As I recently told the NRC Commissioners,\1\ UCS believes that the NRC has a good plan. However, the plan's quality is not the most important factor in determining whether the NRC succeeds in its oversight mission. What matters most is how well the agency implements its program. Too often, the NRC fails to follow its plans and does not regulate in a consistent, timely manner.
The NRC's Inspector General recently reported\2\ to Senator Lieberman that the agency failed to properly discharge its responsibilities to the people of Connecticut and to workers at the Millstone Nuclear Power Station. These failures are particularly troubling because they involved the highest profile nuclear facility in the country. The NRC created a Special Projects Office with responsibility for only one site -- Millstone. The Inspector General documented numerous regulatory failures involving that office despite its singular focus. The Inspector General also reported that many failures were caused by the NRC not following its own procedures and policies.
The Inspector General's report is the latest example in a long history of the NRC's failing to follow through on its plans. Let me cite fire protection as an old, yet still ongoing, example. The NRC created Appendix R, the fire protection rule, to 10 CFR Part 50 in January 1980 to address safety concerns following the serious fire at the Browns Ferry Nuclear Plant in Alabama in March 1975.
Nineteen years later, the majority of nuclear plants do not now, and never have, satisfied the Appendix R regulations. When Rep. Markey asked the NRC about this situation in May 1997, the agency replied that it was considering a revision to Appendix R. Nearly two years later, we understand that the agency is still thinking about revising the regulation. Last December, Rep. Markey asked the General Accounting Office to investigate fire safety issues at nuclear plants.
In the meantime, the majority of nuclear plants are operating in violation of fire safety regulations. At the Salem Generating Station in New Jersey, for example, both reactors were shut down from 1995 to 1997 while its owner made extensive repairs to safety equipment. Numerous fire protection deficiencies went uncorrected during this lengthy shut down. The NRC allowed both reactors to restart despite knowing that fire protection requirements were not met. Not only that, but the NRC is content with the owner's plans to leave the problems uncorrected for several more years.
The NRC tolerates violation of its fire protection regulations because plant owners have taken so-called interim compensatory measures. The most common of these measures involves workers, called fire watches, walking through the plants looking for smoke or flames. Such "interim" measures have been used for more than six years at many plants.
But interim measures are not a substitute for permanent solutions. When I get a flat tire, I replace it with a spare mini-tire. That's an interim measure I can use until I get the flat fixed or a new tire. It would be irresponsible for me to undertake a cross-country trip on that interim mini-tire. My poor judgement would place myself and other travelers at an undue risk. Likewise, it is irresponsible for the NRC to rely on fire watches indefinitely. This poor decision places millions of Americans living around nuclear plants at undue risk.
The risk from fire is real. The NRC reported\3\ that fire represents 7 to 50 percent of the overall reactor core damage risk at nuclear plants. According to this data, there is a plant where the fire risk equals the risk from all other accident scenarios combined. So, the fire risk is real. And the regulations created to properly manage that risk are essentially being ignored by the NRC.
We respectfully request that this Subcommittee compel the NRC to resolve the fire protection problems. The NRC must either enforce or revise its fire safety regulations. Continued neglect, predicated on "interim" compensatory measures and "considerations" of rulemaking, must end. The regulations were promulgated in direct response to the serious Browns Ferry fire. If another serious nuclear plant fire were to occur, the American public would be very distressed to learn that these fire protection regulations had not been enforced.
Your Subcommittee's oversight hearings have accelerated the NRC's change process. We sincerely appreciate the Subcommittee's efforts in this regard. We trust that this Subcommittee will not judge the NRC solely on its plans. We hope that you will evaluate the results from these new and improved processes, even though this data will not be available until late this year. We respectfully request that this Subcommittee continue these efforts to ensure that the NRC reaps the maximum benefits from its plans.
References:
1) Union of Concerned Scientists, Presentation to NRC Commissioners, "Looking for Golidlocks: The NRC's Inspection, Assessment, and Enforcement Programs," January 20, 1999. (Attached)
2) Nuclear Regulatory Commission Inspector General,. "NRC Staff's Handling of Harassment and Intimidation Complaints at Millstone (Case No. 99-01S)," December 31, 1998.
3) Nuclear Regulatory Commission, NUREG-1150, "Severe Accident Risks: An Assessment for Five U.S. Nuclear Power Plants," June 1989.
The NRC staff is to be commended for the comprehensive and thorough reactor oversight process improvement recommendations detailed in SECY-99-007. They faced a daunting challenge while seeking a "Goldilocks" oversight process -- one that is not too stringent or too lax, but just right. A large number of our concerns have been addressed. On paper, this process appears fundamentally sound and capable of successfully meeting the stated expectations. However, it must be noted that, on paper, so was the old process. It's not the process that will make or break this effort, it's the implementation.
The process was developed with an objective of increasing public confidence in the NRC's regulatory function. Key elements of the new process are these seven cornerstones of plant safety:
Inspection Process
The NRC's limited inspections provide it with a very small slice of the overall safety picture at nuclear plants. It is important that the NRC properly characterize its findings. Based on my experience prior to joining UCS, it appeared that inspection findings were graded on a curve because the threshold for a non-conforming condition seemed lower at a plant which the staff believed to have performance problems than it was at a plant that the staff believed was doing okay. The staff's feelings towards licensees must not influence inspection findings to prevent a self-fulfilling prophecy situation.
Since the proposed baseline inspections will concentrate on areas not covered by performance indicators, there will be little chance to confirm or refute inspection findings. Findings that are too positive or too negative will likely pass through to the assessment process unchallenged. Findings that are "just right" are very important.
The NRC's Inspection Manual tells inspectors what to examine and how often, but does not provide much assistance in the form of objective acceptance criteria. As a result, inspectors are being asked to evaluate the condition of a plant component or a plant owner's process without benefit of an "answer key." As the Inspection Manual chapters is revised, the NRC should provide objective acceptance criteria whenever possible. Otherwise, inspections findings may continue to be influenced by the staff's perceptions of licensee performance.
The NRC should post all the inspection reports issued for each operating nuclear plant within the past year on its internet website.
Assessment Process
In Table 4.1, the staff listed the following success criterion for the assessment process:
Number of executive over-rides (cases where the outcome is something different than the input) at end-of-cycle review is less than 5%
Since there are about 25 plants in each region, this would consider one plant in each region or five plants in one region being handled subjectively as a success. That is wrong. That would be carrying over a major flaw of the SALP process into the new process. The appropriate goal should be no executive over-rides. None. The staff, on occasion, may be justified in over-riding the assessment results with its judgement, but that situation should not be considered to be a successful outcome of the assessment process. Executive over-rides should be used very sparingly and not routinely accepted.
The proposed assessment process relies heavily on performance indicators. We have the following concerns regarding the use of these indicators:
The Reactor Coolant System specific activity PI has a green-to-white threshold of >50% of the Technical Specification limit. This PI is intended to monitor the integrity of the fuel cladding barrier. In April 1998, UCS provided a technical report to the NRC on our research which concluded that it is illegal and potentially unsafe for nuclear plants to operate with any fuel leakers. We have since submitted 2.206 petitions on the River Bend and Perry plants because these plants are operating with known fuel leakers. In our report and our petitions, we have challenged the bases for the RCS specific activity Technical Specification. We respectfully request that the NRC staff answer our nuclear safety concerns raised nearly a year ago before adopting this PI.
The Containment Leakage PI has a green-to-white threshold of >100% LA. Containment leakage is a virtually meaningless indicator. The containment integrated leak rate tests are performed every 18 to 24 months with the plants shut down. If leakage is exceeds 100% LA the plant will not restart. Thus, it is highly unlikely that this PI will be anything but green. A PI that doesn't ever change color is worse than useless because it provides a false sense of security.
Speaking of false senses of security, the containment leakage PI would have been in the green at DC Cook even though that facility's containment spray and ice condenser systems were severely degraded. The appropriate containment PI would be the reliability of the containment heat removal systems.
The Safety System Performance Indicators suffer from the same problem that afflicts the probabilistic risk assessments -- they do not properly account for system degradation caused by passive design problems, or "blunders" to use Dr. Thadani's term. For example, the Emergency Power PI has a green-to-white threshold of >0.025. In NRC Inspection Report 50-213/96-201, dated July 31, 1996, the staff concluded:
The most significant issue noted by the team was the failure of the [Haddam Neck] licensee to appropriately consider design-basis scenario loads on the Class 1E station batteries sizing calculations. Specifically, the licensee's calculation did not account for all of the loads associated with a LOCA coincident with loss-of-offsite-power, and did not demonstrate that the battery voltage would remain above the minimum level required for operation of equipment.
In other words, the safety-related batteries at Haddam Neck, which passed the Technical Specification surveillance tests for years with flying colors, would have failed in event of an accident due to a design problem. Yet, design problems like these do not count against the system reliability numbers. The PI's must reflect that reality or they will provide misleading signals.
Curiously, while the Haddam Neck inspection report documented numerous problems with systems intended to protect public health and safety, the systems needed to generate electricity worked well.
For the Physical Protection cornerstone, both the personnel screening process performance and the personnel reliability program performance indicators have a green-to-white threshold of 3-5 reportable events. The time period is not specified. We assume it is one year, but it should be clearly defined.
Attachment 1, Table 5 provides the action matrix proposed for assessment program results. The second column states actions the NRC just might take for one or two inputs in the white. Is it one or is it two? The appropriate threshold should be clearly defined.
The fourth column on Attachment 1, Table 5 states actions the NRC may take for repetitive degraded cornerstones et al. While the proposed response are prudent, the trigger for this response is too subjective. Since this response level is that level which failed at Millstone, Salem, DC Cook, etc., it is vital that it be as clearly defined as possible to prevent repeating those oversight errors.
Enforcement Process
The current Enforcement Policy is rumored to be non-punitive. In the current Enforcement Policy, the staff identifies an apparent violation of federal safety regulations and provides the licensee an opportunity to explain its case at a pre-decisional enforcement conference. The staff can then impose a civil penalty on the licensee. The licensee can pay the fine or appeal it.
"Punitive" is defined as "inflicting, involving, or aiming at punishment."
"Punishment" is defined as "a penalty inflicted on an offender through judicial procedure." Thus, it seems reasonable to conclude that a process which collects multi-million dollar civil penalties from offending licensees through an administrative process which affords the opportunity to both contest the violation and appeal the penalty is, in fact, punitive. If it waddles like a duck, swims like a duck, and quacks like a duck, it's a duck. Let there be no mistake -- the Enforcement Policy is punitive. The good news is that it's supposed to be punitive.
The bad news is that enforcement actions are so randomly applied that the policy is totally ineffective. While there are plenty of examples to illustrate arbitrary and capricious enforcement actions, the classic cases are those associated with the duration of the non-conforming condition. The statute permits the NRC to assess a penalty of up to $110,000 per violation per day that the violation existed. The staff rarely invokes this provision. In 1996, the NRC fined the LaSalle licensee for about 20 days of a problem. In 1998, the NRC did not fine the DC Cook licensee a problem lasting about the same duration. The staff must develop the means to consistently and meaningfully apply the per day provision of the statute.
General Observations
The staff went to considerable effort to identify how the outcome from the inspection, assessment, and enforcement processes will be communicated to stakeholders. It is also necessary to complement these communications with much better documentation of staff decisions that produced the outcome. The nuclear industry is required by NRC regulations to provide a paper trail for decisions affecting nuclear safety. The staff's decisions have the same importance as those made by licensees, yet the documentation standards are significantly less rigorous. The staff must, as a minimum, meet the industry's standards.