State-of-the-art report on the current status of methodologies for seismic PSA

There has never been an earthquake sufficiently damaging to any operating U.S. nuclear power station to cause safety concerns. In the past decade, there have been two very destructive earthquakes that were near enough to nuclear power stations to have caused at least temporary world-wide concern: the large Armenian earthquake of November, 1988, and the large Japanese earthquake near Kobe in January, 1995. However, in each case, these earthquakes produced only minor ground motion at the operating nuclear stations located close enough to merit some examination: in each case, inspections revealed essentially no damage, because the actual ground motions at the nuclear-plant sites were in both cases quite small and well within the design basis. Nor has there ever been any other earthquake known to have damaged a nuclear power station. Therefore, the published historical record is not adequate for the analysis discussed here. The frequency of potential earthquake-initiated core-damage accidents at a given nuclear-power station can only be known from calculations, using a combination of real-earthquake data, test data, models of various phenomena, and systems analysis. Despite the lack of actual earthquake experience at nuclear-power stations, almost every full-scope probabilistic safety analysis (PSA) that has examined earthquake-initiated accidents at nuclear power stations has found that this category represents one of the important initiator groups. Occasionally, one of the earthquake-initiated sequences is among the few largest contributors to calculated core-damage frequency and/or off-site risk. Of course, it is important to keep in mind that PSAs calculate a residual risk, which is judged to be acceptable by the regulatory authorities, and that the seismic contribution to this residual risk is what is being discussed here. Usually the potential accident sequences identified in the PSAs are very plant-specific in character, such that the specific vulnerability is unlikely to exist at any other plant, even another similar plant. Sometimes the issue is site-related, and sometimes it is design-related. Also, even in areas where earthquakes are very uncommon phenomena, these types of accident sequences often appear as important contributors to the residual risk, typically because in such areas the attention given to designing nuclear stations against earthquakes is much less than in earthquake-prone areas. Given this background, it is obvious that no full-scope PSA can be considered complete without an examination of earthquakes. This report is a review of the methodology for conducting a seismic-PSA at a nuclear power station. The objective of this review is as follows: To provide an up-to-date review of the state-of-the-art of the various sub-methodologies that comprise the overall seismic- PSA methodology for addressing the safety of nuclear power stations, plus an overview of the whole methodological picture. In preparing this review, the author has had in mind several categories of readers and users: policy-level decision-makers (such as managers of nuclear power stations and regulators of nuclear safety), seismic- PSA practitioners, and PSA practitioners more broadly. The review will concentrate on evaluating the extent to which today's seismic-PSA methodology produces reliable and useful results and insights, at its current state-of-the-art level, for assessing nuclear-power station safety. Also, this review paper will deal exclusively with seismic-PSA for addressing nuclear-power-station safety. Because the author is based in the U.S., it is natural that this review will contain more emphasis on U.S. experience than on experience in other countries. However, significant experience elsewhere is a major part of the basis for this evaluation. In summary, this report is an up-to-date review of the state-of-the-art of the methodologies for conducting a seismic- PSA at a nuclear power station, including the six sub-methodologies that comprise the overall methodology. The review concentrates on evaluating the extent to which today's seismic-PSA methodology produces reliable and useful results and insights. The evaluation covers six sub-methodologies that must be combined to produce an overall full-scope seismic PSA: the seismic-hazard methodology; the local-ground-motion and building-motion methodology; the walk-down methodology; the failure-mode and fragility methodology; the systems-analysis methodology; and the consequence/release methodology. The report finds that all of these sub-methodologies are both reliable and useful, and that when combined together the overall seismic-PSA methodology can provide important engineering insights about how nuclear power-plants respond to earthquakes. This is true even though the numerical uncertainties in the bottom-line results can be large (plus-or-minus more than one order of magnitude or more is common.) However, a number of areas within the various sub-methodologies can be applied properly only if special expertise is available. The report describes the technical issues in detail, and outlines the approaches that have proven to be most successful, based on world-wide experience with about a hundred seismic PSA studies