In assessing the seismic safety of nuclear power plants, it is essential to analyze the structures using the observed ground motion. In particular, spatial variation in which the characteristics of the ground motion record differ may occur if the location is different within the site and even if the same earthquake is experienced. This study analyzed the spatial variation characteristics of the ground motion observed at the structure and site using the earthquake records measured at the Hamaoka nuclear power plant. Even if they were located on the same floor within the same unit, there was a difference in response depending on the location. In addition, amplification was observed in Unit 5 compared to other units, which was due to the rock layer having a slower shear wave velocity than the surrounding bedrock. Significant differences were also found in the records of the structure’s foundation and the free-field surface. Based on these results, the necessity of considering spatial variation in the observed records was suggested.

According to several seismic design standards, a ground motion time history should be selected similar to the design response spectrum, or a ground motion time history should be modified by matching procedure to the design response spectrum through the time-domain method. For the response spectrum matching procedure, appropriate seed ground motions need to be selected to maintain recorded earthquake accelerogram characteristics. However, there are no specific criteria for selecting the seed ground motions for applying this methodology. In this study, the characteristics of ground motion time histories between seed motions and spectral matched motions were compared. Intensity measures used in the design were compared, and their change by spectral matching procedure was quantified. In addition, the seed ground motion sets were determined according to the response spectrum shape, and these sets analyzed the response of nonlinear and equivalent linear single degrees of freedom systems to present the seed motion selection conditions for spectral matching. As a result, several considerations for applying the time domain spectral matching method were presented.

The design response spectrum presented in the seismic design standard reflects the characteristics of the tectonic environment at a site. However, since the design response spectrum does not represent the ground motion with a specific earthquake magnitude or distance, input ground motions for response history analysis need to be selected reasonably. It is appropriate to use observed ground motions recorded in Korea for the seismic design. However, recently recorded ground motions in the Gyeongju (2016) or Pohang (2017) earthquakes are not compatible with the design response spectrum. Therefore, it is necessary to convert the recorded ground motion in Korea to a model similar to the design response spectrum. In this study, several approaches to adjust the spectral acceleration level at each period range were tested. These are the intrinsic and scattering attenuation considering the earthquake environment, magnitude, distance change by the green function method, and a rupture propagation direction's directivity effect. Using these variables, the amplification ratio for the representative natural period was regressed. Finally, the optimum condition compatible with the design response spectrum was suggested, and the validation was performed by converting the recorded ground motion.