish their horizontal continuity. That
20
explains why there is little correlation between
performance and reported thickness, as shown in Figure 6. 0

• The larger the oil content, the better the performance. This 0 10 20 30 40
Recovery Factor (%)
50 60 70

point is as simple as the more oil is present in the Foster - A Foster - B Foster - C Foster - D OSR=0.2 OSR=0.3
reservoir, the less heat is wasted heating the oil sands (oil, OSR=0.4 OSR=0.5

water and sand). Thus, the SAGD projects in Athabasca Figure 7. Pad performance in Foster Creek A,B,C and D from start
will in general always be better than SAGD in Peace until December 2005
River or Cold Lake as the average saturation in Athabasca
is around 80% and porosity 30-32%, while saturation and Pad B, containing 6 well pairs, has been operated for 51
porosity in Peace River or Cold Lake are around 70% and months up to December 2005. The center well pair (both
28-30%, respectively. injector and producer) was on and off constantly early on and
• Vertical permeability plays a relative role. Shale was shut down after 20 months for 10 months. Injection
laminations within a sand will slow down or even stop the resumed at higher rate 30 months after startup, but it was shut
vertical growth of the steam chamber and the drainage of down again at about 40 months of operations. Thus, the pore
both bitumen and condensate. On the other hand, there is volume of steam injected into this well pair is fairly small
not enough evidence that kv/kh plays a significant role as (8.3% HCPV, which is 10% of the total in the pad). It is
long as the vertical continuity exists and permeability is therefore very likely that the steam chambers did not coalesce
6 IPTC 12860

leading to a decay in performance as the pad was split into a 3 River is only 125 m deep. This means that the operational
well pair pad and another 2 well pair pad, without the pressure gradient is 14 kPa/m, which is above the hydrostatic
efficiency and higher recovery that characterize larger pads. pressure gradient of 10 kPa/m. With this gradient, the process
This highlights the need for operational excellence in is trying to take advantage of a small convective flow.
successful SAGD operations, especially in steam availability However, a higher pressure than that may be unsafe from a
and the advantage of having larger pads. caprock integrity point of view because of the shallow depth.
On the other hand, as seen in Figure 7, the performance of On the other hand, Foster Creek operates at 2800 kPa, but
pad A improved significantly after 75 months of operation. its depth is 380 m. Therefore the pressure gradient is almost 8
Pad A has 4 wells, which were operated with ups and downs kPa/m, lower than hydrostatic, which means that the process
during the first 50 months. At this time, a center pair (both relies mainly on conduction for transfer heat. Thus, from a
injector and producer) was shut down permanently for geomechanical point of view, MacKay River is a high pressure
unknown reasons. However, it is likely that at this point the operation while Foster Creek is a low pressure operation.
steam chambers had coalesced so the pad’s performance Figure 8 and Figure 9 show the performance of different
remained the same. After 75 months the other center injector operations according to their injection pressure and pressure
was shut down, but the producer of that pair remained in gradient, respectively. Among the issues that these figures
operation. Thus, by taking advantage of a big single steam highlight are:
chamber and all the heat stored in the unswept bitumen • There is no clear trend between either injection pressures
between well pairs, a significant improvement in performance or pressure gradients and performance. The performance
was achieved by injecting steam into the side wells and of the different MacKay River pads depends more on the
producing oil from the side wells plus a center well. Note that geology than the pressure.
shutting the center injectors in mature pads accelerates the • Both high and low pressure pads each have excellent
recovery of unswept oil between chambers, as was observed in performers and ‘relatively poor’ performers (Figure 8).
pad A at Foster Creek [3]. • High pressure gradient operations are not better than low
Operational Pressure pressure ones (Figure 9). Therefore, a convective
The issue of high pressure versus low pressure is a component in the heat transfer and a small gain in
constant debate for SAGD operators. The advantages and absolute permeability due to compressibility do not seem
disadvantages of low pressure SAGD are: to be important factors.
• At low pressure, the steam has more latent heat and the
operating temperature in the reservoir is lower, so that the 2.7

.2
energy efficiency is better, which leads to a better OSR.

=0
1750 kPa

SR
0.3
• At low pressure, heat transfer from the steam chamber 2.4

O
2250 kPa
R=
2800 kPa OS
4800 kPa
into the formation has to rely mainly on conduction 2.1
Steam Injected (CEW)/OOIP

.4
because the mean effective stress remains unchanged or OS
R=
0

1.8
increases due to the thermal stresses. Therefore, no
advantage is taken of the non-linear compressibility and 1.5
=0.5
geomechanical behavior of oil sands. OSR

• Viscosity is not reduced as much due to the lower 1.2

temperature, which means that the drainage rate is 0.9


smaller. According to Butler [5], the production rate is
proportional to the square root of the mobility: 0.6

k 0.3

v 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
The advantages and disadvantages of high pressure SAGD HCPV Produced
are: Figure 8. Impact of operational pressure at different SAGD
• At high pressure, the latent heat is smaller but the operations
operating temperature in the reservoir is higher, so the
energy efficiency is lower, and so is the OSR.
• At high pressure, the transfer of heat into the formation
has a component of convection due to compressibility,
because the mean effective stress becomes lower. This is
because the pressure front travels ahead of the thermal
front.
• Viscosity is lower, therefore the rate of production is
higher.
Therefore the picture is not as straightforward as just latent
heat. Moreover, the definition of high or low pressure is
relative. MacKay River operates at 1750 kPa, which at first
glance may seem like a low pressure SAGD, but MacKay
IPTC 12860 7

at lower pressures, and compare the efficiency of high versus


2.7 low pressure. Figure 10 shows the change of viscosity at the

.2
UTF Phase B for different pressures, and how it impacts the

=0
7.4 kPa/m

SR
2.4 0 .3

O
12.9 kPa/m R=
14 kPa/m OS time to achieve 55% bitumen recovery (7.5 years at 2250 kPa
15 kPa/m
2.1 16 kPa/m in the field). These times were calculated by adjusting the rate
Steam Injected (CEW)/OOIP

0.4
OS
R= of recovery proportionally to the change of the square root of
1.8
mobility, according to Butler’s model [5].
1.5 11.5
=0.5
OSR
Time to achieve a recovery
1.2 11.0 factor of 55% for different
viscosities, due to different 1500 kPa
0.9 10.5
operational pressures

Viscosity (cP)
0.6 10.0
1750 kPa
0.3
9.5

0.0
9.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
HCPV Produced
8.5
Figure 9. Impact of operational pressure gradient at different 2250 kPa
SAGD operations 8.0
7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0
This raises the question of the SAGD operating pressure of Time (years)

a project as the results show to be pressure independent. In


Figure 10. Change of viscosity at different pressures and its
order to answer that question, the performance of the UTF impact to achieve 55% bitumen recovery at the UTF Phase B
Phase B has been matched to a theoretical model developed by (Athabasca) assuming that the production rate is proportional to
Edmunds and Peterson [6] that predicts the cOSR as a function the square root of mobility
of recovery. This model was calibrated for different
parameters, and a comparison was made with the field data. Figure 11 shows the impact of injection pressure on cOSR
In the Edmunds and Peterson model, the steam at 55% bitumen recovery. The blue line assumes that the
consumption is the ratio between the total heat consumption, change in injection pressure does not change the production
H, which is the amount of heat stored in the reservoir plus the rate; in this case, low pressure is much more efficient than
amount of heat loss to the caprock, and the latent heat of high pressure. On the other hand, the pink line also takes into
vaporization, Hlv: consideration the viscosity changes; it takes longer to achieve
a 55% recovery factor. In this case, the differences in cOSR
H
S= are much smaller between high and low pressures, while the
H lv time to achieve the same recovery is larger. An additional
where parameter such as net present value (NPV) would be needed to
(
H = AΔT Cvr hη s + k t C vo t ) make a final decision, but high pressure looks more promising.
Moreover, gains in absolute permeability and thermal
The cumulative oil production from a fully drained steam efficiency due to high pressure (heat transfer ahead of the
chamber is given by: steam chamber by convection) were not included.
O = Ahη sφ ( S oi − S or ) Nonetheless, the author favors high-pressure injection in the
early stages of steam injection. As the steam chambers
Thus, the cSOR is given by: coalesce and grow along the full length of the wells, a low
S ⎛ ΔT ⎞⎛⎜ kC t⎞ pressure injection might be more efficient, as a large amount
cSOR = = ⎜⎜ ⎟⎟ C vr + t vo ⎟ of heat is already stored in the reservoir, and a significant
O ⎝ H lvφΔS 0 ⎠⎜⎝ hη s ⎟⎠ amount of the recovery comes from unswept areas between
where A is the planar area of the steam chamber, h the well pairs. Moreover, some of the injector wells can be
height above the producer, ηs the sweep efficiency, Cvr and Cvo shutdown.
the heat capacities of the reservoir and overburden,
respectively, kt the overburden thermal conductivity, t the
time, and Soi and Sor the initial and residual oil saturations,
respectively. The model is simple but it captures the essence
of the process based on the recovery factor, geometrical
properties, and reservoir, caprock and steam properties. The
model can only match mature operations because it assumes
that heat losses to the overburden occur from the moment
SAGD starts, so it penalizes the early OSR.
After calibrating the model and knowing that the
operational pressure at the UTF Phase B was 2250 kPa [4], it
was possible to change both the pressure and rate of recovery
8 IPTC 12860

0.46
The improvement in performance of large pads is due to
7.44 years Recovery Factor is 55% for All Cases the efficient drainage of bitumen between well pairs. Such an
0.45
efficient drainage is the result of both a superposition of heat
0.44
7.44 years
sources that warms the oil between well pairs faster, and the
0.43 ability to drive that oil toward the producers in mature pads
where the steam chambers have coalesced. On the other hand,
cOSR

0.42
10.45 years steam chambers can become very flat on the sides, so gravity
0.41
9 years
drainage is limited even if the steam chamber is still growing
0.40 outwards. Thus, in mature pads, the confined wells produce at
0.39 larger rates than the side wells, something observed in almost
7.44 years
0.38
every mature pad analyzed.
1500 1600 1700 1800 1900 2000 2100 2200 2300 Well Spacing and Length
Injection Pressure (kPa) Well spacing and length are both controversial issues that
Pressure Pressure & Adjusted Viscosity (Mobility) the present study was not able to resolve completely as the
Figure 11. Potential impact on the cOSR at 55 % recovery in the information is scattered. The operators, especially for well
UTF Phase B by changing the injection pressure lengths, usually present it as averages for field developments.
There is a clear trend for operators to use 100 m as a typical
PAD Size well spacing, as seen in Figure 13. Figure 13 indicates that
Well pair interaction between pads has been shown to be spacing of 150 m leads to poorer performance. The two
important in the different pads analyzed at EnCana’s Foster operations using 150 m well pair spacing are Firebag, which is
Creek. Therefore, it is expected that pad size could have a too young to make a judgment yet, and Surmont, which has
meaningful impact on the performance of SAGD operations. suffered from a lack of steam. On the other hand, analysis of
Figure 12 shows that most of the best performers are pads with the UTF Phase B shows that a spacing of 90 m seems most
4 or more well pairs. Thus, the UTF Phase B, which is an economical, and was the basis for the UTF Phase D [6]. This
excellent operation with an excellent reservoir quality, is a 3 seems to indicate that 100 m well spacing is a good choice
well pair pad and may not achieve the level of performance of unless recovery needs to be accelerated significantly, which
bigger pads as in MacKay River, which is a nearby operation. will require closer spacing.
Moreover, it implies that single well pair pilots, such as Hilda
Lake, which are good indicators of whether the steam chamber 2.7

.2
can develop and SAGD operations are feasible, are not good

=0
70 m

SR
2.4 0 .3

O
90 m
indicators of the full performance of SAGD operations, and 100 m OS
R=

will underestimate the cOSR and ultimate recovery that can be 2.1
150 m
Steam Injected (CEW)/OOIP

0.4
expected in multi-well pads. The pads with 10 well pairs OS
R=
1.8
belong to Firebag, and they are underperforming due to two
major reasons: 1.5
• lack of steam for the entire project: not all the 10 well OSR
=0.5

1.2
pairs have been steamed in the pads
• distance between well pairs is 150 m (much larger than 0.9
the typical 100 m spacing of other operation): the steam
chamber will take longer to coalesce. 0.6

0.3
2.7
.2

0.0
=0

2 Wells
SR

2.4 0 .3 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
O

3 Wells R=
4 & 5 Wells OS
6 Wells HCPV Produced
2.1 7 Wells Figure 13. The role of well spacing in the performance of SAGD
Steam Injected (CEW)/OOIP

10 Wells 0.4
R= operations
OS
1.8

1.5
On the issue of well length, the picture is a lot more
OSR
=0.5 unclear as steam chambers always grow in 3D, and it takes
1.2 time to take advantage of the full well length. Moreover, if
0.9
insufficient steam is available, it will be difficult to use the full
well length. Thus, there is no point in using very long wells if
0.6 the supply of steam is insufficient: it can lead to dramatic
situations such as the 1000 m well pair at Surmont, which is
0.3
severely underutilized as shown by their seismic images and
0.0 temperature observation wells [7].
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Because of the unique relation between steam pressure and
HCPV Produced
temperature, SAGD projects must be designed for a given
Figure 12. Impact of number of well pairs per pad on the efficiency
of SAGD operations operating temperature or pressure, which will fix the other
variable. Then, 2D simulations can give a first-order estimate
IPTC 12860 9

of how much steam will be required to operate the early stages 60% depending on the shale distribution and bitumen
of SAGD. As mentioned earlier, lower temperatures (and content. Thus, the issue of recovery factor can be seen as:
pressures) can be used once the process is mature. Moreover, o SAGD will have a better and faster recovery than
once the steam chamber is large, it is almost impossible to CSS over a single pad where a steam chamber
maintain the original pressure, so any SAGD project usually can be developed.
ends up operating at a lower pressure than at the early stages. o CSS will be better than SAGD as a single full
SAGD vs. CSS field option because the technology is more
The two main technologies to develop oil sand deposities with robust. Nonetheless CSS recovery is still
viscosities above 104 cp are SAGD and CSS. Both significantly affected by the presence of shale
technologies operate very differently, and rely on different baffles or thief zones.
driving mechanisms (although gravity plays a significant role • It seems that, while SAGD is viable only with thin basal
in mature CSS). Among the bigger differences in performance mobile water and not viable with top water or thick basal
and particularities of each technology are: water, CSS is viable but its performance is diminished
• Recovery factors in SAGD showed a trend towards the substantially. In the case of a gas cap, SAGD is not viable
60-70% as long as the steam chamber develops. On the while CSS is. Imperial Oil has successfully carried out
other hand, recoveries at Imperial Oil Cold Lake are CSS with a gas cap by injecting steam into all the pads
between 10 and 60% depending on shale content and before any production occurred. However, recovery
distribution, and oil content [1]. The full field average for expectations are 30 to 40% lower than for similar pads
CSS at Cold Lake is 38%, while the current recovery is with no gas cap [1].
30% for the full leases in MacKay River and Foster • A tailored approach to different reservoir areas seems to
Creek. However, in these two projects all the wells are be the ideal approach. However, the shortcomings of such
still in operation, and some of them are even fairly new. approach are the quality and detail achievable presently
Similarly, Hangingstone has already achieved a 37% for reservoir characterization, especially when flow
recovery to date with 13 out of 15 well pairs still in barriers at the reservoir scale have shown to impact
operation. All these SAGD operations are much younger negatively operations of both SAGD and CSS, although in
than Imperial Oil’s CSS, and a much higher overall different degrees. Moreover, the issue of thief zones
recovery is expected. Notice that the weight cut-off in remains presently unresolved as CSS, the only viable
OOIP for most of the SAGD operations is 10%, while technology currently under such conditions, performs
Cold Lake uses 8%. Therefore the comparison in recovery more poorly than when there are no thief zones.
terms is not straightforward, but one way to interpret this • Optimization seems more likely in SAGD as the process
difference is that CSS can be applied in leases that are not is based on gravity, which is a stable process. On the other
as rich as the leases where SAGD is being deployed. hand, CSS under fracturing conditions, which is necessary
• cOSR in the best CSS (Imperial Oil’s) is 0.34 (dry steam in bitumen reservoirs, is very unpredictable. Steam
basis) on average over the full field, which is comparable distribution is controlled by both reservoir and stress field
to an average SAGD such as Hangingstone. However, it heterogeneities, which are impossible to quantify with the
cannot compete with the best SAGD projects such as necessary precision and coverage to control the process.
Foster Creek or MacKay River that have cOSR of 0.39 Moreover, fracturing in oil sands is poorly understood, yet
and 0.46, respectively, in their field developments. key to CSS.
However, SAGD can only be deployed in areas of the Conclusions
reservoir with good vertical connectivity, no gas cap, A review of SAGD field performance has been carried out in
small mobile water zones and relatively thick pay zones. order to understand why SAGD has become the preferred
• All of the SAGD operations analyzed except Hilda Lake technology to develop the oil sands in-situ over the last ten
are in the Athabasca deposit, whose in-situ viscosity is years, and to try and gain insights as to what makes a SAGD
around 106 cp, while most of the CSS projects are in project successful. From this study it was established that:
reservoirs with in-situ viscosity around 105 cp or less. It • The key parameter for a successful SAGD process is the
should be noted that the GOR in both Cold Lake and geology. SAGD targets must be reservoir areas with
Peace River bitumen is between 8 and 10 m3/m3, while average pay thickness above 15 m, good vertical
the Athabasca bitumen is mostly a dead oil. In addition, communication and no thief zones. If one of these
gas drive has been identified as a relevant mechanism for conditions is not meet, the performance of a SAGD
recovery in CSS [8]. Therefore, SAGD has been deployed project will be hindered up to the point where it may not
successfully in reservoirs with a dead oil, while CSS is be viable. Moreover, slight differences in geology can
unlikely to work in a dead oil reservoir. easily reveal contrasting performances even over
• Despite the fact that CSS is a more robust process, contiguous pads.
especially in the presence of shale baffles, these baffles • The second most important parameter for a successful
affect the energy efficiency and ultimate recovery as well, SAGD is the operation itself. SAGD operations
and their effect can be significant depending on their performance is badly compromised by mainly lack of
continuity. Evidence of that has been shown by Imperial steam, but also by long boiler shut downs, and by the loss
Oil [1] where recoveries can fluctuate between 10 and of confined injectors early on in the process, which has
led to splitting big pads into small pads.
10 IPTC 12860

• The best and large field operations, such as MacKay Acknowledgements


River or Foster Creek have a cOSR of 0.46 and 0.39, The author thanks Shell International Exploration &
respectively, with recoveries already in the 30% with all Production B.V. for permission to publish this paper.
the pads still in operation and with a significant life ahead. References
Operations with a more average performance, such as 1. Imperial Oil, 2006. Cold Lake Annual Performance Review.
Hangingstone, have a cOSR of 0.31, a recovery of 37%, www.ercb.ca/ November 28, 2006
and the operation is still on-going. Therefore, despite the 2. Petro-Canada, 2006. MacKay River Performance Presentation
Approval No. 8668. www.ercb.ca/ October 27, 2006.
fact that only 2 out of the 32 pads analyzed in this study
3. Encana, 2006. Foster Creek Development. www.ercb.ca/ May 30,
are finished, it is not unreasonable to expect ultimate 2006.
recoveries of SAGD operations around 60 to 70%. 4. Devon, 2005. Dover SAGD Progress Review. www.ercb.ca/
• The dry cOSR of on-going SAGD projects in Canada is April 28, 2005.
between 0.30 and 0.50, with 55% of the pads above 0.35, 5. Butler, R. 1997. Thermal Recovery of Oil and Bitumen, 528 p.
and about 30% of them above 0.40. 6. O’Rourke, J.C., Begley, A.G., Boyle, H.A., Yee, C.T., Chambers,
• If areas of a reservoir are appropriate for SAGD, indeed J.I. and Luhning, R.W., 1997. UTF project status update May 1997.
SAGD must be the preferred technology to develop these 48th Annual Technical Meeting of the Petroleum Society in Calgary,
Alberta, Canada, June 8-11.
areas as its performance can be outstanding, superior to
7. Edmunds, N. and Peterson, J., 2007. A unified model for
CSS, and with less issues such as wellbore and caprock prediction of CSOR in steam-based bitumen recovery. 8th Canadian
integrity, steam production at surface and less sand International Petroleum Conference, Calgary, June 12-14.
production. However, the biggest issue when choosing 8. ConocoPhillips, 2007. Surmont Pilot Performance and Resource
SAGD over CSS is whether the reservoir characterization Management Presentation to the EUB. www.ercb.ca/ April 23, 2007
is good enough to identify the presence of thin shale 9. Batycky, J.P., Leaute, R.P. and Dawe, B.A., 1997. A mechanistic
baffles that are continuous on a reservoir engineering model of cyclic steam stimulation, SPE 37550.
scale, and may halt the development of the steam
chamber.
Among specific issues regarding SAGD operations, it was
concluded that:
• High pressure seems to be more efficient than low
pressure as drainage occurs faster, so high recoveries are
achieved faster although with a slightly smaller cOSR
than low pressure SAGD. The author likes high pressure
SAGD for start-up, and move to a lower pressure once the
pad is mature and the steam chambers have coalesced.
• Larger pads with 5 well pairs or more are in general more
efficient as it is possible to make better use of the heat
trapped between the well pairs, and adapt the operation
when steam chambers coalesce in order to produce the
unswept oil trapped between the well pairs, which is
usually mobile at that point and can be driven towards the
confined producers.
• There is a trend among most of the operators to use a 100
m well pair spacing, which is supported by the
observations from the UTF Phase B. Coalescence of the
steam chambers with larger spacing is possible, but may
take a longer time.
• The choice of well length must be done jointly with the
development of facilities, as longer wells will require
more steam to make use of their full length. Operations
with well lengths between 500 and 750 m have so far
been successful. However, it is difficult at this time to
make a judgment on those that are 1000 m long. They are
either too young or have had steam availability issues.
Appendix
Table A-1. Performance and pad details of the SAGD operations analyzed
Project Operator PAD No. OOIP Bitumen Water RF (%) COSR Steam HCPV Well Average Well Start Latest Months in
Wells (1000m3) Prod. Prod. Injected/ Produced Spacing Thickness Length Production Data Production
(1000m3) (1000m3) OOIP (m) (m) (m)

Christina Lake EnCana A2+A3 2 900 518 42.8 0.48 1.21 0.58 90 25 700 Oct-02 Apr-07 54
A4 1 450 159 35.3 0.38 0.92 0.35 90 25 700 Oct-03 May-06 31
A5+A6 2 683 160 23.4 0.38 0.61 0.23 90 20 450 Sep-04 May-06 20
Foster Creek EnCana A 4 1746 1048 60.0 0.43 1.41 0.60 100 30 700 May-97 May-06 108
B 6 2844 768 27.0 0.29 0.92 0.27 100 27 700 Nov-01 May-06 54
C 6 3114 1183 38.0 0.40 0.95 0.38 100 30 700 Nov-01 May-06 54
D 6 3237 1360 42.0 0.43 0.99 0.42 100 30 700 Nov-01 May-06 54
E 6 3133 1128 36.0 0.36 0.99 0.36 100 25 700 Nov-01 May-06 54
Exp1 6 2690 565 21.0 0.40 0.53 0.21 100 24 700 Nov-03 May-06 30
Exp2 4 1960 235 12.0 0.42 0.28 0.12 100 26 700 Nov-04 May-06 18
F 6 3502 210 6.0 0.50 0.12 0.06 100 31 700 Sep-05 May-06 8
G 6 2574 180 7.0 0.48 0.15 0.07 100 23 700 Oct-05 May-06 7
Surmont ConocoPhillips A+B 2 1128 278 24.6 0.33 0.75 0.25 120 47 300 Oct-97 Mar-07 113
C 1 1214 68 5.6 0.25 0.22 0.06 150 41 700 Jun-00 Mar-07 81
MacKay River Petro-Canada A 7 2073 490 1583 23.6 0.27 0.89 0.24 100 19 700 Nov-02 Sep-06 46
B 7 3578 1171 2522 32.7 0.48 0.68 0.33 100 25 700 Nov-02 Sep-06 46
C 6 4358 1169 2447 26.8 0.53 0.50 0.27 100 34 700 Nov-02 Sep-06 46
D 5 2581 790 1742 30.6 0.36 0.85 0.31 100 29 700 Nov-02 Sep-06 46
Hanginstone JACOS A+B 2 757 555 1848 73.3 0.29 2.52 0.73 100 20 500 Jun-99 Dec-06 90
C+D+E 3 998 460 1420 46.1 0.29 1.59 0.46 100 20 750 Jul-00 Dec-06 77
F+H+I 3 949 469 1299 49.4 0.32 1.52 0.49 100 18 750 Feb-02 Dec-06 58
J+K 2 996 277 776 27.8 0.34 0.82 0.28 100 20 750 Aug-03 Dec-06 40
L+M+N 3 1217 280 662 23.0 0.39 0.60 0.23 100 22 700 Jul-04 Dec-06 29
O+P+Q 2 808 101 298 12.5 0.32 0.39 0.13 100 22 700 Aug-05 Dec-06 16
Dover Petro-Canada Phase B 3 1065 697 1661 65.4 0.42 1.57 0.65 70 16 500 Jan-93 Feb-05 145
Phase D 2 851 440 1019 51.7 0.35 1.47 0.52 90 22 750 Jun-96 Feb-05 104
Phase E 1 504 155 350 30.8 0.35 0.87 0.31 90 22 750 Jun-99 Feb-05 68
Phase F&G 2 930 46 125 4.9 0.27 0.18 0.05 90 22 750 Dec-03 Feb-05 14
Firebag Suncor Pad1 10 19100 2157 6.5 0.32 0.35 0.11 150 44 900 Feb-04 Feb-07 36
Pad2 10 18000 1321 3.8 0.26 0.28 0.07 150 39 900 Feb-04 Feb-07 36
Hilda Lake BlackRock I1P1 1 509 170 702 33.4 0.29 1.15 0.33 - 23 1000 Oct-97 May-07 115
I3P3 1 509 102 399 20.1 0.22 0.91 0.20 - 23 1000 Oct-00 May-07 79

Você também pode gostar