PSI - Issue 44

Annalisa Rosti et al. / Procedia Structural Integrity 44 (2023) 91–98 Annalisa Rosti et al. / Structural Integrity Procedia 00 (2022) 000–000

95 5

5. Identification of macro-categories of masonry churches

Consistently with exposure data, masonry churches were allocated to various macro-categories accounting for construction age (i.e. Before XV, XV-XVI, XVII-XVIII, XIX-XXI) and plan area (i.e. small: <150 m 2 ; medium: 150 400 m 2 ; large dimensions: ≥ 400 m 2 ). Classification based on dimensions accounts for the fact that the structural complexity of a church, and consequently the number of potential local mechanisms, directly correlated to the number of macroelements, increases with the plan area (e.g. Cifani et al. 2005). The available sample resulted to be rather uniformly distributed into predefined categories of construction age. The majority of the available churches (63%) has small dimensions, whereas churches of medium and large dimensions constitute 28% and 9% of the available dataset. Given their limited representativeness, churches of medium and large dimensions were pooled together for statistical elaborations. 6. Damage Probability Matrices (DPMs) Damage probability matrices were derived by statistically processing the dataset of 1843 masonry churches. Global damage levels were defined based on damage data gathered during the first (Fig. 4 a) and last post-earthquake survey (Fig. 4 b), whereas the ground motion range was subdivided into equally-spaced bins of 0.10 g width. In line with existing studies (e.g. Lagomarsino and Podestà 2004b, c; De Matteis et al. 2016; Canuti et al. 2021), empirical damage distributions were fitted by the binomial model, allowing for describing the repartition of damage in the different states by a single parameter, µ D (i.e. the mean damage level of the discrete distribution, see e.g. Rota and Rosti 2017; Rosti and Rota 2017). Comparison of Fig. 4 (a) and (b) points out the tendency of the mean damage to increase from the first to the last post-earthquake survey, due to the cumulated effect of repeated earthquake shaking. Cumulated damage has however a limited impact on resulting empirical damage distributions, also considering that about 90% of the available dataset was inspected only once. Based on this consideration, only DPMs relying on damage data collected during the first post-earthquake survey are reported in the following. DPMs were also derived for various macro categories of masonry churches, accounting for plan area (Fig. 5) and construction age (Fig. 6). As already observed, results show the general tendency of seismic vulnerability of masonry churches to increase with the ground motion severity. Frequency of higher levels of damage significantly increases as a function of the experienced ground shaking. Differently from damage metrics based on the mean level of damage, approaches based on the maximum damage indeed allow for capturing the peaks of damage. By contrast, frequency of lower damage states could be underestimated. Comparison of observed and predicted damage distributions highlights the suitability of the binomial model, in line with existing studies (e.g. Lagomarsino and Podestà 2004b, c).

a

b

Fig. 4. Empirical DPMs for the entire sample of masonry churches. Global damage levels are based on damage information collected during the (a) first and (b) last post-earthquake survey.