GNGTS 2022 - Atti del 40° Convegno Nazionale

330 GNGTS 2022 Sessione 2.2 are RC buildings. The dataset adopts the typological building classification adopted in Rosti et al. (2021a,b). Empirical fragility curves are derived by statistical processing of the damage database for several masonry and RC building typologies representative of the Italian building stock. Fragility curves are calibratedby characterizing the groundmotion intensity at thebuildings locatedwithin the L’Aquila municipality, i.e. the area severely affected by the earthquake, using the broadband shaking scenarios from the PBS. In line with existing literature studies, the cumulative lognormal distribution is adopted for describing the probability of reaching or exceeding a preselected damage level, as a function of the seismic intensity measure. The ground motion IMs selected for the fragility analysis is PGA, which is the reference ground motion intensity measure used in the Italian national platform for seismic risk assessment (Borzi et al., 2021; Dolce et al., 2021). Although not reported here for brevity, other IMs, such as PGV and SAavg (weighted average spectral acceleration in the range 0-1s) were considered for the derivation of the fragility curves. As a verification benchmark, fragility curves are derived, through the same statistical approach, using the latest version of the ShakeMap (v4) according to Michelini et al. (2020). Results. Fig. 2a shows the comparison between the ground shaking scenario, in terms of Peak Ground Acceleration (PGA), obtained from PBS (left) and from the ShakeMap (right). The maximum horizontal component (Hmax) is shown to enable a consistent comparison between the two approaches, since ShakeMaps are released only for the Hmax component. Note that the PBS scenario is computed by spatially interpolating the PGA values computed directly from the broadband acceleration time histories obtained by the simulation approach combined with the ANN2BB procedure. It is found that PBS provides a realistic spatial distribution of PGA values, which reflects the physical features of the source rupture and of local site response, while the ShakeMap provides a smooth pattern with limited spatial variability. Fig. 1 - The case study of the Mw6.2 2009 L’Aquila earthquake: area under study (top left), 3D PBS model by SPEED (top right) and damage database in the detailed study area addressed in this work (bottom).