Professional Documents
Culture Documents
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
CO2 flooding in oil reservoirs to enhance oil recovery is successfully applied in many ongoing projects
in the world. The bulk of industry research and field testing of CO2 has been directed towards miscible
displacement. This method of using CO2 appears to have the greatest potential for recovery of oil that
would not have been recovered using conventional production methods.
The main objectives of this paper are to investigate the most important reservoir and operational
parameters in CO2 flooding and to provide guidelines to maximize CO2 injection performance for
different levels of reservoir heterogeneity. Several compositional simulation models were generated
using actual field data. For each model, reservoir heterogeneity was evaluated using Lorenz and
Dykstra Parsons coefficients. The different compositional simulation models were setup to investigate
the effects of key parameters such as injection rates, injection patterns, well spacing, injection scheme
(Continuous CO2, WAG, and SWAG), WAG cycle length, vertical to horizontal permeability, and
horizontal injection wells on the recovery factor and production of CO2 projects.
Those different heterogeneous models were used to simulate different scenarios over a wide range of
both reservoir and operational parameters and provided a comprehensive data set to evaluate the CO2
flood effectiveness in a variety of conditions. More than 700 simulation runs were designed using
Experimental Design techniques to test those parameters. Several correlations (proxy models) were
also developed and blind tested to check their effectiveness in predicting the performance of
combination of parameters that were not even tested. As a result of this work, it is possible to use
experimental design to test the performance of CO2 injection. It can be performed with acceptable
accuracy by performing a limited number of runs; hence, reducing the time, effort, and money.
It was observed that the use of different heterogeneous models resulted in different overall field
performance. Our results also showed that implementing WAG and horizontal wells enhance the oil
recovery significantly.
Introduction
Carbon Dioxide gas injection has been becoming increasingly important for oil recovery and
environmental considerations in oil field development and CO2 sequestration. The miscible CO2 gas
injection is becoming more and more popular amongst other EOR techniques. This can be related to
the higher oil prices as well as the availability of more CO2 sources considering the global regulations
and restrictions on CO2 emissions.
2 SPE 151871
This paper addresses the simulation of CO2 flooding through different scenarios, considering
heterogeneities in order to study their impact on oil recovery. We Performed sensitivity studies to
investigate the effects of key parameters such as injection rates, injection patterns, well spacing,
injection scheme (Continuous CO2, WAG, and SWAG), WAG cycle length, vertical to horizontal
permeability, and horizontal injection wells on the recovery factor and production performance of CO2
projects. The reservoir simulations for CO2 injection drive mechanism are executed using Eclipse
compositional simulator.1
Also, in this paper we combined reservoir modeling and flow simulation with experimental design
techniques to help provide guidelines and maximize CO2 injection performance for different levels of
reservoir heterogeneity .A correlation (proxy model) was also developed and blind tested to check its
effectiveness in predicting the performance of combination of parameters that were not used in
generating the correlation.
Model Desciption
The model that was used is composed of four formations. The average reservoir depth is 2950 meters.
The dimensions of the model are 3,814 meters long by 2,500 meters wide by 52 meters thick. The
model is a 3-phase (Oil, Gas, and Water) and described on a regular Cartesian grid.
The reservoir2 was originally highly undersaturated with a stock tank gravity of 32 API. Initial reservoir
pressure 340 bar and bubble point pressure of 171 bar at a reference depth of 2910 m and 104 °C. The
CO2 minimum miscibility pressure was determined experimentally to be 271 bar.
Different heterogeneous models were developed using actual Petrophysical data. Then we quantified
the reservoir heterogeneity for the generated models. To quantify the heterogeneity levels in the
models of the present study, the Dykstra-Parsons coefficient, VDP and Lorenz coefficient, LC were
applied.3 Table 1 shows the summary of the different heterogeneous models. The ranges of
Heterogeneity Index are:
1. HI = 0, Ideal homogeneous reservoir
2. 0 < HI < 0.25, slightly heterogeneous
3. 0.25 < HI < 0.5, heterogeneous reservoir
4. 0.5 < HI< 0.75, very heterogeneous reservoir
5. 0.75 < HI < 1, extremely heterogeneous reservoir
Table 1: Classification of the different heterogeneous reservoir models
Dykstra Parsons
Model Lorenz Coefficient Heterogeneity Type
Coefficient
Model (A) 0 0 Homogeneous
Model (B) 0.1 0.09 Slightly heterogeneous
Model (C) 0.2 0.15 Slightly heterogeneous
Model (D) 0.3 0.28 Heterogeneous
Model (E) 0.45 0.497 Heterogeneous
Model (F) 0.65 0.69 Very-Heterogeneous
Model (G) 0.72 0.75 Very-Heterogeneous
Model (H) 0.8 0.81 Extremely-Heterogeneous
SPE 151871 3
Those different heterogeneous models were used to simulate different scenarios over a wide range of
both reservoir and operational parameters and provided a comprehensive data set to evaluate the CO2
flood effectiveness in a variety conditions.
Simulation Results and Discussion
The results of the CO2 injection sensitivity analysis are presented in several sections. The first shows
the effect of injection rate on the recovery factor, the second presents the effect of well pattern and well
spacing, the third section shows the effect of injection scheme, the fourth section presents the effect of
WAG cycle length, the fifth section shows the effect of vertical to horizontal permeability, and the last
section discusses the effect of horizontal injection wells on the CO2 injection performance. Our
simulation case studies considered the CO2 injection for miscible displacement.
CO2 Injection Rate Sensitivity Analysis
One of the most important parameters that affect the miscible displacement process and its efficiency is
the gas injection rate. To investigate the effect of the injection rate on the CO2 flooding process several
sensitivities were run. Injection rates at reservoir conditions of (0.1, 0.2, 0.5, 1, 2, 5, 10, 20, 50, 100,
200, 500 and 1000 MMscf/D) were used. The following features were applied in these cases:
• Continuous CO2 injection will be applied.
• The model was run for 10 years.
• The production wells were controlled by bottom-hole pressure constrained of 271 bar above the
minimum miscibility pressure.
• The vertical to horizontal permeability ratio, kv/kh, is 0.1.
Fig.1 shows the comparison of recovery factor with injection rate for different heterogeneous models.
We found that low injection rate may lead to slow displacement process and hence low recovery at the
designed time of the project. It indicates that when the gas injection rates increase the recovery factor
increases and good oil recovery is observed in all different heterogeneous models. Also, it is clear that
the models with heterogeneity index that lie in the range of 0.2 to 0.45 have the same recovery factor
for injection rate lower than 400 MMscf/D, while for injection rate higher than 400 MMscf/D there is a
slightly difference.
Then we calculated ratio between hydrocarbon pore volume (HCPV, resbbl) and total field gas injected
(FGIT, stb) to generate dimensionless curves for the different heterogeneous models by plotting
recovery factor vs. HCPV/FGIT resbbl/Stb. Fig.2 shows the various dimensionless curves for the
different heterogeneous models.
4 SPE 151871
Fig.3: Oil recovery factor for well patterns sensitivity Fig.4: Oil recovery factor for well spacing sensitivity
analysis analysis
CO2-Continous 70.3
Model (A) SWAG 75.7
WAG 76.4
CO2-Continous 68.2
Model (C) SWAG 74.3
WAG 78.5
CO2-Continous 65.4
Model (E) SWAG 70.4
WAG 73.5
CO2-Continous 54.7
Model (F) SWAG 57
WAG 58.5
CO2-Continous 28.38
Model (H) SWAG 34.5
WAG 38.3
This figure shows the recovery factor with heterogeneity index for different injection scheme. This
difference may be due to in case of continuous CO2 injection higher mobility causes lower reservoir
sweep and early production of CO2, which can lead to a lower oil recovery.
6 SPE 151871
Fig.5: Recovery factor as a function of the heterogeneity index for different injection scheme
Fig.6: Oil recovery factor for WAG cycle length sensitivity analysis
Fig. 7: Recovery factor as a function of the Fig.8: Recovery factor as a function of the
heterogeneity index for different kv/kh with WAG heterogeneity index for different kv/kh with
continuous CO2 injection
8 SPE 151871
Factor Range
Fig. 9 presents oil recovery factor for different heterogeneous models of horizontal well sensitivity
analysis. It indicates that in case of homogeneous reservoir there is no significant change in recovery
factor due to the change of horizontal well location, orientation, and length of horizontal section, while
when the heterogeneity of reservoir increases a noticeable change happens to the recovery factor.
Also, we found that the recovery factor will increase when the ratio of horizontal length section increase
for the different heterogeneous models. In case of heterogeneous reservoirs, maximum recovery is
obtained when drilling horizontal well in the lower section of the reservoir in both orientations X, and Y.
Firstly, we defined the factors that were used to generate the experimental design. Table 7 shows all of
the uncertain parameters and their corresponding range of variation for this study. Based on the
number of factors and their ranges the numbers of required runs to perform this sensitivity without
experimental design are 69120 which are too much and required a lot of effort, cost, and time to do it.
It is seen that there are both categorical and numeric type of variables. The most suitable experimental
design method for these types of mixed variables and ranges is D-optimal design.5
10 SPE 151871
According to the selected experimental design method and the number of factors, the minimum runs
required to perform this sensitivity are 31 runs and the maximum are 69120 runs. In this case we
selected a number of runs between the minimum and maximum to complete this sensitivity analysis. To
perform this experiment we used 80 reservoir simulation cases to generate the observation of desired
response. APPENDIX-A presents the required runs to perform this experimental design.
Then the next step is to run the experiment and generate the empirical model. The generated empirical
model based on the observed data from this process is shown below. Є represents the error term
observed in the response. In this case Є = 1.9. Fig. 11 shows the plot of predicted recovery factor with
calculated one.
0 .
0.1 . continous .
0.2 . WAG‐1 .
0.3 . WAG‐3 .
. . . 0.45 . WAG‐6 .
0.65 . WAG‐12 .
0.72 . SWAG .
0.8 .
1 .
10 .
20 . H‐UPPER .
50 . H‐MIDDLE .
100 . H‐LOWER .
200 . V .
500 .
1000 .
.1 .
5‐Spot . 0 .25 .
9‐Spot . 0.5 .
Line drive . 0.75 .
1 .
.
60 .
є
120 .
SPE 151871 11
Table 8 shows the regression results obtained using the software. Look in the column labeled Prob>F.
This column lists p-values. A low p-value (a value less than 0.05) indicates that results are statistically
significant.
Table 8: Test Effects
Fig.11: 3D view for simple model Fig.10: 3D view for Complex model
For each reservoir model two different scenarios are tested. Table 10 shows the different applied
parameters and the comparison between the predicted recovery values are the ones from empirical
model and the actual recovery factors are the flow simulation results. These results show that there is a
significant difference between the recovery values obtained from the empirical model and the recovery
values obtained from the reservoir simulation.
From these results it is clear that we can use the experimental design as a primary tool to predict
recovery factor and test the effect of different parameters in CO2 injection without spending resources
on a numerical simulation for the model used to generate the empirical model. While it is not applicable
for other reservoir models with different geologic structure i.e. each reservoir model should have its own
empirical model.
SPE 151871 13
Table 10: Predicted model results vs. simulation model results using the different reservoir models
Conclusions
Based on the results and analyses obtained from this study, several conclusions are made as follows:
1. Experimental Design can be integrated with reservoir modeling techniques for studying the impact
of different reservoir and operational parameters for CO2 injection because it reduces the amount of
time and cost needed to analyze the impact of these parameters on the CO2 injection performance.
2. This research demonstrates that D-optimal design can be successfully employed for generating the
empirical model to predict the performance of CO2 injection using different reservoir and operational
parameters and each reservoir model should have its own empirical model.
3. The recovery factor increases with the increase in the injection rate for different heterogeneous
models. The 5-spot pattern gives the maximum recovery for different heterogeneous models. The
recovery factor decreases as the well spacing decrease.
4. In the case of continuous CO2 injection, higher mobility of CO2 causes lower reservoir sweeps and
early CO2 production, hence lower oil recovery. However, the recovery factor increases in the cases
of WAG and SWAG.
5. WAG injection is effective in increasing the sweep efficiency the reservoir, while the effect of WAG
increases as the heterogeneity increases. No significant change in oil recovery was observed using
several WAG cycle lengths in case of homogeneous models. However, when the heterogeneity
increases, the recovery factor increases with a decreasing WAG cycle length.
6. The recovery increases as the kv/kh increase for different heterogeneous models. Therefore, oil
recovery factor for CO2 injection is dependent on the reservoir heterogeneity.
7. Our results show that in case of WAG the effect of change kv/kh will be higher than in case of
continuous CO2 injection.
8. The effect of kv/kh ratio on the recovery performance in heterogeneous reservoirs is more significant
than homogeneous reservoirs. Lower vertical permeability’s resulted in lower oil recoveries due to
lower vertical communication between layers.
9. For horizontal injection wells sensitivity analysis, a significant change in the recovery factor due to
the change of horizontal well location, orientation, and length of horizontal section as the reservoir
heterogeneity increase. The recovery factor increases as the ratio of horizontal length section
increases for the different heterogeneous models. The maximum recovery is obtained when drilling
horizontal well in the lower section of the reservoir in both orientations x, and y.
Recommendations
Even though comprehensive uncertainty analysis was performed on different parameters and factors
affecting CO2 injection, there is also several design parameters not investigated in this study, which
could affect the CO2 injection performance. Further numerical studies should be done on the fluid
property effect and the dynamic parameters.
References
1. ECLIPSE-300 Reference Manual, 2004A-1 Release, Schlumberger Reservoir Technologies,
Houston, 2004.
2. Jean-NoeJaubert.: “A crude oil data bank containing more than 5000 PVT and gas injection data”,
Journal of Petroleum Science and Engineering, 2001.
3. Tarek Ahmed, “Reservoir Engineering”, Second Edition, 2001.
4. Anderson, M.J. and Whitcomb, P.J. 2007. DOE Simplified: “Practical Tools for Effective
Experimentation”, Second Edition, 147-160, New York City: Productivity Press.
5. Fabian Triefenbach, “Design of Experiments the D-Optimal Approach and Its Implementation As a
Computer Algorithm”, Bachelor's Thesis, South Westphalia University, GERMANY, 2008.
SPE 151871 15
Nomenclature
DOE= Design of Experiments
H= Horizontal
HCPV= Hydrocarbon pore volumes
HI = Heterogeneity Index
kv/kh = Ratio of vertical to horizontal permeability
L.C. = Lorenz Coefficient
L= Lower
M= Middle
R.F. = Recovery Factor, %
SWAG = Simultaneous water alternating gas
U= Upper
VDP = Dykstra Parson Coefficient
WAG = Water Alternating Gas
WAG-1 = Water Alternating Gas Every One Month
WAG-3 = Water Alternating Gas Every Three Month
WAG-6 = Water Alternating Gas Every Six Month
WAG-12 = Water Alternating Gas Every Twelve Month
APPENDIX A
In the following section a summary of the experimental design runs are given. This table shows the
required runs to perform this experimental design. The first column indicates the experimental number
and the next seven columns represent the levels of the modeling parameters for each experiment. The
last column represents the obtained recovery factor (response) from each scenario.
Injection CO2 Rate- Well Spacing,
Run No. H.I. Injection Type Injection Well Type Injection Pattern kv/kh R.F. %
MMSCF Acre
1 0.65 WAG-6 1 V 9-Spot 0.75 15 44.36
2 0.2 Continuous 200 V 9-Spot 1 60 63.54
3 0.8 Continuous 200 H-Lower 5-Spot 0.1 120 48.69
4 0.8 Continuous 200 H-Upper 5-Spot 0.1 120 47.70
5 0.2 WAG-3 200 V 9-Spot 0.25 60 70.20
6 0.1 WAG-6 100 V 9-Spot 1 15 70.24
7 0.2 WAG-1 500 V 9-Spot 0.25 120 77.27
8 0.72 WAG-12 50 V Line Drive 0.5 120 37.92
9 0.45 WAG-12 1000 V 9-Spot 0.5 60 82.22
10 0.8 Continuous 10 V Line Drive 0.5 15 69.04
11 0.1 WAG-12 500 V Line Drive 0.25 60 77.34
12 0.45 Continuous 200 H-Upper 5-Spot 0.1 120 63.17
13 0.2 Continuous 200 H-Middle 5-Spot 0.1 120 66.46
14 0.2 WAG-6 10 H-Middle 9-Spot 1 120 42.38
15 0.3 WAG-3 500 V 5-Spot 0.5 15 80.09
16 0.65 WAG-3 50 V 9-Spot 0.1 60 44.89
17 0 WAG-3 1 V Line Drive 1 120 40.56
18 0.2 WAG-6 1000 V Line Drive 1 120 66.37
19 0.65 SWAG 500 V 9-Spot 0.1 15 55.03
20 0.45 Continuous 1 V 9-Spot 1 15 20.85
21 0 SWAG 50 V 5-Spot 0.75 60 79.53
22 0.2 WAG-6 200 V 5-Spot 0.1 15 11.04
23 0.2 Continuous 200 V 5-Spot 0.1 15 66.35
24 0.8 WAG-12 50 H-Lower 5-Spot 0.75 60 22.51
25 0.1 WAG-6 20 V 5-Spot 0.1 60 64.12
26 0.3 WAG-3 20 V Line Drive 0.1 120 58.57
27 0.8 SWAG 100 V 9-Spot 0.75 60 53.84
28 0.45 WAG-3 500 V 5-Spot 1 60 83.91
29 0.3 WAG-6 10 V 5-Spot 0.1 120 66.13
30 0.1 SWAG 10 V Line Drive 0.75 60 73.20
31 0.3 WAG-1 100 V 5-Spot 0.5 120 78.97
32 0.65 WAG-3 100 V 5-Spot 1 15 54.61
33 0.2 SWAG 1000 V Line Drive 0.1 15 82.95
34 0.1 WAG-6 200 V 5-Spot 1 15 74.06
35 0.45 WAG-1 500 H-Lower Line Drive 0.75 60 76.63
36 0.3 SWAG 50 V 5-Spot 0.5 15 84.51
37 0.45 WAG-1 1000 V 5-Spot 1 120 45.78
38 0.72 SWAG 1 V Line Drive 0.5 120 46.28
39 0.2 WAG-6 20 V 9-Spot 0.25 120 62.93
40 0.72 WAG-3 200 V 9-Spot 0.25 120 46.98
41 0 Continuous 10 V 5-Spot 0.1 120 22.44
42 0.72 WAG-12 1 V 5-Spot 0.1 60 18.31
43 0.1 WAG-3 20 V Line Drive 0.75 120 64.10
16 SPE 151871