PETROLEUM SOCIETY
PAPER 2002-026
CANADIAN INSTITUTE OF MINING, METALLURGY & PETROLEUM
CO2 Flooding Performance
Prediction for Alberta Oil Pools
J.C. Shaw
Adams Pearson Associates Inc.
S. Bachu
Alberta Energy and Utilities Board
This paper is to be presented at the Petroleum Societys Canadian International Petroleum Conference 2002, Calgary, Alberta,
Canada, June 11 13, 2002. 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
using the Alberta reserves database. The program
estimates live oil and CO2 viscosities at reservoir
conditions, oil MMP and reservoir heterogeneity based
on the rock type, to predict oil recovery at any specific
pore volume of CO2 throughput.
Attention in CO2 flooding for incremental oil recovery
and greenhouse gas (GHG) sequestration has prompted
the need for screening and ranking Alberta oil pools for
this EOR process. In a previous paper by the same
authors, over eight thousands of Alberta oil pools were
ranked for CO2 EOR suitability using a new parametricranking software which utilized six essential reservoir
properties: oil density, residual oil saturation, minimum
miscibility pressure (MMP), reservoir temperature, net
pay thickness, and porosity. This continuation paper
describes the results of using an advanced method to
estimate production forecasts for numerous candidate
pools in Alberta.
Over 8,000 Alberta pools were first screened for CO2flood suitability, and pertinent reservoir properties were
used for the remaining 4,729 pools to calculate oil
recovery. The predicted recoveries for all pools ranged
from 1.2-13.9%, 6.3-18.7% and 11.8-27.1% at
breakthrough and 0.25 and 0.5 hydrocarbon pore volume
(HCPV) injection respectively. These values compared
well to an average of 13% incremental oil recovery from
the field experience of CO2 floods. More importantly, the
results clearly identify the most suitable Alberta pools for
CO2 flooding.
A Microsoft Excel program with VBA based on the
modified Koval method (1963) by Claridge (1972) has
been developed to predict CO2 flooding performance
INTRODUCTION
The recent high oil price and interest in reducing GHG
(Greenhouse Gas) emissions in response to global
warming may have created new business opportunities to
realize incremental value from depleted oil pools through
CO2 flooding. Not all oil pools in Alberta are suitable for
CO2 flooding. Thus, in a previous paper by the same
authors1, about 8,800 Alberta oil pools were ranked by
using a newly-developed VBA program capable of
retrieving pertinent information from the digitized
Alberta reserve database and perform parametric
technical rankings. Six parameters with different
assigned weightings were used in the technical ranking.
These include API gravity of oil, residual oil saturation,
ratio between reservoir pressure and predicted minimum
miscibility pressure (P/MMP), reservoir temperature, net
pay thickness, and porosity.
Whether miscibility can be achieved under
reservoir conditions;
3.
Mobility of reservoir, solvent and drive fluids;
Pressure distribution in the reservoir;
6.
Flood pattern and well spacing; and
7.
Reservoir heterogeneity.
Secondary recovery (Swi = Swc), where Swi is initial
water saturation, Swc is connate water saturation
Five spot
Trapped oil effect
Gravity stabilisation
Continuous or slug injection
This secondary recovery method is based on the
original work of Koval (1963)4 as modified by Claridge
(1972)5 for areal sweep. The following summarizes the
method by Claridge that is based on the concept of
apparent pore volume. It combines methods of predicting
areal coverage and linear displacement efficiency and is
used to calculate oil recovery for a series of assumed slug
sizes (hydrocarbon pore volume or HCPV) in a five spot
CO2 slug-waterflood pilot test.
The recovery efficiency prediction of CO2 flooding
can be used to provide useful estimates of financial
viability of the project. Ideally, the predictive models
should consider:
Size of solvent slug and type of drive fluid;
5.
Modified Koval Method
Klins (1984)3 summarized the assumptions and
limitations of the modified Koval method as follows:
CO2 FLOODING PREDICTIVE MODELS
2.
Vertical, areal and microscopic sweep
efficiencies;
However, due to the unavailability and uncertainties of
reservoir data and limitations of the analytical methods,
these predictions may be valid only under certain
conditions regardless the level of sophistication of the
models. Since the final objective of this study is to
provide CO2 flooding prediction for a large number of
Alberta oil pools, desktop methods are preferred for their
simplicity and applicability in this situation where there
are only limited reservoir data available from the Alberta
reserves database.
However, the screening software is not capable of
providing production forecasts of CO2 flooding, which is
the motivation of this study. Numerous active CO2
flooding projects in the United States and Canada have
provided valuable theoretical and practical information
on the technology. Desktop engineering prediction tools
such as US DOE CO2 Prophet2 have been developed
for quick technical and economic assessment. These
tools are based on sophisticated analytical equations
derived from theoretical calculations, numerical
simulation and field experience. However, we are not
aware of any tools that are capable of evaluating the
performance of large numbers of oil pools as reported in
this paper.
1.
4.
Koval introduced the Koval Factor in his work and it
is defined as follows:
K = HFE = HF 0.78 + 0.22 o
N p V piBT
1.0 N p
Fi
Effective viscosity or mobility ratio defined by
Koval
E = 0.78 + 0.22 o
1/ 4
1/ 4
1.6
K 0.61
Fi V piBT
1.0 (F )
i BT
1.28
0.26
K
HCPV of solvent injection in an ideal fivespot of unit thickness which will produce an
invaded area whose outer envelope
corresponds to EAFI times the area of the fivespot
Koval factor
Heterogeneity factor
VpiBT
the value of Vpi at solvent breakthrough
1 (homogeneous reservoirs)
Np
HCPV of oil produced, vol/vol
V DP
log10 H =
0.2
(1 V DP )
CO2 FLOODING PREDICTIVE MODELS
Estimation of oil and CO2 Viscosity
The following equations from Beggs and Robinson
(1975)7were used to estimate the dead and live oil
viscosity for all oil pools in Alberta:
(stratified reservoirs)
F
Gravity override factor
1, if assuming no gravity override
where:
Dead Oil Viscosity, od (cp)
o = Oil viscosity, cp
s = CO2 viscosity, cp
od = 10 x 1
VDP = Dykstra-Parsons coefficient
where:
Solvent breakthrough at:
V pvdBT =
1
K
VpvdBT =
invaded pore volumes injected at
breakthrough
E ABT
x = yT 0.601 ,
z = 2.1646 0.033580 ( API )
Live Oil Viscosity, o (cp)
B
o = A od
0.4M
= 1+
(1 + M )
EABT =
y = 10 z ,
where:
A = 12.589 ( Rs + 200) 0.482 ,
B = 1.276 ( R s + 15) 0.090
areal sweep efficiency at breakthrough per
Caudle-Witte correlation6
V piBT = E ABT V pvdBT
where:
VpiBT =
Rs
Solution gas/oil ratio, scf/STB
Reservoir temperature, C
actual pore volumes of solvent injected at
breakthrough
Finally, the oil produced from the miscible injection Np
is calculated by using Claridges simplified equation as
follows:
API =
Oil gravity
>25%, original reservoir pressure >10.3 MPa (1500 psi)
and ideally 1.4 MPa (200 psi) higher than MMP at the
time of CO2 injection. Immiscible CO2 flooding is much
less common; nevertheless it has being applied to heavy
and medium oils (10-25 API) and in-situ viscosities of
The viscosity and density of CO2 at original
temperature and pressure conditions was estimated based
on an interpolation method of two digitized CO2 viscosity
and density figures from the SPE Monograph on Miscible
Flooding by Stalkup (1983)8. It is assumed that the CO2
flooding will be conducted at the original reservoir
pressure to achieve miscibility. The viscosity ratio of
oil/CO2 plays an important role in the performance
prediction.
100 to 1000 cp. Thus, some oil reservoirs will be better
suited, hence more economic, than others, for CO2
flooding. These reservoirs should be the ones used first
for CO2 storage.
A total of 4,729 Alberta oil pools (out of a total of
8,630 pools initially) passed the following screening
criteria. This number was derived by eliminating 54
pools that have been on solvent flooding, 1,161 pools that
have reservoir temperature <31C, 24 pools that have
remaining oil saturation <40%, 336 pools that have no
reservoir pressure data and 2,326 pools that have P/MMP
<0.95. The screening was conducted because:
Estimation of Reservoir Heterogeneity
Since the CO2 flooding performance is also dependent
on the reservoir heterogeneity, it is necessary to provide
general estimations of the degree of heterogeneity. The
heterogeneity factor introduced by Koval requires input
of Dykstra-Parsons coefficient (VDP)9 which is not
available in the Alberta reserve database. It will be
extremely time-consuming to obtain the value by
systematically reviewing core analysis data from each oil
pool. Unfortunately, only very limited numbers of
Dykstra-Parsons coefficients are available in the
literature. In the SPE textbook series for Waterflooding
Willite (1986) 1 0 , a range of these values were
summarized in a plot, which shows a range of 0.5-0.9 VDP
with an average of 0.7 for most rocks. Therefore, two
default values, 0.7 and 0.9, were used to represent VDP of
sandstone and carbonate rocks respectively. The resulting
heterogeneity factors (H), 8.5 and 26.3, were calculated
by using the above-mentioned heterogeneity equation.
The VDP values can be easily changed in this VBA
program to accommodate actual calculated results from
core analysis data.
The predicative model only predicts recovery for
pools that are miscible with CO2 and,
CO2 critical temperature is 31C.
Calculated Fluid Properties
With the built-in program that calculates the CO2 and
oil properties, the output data are used for further
evaluation of recovery factor. For the 4,729 oil pools that
passed the screening, the API gravity ranges 17-72; live
oil viscosities 0.06-44.3 cp; CO2 viscosity of 0.02-0.1 cp
and CO2 density of 0.23-0.93 g/cc.
Recovery Factor
The recovery factors of CO2 miscible flood at solvent
breakthrough, 0.25 and 0.5 HCPV of solvent injection
were calculated by using the Excel VBA program. Of the
4,729-screened pools, the calculated recoveries ranges at
the solvent breakthrough, and 0.25 and 0.5 HCPV
injections are shown in Table 1. The results are compared
to an average of 13% incremental oil recovery from the
field experience of all CO2 floods.11 It is important to
note that the Koval heterogeneity factors used for the
calculations are 7.8 and 26.7 for sandstone and carbonate
respectively. Figure 1 shows the frequency distribution of
CO2 recovery factors at 0.25 and 0.5 PV injections.
A lookup table was designed to differentiate whether
the formation rock is either sandstone or carbonate for all
the producing formations in Alberta, which is included in
the VBA program.
PREDICTED RECOVERY OF CO2 FLOODING
The reservoirs suitable for EOR using CO2 flooding
have various degrees of suitability, depending on the
intrinsic reservoir and oil characteristics. The range of
reservoir and fluid properties suitable for CO2 miscible
injection is quite wide; however, reservoirs should have
an oil gravity API>30, remaining oil saturation (So)
The calculated results are sorted in a descendant order
based on the breakthrough recovery efficiency. It is
interesting to find that, with the assumed heterogeneity
factors, all top-ranked pools are sandstones. The highest
ranked carbonate reservoir is at the 2,568th place. The
previous ranking study considered six reservoir
properties: oil gravity, oil saturation, reservoir pressure,
temperature, net pay thickness and porosity. As shown in
the previous ranking scores in Table 1, most of carbonate
pools were much higher ranked than the present
predictive results mainly due to the higher heterogeneity
factor that is used for these pools. This observation
points out the importance of better knowledge of
reservoir heterogeneity. Figure 2 shows the recovery
distribution with VDP of 0.75 for the carbonate rocks. The
top-ranked carbonate pool is at the 93rd place due to a
lower corresponding H of 9.8.
2.
Light oil;
High initial reservoir pressure;
Sandstone with lower heterogeneities and,
Modest initial reservoir temperature.
Not previously solvent flooded;
Reservoir temperature > 31 C;
Residual oil saturation > 40%;
Available reservoir pressure data; and
CO2 miscibility (P/MMP >0.95).
By using the newly developed Excel VBA program
based on the modified Koval (1963) method by
Claridge (1972), the CO2 flooding performance for
these 4,729 pools are reported at CO2 breakthrough,
0.25 and 0.5 pore volumes CO2 injection.
3 . Reservoir heterogeneity is accounted for by
assigning a default Dykstra-Parsons coefficient for
the rock type (sandstone vs. carbonate) of each
producing pool to calculate the heterogeneity factor.
4. CO2 and live oil viscosities at the initial reservoir
conditions were estimated for all of the screened
pools. In addition, CO2 density was also calculated
that will be useful in calculating storage capacity for
the future work.
The characteristics of the top ranked pools are:
5 . The user of this program can easily adjust the
miscibility cut-off, heterogeneity factor and pore
volume solvent injection to produce desirable results
for various conditions.
It is important to note that the predictive model
assumes that CO2/waterflood flooding is implemented as
a secondary recovery process and there is no provision to
account for historic production. Since all Alberta oil
pools have gone through different stages of depletion, the
reader is cautioned not to draw blanket conclusions about
the CO2 flooding suitability without first checking the
status and other pertinent information of these pools.
This is especially true for pools that have been either
water or solvent flooded.
6 . The predicted recoveries for all pools range from
1.2-13.9% at solvent breakthrough, 6.3-18.7% and
11.8-27.1% at 0.25 and 0.5 pore volume injection
respectively. This is compared to an average of 13%
incremental oil recovery from the field experience of
all CO2 floods.
7.
Due to the availability and uncertainties of reservoir
data and the limitations of the analytical methods, the
predictive results based on the developed desktop model
should only be used as a guide to rank-order large
numbers of oil pools. It is not intended to replace more
detailed laboratory and engineering studies that are
essential to any EOR design.
Due to the availability and uncertainties of reservoir
data and limitations of the analytical methods, the
predictive results based on the desktop model should
only be used as a guide to rank-order large numbers
of oil pools. It is not intended to replace more
detailed laboratory and engineering studies that are
essential to any EOR design.
8 . It is important to note that the predictive model
assumes that alternate CO 2 /water flooding is
implemented as a secondary recovery process and
there is no provision to account for historic field
production. Since all Alberta oil pools have gone
through different stages of depletion, the reader is
CONCLUSIONS
1 . A total of 4,729 from 8,630 oil pools in Alberta
passed the following screening criteria:
cautioned not to draw hasty conclusions of CO2
flooding suitability without first checking the status
and other pertinent information of these pools.
9.
4 . Koval, E. J., A Method for Predicting the
Performance of Unstable Miscible Displacement in
Heterogeneous Media, Soc. Pet. Eng. J. (June 1963)
145-154.
Since reservoir heterogeneity has a strong influence
on the volumetric sweep efficiency of the flood, it is
important to have this parameter better defined. The
present study uses two default values of the DykstraParsons coefficient, 0.7 for sandstone and 0.9 for
carbonate rocks. However, this coefficient can be
better estimated for individual pools based on core
data providing the resources are available.
5.
Claridge, E. L., Prediction of Recovery in Unstable
Miscible Flooding, Soc. Pet. Eng. J. (April 1972)
143-154.
6.
Caudle, B. H. and Witte M. D., Production Potential
Changes During Sweep-Out in a Five Spot Pattern,
Trans. AIME, 1959, Vol. 216, p. 447.
7. Beggs, H. D. and Robinson, J. R., Estimating the
Viscosities of Crude Oil Systems, JPT, (Sept. 1975)
1140-1141.
REFERENCES
8.
1.
Shaw, J. C. and Bachu, S., Screening and Ranking
Alberta Oil Pools for CO2 Flooding and
Sequestration, CIM Paper 2001-157, presented at
the Petroleum Societys Canadian International
Petroleum Conference, Calgary, Alberta, Canada,
June 12 14, 2001.
Stalkup, F. I., Miscible Displacement, Monograph
Series, SPE, Dallas, New York, 1983.
9 . Dykstra, H., and Parsons, R. L., Secondary
Recovery of Oil in the United States, API,
Washington (1950).
10. Willhite, G. P., Waterflooding, SPE Textbook
Series Vol. 3, 1986, p.172.
2. Paul, G. W. (Principal Investigator), Development
and Verification of Simplified Prediction Models for
Enhanced Oil Recovery Application: CO2 (Miscible
Flood) Predictive Model, U.S. DOE Report DEAC19-80BC10327, Intercomp-Denver (1983).
11. Holt, T., Jensen, J.I., Lindeberg, E.: Underground
Storage of CO2 in Aquifers and Oil Reservoirs,
Energy Conv. Mgmt., 1995, 36, 335-338.
3 . Klins, M. A., Carbon Dioxide Flooding; Basic
Mechanisms and Project Design, International
Human Resources Development Corporation,
Boston, MA, 1984.
TABLE 1
SUMMARY OF RECOVERY FACTORS OF 4,729 ALBERTA
POOLS WITH ASSUMED HETEROGENEITY FACTORS
Recovery @
0.25 HCPV
Injection (%)
Breakthrough
Recovery (%)
Recovery @
0.5 HCPV
Injection (%)
Minimum
1.2
6.3
7.0
Maximum
Average
13.9
5.1
18.7
11.1
27.2
14.9
Oil Recovery @
0.25 PV CO2
Injection (%)
Oil Recovery @
0.50 PV CO2
Injection (%)
Break-through
Recovery of CO2
(%)
TABLE 2- SUMMARY OF PREDICTED RESULTS OF THE TOP 20 RANKED POOLS
STRACHAN
SECOND WHITE SPECKS A
43.97
47.38
399
82
0.97
30782
10680
2.88
0.18
0.06
2.97
0.74
0.13
0.18
0.26
GARRINGTON
VIKING CC MANV B &
LOWER MANNVILLE VVV
48.22
46.08
385
68
0.93
32119
10680
3.01
0.22
0.07
3.00
0.82
0.13
0.18
0.26
CAROLINE
BASAL MANNVILLE MU #3
47.60
42.97
483
92
0.82
30916
10680
2.89
0.21
0.06
3.69
0.70
0.12
0.17
0.25
PECO
CARDIUM G
45.20
47.16
210
77
0.93
31372
10680
2.94
0.24
0.07
3.73
0.78
0.12
0.17
0.25
RICINUS
CARDIUM 000
49.46
46.26
323
78
0.95
26853
10680
2.51
0.22
0.06
3.74
0.72
0.12
0.17
0.25
KAKWA
C CARDIUM B
49.38
47.61
268
55
0.92
20632
9650
2.14
0.25
0.06
3.82
0.77
0.12
0.17
0.25
CAROLINE
CARDIUM K
53.20
44.93
312
74
0.90
27372
10680
2.56
0.25
0.06
3.96
0.75
0.12
0.17
0.25
Field Name
Pool Name
Previous
1
Scores
API
Gravity
Initial
Solution
GOR
Temp
(C)
So
(%)
Initial
Pressure
(kPa)
MMP
(kPa)
P/MMP
Live Oil
Viscosity
(cp)
CO2
Viscosity
(cp)
Oil/CO2
Viscosity
Density
CO2
(g/cc)
46.81
46.71
254
52
1.00
21702
9650
2.25
0.27
0.07
3.98
0.80
0.12
0.17
0.25
RICINUS
A CARDIUM A
PRIMARY AREA
CARDIUM E & KKK
55.85
45.08
323
78
0.95
26389
10680
2.47
0.23
0.06
4.07
0.71
0.12
0.17
0.25
BRAZEAU RIVER
CARDIUM I
54.07
46.93
240
76
0.86
25567
10680
2.39
0.24
0.06
4.13
0.71
0.12
0.17
0.25
WESTPEM
OSTRACOD F
40.90
42.76
260
96
0.84
38634
11720
3.30
0.27
0.07
4.19
0.75
0.12
0.17
0.25
RICINUS
CARDIUM III
55.49
43.84
363
75
0.99
25817
10680
2.42
0.25
0.06
4.20
0.72
0.12
0.17
0.25
BRAZEAU RIVER
LOWER MANNVILLE I
44.22
45.52
280
90
0.93
27664
10680
2.59
0.23
0.05
4.23
0.67
0.12
0.17
0.25
BRAZEAU RIVER
CARDIUM K
51.04
46.26
245
76
0.95
25949
10680
2.43
0.25
0.06
4.23
0.71
0.12
0.17
0.25
BRAZEAU RIVER
LOWER MANNVILLE P
37.60
43.4
248
96
0.97
36874
11720
3.15
0.27
0.06
4.24
0.74
0.12
0.17
0.25
PECO
CARDIUM D
51.83
47.38
200
74
0.91
25145
10680
2.35
0.25
0.06
4.28
0.71
0.12
0.17
0.25
KAKWA
MAIN CARDIUM A
47.96
47.61
192
53
0.95
20561
9650
2.13
0.28
0.07
4.32
0.78
0.12
0.17
0.24
CAROLINE
CARDIUM F
52.18
45.08
246
77
0.88
28125
10680
2.63
0.26
0.06
4.32
0.74
0.12
0.17
0.24
CAROLINE
CARDIUM G
56.62
45.08
312
69
0.98
22181
10680
2.08
0.25
0.06
4.42
0.69
0.12
0.17
0.24
SYLVAN LAKE
SECOND WHITE
SPECKS D
48.33
46.71
177
47
0.99
22607
8270
2.73
0.33
0.07
4.43
0.84
0.12
0.17
0.24
KAKWA
FIGURE 1-FREQUENCY DISTRIBUTION OF CO2 RECOVERY FACTOR AT 0.25 AND 0.5
HCPV INJECTION ASSUMING VDP OF 0.7 (SANDSTIONE) AND 0.90 (CARBONATE)
FREQUENCY DISTRIBUTIONS
4,500
4,000
3,500
Count
3,000
2,500
2,000
1,500
1,000
500
0
3%
6%
9%
12%
15%
18%
21%
24%
27%
30%
Recovery Percentage
Oil Recovery @ 0.25 PV CO2 Injection
Oil Recovery @ 0.50 PV CO2 Injection
FIGURE 2-FREQUENCY DISTRIBUTION OF CO2 RECOVERY FACTOR AT 0.25 AND 0.5
HCPV INJECTION ASSUMING VDP OF 0.7 (SANDSTIONE) AND 0.75 (CARBONATE)
FREQUENCY DISTRIBUTIONS
4,500
4,000
3,500
Count
3,000
2,500
2,000
1,500
1,000
500
0
3%
6%
9%
12%
15%
18%
21%
24%
27%
Recovery Percentage
Oil Recovery @ 0.25 PV CO2 Injection
Oil Recovery @ 0.50 PV CO2 Injection
30%
