Research news

Research dealing with the seismic safety of Krško NPP2

Photo: © 2018 Katja143, CC BY-SA 4.0 Mednarodna

Publish Date: 04.06.2021

Category: Our contribution to sustainable development goals

Sustainable development goals: 11 Sustainable cities and communities (Indicators)

Development of a new non-ergodic ground motion model for seismic hazard analysis at location of the second block of the Krško Nuclear Power Plant (Krško NPP2).

Norman Abrahamson from the University of California, Berkeley, and researchers from the Earthquake Engineering research programme of the Faculty of Civil and Geodetic Engineering of the University of Ljubljana, Anže Babič and Matjaž Dolšek, have developed a new non-ergodic ground motion model based on recorded ground motions from the Seismic network of Republic of Slovenia (SNRS) established by Slovenian Environment Agency (ARSO).

The new probabilistic ground motion model enables evaluation of spectral accelerations depending on earthquake characteristics, seismic source coordinates and building site coordinates. An independent international peer review team panel recently issued a positive opinion to include the non-ergodic ground motion model in the probabilistic seismic hazard analysis for earthquake-resistant design of the new NPP (Krško NPP2), which is the first application of such a ground motion model in the world.

The non-ergodic ground motion model is based on recorded ground motions of small magnitude earthquakes occurring in the vicinity of Krško over the last 20 years. More than 1,000 ground-motion recordings were used from the SNRS ground-motion database precisely managed by the Slovenian Environment Agency (ARSO). By taking into account local earthquakes, it was possible to assess the impact of seismic source locations, regional seismic wave path and that of the site (building) locations on the ground-motion intensity relative to the conventional ground-motion model developed by a conventional seismic hazard analysis approach for the nuclear building site. In developing the non-ergodic ground motion model, machine learning methods based on the Gaussian process theory combined with the Markov chain Monte Carlo sampling were applied.

The results of the study show that the ground motion at bedrock in the region concerned is lower than the ground motion assessed by using conventional approaches. The application of the non-ergodic ground motion model is thus reasonable, as conventional models are based on ground motion recordings from the broader geographic area, which are not necessarily typical of specific sites. The figure below shows the influence of a non-ergodic ground motion model relative to a conventional ground motion model. Presented are effects of earthquakes of magnitude 6 with different source locations. The contours in the figure show the relative influence of ground motion at the facility's site bedrock (red dot) for spectral acceleration at 5 Hz. It can be observed that the ground motions at the facility's site due to earthquakes near the nuclear power plant are practically equal to the ground motions from the conventional model, while the ground motion at the facility's site due to earthquakes with epicentre north and west of the nuclear power plant is reduced with respect to that from the conventional model.

Moreover, the use of local data also reduces the standard deviation of the ground motion intensity, having a favourable impact in calculating seismic hazards for long return periods, which is essential for designing the seismic safety of nuclear power plants.


The non-ergodic ground motion model impact relative to the conventional ground motion model. The contours show the relative difference of the two ground motion models at the facility's site bedrock (red dot) based on natural logarithms of spectral acceleration at 5 Hz and earthquakes with magnitude 6 if they occurred at different locations.


The research is financed by GEN energija d.o.o.

back to list