GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 2.2 GNGTS 2024 measures (IM) and engineering demand parameters (EDP)(Jalayer & Cornell, 2009). This linear relaConship, expressed as: is determined by parameters a and b through linear regression. Assuming a lognormal distribuCon, the fragility funcCon can be expressed as: Here, is the standard normal cumulaCve distribuCon and the logarithmic standard deviaCon of linear regression. The methodology can accommodate aleatory and epistemic uncertainCes, considering variaCons in structural response, mechanical properCes, capacity thresholds, and model parameters. Aleatory uncertainCes, linked to record-to-record variability, mechanical properCes, and capacity thresholds, contribute to fragility funcCons. Record-to-record variability is naturally considered by the iniCal data cloud . Furthermore, epistemic uncertainty in model parameters is addressed through a bootstrap procedure involving resampling with replacement, resulCng in n different realizaCons of , leading to n fragility curves and regressions. The report adopts n=1000 bootstrap samples, providing a comprehensive approach to assessing fragility curves that considers mulCple uncertainCes and enhances the understanding of probabilisCc seismic vulnerability. Modeling choices Ground acceleraCons are influenced by source, path, and site effects. Key source modelling parameters include magnitude (M w ), source rupture process (nucleaCon point, rupture front evoluCon, slip distribuCon, depth), path effects, site straCgraphy, receiver-source distance (R), and source-receiver angle. The seismic source, modeled aper the "Medea" fault, undergoes one hundred different rupture process realizaCons (Fig.1). Parameters from the DISS database and source funcCons proposed by Magrin et al. (2016) are employed. Crustal models (CR1 and CR2) and local straCgraphies (L1 and L2) are used to capture the source-to-site path effects. The physical properCes of the crustal models are extracted from literature (Brandmayr et al., 2010). These cellular structures represent realisCc configuraCons present in the Italian territory and were obtained through a nonlinear opCmized inversion of the dispersion curves of surface waves. Both local straCgraphies belong to category B according to Eurocode 8 classificaCon, but L1 can be considered moderately faster (V30= 695 m/s), and L2 slower (V30= 366 m/s). As a result, in common pracCce, all the presented outcomes would represent equivalent scenarios in the selecCon of real signals to be used in nonlinear dynamic analyses. The receiver was placed approximately 26 km from the fault center.