PSI - Issue 84

Martina Caruso et al. / Procedia Structural Integrity 84 (2026) 143–150

149

Fig. 6. Total road length per province, with a detailed view of the selected area.

Lastly, extensive efforts have been made to compile and integrate fragility and consequence functions for buildings and infrastructure, with emphasis on Italian-specific studies. Vulnerability mapping links exposure classes to appropriate fragility and consequence models, either one-to-one or shared across classes with similar seismic performance. At the national scale, building classes are mapped to GEM typological functions. For building-by building exposure models, fragility functions are assigned either by extending the national residential mapping or by applying a simulated design workflow developed at La Sapienza University, currently applicable to pre-2003 reinforced concrete buildings. For roads and bridges, vulnerability mapping is not yet implemented; instead, a simplified spatial-overlap approach is used to support scalable national analyses. 3.2. Hazard inputs for earthquake scenarios Earthquake scenario analysis combines rupture models, ground-motion models, and site conditions with exposure, fragility and consequence models, and may also include secondary hazards such as liquefaction and landslides. The platform incorporates a catalogue of historical and hypothetical earthquake scenarios. Historical events are selected drawing on the GEM Global Earthquake Impact Database (GEID) (Karimzadeh et al., 2025; Yepes-Estrada et al., 2025) and Global EarthquakE ScEnarios (GEESE) (Brooks et al., 2025) tools, and include major Italian earthquakes from 1905 to 2016. When available, ground motion recordings from the Engineering Strong Motion (ESM) database are used to condition simulations; otherwise, spatial and cross-correlations are accounted for. Hypothetical scenarios are derived from a stochastic event set (SES) from the ESHM20 model (Danciu et al., 2022). Epistemic uncertainty in ground-motion characterisation is addressed through a logic tree of ground-motion models for active shallow crustal regions, with weights of 0.4 assigned to LEA19 (Lanzano et al., 2019), 0.3 to KEA20 (Kotha et al., 2020), and 0.15 each to CEA15 (Cauzzi et al., 2015) and BSSA14 (Boore et al., 2014), based on performance against Italian ground-motion data. Secondary hazards are assessed using geospatial models: liquefaction probabilities are estimated following Allstadt et al. (2022), while landslides are evaluated using the slope-displacement approach of Rathje and Saygili (2009). Local site conditions are represented through a Vs30-based site model derived from the Italian seismic microzonation dataset (Mori et al., 2020). Dedicated site models support liquefaction and landslide analyses by integrating hydrological, geomorphological, and lithological parameters from global and national datasets.