PSI - Issue 44

Santa Anna Scala et al. / Procedia Structural Integrity 44 (2023) 267–274 Santa Anna Scala et al. / Structural Integrity Procedia 00 (2022) 000–000

271

5

Fig. 3. ShakeMap in terms of PGA (Michelini et al., 2020) and discretization in iso-seismic areas.

3.2. Fragility curve derivation Fragility curves have been derived using the Maximum Likelihood Estimation (MLE) optimization technique, obtaining as lognormal parameters those that most likely have produced the observed data. In this work, it is assumed that the number of buildings in the different DSs is described by a multinomial distribution function (e.g., Charvet et al. 2014), given the IM value. Moreover, the adopted approach is based on the following assumptions: • fragility curve is a lognormal distribution, completely described by six lognormal parameters, i.e., median PGA values θ DSi (with i = 1, …, 5) and logarithmic standard deviation β ; • logarithmic standard deviation β is common to all building classes with the same structural types. In Fig. 4 and Fig. 5 fragility curves for HR and LR building classes are respectively shown, whereas the corresponding lognormal parameters are provided by Table 1. Note that in Table 1, θ DS5 are not provided for some classes of GQ-HS2 structural type, since no buildings in DS5 belong to such classes. Seismic fragility decreases with the construction age (i.e., going from <1919 to >2001) and increase with the number of storeys (i.e., going from low-rise to high-rise buildings). Such results are easy to deduce from parameters of Table 1: in fact, assuming a common β for all classes with the same structural type, the comparison between curves become the comparison between θ DSi values. Thus, for each structural types, θ DSi increases going from HR to LR classes, given the construction age, and increase increasing the construction age, given the class of number of storeys. Moreover, for each structural type, the maximum θ DSi variation due to the construction age (about 40-70%) is typically greater than the one caused by a different number of stories (about 30-40%). The damage trend with the construction age (e.g., Scala, 2022) could be due to several issues, such as improvements in the normative contents, building materials and construction practices, in addition to the minor material degradation for more recent buildings. Conversely, the number of stories (being tied to the building’s heigh and so to the period of vibration) could affect both structural capacity and seismic demand, leading to a greater amplification of the out-of-plane acceleration at higher floors. 4. Conclusions In the present work, empirical seismic fragility curves of unreinforced masonry buildings are provided based on data collected after L’Aquila 2009 earthquake. The row data deriving from the surveys has been revised, selected, and integrated in order to guarantee its completeness, and then classified in several building classes. The adopted taxonomy considers four unreinforced structural type, eight construction ages and two classes of the number of storeys. Thus,