Appendix A - Key data assumptions
This report uses estimated probabilities and magnitudes of risks as reported in the FutureGen 1.0 risk assessment for the Jewett, Texas project (Jewett). As indicated in the report, there are certain instances where the FutureGen risk assessment (FGRA) for Jewett either insufficiently characterizes or omits the estimate of probability and/or magnitude of certain risks. The sponsor group determined that the missing elements were important to test the model, gain a sense of the order of magnitude of potential impacts, and to gain insight into the full array of potential damages that could arise from a CCS project. Given limited resources, the sponsor group undertook a review of missing information and developed best estimates based on additional written papers and industry experience. Under normal conditions for a ‘real’ project, additional effort would be needed to develop actual site specific parameters for the missing data.
This Appendix describes the missing data to be assessed, the process for developing each assumption, and the final assumptions of probability or magnitude of specified risks used in the analysis. These assumptions are presented to ensure transparency. It is important to note that the assumptions in this Appendix have not been reviewed by FutureGen and are not put forward as assessments for the Jewett site. Rather, they are assumptions that appear to be in a reasonable range of likelihood and are acceptable to the sponsor group for the sole purpose of testing the model and exploring potential results.
Background
Based on the preliminary review of readily available information, IEc consulted with the sponsor group to estimate values for six parameters that were either insufficiently characterized (e.g., reported in qualitative terms) or not addressed in FutureGen Risk Assessment information:
- An annual likelihood that CO2 separation or compression & drying equipment and/or the associated piping would rupture.
- A forecast of future U.S. unit values (i.e., $ per ton) for CO2 through Year 2110.
- An estimated duration and flux of potential releases to groundwater from a deep well release of CO2 prior to its detection (and subsequent action to stop the release).
- An estimated duration and flux of potential releases to the atmosphere from a deep well release of CO2 prior to its detection (and subsequent action to stop the release).
- The carbonate mineral content of the Carrizo-Wilcox or Queen City aquifer matrix in central Texas.
- An estimated cost to stop a release from a deep well extending 10,400 ft below surface (not including potential groundwater treatment costs).
What follows is a description of the sources of information consulted and, where useful, a summary of the discussion points raised in selecting a useable assumption for each area above.
1. An annual likelihood that CO2 separation or compression & drying equipment and/or the associated piping would rupture.
The sponsor group sought publicly available reports indicating the annual percent chance of rupture and asked industry members within the group to validate that reported ranges were reasonably consistent with their experience.
The assumptions used in the Carbon Capture Sequestration Valuation Study (‘CCSVS’) were based on a report released through the Health and Safety Executive (HSE) of the United Kingdom in 2006. The report was prepared by Clive Nussey and is entitled: ‘Failure frequencies for major failures of high pressure storage vessels at COMAH sites: A comparison of data used by HSE and the Netherlands.’
The report categorizes ruptures by hole size: small hole, large hole, and catastrophic release. Table 4 on page 10 of the report provides the following HSE estimates of annual probability of failure and these were used in the CCSVS analysis:
| Type of failure | Annual % chance of rupture |
|---|---|
| Catastrophic | 2-6 × 10-6 |
| Large Hole | 5 × 10-6 |
| Small Hole | 55 × 10-6 |
This report can be found online at: http://www.hse.gov.uk/comah/highpressure.pdf.
2. A forecasy of future U.S. unit values for CO2 through year 2110.
The sponsor group felt that projections of future prices for CO2 would be a sufficient proxy for CO2 unit values. Further, the sponsor group did not feel qualified to unanimously endorse any specific projection as being more likely than any other. That said, a majority of sponsor group members were in favor of using projections developed by economist William Nordhaus in 2010 and published in an article entitled: ‘Economic aspects of global warming in a post-Copenhagen environment,’ in Proceedings of the National Academy of Sciences, 107(26): 11721-11726 (and available online at: www.pnas.org/cgi/doi/10.1073/pnas.1005985107).
The article and additional information about the RICE models used by Dr. Nordhaus are also available on the following website: http://nordhaus.econ.yale.edu/RICEmodels.htm. In this use of the model, the economically optimal emission reductions path leads to a peak in atmospheric concentration of CO2 at just under 600 ppm CO2 in around 2080, before stabilizing at around 500 ppm CO2 in 2200. The corresponding years and modeled dollar values for this path were used in the CCSVS analysis and are reprinted in the following table:
| Year | Optimal/600 PPM (low) | Limit to 2 degrees (high) |
|---|---|---|
| $/Tonne CO2 | $/Tonne CO2 | |
| 2015 | 10.28 | 21.49 |
| 2025 | 17.46 | 38.58 |
| 2035 | 24.12 | 61.45 |
| 2045 | 32.21 | 95.50 |
| 2055 | 41.85 | 144.45 |
| 2065 | 53.15 | 201.31 |
| 2075 | 66.30 | 273.04 |
| 2085 | 81.40 | 321.71 |
| 2095 | 98.49 | 253.46 |
| 2105 | 117.43 | 197.30 |
| 2115 | 137.95 | 193.21 |
3. An estimated duration and flux of potential releases to groundwater from a deep well release of CO2 prior to its detection.
4. An estimated duration and flux of potential releases to the atmosphere from a deep well release of CO2 prior to its detection.
The assumptions for both sets of fluxes were developed together and so the method is reported here for both data points.
Step 1: Identify Release Scenarios, Probabilities and Flux.
The FutureGen risk assessment for the Jewett project contained three tables describing site release scenarios, event probabilities by release scenario, and of those events, and the predicted annual flux from those events (the tables have been reproduced below). Since these tables do not utilize the same events or contain similar information for all events, the first step required the development of a list of release scenarios and the associated probability and flux.
Table 6-1. Summary of sequestration site release scenarios
| Release scenario | Exposure duration | Potential volume | Initial release to | Receptors |
|---|---|---|---|---|
| Upward leakage through the caprock due to catastrophic failure and quick release | Short-term | Variable, could be large | Air | Humans Ecological |
| Upward leakage through the caprock due to gradual failure and slow release | Long-term | Small | Air, groundwater | Humans Ecological |
| Upward leakage through the CO2 injection well(s) | Short-term and long-term | Variable, could be large | Air, groundwater | Humans Ecological |
| Upward leakage through deep oil & gas wells | Short-term and long-term | Variable, could be large | Air, groundwater | Humans Ecological |
| Upward leakage through undocumented, abandoned, or poorly constructed wells | Short-term and long-term | Variable, could be large | Air, groundwater | Humans Ecological |
| Release through existing faults due to the effects of increased pressure | Long-term | Variable, could be large | Air, groundwater | Humans Ecological |
| Release through induced faults due to the effects of increased pressure | Long-term | Variable, could be large | Air, groundwater | Humans Ecological |
| Lateral or vertical leakage into non-target aquifers due to lack of geochemical trapping | Long-term | Variable | Groundwater | Humans Ecological |
| Lateral or vertical leakage into non-target aquifers due to inadequate retention time in the target zone | Long-term | Variable | Groundwater | Humans Ecological |
| Gas intrusion into groundwater (with potential release of radon) | Long-term | Low | Groundwater | Humans Ecological |
Table 2. Event probabilities release scenario for jewett, TX site
| Release scenario | Release likelihood (annual) |
|---|---|
| Pipeline Events | |
| Pipeline Rupture | 1-in-200 (0.5%) |
| Pipeline Puncture | 1-in-100 (1.0%) |
| Sequestration Site Events | |
| Wellhead Equipment Rupture | 6-in-100,000 (0.006%) |
| CO2 Injection Well Leak | 3-in-100,000 (0.003%) |
| Other Well Leak | 7-in-100 (7.0%) |
| Rapid Leakage through Caprock | 2-in-10 billion (0.00000002%) |
| Slow Leakage through Caprock | 4-in-100,000 (0.004%) |
| Release through Existing, Induced Faults | 2-in-100 million (0.000002%) |
| Source: FGRA Table 6-11 | |
Table 4. Futuregen risk assessment table 5-8, PP 5-41 to 5-46 with study analysis
| Site | Mechanism | Annual flux ifevent occurs | |
|---|---|---|---|
| Minimum(MT/YR) | Maximum(MT/YR) | ||
| Jewett | Leakage via Upward Migration through Caprock due to Gradual and slow release | 0 | 4,918 |
| Leakage via Upward Migration through Caprock due to catastrophic failure and quick release | NA | NA | |
| Leakage through existing faults due to increased pressure (regional overpressure) | 118 | 3,526 | |
| Release through induced faults due to increased pressure (local overpressure) | 24 | 705 | |
| Leakage into non-target aquifers due to unknown structural or stratigraphic connections | 2,350 | 79,910 | |
| Leakage into non-target aquifers due to lateral migration from the target zone | 28,928 | 867,845 | |
| Leaks due to deep CO2 wells, high rate | 11,000 | 11,000 | |
| Leaks due to deep CO2 wells, low rate | 200 | 200 | |
| Leaks due to deep O&G wells, high rate | 11,000 | 11,000 | |
| Leaks due to deep O&G wells, low rate | 200 | 200 | |
| Leaks due to undocumented deep wells, high rate | 11,000 | 11,000 | |
| Leaks due to undocumented deep wells, low rate | 200 | 200 | |
Six scenarios were used:
- Rapid release through cap rock.
- Gradual release through cap rock.
- CO2 injection wells.
- Oil & Gas wells.
- Undocumented wells.
- New and Induced faults.
Step 2: Assume detection levels and timing
In order to assume a volume of released CO2, the group used an assumption about the amount of released CO2 that could be detected through monitoring. Assumed detection levels were based on the assumed sensitivity of the monitoring equipment and the frequency of monitoring. A default was developed based on anecdotal information collected from project operators and researchers suggesting a detection level of .3TPD is a detection level (meaning smaller amounts of leakage would not be detected).
Step 3: Assume movement to near surface waters and atmosphere
The paper: ‘Detection of CO2 leakage by eddy covariance during the ZERT project’s CO2 release experiments,’ by Jennifer L. Lewicki, George E. Hilley, Marc L. Fischer, Lehua Pan, Curtis M. Oldenburg, Laura Dobeck, and Lee Spangler modeled movement of CO2 released in the subsurface. The model suggested there would be some attenuation of released CO2 in the rock formation as it migrated to the surface. Some of this CO2 would be released into near-surface groundwater and some CO2 would migrate all the way to the atmosphere. This basis approach was applied as a convention for subsurface releases from closed wells or through formations. The convention was to assume that 0.01% of seepage reaches groundwater and 10% of that reaches the atmosphere. Releases from operational wells were assumed to migrate quickly to the surface and then be released to the atmosphere.
Applying these steps to the six release events suggests the following:
1. Rapid release through cap rock.
Because there is no estimate of flux for this scenario in the FGRA, this pathway has not been analyzed. An option would be to use the FGRA probability for some initial period of years (perhaps 20) and assume that 50% (or some other percentage) reaches both groundwater and atmosphere. With no clear basis or consensus regarding such assumptions, it was decided not to develop this scenario.
2. Gradual release through cap rock.
- Years 1-50 (operational period): Assume that there is subsurface monitoring and surface monitoring that takes a 30 day average. Assume detection level of 0.3 T/day and maximum flux of 13.5TPD per FGRA. Assume it take 3 months to correct after detection (for a total of 120 days).Minimum = 120D*0.3T = 36 TMaximum = 120D*13.5T = 1,620 T
- Years 50-100 (Closure period): Assume that there is subsurface monitoring and surface monitoring that takes a 6 month / 180 day average. Assume detection level of 0.3 T/day and maximum flux of 13.5TPD per FGRA. Assume it take 3 months / 90 days to correct after detection (for a total of 270 days).Minimum = 270D*0.3T = 81 TMaximum = 270D*13.5T = 3,645 T
- Groundwater flux: estimate 0.01% reaches groundwater: Operational: Min = 0.01%*36T = 0.004 T // Max = 0.01% *1,620T = .16T Closure: Min = 0.01%*81T = 0.01 // Max = 0.01% *3,645T = 0.36T
- Atmospheric flux - assume 10% of GW impact reaches the atmospherePotential for greater movement through wells, less impact on water
3. CO2 injection wells
- Operational wells – assume that there are sensitive detection levels, continuous monitoring, max 30 days to control; Min: use 0.1 T from In Salah report Max: use min flux from FGRA of .55 TPD = 30 D * .55 TPD = 16.5 T
- Closed wells – assume detection sensitive but 6 months (180D) to control; Min: use 0.1 T from In Salah – 180D * 0.1TPM = 0.6 T Max: use min flux from FGRA of .55 TPD = 180 D * .55TPD = 99 T
4. Oil & Gas (O/G) wells
Assume same leakage / detection volume as CO2 injection wells but that all O/G wells in project area are like closed wells. FGRA gives this class of wells a higher probability of occurrence – This should perhaps be altered given the regulatory requirements for AOR but was not for the analysis.
Min: use min flux from FGRA of 180D * .55 TPD = 99 T
Max: use max flux from FGRA of 180 D * 30 TPD = 5,400 T
5. Undocumented wells
These wells are covered in the other well category which has a higher probability of occurrence. Since they are undocumented, there may not be as rapid detection. However, since they are undocumented and therefore difficult to make accurate assumptions about, and given AOR requirements, it was decided to omit these wells from the analysis.
6. New and Induced faults
FGRA assigns these a very low probability. The group agreed with an assumption that such releases would be found in the first years of operation and should not be a problem after injection ceases. Assume quick detection but up to 6 months to control.
Min: since FGRA min flux is 0.07 TPD – a level we believe is undetectable; use
Min detectable flux (GCCSI) of .3 TPD * 180 D = 55 T
Max: use max flux from FGRA of 9.6 TPD * 180 D = 1,728 T
5. The carbonate mineral content of the carrizo-wilcox or queen city aquifer matrix in central texas.
This value relates to the potential buffering capacity of the rock formation into which potentially leaked CO2 would migrate. The buffering capacity provides insight into the amount of potentially leaked CO2 that could migrate to the surface, impacting near surface ground water (including drinking water) or being released to the atmosphere. This value is determined regionally or for geologic formations.
Expert geologists were consulted regarding the geologic area near Jewett and they pointed to a paper published by El Sevier in 1987, entitled ‘Diagenetic evolution of Cenozoic sandstones, Gulf of Mexico sedimentary basin.’ This paper provides a peer reviewed estimate of range of the mineral content in the area of 0% to 4%. It was determined that 2% would be acceptable for purposes of the CCSVS analysis.
The paper is available through El Sevier:
‘Diagenetic evolution of Cenozoic sandstones, Gulf of Mexico sedimentary basin’
Lynton S. Land, Kitty Lou Milliken, Earle F. McBride
Department of Geological Sciences, University of Texas at Austin, Austin, TX 78713
U.S.A.
Received 27 October 1986. Available online 22 April 2003.
http://dx.doi.org/10.1016/0037-0738(87)90033-9
It can also be found online at:
http://www.sciencedirect.com/science/article/pii/0037073887900339
6. An estimated cost to stop a release from a deep well extending 10,400ft below surface (not including potential groundwater treatment costs).
The sponsor group consulted the well workover cost assumptions in a report entitled: ‘2008 – The Economic Impact of New Mexico’s Oil and Gas Industry,’ by Jay Lillywhite and C. Meghan Starbuck (available online at: http://www.energyadvancesnewmexico.com/files/NM_Economic_Impact.pdf) to develop a range of estimates that were further tested by informally consulting industry experts. The report suggested an average annual cost of roughly $15/foot for subsurface repair in wells that were about 8,000 feet deep. In consulting with industry, we further developed this estimate to account for the age of the well, and for normal and rare damage. The final estimate is included in the table below:
| Relative damage | Dollar estimate | Incidence |
|---|---|---|
| Average Damage | $50,000 – 300,000 | 90% |
| High End | $2M – 3.65M | 10% |
| Summary of proposed CCS valuation parameters | ||||
|---|---|---|---|---|
| Parameter description | Unit | Value | Primary data source | |
| 1 | Annual likelihood that CO2 Separation or Compression & Drying equipment/associated piping would rupture | % chance of rupture, annually | 2-6 × 10-6 for catastrophic5 × 10-6 for large hole55 × 10-6 for small hole | U.K. HSE Report on High Pressure Vessel Failure |
| 2 | Forecast of future U.S. unit values for leaked CO2 through Year 2110 | $/ton | Use the 600 ppm scenario with prices ranging from $10.28 to $137.95 over 100 yrs | William Nordhaus // RICE Model |
| 3 | Estimated duration and flux of potential releases to groundwater from a deep well release of CO2 prior to its detection (could be a point estimate, range or a distribution) | Flux (tons/unit time) and Duration (time) | See discussion above | See discussion above |
| 4 | Estimated duration and flux of potential releases to atmosphere from a deep well release of CO2 prior to its detection (could be a point estimate, range or a distribution) | Flux (tons/unit time) and Duration (time) | See discussion above | See discussion above |
| 5 | Carbonate mineral content of Carrizo-Wilcox or Queen City aquifer matrix in central Texas (~60 miles east of Waco, TX) | % of matrix | Range: 0% to 4%Use 2% for analysis | Report: Diagenetic Evolution Of Cenozoic Sandstones, Gulf Of Mexico Sedimentary Basin |
| 6 | Estimated cost to stop a release from a deep well extending 10,400 ft below surface (not including potential groundwater treatment costs, which will be addressed separately) | $ per well release | Average case: $50-300KHigh end (rare cases) $2-3.65M – per industry reviewPer discussion: Assume 90% occurrences are average and 10% are high cost | Consult with industry members of sponsor group and review report:The Economic Impact of New Mexico’s Oil and Gas Industry |
