How to eliminate non-damaging earthquakes from the results of a probabilistic seismic hazard analysis (PSHA)—A comprehensive procedure with site-specific application

Abstract

The results of probabilistic seismic hazard analyses are frequently presented in terms of uniform hazard spectra or hazard curves with spectral accelerations as the output parameter. The calculation process is based on the evaluation of the probability of exceedance of specified acceleration levels without consideration of the damaging effects of the causative earthquakes. The same applies to the empirical attenuation equations for spectral accelerations used in PSHA models. This makes interpreting and using the results in engineering or risk applications difficult. Uniform hazard spectra and the associated hazard curves may contain a significant amount of contributions of weak, low-energy earthquakes not able to damage the seismically designed structures of nuclear power plants. For the development of realistic engineering designs and for realistic seismic probabilistic risk assessments (seismic PRA) it is necessary to remove the contribution of non-damaging earthquakes from the results of a PSHA. A detailed procedure for the elimination of non-damaging earthquakes based on the CAV (Cumulative Absolute Velocity)-filtering approach was developed and applied to the results of the large-scale PEGASOS probabilistic seismic hazard study for the site of the Goesgen nuclear power plant. The procedure considers the full scope of epistemic uncertainty and aleatory variability present in the PEGASOS study. It involves the development of a set of empirical correlations for CAV and the subsequent development of a composite distribution for the probability of exceedance of the damaging threshold of 0.16 gs. Additionally, a method was developed to measure the difference in the damaging effects of earthquakes of different strengths by the ratio of a power function of ARIAS-intensity or, in the ideal case, by the ratio of the square roots of the associated strong motion durations. The procedure was applied for the update of the Goesgen seismic PRA and for the confirmation of a revised safe shutdown earthquake for the Goesgen nuclear power plant. The application of the procedure leads to results which are in reasonable compliance with evaluations based on the macroseismic method using European macroseismic intensities and associated vulnerabilities. The paper is an extended version of the paper #1142 presented at the 19th SMIRT conference in Toronto, 2007.