In this article, a novel strategy to generate broadband earthquake ground motions from the results of 3D physics‐based numerical simulations (PBSs) is presented. Physics‐based simulated ground motions embody a rigorous seismic‐wave propagation model (i.e., including source, path, and site effects), which is however reliable only in the long‐period range (typically above 0.75–1 s), owing to the limitations posed both by computational constraints and by insufficient knowledge of the medium at short wavelengths. To cope with these limitations, the proposed approach makes use of Artificial Neural Networks (ANNs), trained on a set of strong‐motion records, to predict the response spectral ordinates at short periods. The essence of the procedure is, first, to use the trained ANN to estimate the short‐period response spectral ordinates using as input the long‐period ones obtained by the PBS, and, then, to enrich the PBS time histories at short periods by scaling iteratively their Fourier spectrum, with no phase change, until their response spectrum matches the ANN target spectrum. After several validation checks of the accuracy of the ANN predictions, the case study of the 29 May 2012 $Mw$ 6.0 Po Plain earthquake is illustrated as a comprehensive example of application of the proposed procedure. The capability of the proposed approach to reproduce in a realistic way the engineering features of earthquake ground motion, including the peak values and their spatial correlation structure, is successfully proven.

Ground motion selection is one of the most important phases in the derivation of fragility curves through non-linear dynamic analyses. In this context, an easy-to-use software, namely S&M—Select & Match, has been adopted for the selection and spectral matching of recorded ground motions approaching a target response spectrum in a broad period range. In this paper, after a brief description of the key features of the S&M tool, two sets of 125 accelerograms, separately for stiff (i.e. site classes A and B according to the Italian code) and soft soil (i.e. site classes C and D) conditions, have been selected on the basis of the elastic design spectra of the Italian seismic code defined for different return periods. The selected ground motions have been analysed and used for non-linear dynamic analysis of a case study representative of a common Italian RC building type designed only to gravity loads. Results have been analysed in order to check the capability of the considered signals to adequately cover all the damage levels generally adopted in seismic risk analyses, as well as the effects on seismic response due to the selection criteria permitted by the proposed tool.

In this article, the outcomes of a research cooperation between Politecnico di Milano, Italy, and Munich RE, Germany, aiming to improve ground‐motion estimation in the Istanbul area through 3D physics‐based numerical simulations (PBSs), are illustrated. To this end, 66 PBSs were run, considering earthquake scenarios of magnitude ranging from $Mw$ 7 to 7.4 along the North Anatolian fault (NAF; Turkey), offshore Istanbul. The present article focuses on the detailed introduction of the simulated scenarios comprising: (1) the setup of the 3D numerical model, (2) the validation of the model with recordings of a recent earthquake, (3) the PBSs results, (4) a parametric study on the effect of different features of the seismic source, and (5) a comparison with well‐established ground‐motion prediction equations to highlight the main differences resulting from the use of a standard empirical approach as opposed to physics‐based “source‐to‐site” numerical simulations. As a main outcome of this study, we observed as, for magnitude $Mw$ 7 and 7.2, PBSs are in agreement with empirical prediction models whereas, for magnitude $Mw$ 7.4, PBSs provide higher ground‐motion estimates, as a consequence of directivity effects, amplified by the specific geometry of the portion of the NAF facing Istanbul.