Sunteți pe pagina 1din 10

PETROLEUM SOCIETY

CANADIAN INSTITUTE OF MINING, METALLURGY & PETROLEUM PAPER 2007-154

A History Match of CSS


Recovery in the Grosmont
J. NOVAK, N. EDMUNDS, M. CIMOLAI
Laricina Energy Ltd.

This paper is to be presented at the Petroleum Society’s 8th Canadian International Petroleum Conference (58th Annual Technical
Meeting), Calgary, Alberta, Canada, June 12 – 14, 2007. Discussion of this paper is invited and may be presented at the meeting if
filed in writing with the technical program chairman prior to the conclusion of the meeting. This paper and any discussion filed will
be considered for publication in Petroleum Society journals. Publication rights are reserved. This is a pre-print and subject to
correction.

Abstract Introduction
The Upper Devonian Grosmont Formation, located in the In 1977 the Union Oil company initiated what would
Alberta oil sands region, contains an estimated 300 billion eventually be a decade of steam drive, combustion floods, and
barrels of bitumen. The reservoir in the Grosmont is steam stimulation tests in the Grosmont carbonate formation.
characterized by heavily karsted dolomite. The rock is fully Classified as the Buffalo Creek operations, this paper will focus
charged with oil and has potentially very high bulk on the single well cyclic steam stimulation tests (“Huff and
permeability. These properties make the Grosmont prospective Puff”) performed from 1980-1981. This period consisted of 5
for steam stimulation and extraction of its bitumen reserves. sequential injection and production cycles executed in well
Throughout the 1970’s and 80’s, a variety of pilots were 10A–5–88–19W4. Drilled to a depth of 300m, well 10A was
operated in the Grosmont. The Buffalo Creek pilot, operated by perforated in the interval 290mKB – 300mKB, a highly porous
the Union Oil company from 1980 – 86, utilized a cyclic steam (25-30%), bitumen saturated (85%) streak in the Grosmont 2. 4
stimulation (CSS), or “huff and puff”, injection scheme. The observational wells were also drilled within a 30m radius of
pilot achieved peak oil rates of 70m3/d and a low cycle SOR of well 10A in an effort to monitor the advance of the steam front
3.6 (see Figure 1). Temperature response was observed with
The first 2 years of the Buffalo Creek pilot were history thermocouples placed at several intervals between the depths of
matched, consisting of 5 injection/production cycles. The 285 and 300m in each well.
primary evaluated data were daily produced oil rate and
producing BH temperature. After analyzing the pilot’s Figures 2 and 3 depict the producing day oil and water rates
performance and simulated reservoir model behavior, it has and the producing BH temperature in the field for each of the 5
been concluded that the reservoir in the Grosmont can indeed cycles from 1980-1981. A comparison of simulated production
be characterized with high bulk permeability. This aspect of the data with these data sets will be the primary criteria in
formation was the catalyst of the production volumes achieved determining the accuracy of the simulator reservoir model.
in the Buffalo Creek pilot and is a key feature that must be
considered in any future recovery operations.

1
Steam Chamber Profile of the Pilot Several other reservoir features remain uncertain and will be
tested in the history match models; such as the top depth
The observational wells drilled in an equilateral pattern reached by the steam chamber, reservoir compressibility,
around injection well 10A were intended to determine the speed relative permeability and bulk permeability.
and directional penetration of the steam front via temperature
response in each well. First cycle steam infiltration occurred The greatest question mark of the aforementioned reservoir
rapidly at observation well 10D and then at 10B during the first features is the bulk permeability. It is understood that the
few days of injection. The initial response at wells 10D and geology in the carbonates contains average porosity with large
10B were over one narrow interval 288 to 289.5 mKB of vugs and fractures throughout. Core analysis permeability of
porosity 35-40%. Temperature response at wells 7E and 10C carbonates is often insufficient as it is restricted to measuring
did not occur until the 22nd day of injection. After 2 cycles of the permeability of the single core sample. In a more
injection, the temperature distribution in the Grosmont 2 homogeneous reservoir, this would be a sufficient calculation of
depicted a slight steam override existed toward wells 10C and bulk permeability. However, carbonate reservoirs, such as this
10D (North-Northwest) while toward 7E (South) steam travel one, are typically highly heterogeneous and contain many
was downward. After 4 cycles the temperature response from connected vugs and extended fractures. These features
surrounding observation wells confirmed that the heat potentially contribute several orders of magnitude to the overall
distribution was multidirectional. However; well 7E still permeability of the reservoir. If one includes lost cores as
showed no signs of communication with the hot water zone and possible large vugs or “caves” in the reservoir, 101 - 104+ Darcy
only slight conduction heating was evident here. After 5 cycles levels of permeability tend to exist in extensive regions within
of injection, the temperature response in the observation wells carbonate reservoirs.
indicated that the steam front never reached any of the wells.
However; wells 10B, 10C, and 10D were located in the hot Relative permeability in any reservoir is generalized at best,
water zone. Well 7E only ever showed slight conduction with few correlations to pick from based on reservoir specific
heating. lithologies and critical saturation values. It is known that this
particular reservoir has a dolomite lithology and is oil-wet;
We can conclude from the described temperature profile and however, critical water saturations and residual oil saturations
recorded BH pressure data that the reservoir is highly remain unknown. Reservoir models used for the history match
heterogeneous in porosity and especially permeability. While will utilize Honarpour correlations with variable SWC and
the porosity seems to be fairly uniform in a radial direction, as SORW values.
well logs of the observational wells would suggest, the vertical
porosity distribution spans over a large range of values.
However; considering that the temperature distribution was
neither radially or vertically uniform, we can assume that the The History Match Model
permeability in the reservoir is both radially and vertically Several parameters from the Buffalo Creek field operations
heterogeneous. The initial temperature response in wells 10D were constrained in the history match simulations. Steam
and 10B was most likely through a vug that spanned between injection rates and production liquid rates were controlled to be
the observation wells and well 10A. We can also assume that equal to those achieved in the field. Initial model injection rates
observation well 7E never showed anything higher than also had the injection BH quality set to the averaged reported
conductive heating because it is located out of the localized well head qualities for each cycle; however, updated models
interconnected fracture and vuggy permeability “network”. A
now account for heat loss in the well during injection. The heat
lack of thermocouple data at shallower depths than 285m in any
loss during steam injection was accounted for in updated
of the drilled wells prevents us from knowing how high the
reservoir models because it is believed to be substantial enough
steam chamber penetrated into the Grosmont D. The
comparatively low injection BH pressure data also supports that to cause a decrease in steam quality, significantly affecting the
this reservoir can be characterized with a massive reservoir amount of heat being injected into the reservoir. Current
permeability in all three dimensions. reservoir models include bottom hole steam quality as an input
variable within reasonable bounds. Further work is planned to
more accurately evaluate well bore heat losses.
Reservoir Description The history match itself will consist of comparing producing
From the Buffalo Creek progress reports, well logs, core oil rates and producing bottom hole temperatures between those
analysis and engineering calculations the ensuing petrophysical generated from the reservoir simulation and the data attained in
data have been determined for the reservoir and are summarized the pilot. A match of these parameters within acceptable error
in Table 1. should produce a representative model of the Buffalo Creek
reservoir.
The rock matrix in this reservoir is primarily dolomitic.
Digitized well log data that was correlated against a limestone Reservoir Model 1
scale was adjusted for the dolomite grain density in the Several reservoir models were simulated with varying
reservoir from which we were able to calculate an average relative permeability curves and uniform vertical and horizontal
porosity. Similarly, from the reservoir grain density we were permeability descriptions. Initial models determined that
able to determine thermal conductivity and heat capacity values. considering the steam front never reached any of the
The bitumen insitu is exceptionally viscous, categorized as 7 observational wells, for the Buffalo Creek operations to achieve
API. Figure 4 is a plot of the well logged porosity, adjusted for the peak oil rates that they did the steam chamber must of
a dolomite matrix, vs. depth of the injection/production well penetrated the tight zone separating the Grosmont C and D.
10A-5-88-19W4. From the plot we can observe there is little This assertion is neither confirmed or refuted by the progress
uniformity in the porosity over the entire depth of the reservoir. reports due to the lack of thermo couple data above the tight

2
zone. So with this knowledge in mind we were able to The reservoir parameters and well flow data summarized in
determine a representative reservoir size and produce our best Table 3 were selected as input variables for the program
history match with generalized permeability and reservoir because of their uncertainty and influence they may have had on
characteristics, listed in Table 2. the Buffalo Creek field production. Based on low and high end
values of rock compressibility’s we knew existed for similar
An analysis of Figure 5 reveals that, while it isn’t an reservoir descriptions, we were able to place bounds wherein
overwhelming success, the general production trends in the field the software would select a value that would fairly accurately
are being re-produced in Reservoir Model 1. With some represent the compressibility of the Buffalo Creek reservoir.
adjustment to relative permeability and cycle injection qualities, The critical water saturation and residual oil saturation remain a
the simulated oil rate could be fine-tuned to even less error. An mystery and could conceivably be any number physically
examination of the BH temperature match in Figure 6 portrays a possible for the in-situ geological description. Therefore, the
less optimistic situation. ranges for both saturations encompass all possible values
physically possible from which the program will select and
The plotted BH temperature data recorded in the Buffalo generate relative permeability curves from Honarpour
Creek pilot depict a sharp cooling trend during production. This correlations. Once more is understood about the reservoir, the
suggests that the majority of the heat dispersed away from the limits may be adjusted. Wellhead steam qualities are available
well bore during injection. In comparison with the simulated for each day of injection; however, heat losses from the
producing BH temperature, we can see a lot of the heat wellhead to the well bore will reduce the quality of the steam
remained around the well bore during injection and was present upon injection. While a well bore heat loss model is currently
in production cycles. An investigation of the steam chamber being constructed separately, the heat loss in the well is being
morphology in the simulator does show the steam front accounted for by constraining the algorithm to select cycle
penetrated through the tight zone separating the two Grosmont qualities between reasonable minimum values, respective to
formations and reached the top of the 37m reservoir during first each cycle, and the reported average well head quality for each
cycle injection. However, because the permeability is pretty cycle.
much uniform throughout the model, the steam chamber itself is
fairly uniform from the bottom of the perforations to the top of Equations (1) and (2) in Table 3 represent the vertical and
the pay zone. This uniformity leads the well bore to remain at a horizontal matrix permeability for the reservoir model. To
fairly constant temperature in the simulator throughout the account for this reservoir potentially ranging in permeability
injection and production cycles. from milidarcies to several thousands of darcies, an exponential
relationship between permeability and porosity was developed.
After several simulations with various isotropic vertical and The coefficients of these correlations (av, bv, ah, bh) are what
horizontal permeability, it became apparent that an accurate BH will be manipulated by the algorithm. Limits were chosen for
temperature match wasn’t going to be achieved using a each coefficient based on the resultant range of permeability’s
reservoir model with a homogeneous permeability distribution. generated for respective reservoir porosity values. While this
There were also several other variables in the simulator model correlation suggests that there is a direct relationship between
that had yet to be addressed and could have a large effect on the porosity and permeability in the reservoir, we know that is most
temperature profile in the reservoir. Dealing with these likely not the case. The purpose of the correlations is to serve
solely as a vehicle for generating a heterogeneous permeability
parameters and introducing heterogeneous reservoir
distribution over a large range of values. Figure 7 is a visual
permeability into the simulator model would be too inefficient
representation of the generated permeability correlation,
with current manual adjustment methods. Therefore a genetic
complete with the set lower and upper limits.
algorithm was implemented in Java for iterative adjustment of
all variables and reservoir parameters at once. Preliminary K_barrier in Table 3 represents the permeability of the tight
simulations with this method have produced some remarkably zone that exists between the Grosmont C and the Grosmont D
promising results. formations. The permeability of this zone is important because
it will be a critical factor in determining the vertical penetration
Reservoir Model 2 of the steam chamber and the resultant heat distribution away
The Java program that was developed for the second from the well bore. The permeability of such a zone is largely
reservoir model is patterned after species evolution. Initially unknown even though the area has extremely low porosity.
the program was created for reservoir optimization, but with This may lead to believe an associated low permeability as well;
some input file and simulator alterations the program can now however, one can assume the vertical fractures that exist in the
also be utilized for history matching. Essentially, the algorithm region are present in the barrier as well. The range of the
starts by generating values for pre-defined input variables permeability for this variable in the algorithm is from 0.1md to
within maximum and minimum limits that are also classified by 100D and is programmed to be selected in logarithmic
the user. Then the model, or “child”, is simulated with the partitions.
randomly generated parameters and scored based on some
objective function. In this case, a suitable scoring function Now that all of the input variables have been defined, the
would be a measure of the error between the simulated and field model is ready to be simulated. Figures 8 and 9 are output of
oil rates and BH temperatures over a designated period. After the simulation model, with the input data located in Table 3,
several models (“children”) have been simulated and scored, the matched against the Buffalo Creek field production.
program will pick variables at random from the reservoir
models with the highest scores (lowest error or, “strongest An analysis of Figure 8 reveals that this reservoir model is
children”) and repeat the process. Table 3 is an example of a not an accurate representation of the Buffalo Creek reservoir in
reservoir model with parameters generated by the program. terms of produced oil rate. The cycle 5 match contains the
largest discrepancy, with largely varying peak oil rates and
overall production profile. The match of cycle 3 is the most

3
accurate, with a relatively close peak oil rate match compared to permeability values for each individual grid block and injection
the other cycles and a similar overall production profile to that qualities calculated from well heat loss calculations. This
achieved in the field. Of more interest generated from this degree of resolution and well flow accuracy combined with a
model is the match of BH temp simulated and the BH temp completed and improved genetic algorithm should resolve the
produced in the field. Buffalo Creek reservoir and provide valuable insight for future
An examination of Figure 9 reveals a strikingly accurate bitumen recovery methods in carbonate formations.
match between the Buffalo Creek BH temperature data and the
BH temperature data simulated from the reservoir model
generated from the genetic algorithm. Each cycle shows the Acknowledgements
same declining temperature profiles and similar final
temperatures for three of the four cycles. The simulated We would like to thank Laricina Energy Ltd. for the
reservoir never quite reached the initial producing temperature permission of publishing this work at the Petroleum Society’s
attained in each Buffalo Creek cycle; however, this is a small 8th Canadian International Petroleum Conference, Calgary
blemish in an exceptionally accurate BH temp match. The 2007.
accuracy of this match strongly supports the theory of an
extensively heterogeneous permeability distribution exists in NOMENCLATURE
this reservoir and is in all probability a regional feature.
BH = Bottom hole
SWC = Critical water saturation
Conclusion SORW = Residual oil saturation
kV = Vertical permeability
The Buffalo Creek steam stimulation operations from 1980-81 kH = Horizontal permeability
in the Grosmont formation produced unexpectedly large φ = Porosity
bitumen volumes considering the injection scheme that was bmin/bmax = Permeability coefficient limits
utilized. These high recoveries achieved with an unrefined amin/amax = Permeability coefficient limits
exploitation design suggest the zone contains large volumes of bh = Horizontal permeability coefficient
bitumen readily accessible with present recovery methods. bv = Vertical permeability coefficient
Before any stimulation methods may be considered for bitumen ah = Horizontal permeability coefficient
production from this formation, a representative reservoir av = Vertical permeability coefficient
description is required. Attaining this reservoir description via bmin/bmax = Permeability coefficient limits
history matching; however, has proved to be just as complex as amin/amax = Permeability coefficient limits
the geology in the reservoir itself.
The biggest hurdle in modeling the reservoir in the Grosmont
has been estimating the bulk effective permeability. While core REFERENCES
testing can determine matrix permeability, large vug and
1. CORDELL, G.M., Reservoir Simulation of a Grosmont
fracture networks within the formation contribute several
Carbonate Pilot; paper presented at the AOSTRA third
magnitudes of Darcy’s to the overall permeability of the
annual conference, Calgary, AB, 10-11 June 1982.
reservoir. A reservoir model with a uniform 5 Darcy 2. DEMBICKI, E.A., MACHEL, H.G., Recognition and
permeability profile provided the necessary medium for Delineation of Paleokarst Zones by the Use of Wireline
sufficient steam penetration to attain the oil rates produced in Logs in the Bitumen-Saturated Upper Devonian
the field, but a comparison of simulated and field BH Grosmont Formation of Northeastern Alberta, Canada;
temperature data was not accurate. Consequently, the reservoir in AAPG Bulletin, V.80, No.5, pp. 695-712, May 1996.
model was updated into a second phase with the goal of 3. HARRISON, R.S., Geology and production history of
avoiding the heat containment problems encountered with the the Grosmont Carbonate Pilot Project; AB, 1982 in R.F
first model. Meyer, J.C Wynn, and J.C. Olson, eds., The Future of
The intention of Reservoir Model 2 was to achieve a similar, Heavy Crude and Tar Sands: Second International
if not better, oil rate match and improve the BH temperature Conference on Heavy Crude and Tar Sands, v. 1,
match of Reservoir Model 1. There were also several unknown pp.199-204.
reservoir parameters that needed to be resolved and were also 4. OUTTRIM, C.P., EVANS, R.G., Alberta’s Oil Sands
considered in the design of Reservoir Model 2. To meet the Reserves and their Evaluation; Energy Resources
objectives of Reservoir Model 2, “manual” history matching Conservation Board, Calgary, Alberta pp. 38-42 in
was replaced by a more ideal automated history matching Redford, D.A, Winestock, A.G., The Oil Sands of
process utilizing a genetic algorithm. While the algorithm was Canada-Venezuela 1977, The Canadian Institute of
not yet complete at the time of this report, an example reservoir Mining and Metallurgy, CIM Special Volume 17, 1977.
description is represented outlining the history matching 5. UNION OIL COMPANY OF CANADA., Buffalo
procedure of the Java program. Even though the example Creek Test Site: Progress Report ER-81-17; ERCB
reservoir description generated was far from being an accurate Approval No. 2367C, AUB, EUB, AB, 1981.
representation of the Buffalo Creek reservoir, the model did 6. UNION OIL COMPANY OF CANADA., Buffalo
showcase a heterogeneous permeability distribution provided Creek Test Site: Progress Report ER-81-38; ERCB
the necessary heat disbursement away from the bottomhole. Approval No. 2376C, AUB, EUB, AB, 1981.
Ultimately we can conclude that the in-situ scenario at the 7. UNION OIL COMPANY OF CANADA., Buffalo
Buffalo Creek test site is a diversely large permeability Creek Test Site: Progress Report ER-82-08; ERCB
distribution, independent of porosity. This assertion will lead to Approval No. 2367C, AUB, EUB, AB, 1982.
the next phase in reservoir modeling for carbonate reservoirs
and the development of the genetic algorithm. The future
reservoir model will contain individual vertical and horizontal

4
Figure 1

Figure 2
Buffalo Creek Production from 1980-1981

140

120
Oil
Water
100
Rate (m^3/d)

80

60

40

20

0
Ja

Au

Ju

Se

D
ov

ec
ay

ar
n-

n-
g-

p-
-8
-8

-8
-8
80

81
80

81
1-
0-

1-
0-
-3

-1
-1

-2
05
25

30
09
0

3
7

Time

5
Figure 3
Buffalo Creek Producing BH Temperature from 1980-81

300

250
Temperature (deg C)

200

150

100

50

0
30

09

17

25

05

13

21

30
-J

-M

-A

-N

-M

-J

-S

-D
an

un
ug

ep
ov

ec
ay

ar
-8

-8
-8
-8

-8
-8

-8
-8
0

1
1
0

1
0

Time

Figure 4
10A-5-88-19-W4 (Injector/Producer) Well Porosity vs. Depth

50

45

40

35

30
Porosity (%)

25

20

15

10

0
260 265 270 275 280 285 290 295 300
Depth (m)

6
Figure 5
Oil Production History Match

140

120

Buffalo Creek
100 Reservoir Model 1
Rate (m^3/d)

80

60

40

20

0
Ja

Au

Ju

Se

D
ov

ec
ay

ar
n-

n-
g-

p-
-8
-8

-8
-8
80

81
80

81
1-
0-

1-
0-
-3

-1
-1

-2
05
25

30
09
0

3
7

1
Time

Figure 6
Producing BH Temperature History Match

300

Buffalo Creek
250
Reservoir Model 1
Temperature (deg C)

200

150

100

50

0
30

25
09

17

05

13

21

30
-J

-M

-A

-N

-M

-J

-S

-D
an

un
ug

ep
ov

ec
ay

ar
-8

-8
-8
-8

-8

-8

-8
-8
0

1
1
0

1
0

Time

7
Figure 7
Permeability Correlation

100000000

10000000

1000000 amin = 0.001, bmin = 23


amax =0.030, bmax = 41
100000
ah = 0.0122, bh = 35
10000 av = 0.016, bv = 26

1000
Darcys

100

10

1
0 5 10 15 20 25 30 35 40 45 50
0.1

0.01

0.001
Porosity (%)

Figure 8
Oil Production History Match

140

120

Buffalo Creek
100 Reservoir Model 2
Rate (m^3/d)

80

60

40

20

0
Ja

Au

Ju

Se

D
ov

ec
ay

ar
n-

n-
g-

p-
-8
-8

-8
-8
80

81
80

81
1-
0-

1-
0-
-3

-1
-1

-2
05
25

30
09
0

3
7

Time

8
Figure 9
Producing BH Temperature History Match

300

Buffalo Creek
250
Reservoir Model 2
Temperature (deg C)

200

150

100

50

0
30

25
09

17

05

13

21

30
-J

-M

-A

-N

-M

-J

-S

-D
an

un
ug

ep
ov

ec
ay

ar
-8

-8
-8
-8

-8

-8

-8
-8
0

1
1
0

1
0

Time

9
Table 1 – Buffalo Creek Petrophysical Reservoir Data

Average Porosity (%) 25.4


Average Water Saturation (%) 15
Initial Formation Pressure (kPa) 1000
Bitumen Gravity (API) 7
o
Reservoir Thermal Conductivity (kJ/D-m- K) 216
Reservoir Heat Capacity (kJ/m3/oK) 2600
Matrix Permeability (mD) 10-200

Table 2 – Reservoir Model 1 Characteristics

Size of Zone (m) 37


Top Depth (m) 263
Relative Permeability Curves (Honarpour) :
SWC (%) 25
SORW (%) 15
kV (Darcy) 5
kH (Darcy) 5
k – 2m “tight” streak separating Grosmont C and D, vertical (Darcy) 1

Rock Compressibility (v/v/kPa) 3E-04


Injection BH steam Qualities Wellhead Qualities

Table 3 – Reservoir Model 2 Characteristics

Size of Zone (m) 37


Top Depth (m) 263
Rock Compressibility (v/v/kPa) 3.031215E-04
Relative Permeability Curves (Honarpour):
SWC (%) 9.375
SORW (%) 21.25
Cycle 1 Steam Quality (%) 27.5
Cycle 2 Steam Quality (%) 70
Cycle 3 Steam Quality (%) 40
Cycle 4 Steam Quality (%) 50
Cycle 5 Steam Quality (%) 40
(1) kH = ah * exp ^ ( * bh) (Darcy) ah = 0.0122
bh = 35
(2) kV = av * exp ^ ( * bv) (Darcy) av = 0.016
bv = 26
k_barrier, vertical (Darcy) 32.5

10

S-ar putea să vă placă și