Chapter 9 - Conclusions and uncertainties

The Monte Carlo approach used in this study provides an analytically rigorous foundation for understanding and managing prospective risks that could arise from CCS projects. The application of this model to a ‘realistic’ site-specific project, e.g., Jewett, TX FutureGen site, provides successful proof of concept that estimates of monetized damages with associated probabilities can be developed on a site-specific basis for CCS projects.

A key consideration that is reaffirmed by the results from this study is the importance of site location. Specifically, estimated damages are driven by the composition of the CCS plume, the operating structure of the plant, the integrity of the pipeline and site-specific geology. Notably, changes in these risk categories can vary damages estimates by orders of magnitude. Accordingly, results from this study suggest that well-sited, well-operated CCS projects have a relatively small potential for damages, but sound site selection and site-specific monitoring are essential.

Deriving estimates of prospective damages for the Jewett, TX CCS project necessitates that this study impute a variety of assumptions, each of which is associated with varying degrees of uncertainty. Specifically, the damages estimates discussed herein rely in substantial part on the FutureGen Risk Assessment and EIS analyses, both of which were subject to significant public review and comment. In general, the FutureGen analyses apply conservative assumptions, which are more likely to overstate the potential for damages when applied without modification in this study.

As described throughout this document, FutureGen analyses do not provide all of the inputs needed to estimate potential financial consequences if one or more releases occur at the plant, pipeline and/or sequestration portions of the project. We relied on publicly available information to address these additional data needs wherever possible, supplemented by expert professional judgment, as well as communications with industry and trade association experts.

With respect to the key conclusions and drivers of this analysis and associated results, we make the following observations, specific to the Jewett, TX site evaluated as proof of concept:

• Overall, estimated total damages are approximately $8.5 million (50^th percentile) and $18.6 million (95^th percentile) – These estimates include all potential adverse events over the 100-year analysis period, and are expressed in 2010$. These estimates translate to approximately $0.17 (50^th percentile) and $0.37 per tonne of CO₂ sequestered (50 million tonnes of CO₂ are expected to be sequestered at the Jewett, TX site);
• Potential releases at oil & gas/other wells in the sequestration site area are responsible for over 95% of estimated total damages – At the Jewett, TX site there are believed to be dozens of deep oil & gas wells and potentially other old wells. FutureGen analyses assign release probabilities to those wells (i.e., 7-in-100 annual chance of a release incident) that are multiple orders of magnitude higher than any other type of release event at the plant or sequestration site, contributing substantially to the prominence of this potential damages category.¹⁹ Several types of potential damages contribute substantially to this total, including the cost of actions to: stop the release; address groundwater contamination; pay for offsets to address CO₂ leakage; compensate for human health damages; and (much less significantly) address habitat/other damages.
• Potential damages associated with other types of sequestration site events are negligible – Estimated damages for aboveground wellhead incidents and injection well incidents are $0 for both 50^th and 95^th percentile cases, reflecting their extremely low event probabilities, i.e., 3-in-100,000 annual probability for injection well releases and 6-in-100,000 annual probability for aboveground wellhead releases;
• Potential damages associated with plant site events are negligible – Estimated damages for plant site releases are $0 for both 50^th and 95^th percentile cases, reflecting their extremely low event probabilities, i.e., 5.5-in-100,000 annual probability;
• Potential damages associated with pipeline events are low – Estimated damages for pipeline rupture events are $0 for 50^th percentile and approximately $0.3 million for 95^th percentile cases. Estimated damages for pipeline puncture events are $0 for 50^th percentile and approximately $0.2 million for 95^th percentile cases. Although the likelihood of pipeline rupture and puncture events are reasonably likely to occur during the 50 year operational period (annual probabilities of 1-in-200 and 1-in-100, respectively), if an event occurs the extremely rural setting severely limits potential human health damages, the amount of CO₂ in the affected 5 mile section of pipeline (in between safety shutoff valves) is relatively low (1,290 tonnes) and the potential for ecological/other damages is limited.
• If there were no H₂S in the sequestration stream, total estimated damages would be approximately 10% - 15% lower – Although H₂S is the primary driver of human health risks in this analysis, the extremely rural location for plant, pipeline and sequestration operations severely limits potential human health damages, minimizing the added impact of H₂S.

With respect to key uncertainties, we make the following observations specific to the Jewett, TX site evaluated:

• Estimated damages associated with CO₂ leakage are highly uncertain – The potential volume of CO₂ released to the atmosphere if an adverse event occurs, and the potential cost ($ per tonne) associated with such releases, are both highly uncertain. For uncertainty analysis purposes, we note that the ‘base case’ Monte Carlo model CO₂ leakage damages estimates at the sequestration site are approximately $2.0 million (50^th percentile) and $4.0 million (95^th percentile).To address CO₂ offset pricing uncertainty, the Monte Carlo model includes a relatively broad range of unit costs for CO₂ leakage; that is, between approximately $10 and $320 per tonne depending on the year. Based on currently available information and forecasts, it is difficult to envision offset prices exceeding this range. However, it is possible that CO₂ emissions remain unregulated in the future, resulting in no cost to project proponents for CO₂ releases to the atmosphere. In such a scenario, there would be no CO₂ leakage damages.With respect to volume uncertainty, in the ‘base case’ Monte Carlo model, potential leakage volumes at the sequestration site range between 99 tonnes and 5,400 tonnes. For sensitivity analysis purposes, if the model were changed to assume that each sequestration site leakage event released 50,000 metric tonnes to the atmosphere (instead of a randomly selected volume between 100 and 5,400 tonnes) 50^th and 95^th percentile damages estimates for CO₂ leakage would increase to approximately $36.6 million and $67.7 million, respectively using ‘base case’ Monte Carlo CO₂ offset pricing assumptions.
• Potential groundwater damages are uncertain, but likely constrained by site-specific factors – This study includes a series of bounding calculations designed to broadly assess potential groundwater damages arising from CCS at the Jewett, TX site. A more detailed assessment of site-specific damages arising from potential impacts to groundwater would require data that are not readily available for the Jewett, TX site, and a commitment of resources beyond the scope of this analysis.Nevertheless, readily available information indicates that there are several site-specific factors that act to limit the potential magnitude of damages at the Jewett, TX sequestration location, including but not limited to:
  • Prior to the sequestration project moving forward, project sponsors likely would exercise their existing options to purchase groundwater rights in the sequestration area;
  • Groundwater surrounding the Jewett, TX sequestration site currently is not used for commercial, industrial, or other purposes;
  • The sequestration site and surrounding area has very few residences;
  • Publicly available groundwater sampling data from the area suggest a high likelihood that carbonate minerals in the matrix exceed 1%, providing a natural buffering capacity that substantially limits the potential for pH-related impacts; and
  • Inexpensive point-of-use treatment technologies are readily available for the types of impacts that might occur.

  In light of these factors, readily available site information and expert professional judgment, the Monte Carlo analysis assigns damages to ‘other well’ release events in the following manner: $50,000 (75% probability); $500,000 (20% probability); or $5,000,000 (5% probability). Given the considerations noted above, it is difficult to envision ‘per-event’ damages exceeding the range utilized. However, the percentages assigned to each portion of the assigned damages range also are uncertain. For sensitivity analysis purposes, changing the damages distribution percentages to 50% (damages of $50,000), 30% (damages of $500,000) and 20% (damages of $5,000,000) increases sequestration site groundwater damages from $1.3 million to $6.7 million (50^th percentile) and from $7.9 million to $20.2 million (95^th percentile).

• FutureGen use of averaged atmospheric conditions could understate potential ‘upper end’ estimates of human health effects – The Monte Carlo analysis uses FutureGen estimates of human health impacts ‘as is’, and as previously noted those estimates reflect averaged atmospheric conditions. While averaged conditions likely provide a reasonable estimate of ‘most likely’ impacts, that approach can lead to understatement of ‘upper end’ effects if a subset of atmospheric conditions could lead to higher plume concentrations, longer plume durations and/or plumes directed towards areas with higher populations densities. Publicly available information from FutureGen analyses is not sufficient to plausibly quantify the potential significance of this factor, although the highly rural setting for the Jewett, TX plant, pipeline and sequestration site likely limits the potential magnitude of understatement.
• FutureGen estimates of sequestration site event probabilities are used ‘as is’ in the Monte Carlo model, and FutureGen does not quantify the potential magnitude of underlying uncertainties – FutureGen estimates of sequestration site event probabilities are based on evaluation of site-specific data, industrial and natural analogs and expert judgment. While the underlying uncertainty in these estimates is not quantified, in some cases FutureGen identifies ranges for event probabilities (e.g., other well releases). In such cases, the Monte Carlo model conservatively uses the highest event frequency within the range (rather than the midpoint or low end) and therefore may be more likely to overstate than understate potential damages.
• FutureGen estimates of pipeline event probabilities are used ‘as is’ in the Monte Carlo model, and likely have a relatively low level of uncertainty – FutureGen estimates of pipeline failure frequencies are based on a substantial database of pipeline failure incidents maintained by the U.S. Office of Pipeline Safety. The existence and direct relevance of this event frequency data suggests a relatively low level of uncertainty for pipeline event probabilities.
• FutureGen does not estimate plant event probabilities, estimates used in the Monte Carlo model are based on information in published technical literature and expert judgment – While there is uncertainty associated with this parameter, failure of plant equipment like that used for CO₂ removal purposes is rare. Given the low population density around the Jewett, TX plant, the event failure frequency used in the Monte Carlo model (5.5-in-100,000 annual chance) would have to understate event frequency by at least a factor of 5 before potential plant site event damages would become something other than a negligible contributor to ‘most likely’ and ‘upper end’ damages for the Jewett, TX plant.

¹⁹ Pipeline punctures or ruptures are the next most common type of release (annual probabilities of 1-in-100 and 1-in-200, respectively), but for reasons described elsewhere are associated with much smaller potential monetary damages.