Union of Concerned Scientists
breakdowns that led to broken equipment and, ultimately, extended outages. The agency’s generic communications have been too narrowly focused on hardware problems; the lessons of non-hardware problems must also be communi- cated so that they can be incorporated into pro- cedures and training at all nuclear power plants.
Recommendation #3: The NRC must expand the scope of its oversight efforts when programmatic break- downs at a plant are identified (i.e., when Manual Chapter 0350 is invoked). Specifically, the NRC must ensure that other plants operated by the same company do not suffer from similar problems or experience declining performance while the company focuses on restarting the troubled plant.
The programmatic breakdowns that caused most of the 51 year-plus outages typically led owners of plants with multiple reactors to shut down all the reactors at the affected site until the many problems were corrected. But such breakdowns seldom inspired the NRC to look at reactors operated by the same owner at other sites.
A programmatic breakdown at one plant does not necessarily mean the same breakdown exists at a poorly managed company’s other plants, but it certainly represents a possibility that the NRC must investigate. Furthermore, even when a programmatic breakdown is con- fined to just one of a company’s plants, the significant attention and resources devoted by management to restarting that facility could contribute to declining performance at other sites. Thus, when programmatic breakdowns are identified at a plant, the NRC must take tan- gible steps to (1) determine whether other plants operated by the same company have the same problems, and (2) ensure that performance does not deteriorate at those plants while the company focuses on restarting the troubled facility.
Recommendation #4: When longstanding problems are identified at a plant, the NRC must require the owner to (1) determine why its testing and inspection programs failed to find the problems earlier and (2) address those failures. A review of the 51 year-plus reactor outages reveals the bizarre fact that numerous safety problems that were invisible to plant workers and NRC inspectors during years of tests and inspections magically became visible after an extended outage caused a sea change in how that plant was perceived. The workers and inspectors all suffered from a bias that steered test results away from finding problems.
The author of this report, prior to joining UCS, experienced an effective way of properly focusing inspections as a member of a team of consultants hired to inspect four systems at the Salem nuclear power plant in New Jersey. The team leader’s directive was to assume that the systems were broken and uncover the evidence. My ini- tial reaction to these instructions was, “I’ve been doing this for 17 years; I know how to evaluate systems,” but I soon came to the realization that all of the safety assessments I had conducted in those 17 years were biased toward documenting the number of links in the plant’s safety chain. The Salem inspection, on the other hand, was biased toward finding the weakest link in the chain and testing its soundness.
However the task is accomplished, the NRC must break the longstanding pattern of NRC in- spectors and plant workers repeatedly overlooking safety problems until operations grind to a halt.
Recommendation #5: The NRC must develop a cen- tral repository for all current information about plant safety levels, potential safety problems, and generic safety issues so that all agency employees have access to the same data when making regulatory decisions.