A novel method of estimating earthquake durations for the analysis of floor vibrations of nuclear power plants

• A new method is developed to estimate the duration of ground motions. • It is suitable for floor vibration analyses of nuclear facilities on hard rock. • Duration is driven by the linear part of energy accumulation. • For rupture distances <100 km duration seems to be insensitive of moment magnitude. • The method predicts shorter durations than 5%-75% accumulation of Arias intensity. Many low-seismicity countries such as Finland have adopted IAEA requirements and recommendations for seismic design of new and existing nuclear power plants (NPPs). In low seismic regions, the structural seismic design is associated with floor vibration of NPPs. The floor vibration analysis is usually conducted in the time domain for which maximum amplitudes are retrieved from design spectra while the duration of ground motion is estimated as an interval between 5% and 75% of accumulation of the Arias intensity. As this method was developed for active seismic regions, it often overestimates the duration for the regions with low seismicity. The present article introduces a new twofold method for estimating the duration. First, the Arias intensity is calculated for a complete and consecutively reduced accelerograms resulting in a deviation curve. Second, this curve is simplified by a piecewise linear regression fitting. The simplified deviation curve has a linear time frame that includes the most significant part of the Arias intensity. The length of the time frame defines the effective duration of a specific ground motion. This implies that the effective duration depends directly on the ground motion instead of predefined percentiles of the Aries intensity. In this study, the method was applied to a set of ground accelerations adopted from eastern Canada, which is geologically similar to the Fennoscandian Shield where appropriate recordings are absent. The results showed that the durations depend on distance, but they were insensitive of magnitude for short rupture distances. This indicates that smaller events can also be useful for estimating the durations even though they do not meet the requirement of design basis earthquake in terms of the peak ground acceleration. The durations obtained with the proposed method were typically shorter than those based on the 5%–75% criterion. The durations received can be used to generate the acceleration time histories compliant with the design response spectra. We also propose durations with different rupture distances for the seismic design of the structures, systems, and components of nuclear facilities in Finland. In a feasibility study, we calculated floor vibrations of a generic reactor building using 3D finite element analysis. The results show that floor accelerations are very similar, when the base accelerogram is complete or shortened to the length proposed in this study.