economic incentives for co2 sequestration in oil ...

3 downloads 0 Views 165KB Size Report
production based credits, and modified accounting rules (e.g. accelerated depreciation). Here, the analysis is designed to estimate a CO2 storage credit that ...
ECONOMIC INCENTIVES FOR CO2 SEQUESTRATION IN OIL RESERVOIRS Yousef Ghomian* Mehmet Urun Gary A. Pope Kamy Sepehrnoori Christopher Jablonowski The University of Texas at Austin * Lead Author, now with Chevron USA, 713-372-1373, [email protected]

ABSTRACT CO2 capture and storage (CCS) is receiving significant analytical attention from scientists, engineers, and economists as a carbon management option. In addition to technical feasibility studies, detailed models of the performance and economics of CCS projects are important and can serve as the basis for policy formulation and regulation. To this end, this study provides a bottom-up analysis of a coupled enhanced oil recovery (EOR) and CCS project in a mature oil reservoir. The core of the bottom-up analysis is a compositional reservoir simulator that accounts for geological setting, reservoir physics and injection methods, and the configuration of wells. The simulator provides accurate and consistent estimates of reservoir performance over time. Reservoir simulation results are used in conjunction with experimental design and a project economic model to develop estimates of project performance and economics. Regression analysis is used to generate simplified response surfaces for key project variables and these are used to carry out a probabilistic analysis. We find that coupled EOR and CCS projects are unlikely to be initiated in a low oil price environment unless some form of CO2 credit is provided; coupled projects are economically unattractive because of the large probability of negative NPV outcomes. In a high price environment, coupled EOR and CCS projects are more economically attractive, but such projects are unlikely to be initiated on their own merits; operators in these cases are likely to pursue conventional EOR projects and would need to be compensated for CO2 stored beyond the normal economic limit. The technical performance and economics of these projects is complex, but it appears clear that any design of a CO2 credit scheme includes (at the least) differentials based on reservoir characteristics, CO2 injection method, and the configuration of wells.

INTRODUCTION CO2 concentration in the atmosphere has increased significantly from the pre-industrial age to today. This increase is of interest because CO2 is a greenhouse gas and may cause global warming, other changes in the environment, and economic damages. It is possible that reducing emissions may be economically optimal. Therefore, technical and economic studies of CO2 capture and storage (CCS) are needed to inform the policy debate and to determine the optimal course of action. The global oil and gas industry may have an important role to play if CCS projects are pursued. The industry is experienced in planning and operating large and complex injection projects. Geological storage is possible in mature oil reservoirs, deep saline aquifers, coal seams, and other settings. Over 35 million tons of CO2 have been injected into oil reservoirs for the purpose of enhanced oil recovery (EOR) (Moritis, 2002). Therefore, it is likely that coupled EOR and CCS projects will be part of a global CO2 storage solution if one is needed.1 This paper develops a bottom-up analysis of a coupled enhanced oil recovery (EOR) and CCS project in a mature oil reservoir. A coupled EOR and CCS project is defined here by the following two features: (i) CO2 is injected into the reservoir until original reservoir pressure is attained (that is, more CO2 than would be injected in a conventional EOR-only project), and (ii) the reservoir is not in contact with an aquifer nor has it been flooded in secondary or tertiary oil recovery. In a conventional EOR project, the cost of CO2 greatly affects the quantity of CO2 purchased and therefore the quantity of oil recovered. The implication of this relationship for a coupled EOR and CCS project is that the cost of CO2 will have to be lower to provide a private incentive for operators to inject CO2 beyond the normal economic limit, all things being equal. The core of the bottom-up analysis is a compositional reservoir simulator that accounts for geological setting, reservoir physics and injection methods, and the configuration of wells. The simulator provides accurate and consistent estimates of reservoir performance over time. Reservoir simulation results are used in conjunction with experimental design and a project economic model to develop estimates of project performance and economics. Regression analysis is used to generate simplified response surfaces for key project variables and these are used to carry out a probabilistic analysis. The purpose of this research is to develop a general understanding of coupled EOR and CCS project economics, and to estimate the level of direct economic incentives that may be required to motivate these projects. Many types of incentives are possible, including general tax credits, storage based credits, production based credits, and modified accounting rules (e.g. accelerated depreciation). Here, the analysis is designed to estimate a CO2 storage credit that would be paid to operators who engage in coupled EOR and CCS projects. The credit is computed in units of dollars per thousand standard cubic feet of CO2 stored ($/MSCF). In summary, this is accomplished by computing the CO2 credit that makes project net present value (NPV) equal to zero.

CO2 CAPTURE, COMPRESSION, AND TRANSPORTATION Stationary facilities such as fossil fuel electric power plants and petroleum refineries are significant sources of CO2 emissions. If carbon capture projects are pursued, these single point sources are likely to be targeted first. Capturing, compressing, and transporting CO2 from these facilities is currently complex and costly. Over time, these costs should decline. The summary of costs presented here is used as the basis for the specification of costs in the economic analysis to follow.

1

Currently, the Weyburn project is the only on-going commercial coupled EOR and CCS project (Malik and Islam, 2000). Carbon dioxide is transported from a North Dakota coal-gasification plant through pipelines and is injected into the Weyburn oil field. Other forms of geological CO2 storage have been studied, but to date, very few have been implemented.

Capture and Compression There are different methods for capturing CO2 from an emissions stream. In chemical adsorption/stripping, CO2 reacts with solvents to form a weak-bonded intermediate compound which then is broken utilizing heat, regenerating the original solvents for reuse and producing a CO2 stream. Commonly used chemicals are alkanolamines such as monoethanolamine (MEA) which can reduce the CO2 concentration to as low as 100 ppm at low pressure (200 psi) (Nguyen, 2003).2 After CO2 is captured, it must be compressed to pressures above 1200 psi for transportation to ensure single-phase flow. Ennis-King and Paterson (2002) showed that the isothermal work required to compress CO2 at 35 °C from an initial pressure of 14.7 psi to 1750 psi requires 0.275 MJ. Jassim and Rochelle (2006) showed that in terms of energy costs, energy consumption of a simple MEA absorption and stripping process along with CO2 compression for transportation could consume 38% of an electric power plant’s energy. For example, a 500 MW Integrated Gasification Combined Cycle (IGCC) power plant, operating at 80 percent capacity, would generate about 7300 tons of CO2 per day or 141 million standard cubic feet per day (MMSCF/day) or 51.4 trillion standard cubic feet per year (TSCF/yr) (Heddle et al., 2003). Rochelle et al. (2005) have estimated the total costs of capture and compression (to 2200 psi) for this size power plant to be $45/ton CO2 which is equivalent to $2.60 per thousand standard cubic feet (MSCF). Transportation Transportation of CO2 through pipelines is the most common mode for EOR projects. The cost of pipeline transportation is a function of pipeline diameter, length, operating pressure, and flow rate. Heddle et al. (2003) estimated the total annual cost per ton of CO2 transportation; we use their figures to estimate the transportation cost for the 500 MW power plant. If we assume that CO2 is transported to the reservoir from sources 200-400 miles away (e.g., from Houston to oil fields along the Gulf Coast), the transportation cost will vary from 0.5 to 1.2 $/MSCF of CO2.

CO2 STORAGE To model the injection and storage components of a coupled EOR and CCS project, we use a compositional reservoir simulator. The simulation forms part the analytical workflow that includes experimental design, economic analysis, response surface modeling, and probabilistic analysis. The workflow is depicted in Figure 1 at the end of this section. Reservoir Simulation As stated in the introduction, a coupled EOR and CCS project is defined here by the following two features: (i) CO2 is injected into the reservoir until original reservoir pressure is attained, and (ii) the reservoir is not in contact with an aquifer nor has it been flooded in secondary or tertiary oil recovery. The first condition results in more injected CO2 than would be observed in a conventional EOR-only project. That is, in the latter stages of most of the simulations, producing wells are shut-in and the only activity is CO2 injection. The simulation of coupled EOR and CO2 sequestration projects is complex. Reservoir simulators must account for geological setting, reservoir physics and injection methods, and the configuration of wells. For example, different reservoir types such as carbonate and sandstone reservoirs have different responses to CO2 flooding based on reservoir characteristics. The injection method is also important. Water-alternatinggas (WAG) injection methods can enhance sweep efficiency and increase oil recovery compared to continuous injection, but the methods result in different volumes of sequestered CO2. Tighter well spacing increases sweep efficiency but also causes CO2 to be recycled earlier in the project (increasing costs).

2

CO2 injected into a reservoir for EOR must be free of contaminants. Impurities can be detrimental to CO2 miscibility with oil. This requirement will add additional costs to the capture process.

In this study, we performed detailed 3D compositional reservoir simulation using CMG's GEM simulator based on the following three categories of assumptions: • • •

Geological setting: carbonate or sandstone reservoirs; Injection method: WAG or continuous CO2 injection; Well spacing: 20 or 40 acres; these choices were based on a preliminary study of optimal well spacing.

Table 1 provides a summary of reservoir attributes for both geological settings. Combining the elements from these categories results in 8 unique reservoir scenarios. Table 1. Major Assumptions for Carbonate and Sandstone Reservoirs Carbonates Sandstones Permeability distribution

Layered

Stochastic

Vertical to horizontal permeability ratio

0.01

0.1

Average porosity, fraction

0.11

0.23

Reservoir temperature, °F

110

150

0.58

0.33

0.32

0.18

Remaining oil saturation before flood, fraction Remaining oil saturation after flood, fraction

Uncertainty and Design of Experiments Within each of the 8 reservoir scenarios there is technical and economic uncertainty. A goal of this study is to perform a probabilistic analysis of these uncertainties to depict a range of possible outcomes of performance and economics. One approach would be to specify inputs to the reservoir simulator as random variables and simulate numerous outcomes. But if the number of uncertain variables is large, and the solution time for each case if long (as it is here), then this approach is not feasible in practice. Instead, we use experimental design to reduce the number of simulations to be run. The following 6 variables were selected for the uncertainty analysis: • • • • • •

Oil price (A) CO2 cost (B) Flood performance (C) Drilling cost (D) Operation cost (recycling, lifting, and general) (E) Discount rate (F).

Two sets of assumptions for these 6 variables were examined, hereafter referred to using the variable names A-F and case names of Low Price and High Price; the assumptions are summarized in Tables 2 and 3. A fractional 3-level factorial experimental design method was used to reduce the number of simulations. This approach results in 78 cases for each of the 8 reservoir scenarios. A broader discussion of this analysis is available in Ghomian (2008). Project Economic Model The results from each case run in the reservoir simulator are inputs into a project economic model. The economic model accounts for all major cash flows and non-cash items. This includes revenue, royalty, costs, depreciation, taxes, EOR tax credits, etc. and computes the net present value (NPV) for the project. For each case, we compute the CO2 credit that makes project NPV equal to zero. It is unclear at this stage

how such a credit scheme would be implemented in practice. For this research, we assume that the credit is paid during the operation of the project, and that it is paid as a constant nominal credit. This approach is used so that the credit can be easily compared to the market price assumption.3 Table 2. Specification of Variables for Sensitivity Analysis: Low Price Median Factor Low (-1) High (1) (0) Oil price, $/bbl

15

35

55

CO2 price, $/MSCF

1

2.5

4

Flood performance, MSCF/bbl

7

12

20

0.75

0.9

1.05

Fixed operational costs, $mm/month

0.0192

0.024

0.029

Recycle costs, $/MSCF

0.56

0.7

0.84

Lift costs, $/bbl

0.32

0.4

0.48

10

15

20

Drilling cost, $mm/well Operational costs

Discount rate, %

Table 3. Specification of Variables for Sensitivity Analysis: High Price Median Factor Low (-1) High (1) (0) Oil price, $/bbl

55

75

95

CO2 price, $/mcf

1

2.5

4

Flood performance, MSCF/bbl

7

12

20

Drilling cost, $mm/well

1

1.2

1.4

Fixed operational costs, $mm/month

0.0224

0.028

0.032

Recycle costs, $/MSCF

0.67

0.84

1.0

Lift costs, $/bbl

0.4

0.5

0.6

Discount rate, %

10

15

20

Operational costs

3

Other computations are of course possible. For example, one could set CO2 price to zero and compute the CO2 price that makes NPV equal to zero.

In cases where project NPV is positive, the credit is also positive, indicating that a CO2 credit is not required. We capture this value as-is for the analysis, but recognize that this does not mean that such a coupled EOR and CCS project would proceed on its own merits. In a case like this, a conventional EOR project is likely to be profitable (and some CO2 will be stored), but this does not mean that the operator will elect to store additional CO2 beyond the normal economic limit. Again, we do not make any adjustments in these cases, the CO2 credit is recorded assuming a coupled project. A second observation in cases where the credit is positive is that the credit is suggestive of the willingness to pay for CO2 above and beyond the market price assumption. In contrast, when the credit is negative, it indicates that a credit would be required to motivate the coupled project. Response Surface Modeling Response surface modeling refers to the process of using regression analysis to relate the results for key project variables from the reservoir simulator and project economic model to the uncertain variables. Specifically, we specify the dependent variable as the required CO2 credit as computed in the economic model (positive or negative), and specify the independent variables as the 6 uncertain variables defined in Tables 2 and 3. A model is specified for each of the 8 reservoir scenarios. The sample size is determined by the simulations selected for experimental design; in this case there are 78 observations for each scenario. We specify a simple linear regression as follows: y = α + Xβ + ε , where y is n x 1 vector of observations on the dependent variable, α is an intercept term, X is a n x k matrix of observations on the independent variables (with an intercept term), β is a k x 1 vector of coefficients (to be estimated), and ε is a random

(

)

error term, ~ iid 0, σ . Tables 4 and 5 provide the final specifications for the response surfaces. Only statistically significant variables are retained (95% criterion). 2

Probabilistic Analysis In the last step of the analysis, the response surfaces are used to develop a PDF of the CO2 credit for each case. The 6 variables are specified as PDFs and a Monte Carlo simulation is implemented. We assume the variables are statistically independent. For each realization, the response surface is used to estimate the CO2 credit. The variables for the Low Price case were specified as depicted in Figure 2; for the High Price case, the oil price distribution is shifted so that the mean equals 75 $/bbl. Table 4. Response Surface Equations for CO2 Credit: Low Price Response Equation for CO2 Credit Well Injection Reservoir as $/MSCF Spacing Scheme Rock Type

40 Acre Spacing

20 Acre Spacing

CO2 Injection WAG Injection CO2 Injection WAG Injection

Carbonates

2.65+0.13A-0.29B-0.31C

Sandstones

0.81+0.1A-0.33B-0.25C+10.2F

Carbonates

4.53+0.08A+-0.52B-0.18C-1.9D-16F

Sandstones

1.02+0.2A-B-0.4C+20.75F

Carbonates

-0.13+0.23A-0.5B-0.5C+30F

Sandstones

0.65+0.1A-0.4B-0.25C+11.6F

Carbonates

1.72+0.38A-1.18B-0.92C+57.8F

Sandstones

-0.38+0.35A-1.35B-0.86C+48.5F

Table 5. Response Surface Equations for CO2 Credit: High Price Well Injection Reservoir Response Equation for CO2 Credit, $/MSCF Spacing Scheme Rock Type -12.1+0.35A-2.52C+216F Carbonates CO2 Injection Sandstones -0.36+0.1A-0.5C+42F 40 Acre Spacing -3.44+0.22A+-1.25C+110F Carbonates WAG Injection Sandstones 0.7+0.1A-0.6C+32.8F

20 Acre Spacing

CO2 Injection WAG Injection

Carbonates

3.05+0.08A-0.5B-0.43C

Sandstones

-8.9+0.33A-2.3C+185F

Carbonates

8.09+0.15A-0.37B-0.82C

Sandstones

-0.76+0.2A-0.8B-1.1C+79.6F

Figure 1. Analytical Workflow 1.

Define k Cases (8 total) a. Continuous or WAG injection b. Sandstone or carbonate c. 20 or 40 acres

2. Run n reservoir simulations for each case k based on the combinations of 6 uncertain parameters (n=78); combinations selected using DoE—fractional 3-level factorial

3. For each simulation result i (i=1..n): a. Compute project NPV b. Compute CO2 credit required to make NPV=0

4. 5. For each case k, use the final regression equation to perform a probabilistic analysis on the 6 uncertain parameters by specifying them as RV’s; using Monte Carlo analysis, develop a PDF for the CO2 credit

For each case k: a. For n observations, regress CO2 credit on a linear combination of the 6 uncertain parameters b. Retain significant variables and reestimate

Figure 2. Probabilistic Inputs: Low Price

Panel A. Oil price

Panel D. Operating costs

Panel B. CO2 price

Panel E. Flood performance

Panel C. Discount rate

Panel F. Drilling costs

Results and Conclusions Tables 6 and 7 provide the primary results of the probabilistic analysis. Examples of the PDFs for CO2 credit for 4 of the 16 cases are provided in Figures 3, 4, 5 and 6. For all of the Low Price cases, the probability of the need for a CO2 credit ranges from 21.4 to 94.9 percent. Within each reservoir type, the probabilities are consistently higher for the WAG cases relative to continuous injection; for the 40 acre cases this effect is exaggerated. This result indicates that the WAG injection method is more likely to result in a negative NPV. Also, we observe that less CO2 is stored during WAG injection than during continuous injection. For the 40 acre case and for each type of injection method, sandstones are less likely to require a credit than carbonates. For the 20 acre case and for each type of injection method, this result is reversed, and carbonates are less likely to require a credit than sandstones. In all but one case, the credit is positive, indicating that project NPV is positive and a credit is not required. Recall from the previous discussion that this result does not mean that such a coupled EOR and CCS project would proceed on its own merits. In these cases, a conventional EOR project is profitable (and some CO2 will be stored), but this does not mean that the operator will elect to store additional CO2 beyond the normal economic limit. In the 40 acre case, the sample means indicate that continuous injection is preferred in carbonates and sandstones. In the 20 acre case, this is reversed, and WAG injection is preferred

in both reservoir types. The standard deviation of the credit is generally higher when utilizing WAG. These points notwithstanding, given the significant probability for negative NPV outcomes, and the high standard deviations relative to the means, these projects are unlikely to be competitive for capital. Although we do not verify it here, it is probable that these projects would not be pursued even as conventional EOR.

Table 6. Summary of Probabilistic Analysis for CO2 Credit: Low Prices 40 Acre Well Spacing 20 Acre Well Spacing Carbonates Sandstones Carbonates Sandstones WAG Continuous WAG Continuous WAG Continuous WAG Continuous Probability of need for 94.90 CO2 credit,%

28.70

44.60

23.40

30.10

21.40

33.80

31.20

Sample mean of CO2 credit, $/MSCF

-2.89

1.42

0.75

1.43

4.17

3.51

3.21

1.13

Sample standard deviation of CO2 credit, $/MSCF

1.73

2.53

4.10

2.20

7.61

4.80

7.30

2.10

Table 7. Summary of Probabilistic Analysis for CO2 Credit: High Prices 40 Acre Well Spacing 20 Acre Well Spacing Carbonates Sandstones Carbonates Sandstones WAG Continuous WAG Continuous WAG Continuous WAG Continuous Probability of need for 34.00 CO2 credit,%

6.90

6.40

4.00

9.70

2.80

9.80

3.30

Sample mean of CO2 credit, $/MSCF

0.60

6.08

10.12

6.34

18.50

15.95

17.34

6.71

Sample standard deviation of CO2 credit, $/MSCF

7.24

4.27

6.47

3.29

15.03

7.61

13.56

3.33

The results are quite different for the High Price cases. The probability of the need for a CO2 credit ranges from 2.8 to 34 percent; these projects are more likely to be profitable. Within each reservoir type, the probabilities are consistently higher for the WAG cases relative to continuous injection. This result indicates that the WAG injection method is more likely to result in a negative NPV. For the 40 acre case and for each type of injection method, sandstones are less likely to require a credit than carbonates. For the 20 acre case and for each type of injection method, this result is reversed, and carbonates are (slightly) less likely to require a credit than sandstones. The NPVs are significantly higher and all of the credits are positive, indicating that credits are not required. In fact, these results suggest a significant willingness to pay for CO2 above current market prices. Again, it is important to stress that this result does not mean that such a coupled EOR and CCS project would proceed on its own merits. In the 40 acre case, the sample means indicate that continuous injection is preferred in carbonates, but that WAG is preferred in sandstones. In the 20 acre case, WAG is preferred in both carbonates and sandstones. The standard deviation of the credit is higher when utilizing WAG. In contrast to the Low Price cases, it is probable that these projects would be pursued (without credits) as conventional EOR given the probability of positive NPVs.

Figure 3. Low Price, 20 Acre, Carbonate, Continuous

Figure 4. Low Price, 40 Acre, Carbonate, WAG

Figure 5. High Price, 40 Acre, Sandstone, Continuous

Figure 6. High Price, 20 Acre, Carbonate, Continuous

Based on these results, we conclude the following: • •

• •

Coupled EOR and CCS projects are unlikely to be viable in a Low Price environment unless some form of CO2 credit is provided; these projects are judged to be unlikely because of the likelihood of negative NPV outcomes. In a high price environment, coupled EOR and CCS projects are more economically viable, but such projects are unlikely to be initiated on their own merits; operators in these cases are likely to pursue conventional EOR projects (and some CO2 would be stored as a result) but would need to be compensated for CO2 stored beyond the normal economic limit. The technical performance and economics of these projects is complex, but it appears clear that any design of a CO2 credit scheme includes differentials based on reservoir characteristics, CO2 injection method, and the configuration of wells. Additional analysis is desirable on the following issues: o For economically attractive conventional EOR projects, estimate the credit required to motivate incremental investment in coupled EOR and CCS. o Increase the number of scenarios and the number of uncertainties for probabilistic analysis. o Implement the probabilistic analysis with jointly distributed variables. o Estimate global CO2 storage capacity for coupled EOR and CCS projects. o Evaluate other financial incentive structures for coupled project incentives (e.g. accelerated depreciation).

ACKNOWLEDGEMENTS We would like to gratefully acknowledge the funding for this study provided partially by Ministry of Science, Research, and Technology of I. R. of Iran and by the sponsors of the Geological CO2 Storage Joint Industry Project at UT-Austin (BP, Chevron, CMG, ENI, ExxonMobil, Luminant and Shell) and CMG, Ltd. for the use of their reservoir simulation software.

REFERENCES Bachu, S., and Stewart, S., "Geological Sequestration of Anthropogenic Carbon Dioxide in the Western Canada Sedimentary Basin: Solubility Analysis," Journal of Canadian Petroleum Technology, Vol. 41, No. 2, p. 32-40, 2002. Bryant, E.: Climate Process and Change, Cambridge University Press, Cambridge, UK, p. 209, 1997.

Davison, J. E., Freund, P. and Smith, A.: Putting Carbon Back in the Ground, IEA Greenhouse Gas R&D Program, Cheltenham, U.K., ISBN1 89837328, 2001. Ennis-King, J., Paterson, L., "Engineering Aspects of Geological Sequestration of Carbon Dioxide," paper SPE 77809 presented at the SPE Asia Pacific Oil and Gas Conference and Exhibition held in Melbourne, Australia, 8-10 October, 2002. Ghomian, Y., "Reservoir Simulation Studies for Coupled CO2 Sequestration and Enhanced Oil Recovery,” Ph.D. Dissertation, University of Texas at Austin, 2008. Heddle, G., Herzog, H., and Klett, M., "The Economics of CO2 Storage," Report No. MIT LFEE 2003-003, Laboratory for Energy and the Environment, Massachusetts Institute of Technology, 2003. Jassim, M. S., Rochelle, G. T., "Innovative Absorber/Stripper Configuration for CO2 Capture by Aqueous Monoethanolamine," Ind. Eng. Chem. Res., Vol. 45, p. 2465-2472, 2006. Malik, Q. M., and Islam, M. R., " CO2 Injection in the Weyburn Field of Canada: Optimization of Enhanced Oil Recovery and Greenhouse Gas Storage With Horizontal Wells," paper SPE 59327 presented at the SPE/DOE eighth Symposium on Improved Oil Recovery, Tulsa, Oklahoma, April 3-5, 2000. McLean, B., "Initial Thoughts on Safe and Effective Deployment of Geologic Sequestration," Proceeding of third Annual Conference on Carbon Capture and Sequestration, DOE/NETL, Alexandria, VA, 2004. Moritis, G., "California Steam EOR Produces Less; Other EOR Continues," Oil & Gas Journal, Vol. 100, No. 15, p. 43, April 15, 2002. Nguyen, D. N., "Carbon Dioxide Geological Sequestration: Technical and Economic Reviews," paper SPE 81199 presented at the SPE/EPA/DOE Exploration and Production Environmental Conference held in San Antonio, Texas, 10-12 March, 2003. Rochelle, G., Jassim M., Beitler, C., Rueter, C., and Searcy, K., "Integrating MEA Regeneration with CO2 Compression and Peaking to Reduce CO2 Capture Costs," U.S. Department of Energy Report No. DEFG02-04ER84111, June 2005.