Home > Electronic Reading Room > Document Collections > Commission Documents > Commission Papers (SECY) > 2008 > SECY-08-0029

POLICY ISSUE
NOTATION VOTE

SECY-08-0029

March 4, 2008

• PURPOSE
• SUMMARY
• BACKGROUND
• DISCUSSION
• RECOMMENDATIONS
• COORDINATION

PURPOSE:

The purpose of this paper is to obtain Commission approval of the staff-recommended methodology for projecting latent cancer health effects (latent cancer fatalities (LCFs)) in the state-of-the-art reactor consequences analyses (SOARCA).  The staff is also seeking approval to proceed with an external peer review of the methods and technical issues associated with assessing offsite health consequences.  This paper does not address any new commitments or resource implications.

SUMMARY:

The staff provided its proposed plan to perform an updated realistic evaluation of potential severe reactor accidents and their offsite consequences in SECY-05-0233, "Plan for Developing State-of-the-Art Reactor Consequence Analyses," dated December 22, 2005.  In its effort to provide more realistic information to the public and other stakeholders, including Federal, State, and local authorities, on potential nuclear power plant consequences, the staff is using the best available data on plant design and operation; state-of-the-art information on accident progression, source term, and containment performance; and the best practices in emergency preparedness.  For the Peach Bottom and Surry accident scenarios studied in the SOARCA, the staff identified those dominant events as scenarios that have a core damage frequency of 10^-6 or greater per reactor year or containment bypass scenarios that have a frequency of 10^-7 or greater per reactor year.  These core damage frequencies were based on existing Standardized Probabilistic Analysis Risk (SPAR) models.  The accident progression analyses for these sites considered additional mitigation measures that were not credited in previous models and analyses.  The staff also performed sensitivity studies to understand the value of those mitigation measures.

In considering how to communicate any results that may be obtained, the staff has concluded that the modeling methodology for projecting LCFs from offsite radiological releases described in SECY-05-0233 may not be the best approach to achieve the Commission's objectives for SOARCA.  This is because the prediction of the LCFs is problematic, in that computation of potential LCFs for some dose response models aggregates all exposures, including trivial exposures to large populations.

The staff evaluated six different methodologies for projecting LCFs, including the methodology of using multiple dose thresholds for health effects from zero (linear, no threshold (LNT)) to 0.05sievert (Sv) (5 rem) as described in SECY-05-0233.  This evaluation considered the views and opinions of the senior health physicists from across the various U.S. Nuclear Regulatory Commission (NRC) offices, as well as the recommendations of various professional organizations, including the International Commission on Radiological Protection (ICRP), the National Commission on Radiation Protection and Measurements (NCRP), and the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR).  The goal of this evaluation was to identify a methodology that could be in conformance with the views within the scientific community and could be communicated most effectively to the NRC's internal and external stakeholders.

The staff developed a recommendation, supported by the SOARCA Steering Committee, for Commission consideration.  The staff recommends that potential LCFs be expressed as the mean likelihood of LCF for a population-weighted, age and gender averaged, individual living within various distances from the facility.  A linear model would be used without a threshold (LNT) and with a truncation at 100 microsieverts (µSv) (10 millirem (mrem)).

BACKGROUND:

In SECY-05-0233, the staff provided its proposed plan to perform an updated realistic evaluation of severe reactor accidents and their offsite consequences.  The plan describes the staff's intention to incorporate the significant improvements in understanding and modeling of severe accident phenomenology that have been developed through research conducted by the NRC and foreign entities over the last 20 years.  The plan describes the staff's intent to include in its analysis both the plant improvements (e.g.,systems, training, and procedures) that have significantly lowered the likelihood of a severe accident as well as offsite emergency response capabilities that have lowered the offsite health consequences of such events.  The study is also designed to effectively communicate the results to all stakeholders based on realistic estimates of the more likely outcomes.

The SOARCA development plan focuses on dominant events in order to provide proper risk perspective.  The staff identified those dominant events as scenarios (i.e., groups of similar sequences) that have a core damage frequency equal to or greater than 10^-6 per reactor year or containment bypass scenarios that have a frequency equal to or greater than 10^-7 per reactor year.  These frequencies were based on guidance from the Commission, general consistency with risk-informed criteria for assessing plant changes, and insights from the initial SOARCA examinations.  For the Peach Bottom and Surry accident scenarios studied in the SOARCA, the accident progression analyses considered additional mitigation measures that were not credited in previous models and studies.  The staff also performed sensitivity studies to understand the value of these mitigation measures.

When considering ways to perform the offsite consequence analysis more realistically, the staff originally proposed to use a linear dose response model and a range of dose threshold values (i.e.,dose truncation values) varying from 0 to 0.05 Sv (5 rem) for estimating offsite LCFs.  More recently, in a memorandum to the Commission, "Revised Communication Plan and Project Plan for State-of-the-Art Reactor Consequence Analysis," dated July9, 2007, the staff stated that it was reviewing several options to develop a more realistic estimate of the health effects risk that the agency could effectively communicate to its stakeholders.  Since then, the staff has discussed at length the appropriate truncation values and the means to communicate SOARCA results.  These considerable discussions have caused the staff to consider additional alternatives.  Issues, alternatives, and the recommended approach are discussed below and in more detail in the enclosure .

DISCUSSION:

As the SOARCA study proceeded during the last year, staff members have had a high level of interest in as well as disparate views regarding the best way to calculate the health consequences of nuclear power plant accidents, particularly as they relate to the dose model and the range of dose thresholds (i.e.,dose truncation values), and the best method to communicate the results to stakeholders.

With regard to the dose response model and dose truncation values to use in calculating the health consequences of a severe reactor accident, previous studies have traditionally used a LNT model and have aggregated doses over all individuals projected to receive any exposure.  Indeed, several scientific organizations (e.g., ICRP, the U.S. National Academies, and UNSCEAR) have each indicated that the current scientific evidence is consistent with the hypothesis that an LNT dose response relationship exists between exposure to ionizing radiation and the development of cancer in humans.  However, some of those same organizations have also stated that aggregation of small or trivial doses should not be used for risk projection.  For example, in ICRP Report 103, "The 2007 Recommendations of the International Commission on Radiological Protection," approved March 2007, ICRP stated that risk projection of cancer deaths using the LNT model, which is based on collective effective doses involving trivial exposures to thousands of people, is not reasonable and should be avoided.

The question remains regarding the dose response model and dose truncation values that the SOARCA should use.  The staff believes that, as a matter of policy, the NRC could use different approaches for different applications.  Regardless of what model is used as the state of the art for projecting population LCFs, the use of a linear dose response model remains an appropriate technical foundation for the agency's regulatory defense-in-depth approach to protect public health and safety.  Because the SOARCA is not a regulatory analysis, but rather a more realistic representation of the consequences of severe reactor accidents, the use of a single truncation dose and the exclusion of low radiation exposures to members of the public may be seen as more appropriate and may more realistically describe the offsite health consequences.

With regard to communicating the risk to stakeholders, the staff discussed the advantages and disadvantages of the approach described in SECY-05-0233 for presenting the SOARCA results in terms of multiple offsite health consequences for each accident sequence analyzed, using a range of dose truncation values.  The staff was concerned that this approach would not represent effective risk communication in that it might be difficult to provide a context for multiple LCF estimates and might hamper the ability to articulate which of the estimates reflects the staff's best estimate of the more likely outcome.  For example, even though the 1982 Sandia siting study (NUREG/CR-2239, "Technical Guidance for Siting Criteria Development," issued December 1982) includes the results of consequence analyses for several scenarios with moderate consequences, advocacy groups rarely mention these results and often quote the most severe (and most unlikely) outcomes of a highly improbable scenario.  It can be argued that presenting different values would make it difficult for stakeholders to have a common understanding of the results and could lead to selective misinterpretation.

With this information in mind, the staff assessed several alternatives for calculating the offsite health consequences to determine an appropriate SOARCA dose model that would eliminate the concern about using collective effective doses in risk projection and ensure that the agency can effectively communicate the results to stakeholders.  The staff considered several linear models with different truncation criteria.  In addition, the staff considered an alternative (alternative (6) described below) that does not report consequences in terms of absolute values of early fatalities and LCFs.  In this alternative, the consequences would be expressed as the likelihood of an individual's death given a severe reactor accident.  The mean probability of an individual's LCF from an accident would be computed for an average person; that is, the probability would be population weighted and age and gender averaged.  (Consequence analyses routinely use this averaging process.)  The mean individual risk would be calculated using LNT and a truncation dose of 100 µSv (10 mrem).  The calculation would be performed for three distances:  (1) 0 to 16.1 kilometers (km) (10 miles) (comparable to the safety goal); (2)0to 80.5 km (50 miles) (comparable to cost-benefit analysis); and (3) 0 to 161km (100miles).  This metric is not new; past Environmental Impact Statements have used it (see, for instance, NUREC-0537, "Final Environmental Impact Statement Related to the Operation of Midland Plant, Units 1 and 2," issued July 1982).  Furthermore, this approach is similar to the one the Commission used in establishing its Safety Goals.

Alternatives for Addressing Offsite Latent Health Consequences

Given the above, the potential policy implications, and the high level of stakeholder interest, the staff identified the following six alternatives and a recommendation for Commission consideration.

Upon receipt of Commission direction, and in conjunction with continuing SOARCA activities, the staff proposes to submit the methodologies and approaches used for the first two plants to a peer review.  This review will enable the staff to have confidence that its methodologies are adequate for the objectives of the program.

Furthermore, the staff plans to discuss some details regarding the project methodology and some of the preliminary insights and findings from the initial assessments at the March 2008 Regulatory Information Conference.  In accordance with Commission direction not to make the results publicly available until all SOARCA analyses have been completed, the staff will not discuss the preliminary results.

Conclusion

The staff considered a range of options for calculating LCFs and believes that the best approach is to use alternative (6).  The calculation of the mean likelihood of a population-weighted, average individual dying from cancer conditional on the occurrence of a severe reactor accident provides a more direct comparison with the occurrence of cancer fatalities in the general population from causes other than a reactor accident.

In its letter dated February 22, 2008, the Advisory Committee on Reactor Safeguards (ACRS) recommends that consequences be expressed in terms of ranges calculated using 5 rem and some lower threshold, and that a calculation with LNT should also be performed to facilitate comparison with historical results.  This recommendation is encompassed within alternative 1 in this paper.  After a review of the strengths and weaknesses of each alternative, the staff believes that alternative 6 provides the best option for projecting latent cancer health effects in SOARCA.  The staff's formal response to the ACRS will provide the Committee the basis for this approach, and address the additional recommendations provided by the ACRS in the February 22, 2008, letter.

The staff will need to address both scientific and communication issues in the presentation of the SOARCA results to stakeholders.  The staff recognizes that a calculation of LCF using a linear model without dose truncation will be performed, if not by the NRC staff, then by others.  If an NRC stakeholder rather than the NRC itself performs these calculations, the agency may need to expend significant resources in an attempt to explain what the NRC provided and why it was the most appropriate presentation.  The staff will have to make clear that the presentation of any results based on dose truncation does not discount the possibility of LCFs at low doses and thereby may be an incomplete estimate of LCFs.  That is, the staff recognizes that neither the use of a LNT model nor the selection of any particular truncation value for assessing LCF is supported by any specific scientific information regarding cancer induction.  The staff will discuss the reasoning behind the selection of the 100 µSv (10 mrem) truncation value alongside the argument for considering a threshold in the portrayal of risk.

RECOMMENDATIONS:

The staff recommends that the Commission take the following two actions:

COORDINATION:

The Office of the General Counsel reviewed this package and has no legal objection.  The Office of the Chief Financial Officer reviewed this package and determined that it has no financial impact.

Privacy Policy | Site Disclaimer
Friday, March 21, 2008