Probabilistic Approach to Design Earthquake Ground Motions Based on Fault Rupture Scenarios

In the conventional earthquake resistant design of nuclear power plants in Japan, the input ground motions were generated according to the design spectra specified with the magnitudes and the distances. Neither the fault parameters nor the exceedance probability were explicitly indicated in the design spectra. Recent seismological developments have enabled us to estimate the parameters of the fault rupture scenarios and also the occurrence probability of each scenario for major active faults in Japan. The new regulatory guide for aseismic design of nuclear power reactor facilities, revised by the Atomic Safety Commission this September, requires to take into account of these parameters for the design earthquake ground motions. The purpose of this study is to propose a procedure to generate design earthquake motions based on fault rupture scenarios for any life-time exceedance probability in particular in the range of 10-2 to 10-3. At first, we categorized the earthquakes in and around Japan as follows: Category I: inland earthquakes caused by active faults recognized on the surface. Category II: large plate-boundary earthquakes in the subduction zones. Category III: large intra-slab earthquakes in the subducting plates. Category IV: inland earthquakes caused by blind faults not recognized on the surface. Category V: moderate and small plate-boundary earthquakes in the subduction zones. Category VI: moderate and small intra-slab earthquakes in the subducting plates. Next, we chose the KiK-net site of Shizugawa, Miyagi Prefecture, as a test site because the National Research Institute for Earth Science and Disaster Prevention had proposed the earthquake occurrence model in and around Miyagi Prefecture and because good records were available at this site to get the high accuracy of the strong motion prediction. After preliminary seismic hazard analysis based on the empirical relation of peak ground velocities with the magnitudes and the distances, we found that the earthquakes of the Categories II, IV, V, and VI had predominant influence on the site. So, we assumed 5,333 fault rupture scenarios for those earthquakes, predicted strong ground motions due to the rupture scenarios, evaluated the occurrence probability of each scenario, and obtained the seismic hazard curves for the peak accelerations, the peak velocities, and the response spectra at various natural periods of structures. Finally, in order to show some examples of design earthquake ground motions, we selected 64 predicted motions with the 50-year exceedance probability of 10-2 to 10-3, averaged their response spectra, and generated 64 artificial motions with the same response spectrum as the averaged one and with the same phase as that of the 64 selected motions. (authors)