PSI - Issue 78

Giovanni Smiroldo et al. / Procedia Structural Integrity 78 (2026) 1585–1592

1586

Even with the growing availability of strong-motion recordings in modern databases, it is still challenging to find recordings that simultaneously capture both the intensity of shaking that is typically associated with large-magnitude, near-field earthquakes and the specific geological and tectonic context of a given site. To this end, Physics-Based Simulation (PBS) of ground motions has emerged as a powerful and increasingly adopted alternative to recorded time histories. The simulations can accurately capture the physical processes of earthquake generation and wave propagation by including the unique characteristics of the seismic source, the source to-site path, and the local site conditions. However, to be reliably used in engineering practice, PBS-generated datasets must be rigorously validated. Several studies have emphasized the importance of such validation efforts (i.e., Petrone et al., 2021; Smerzini et al., 2024; Smiroldo et al., 2025). In the present study, the validation strategy involves the comparison between ground motion intensity measures derived from PBS simulations with those inferred by empirical Ground Motion Prediction Equations (GMPEs). This allows for a robust assessment of how well the simulations replicate observed seismic behaviour with regards of intensity measures. 2. Simulation methodology The Physics-Based Simulation (PBS) method used in the current study simulates both the generation of seismic waves – considering rupture processes along an extended fault – and their propagation (e.g. Magrin et al., 2016; Panza et al., 2012). Ground motions are calculated as the tensor product of the seismic source tensors and the Green's function of the medium, i.e., the dynamic response of the Earth's crust, both in shallow and deep layers. This process enables a realistic simulation of seismic wave propagation from the source to the site of interest. The PBS technique generates complete time-series seismograms that are directly applicable to structural analysis. By explicitly simulating the rupture process – particularly important for near-source scenarios – the method naturally incorporates source, path, and site effects. In addition, it has the capability to simulate intra-event and inter-event ground motion spatial variability, which remains a challenge for conventional Ground Motion Prediction Equations (GMPEs). The generation of PBS seismograms for a specified hypothetical seismic event includes two main steps:  Simulation of the rupture process and of seismic wavefield generation, obtained by the Extended Source model;  Simulation of wave propagation and calculation of PBS seismograms at the points of interest, obtained by Discrete Wavenumber Method (DWN). The ES model simulates earthquakes by distributing a relative slip field over the fault plane, defined as a rectangular surface discretized using a grid of point sub-sources. Each sub-source contributes to the overall seismic moment and is simulated as an instance of non-stationary random processes. The effect is a stochastic uncertainty and better simulated ground motion. Such an approach is able to produce records of the kinematics of rupture, along with directivity effects, that are extremely significant in near-fault areas. The ES model allows for the simulation of both the spectral characteristics (amplitude and phase) and stochastic properties of real accelerograms, such as their envelope and peak values. Considering the unpredictability of future fault behaviour, rupture process is addressed statistically. For this study, several rupture scenarios are generated according to a Monte Carlo approach (Gusev, 2011). All realizations are different in slip distribution, nucleation site, and rupture propagation, providing a large sampling of source variability for the selected scenario, as illustrated in Fig. 1.