BOSTON — FOR the last three years, John Darby has been helping the Nuclear Regulatory Commission evaluate safety risks at nuclear power plants.
But now, the nuclear engineer is casting doubt on the agency's effort. In an 11-page memo to the NRC recently made public, Dr. Darby details his concerns over the agency's independent plant examination (IPE) program. The NRC's lack of support, he writes, has pushed him to ''the point where in good conscience I must cease to work on your review program until my concerns are addressed.''
His concerns center on the NRC's rules that keep him from resolving problems he spots, including major discrepancies between what plant operators say in their safety reports and what on-site inspections reveal. ''I have been told in no uncertain terms to 'don't look so hard and don't ask so much,' '' he writes.
Questions over the accuracy of IPEs aren't new. But rarely has an engineer so directly involved with a safety review process publicly aired his views. He has sent copies to the NRC, as well as to the American Nuclear Society, his field's professional society.
The criticism comes at a time when the federal agency is considering wider use of risk assessments in drawing up regulations and setting license conditions.
Under the IPE program, utilities must file reports estimating the probability of accidents that could damage a reactor's radioactive core. The program began in 1988, as a way to help utilities focus their safety efforts in the most efficient, cost-effective way, says Mark Cunningham, chief of the probabilistic risk-assessment branch of the NRC's Office of Nuclear Reactor Regulation.
The NRC sent instructions to utilities in 1988 saying it would review each report to ''obtain reasonable assurance that the licensee has adequately analyzed the plant design and operations to discover instances of particular vulnerability to core melt.... Further, the NRC will assess whether the conclusions the licensee draws from the IPE regarding changes to the plant system, components, or accident management procedures are adequate.''
Based on his review of 25 reports over three years, Dr. Darby says that they cannot provide that assurance. In his memo, he writes: ''I am not saying the reports are right or wrong. I am saying that under the current process, I cannot resolve many technically important issues.''
Robert Pollard, a nuclear-safety engineer at the Union of Concerned Scientists in Washington, is more blunt.
The reports are fine ''if you want to wave a six-inch notebook at an inquiring reporter or a congressional committee. As engineering analyses, he's saying, they're worthless.''
The NRC itself has questioned some IPE reports. In 1992, NRC officials reportedly sent back an IPE submitted by the New York Power Authority regarding the James FitzPatrick plant at Scriba, N.Y. The officials said the report characterized the plant as ''absurdly safe,'' in contrast to a recent NRC evaluation that uncovered major problems at the plant.
Darby's own complaints arise from his experiences comparing the IPE data from utilities with information he gleaned while visiting the plants as part of NRC evaluation teams unrelated to the IPE process. Among other examples, he cites an onsite evaluation of a plant with a failure rate for critical valves that was 100 times greater than recorded in its IPE.
Despite the value of visits to the plants, he continues, under the IPE process he is not allowed to visit the plants whose examination he is reviewing. All technical questions must go through the NRC, which decides which questions to relay to the utility. ''...In some cases many questions that I feel are significant are not asked, and some of my original questions have been so changed and diluted they do not address the issues of concern,'' he writes, concluding, ''In my opinion, the review process being used by the NRC for the IPEs is insufficient to provide a technically supportable basis for the use of risk-based regulations.''
The NRC is reviewing Darby's memo. Regarding Darby's point about on-site visits, Mr. Cunningham says, ''Decisions on interactions with licensees involve staff. Contractors are a step removed from that.''
The use of risk assessments in the nuclear industry is good, says Darby, up to a point.
''You can have great methodology, but if it's not applied correctly or rigorously, you'll get answers that aren't as good as they should be,'' he says in an interview.
''What the NRC is saying is that these reviews are only cursory and that it will make more detailed analysis later. But ... if you find things, you ought to resolve them. Otherwise changes [in regulations or operating procedures] could be made in the future that compromise safety.''