PSI - Issue 44

Marco Gaetani d’Aragona et al. / Procedia Structural Integrity 44 (2023) 1760–1767 Marco Gaetani d’Aragona et al./ Structural Integrity Procedia 00 (2022) 000–000

1764

5

horizontal roofs, and 1.5 tons/m 2 for buildings with inclined RC roofs. In the latter case, the total number of stories is considered equal to n-1 and the presence of the additional story is considered only in terms of equivalent mass. To simulate damage and losses from the 6 th April 2009 earthquake, real ground motion records from the L’Aquila 2009 event recorded by five different accelerometric stations located in proximity with the analyzed set of buildings were used for the analyses (i.e., five ground motion records for each direction of each building, see Gaetani d’Aragona et al. (2022a). The Shake-Map in terms of PGA for the event 2009 L’Aquila earthquake obtained in (Michelini et al., 2020) is used to scale the selected accelerograms depending on the building position. Finally, based on the obtained simplified geometry, the orientation of the building main directions with respect to relevant orientation axes (i.e., N-S and W-S) is retrieved to rotate ground motion components according to the longitudinal and transverse main directions of the building according to Somerville (2002). 3.2. Structural analysis Based on the geometrical features derived according to the procedure outlined in §3.1, for each one of the 120 buildings from the database, a Stick-IT model is generated in each main building direction, considering the total frame length in the analyzed direction and the height. In this study, median values of regression parameters reported in Gaetani d’Aragona et al. (2022a) are considered to derive Stick-IT parameters. Then, each model is analyzed via NTHA for each one of the five ground motion records, by rotating and scaling the ground motion components in the two separate directions (longitudinal and transverse). Note that for each record, the rotated component with the maximum PGA is scaled to match the value of PGA from the Shake-Map, while the other component is scaled by the same factor. The Stick-IT model can be adopted to predict the intensity and distribution of EDPs such as IDRs and PFAs. However, in this study, only IDRs are evaluated, since only damage experienced by drift-sensitive components is considered. Thus, the maximum value of engineering demand parameters (IDR max ) is recorded separately for each record, story, and direction. 3.3. Damage and Loss functions In this study, damage and loss analysis is performed considering only those components typical of the Mediterranean area for which specific damage and repair cost functions were developed. Since for residential buildings the largest amount of repair costs is ascribable to damage to infills, partitions and integrated components , only these components are considered. In particular, damage fragilities proposed by Del Gaudio et al. (2019) and cost functions by Del Vecchio et al. (2020) for infills and interior partitions are considered in this study. 3.4. Probabilistic damage and loss analysis A Monte Carlo simulation procedure is employed to account for uncertainties in the definition of damageable components, damage states, and unit repair costs. In particular, due to the variability of possible architectural layouts at the large scale, the definition of damageable components is of difficult definition. To this end, Gaetani d’Aragona et al. (2021a) proposed a simulated design procedure to obtain a number of CALs quantifying the extent of infills and partitions with or without openings based on the in-plane building surface area and shape ratio. Note that the simulation of the architectural layout does not influence the building structural response (Gaetani d’Aragona et al., 2019a). To account for the uncertainties in damage state definition and unit repair cost, fragility functions proposed in §3.3 can be adopted. For each building the value of peak EDP values is firstly determined with the NTHA (§3.2) by adopting the selected ground motion bin (§3.1). These EDPs (separated for direction, story, and record) are then adopted to characterize a statistical distribution, and to generate a larger number of artificial EDP vectors by adopting the procedure by Yang et al. (2009) (fig. 4 (a)). For the generic building, a set of CALs (Gaetani d’Aragona et al., 2021, fig. 4 (b.1)) is generated. For each CAL, the uncertainties in the extension of exterior infill panels, interior partitions, and the percentage of panels with and without openings are simulated considering given probability distributions according to provisions reports (Gaetani d’Aragona et al., 2021a). This way, the variability of architectural layout and the quantification of