NRC's Risk-informed, Performance-Based Regulatory Approach Distorts Safety Assessment Of A Potential Yucca Mountain High-Level Nuclear Waste Repository
April 16, 2001
Yucca Mountain, Nevada, about 80 miles northwest of Las Vegas, is being considered by the U.S. Department of Energy (DOE) as the only site for disposal of the highly radioactive wastes from the nation's nuclear power reactors and from the federal nuclear weapons complex. As more than two decades of site studies near completion, the U.S. Nuclear Regulatory Commission (NRC) is proposing new, site-specific regulations for licensing this deep geologic high-level nuclear waste repository. The repository, if approved, would consist of 50 to 100 miles of tunnels in volcanic rock in which 77,000 tons of long-lived highly radioactive waste would be entombed. Because of the longevity of the radiation hazard associated with these wastes, the risk analysis for the repository must include projections of radiation doses to humans extending an unprecedented 10,000 to as many as 1 million years into the future. NRC is proposing to apply its recently formulated, and not yet fully implemented, risk-informed, performance-based regulatory policy for nuclear reactor licenses to its consideration of a license for this first-of-a-kind high-level nuclear waste repository. The approach is inherently problematic when it is applied to low probability, high-consequence events.
The NRC's policy of risk-informed, performance-based regulation is implemented in its proposed site specific Yucca Mountain high-level nuclear waste repository licensing rule (10 CFR Part 63). The rule, although not finally promulgated, is being applied by the DOE in its suitability and safety analyses of a potential Yucca Mountain repository, and is being used by the NRC staff in its review of the sufficiency of DOE's work for a license application. The proposed rule requires that calculation of the expected annual dose, due to repository releases, to an average individual in the critical group near the repository be carried out through the use of a total system performance assessment (TSPA). The TSPA is constructed as a probabilistic analysis, in keeping with the NRC's policy of risk-informed, performance-based regulation.
The NRC has stated in a staff White Paper that risk-informed, performancebased regulation is "an approach in which risk insights, engineering analysis and judgement, and performance history are used, to (1) focus attention on the most important [to safety] activities, (2) establish objective criteria based upon risk insights for evaluating performance, (3) develop measurable or calculable parameters for monitoring system and licensee performance, and (4) focus on the results as the primary basis of regulatory decision making." (1) The White Paper also notes that this approach is appropriate for nuclear waste repository licensing regulation.
NRC Commissioner Nils Diaz described risk-informed regulation in a broader sense: "Risk-informed regulation is a set of deterministic criteria, operating experience, defense-in-depth, engineering judgements and probabilistic risk assessments that qualitatively and quantitatively increases the knowledge base and is conducive to safety-focused decision making." (2)
Risk, in both of the above statements, has the traditional meaning of the consequence of events and processes (in the repository case, the radiation dose), weighted by the probability that the event or process will occur. In its mathematical form: risk=consequence X probability.
In practice, the TSPA is a set of calculations of projected annual dose to an average individual in the vicinity of the repository plotted against the number of years after repository closure. A large number of calculations is performed, based on the variability of parameters, eg., infiltration of precipitation due to the projected range of climate change scenarios, and a staiistical sampling of values within a range that bounds certain parameters, such as rock permeability, which among other things, affects the infiltration rate. Once a statistically defensible number of dose versus time curves are generated, curves for the mean, median, 95th and 5th percentiles, etc. can be calculated. The resulting statistical curves have been modified by probability, because of the probabilistic sampling of parameter values used in their generation. This is a reasonable approach if the range of parameter variability and distribution of values is strongly supported by physical data. In addition, it provides a quantitative basis for understanding the uncertainty associated with the result. However, the NRC's implementation of the risk-informed, performance-based regulatory approach has distorted the presentation of the TSPA result by selecting the mean, rather than the 90th or 95th percentile of a broad range of calculated results as the value to represent the "expected annual dose."
DOE's TSPA indicates about five orders of magnitude of uncertainty in its annual dose projections during the period of time after waste container failure allows radionuclide releases to begin The majority of the uncertainty is associated with the projected waste container lifetime. Nonetheless, DOE's current TSPA relies on the projection that none of the approximately 12,000 waste containers in a Yucca Mountain repository would fail due to corrosion during the 10,000 year compliance period. Therefore, the only failure mode for the repository during the 10,000 year compliance period would be the occurrence of some low probability event with consequences that could significantly de grade the performance of the repository.
The NRC's proposed repository lisensing rule sets a lower limit of probability, beyond which events are not requires to be quantitatively analyzed in the TSPA If an event has a probability of occurrence of less than one chance in ten thousand during the ten thousand year period of compliance (equal to a probability of 10-8/year); the event is screened out of the TSPA calculation. For example, rockfall in the repository, caused by seismic shaking, that would damage waste containers and significantly affect repository performance, has been screened out of the TSPA by DOE for this reason.
DOE considers the only event that could result in releases during the 10,000 year compliance period would be volcanic disruption. But, in the analysis of this low probability event, when the risk-informed, performance-based approach is applied, the TSPA calculation is distorted to the extent that it is meaningless and misleading.
The NRC staff has explained how the TSPA calculation is made "risk-informed" relative to potential volcanic disruption of a Yucca Mountain repository, which NRC views to have a probability of occurrence of one chance in ten million per year (10-7/year), during the 10,000 year compliance period. The NRC has stated, "Under the proposed 10 CFR Part 63 rule, the expected annual dose is used to determine compliance with proposed performance objectives. Expected annual dose is the dose weighted by probability of event occurrence (i.e. risk), with the maximum annual risk (ital. added) during the post-closure period used to determine compliance."(3) In order to "risk inform" the expected annual dose (consequence), it is multiplied by the probability of occurrence, i.e. 10-7. NRC's sample calculation of the consequence of volcanic disruption, under certain assumed conditions, indicates a peak probability modified "expected annual dose" (risk) of about 1 millirem (.001 rem) per year at the time of the volcanic disturbance. If the risk modifying factor, the 10-7 probability of occurrence, is removed, the expected annual maximum dose is 10,000 rems (104 rems). The Environmental Protection Agency has proposed an all pathways maximum individual dose (not risk) of 15 millirems (.015 rem) per year for a Yucca Mountain repository compliance limit, which is to be applied in NRC's consideration of a Yucca Mountain repository license application.(4)
DOE has followed this calculation approach in analyzing the effects of volcanic disturbance of a Yucca Mountain repository. DOE believes the probability of occurrence to be smaller (1.6X 10-8/year) than the NRC's I0-7/year. But, taking into account some other calculational differences, DOE's result is similar, with an expected maximum annual dose from a volcanic event of about 1,000 to 10,000 rems per year, after the probability has been factored out.(5)
Both NRC and DOE present the results of their performance Rockfall in the repository, caused by seismic shaking, that would damage waste containers and significantly affect repository performance, has been screened out by DOE. assessment relative to volcanic disruption as expected annual dose, when, in fact they are calculating the annual risk (dose times probability). In the case of low probability events, it is misleading to interchange dose and risk, because it is not the risk that is the first concern to the public - it is the magnitude of the consequence, should it occur. Public decisions regarding levels of acceptable risk must first be informed by a rigorous and credible analysis of the consequences.
In the case of volcanic disruption of a Yucca Mountain nuclear waste repository, application of the risk-informed, performance-based approach, through manipulation of the meaning of dose and risk, results in the appearance that regulatory compliance is achieved, when, in fact, if the reference event were to take place, the compliance limit would be exceeded by orders of magnitude. Therefore, because of the very large doses involved, the difference between risk and dose values is literally the difference between life and death for the victims of the reference volcanic event.
(1) U.S. Nuclear Regulatory Commission, 1998. SECY-98 144, White Paper on Risk-Informed, Performance -Based Regulation.
(2) Diaz, N. J., 1999. Benefits of Safety-Focused Regulation. Speech to the 1999 ANS Winter Meeting, Long Beach, CA.
(3) U.S. Nuclear Regulatory Commission, July 1999. Issue Resolution Status Report, Key Technical Issue: Igneous Activity, Revision 2.
(4) U.S. Environmental Protection Agency, 1999. Environmental Radiation Protection Standards for Yucca Mountain, Nevada; Proposed Rule, 40 CFR Part 197.64 FR 166, Aug. 27, 1999, pp.46976-47016.
(5) Swift, P. and M. Sauer, 2001. Total System Performance Assessment - Site Recommendation Rev. 00 - Igneous Activity Analysis. Presentation to DOE/NRC Technical Exchange: Total System Performance Assessment - Site Recommendation
Briefing, January 23, 2001.
RISK POLICY REPORT - April 16, 2001 -- Volume 8, Number 4