PSI - Issue 44

Carlo Del Gaudio et al. / Procedia Structural Integrity 44 (2023) 259–266

260

2

Carlo Del Gaudio et al. / Structural Integrity Procedia 00 (2022) 000–000

1. Introduction Seismic vulnerability can be assessed by alternative approaches, commonly classified as empirical (e.g. Braga et al. 1982; Rota et al. 2008; Rossetto et al 2013; Del Gaudio et al. 2017a; Rosti et al. 2018; Rosti et al. 2020a; Rosti et al. 2021a; Rosti et al. 2022) and analytical/mechanical (e.g. Lagomarsino and Giovinazzi 2006; Borzi et al. 2008; Del Gaudio et al. 2015; Del Gaudio et al. 2016; Del Gaudio et al. 2017b; Del Gaudio et al. 2018). Resting on statistical processing of damage data from past earthquakes, empirical procedures are appropriate for territorial seismic vulnerability and risk applications (e.g. Rota et al. 2011; Rota and Rosti 2017; Rosti and Rota 2017; Del Gaudio et al. 2019; da Porto et al. 2021; Dolce et al. 2021). However, accurate processing and careful consideration of the different sources of uncertainty associated with the acquired data are essential for a reliable quantification of the seismic vulnerability (e.g. Rossetto et al. 2013). This study presents a comprehensive fragility model for Italian residential RC buildings to be used for territorial seismic vulnerability and risk applications, by exploiting the 1980 Irpinia and 2009 L’Aquila post-earthquake damage data (Dolce et al. 2019). The selected post-earthquake databases are first critically examined and then enlarged by exposure data for suitably characterizing the “negative evidence” of damage (i.e., the absence of evidence that is reasonably assumed as evidence of absence) in the lower ground motion range. An intermediate process of data interpretation, involving seismic input definition, classification of the observed seismic damage and identification of representative building types, is conducted. Empirical fragility curves are derived as a function of the peak ground acceleration, for five damage levels of the EMS-98 (Grünthal et al 1998) and for several building types accounting for design level, construction age and building height. Although the general tendency of the observed seismic vulnerability to reduce with design level and age of construction and to increase with the building height, some anomalies emerged in scarcely populated building typologies. Considering the importance of accurate and reliable seismic vulnerability estimates, two empirically based procedures are developed allowing for resolving issues related to data scarcity and leading to a robust fragility model. 2. Post-earthquake damage databases This study derives fragility curves for residential RC buildings by statistically processing post-earthquake damage data gathered after the Irpinia (1980) and L’Aquila (2009) seismic events (Dolce et al. 2019). Use of the 1980 Irpinia and 2009 L’Aquila damage databases is motivated by the considerable number of completely-surveyed municipalities and of inspected RC buildings and by the availability of damage information on both structural and non-structural (i.e. masonry infills/partitions) building components, essential for the definition of global damage levels (e.g. Del Gaudio et al. 2017a; Rosti et al. 2018). A further advantage related to the use of the L’Aquila database is the opportunity of characterizing the negative evidence of damage, avoiding any bias in the subsequent derivation of fragility functions (e.g. Rosti et al. 2021a, b). Focusing on the completeness of the selected post-earthquake databases, all the Irpinia municipalities were completely surveyed (Braga et al. 1982). In case of the L’Aquila damage data, positive evidence of damage was accounted for by selecting municipalities with completeness ratio (i.e. number of surveyed buildings over the total number of buildings, estimated from census data) exceeding 90%, which were reasonably assumed to be completely surveyed (Rosti et al. 2021a). The post-earthquake damage database was then enlarged by undamaged buildings sited in the L’Aquila non-inspected or partially-inspected (i.e. completeness ratio <10%) municipalities for suitably accounting for the negative evidence of damage in the territories less affected by the ground shaking (Rosti et al. 2021a). The number of undamaged buildings was retrieved from national census data (ISTAT 2001). A global level of damage was associated with each inspected building by using literature damage rules for converting the damage description of the survey form into discrete damage levels of the EMS-98. In case of the Irpinia damage data, the damage conversion rule by Braga et al. (1982) and Dolce et al. (2019) was adopted. In case of the L’Aquila damage data, the damage rules by Rota et al. (2008) and Del Gaudio et al. (2017a) were respectively used for structural and non-structural building components. The need of resorting to different damage rules is due to the fact that different post-earthquake survey forms were used in the aftermath of the two considered seismic events. After evaluating damage levels on individual building components, a global level of damage was associated with each