Nuclear Energy Agency Committee on the Safety of Nuclear Installations

There are several significant seismological issues to consider in the analysis of soil structure interaction (SSI) effects of nuclear power plants (NPP). Typically, a seismic hazard analysis will be used to define free-field motions at the plant site; this paper therefore begins with an overview of seismic hazard analysis, with particular focus on stable continental regions such as eastern North America. The free-field motions from a seismic hazard analysis are used as the input to engineering analyses to infer the motions that will be input to the foundation mat (the SSI analysis). Thus the seismological input to the problem is most fundamentally the definition of appropriate free-field motions to input to the SSI problem. These motions are usually defined as three-component time histories that “match” or otherwise represent a target Uniform Hazard Spectrum (UHS) for the site, for the desired probability level. Several general issues arise in defining these time histories: Should they be “real” (previously-recorded) time histories, or are simulated records acceptable? What record characteristics are most important? How many records should be used? How should they be matched to the UHS: tight or loose matching? Matching of the entire UHS with a single broadband record or use of multiple scenarios? There are also specific issues that are crucial for nuclear power plants: How should we model the high-frequency motions expected for plants on rock sites in eastern North America (and similar environments)? How might high-frequency motions be filtered out by considering incoherence across the foundation mat? These are difficult questions for which there is no single “right answer”. Ground motion characteristics for sites in eastern North America are used to illustrate the issues under consideration, including in particular the high-frequency content of eastern motions. Typical UHS shapes and how they arise are described, and the means by which the UHS might be matched are discussed. 1. An Overview of Earthquake Hazard Analysis Over 90% of the world’s seismicity occurs within relatively narrow bands where two or more of the tectonic plates that make up the earth’s lithosphere slide past or collide with each other. In plate-margin regions, seismotectonic processes are relatively well understood. Strain energy is accumulated by the relative motion of the plates, and released by seismic slip along plate boundary faults. For crustal earthquakes (e.g. such as those along the San Andreas fault system), the faulting often ruptures the ground surface during large earthquakes. The magnitudes of observed earthquakes, their rupture dimensions and frequency of occurrence can be directly related to rates of slip and strain energy accumulation. This provides a valuable physical basis for interpreting seismicity. In regions far removed from plate boundaries, including most continental regions, seismicity tends to be more diffuse and infrequent. Nevertheless, large and damaging earthquakes do occur in mid-plate regions, as for example the 1811-1812 New Madrid earthquakes of M>7.5, or the devastating 2001 M7.6 Bhuj, India earthquake (where M is moment magnitude). The causative mechanisms of mid-plate earthquakes are often ambiguous. In general, earthquakes within stable continental interiors relieve long-term internal plate stresses that are driven by distant plate interactions. The locations where stresses are relieved are usually zones of weakness of large crustal extent, typically pre-existing faults left behind by older episodes of tectonism. Previously rifted or extended crust is believed to be of particular importance in localizing seismicity (Johnston et al., 1994). Because the earthquake-generation process is indirect, and