Observational fragility models for masonry buildings using damage data from the 2012 Emilia earthquakes

Seismic fragility models are important elements for risk prediction, loss-estimation and emergency planning, because they define the relationship between different damage levels and the intensity of ground-shaking. These models can be obtained using different methods, among which the most common are (i) nonlinear numerical simulations and (ii) statistical analysis of observational damage data. The present paper presents seismic fragility models for masonry structures, derived using the second approach on damage data collected after the 2012 earthquake that struck northern Italy. These data were obtained from rapid post-earthquake evaluation-forms issued by the Italian Civil Protection Department, which are compiled after seismic events by qualified engineers for assessing damage in buildings. The damage reported in these documents was converted to a five-level scale derived from the European Macroseismic Scale EMS-98. Since data in the rapid evaluation forms may be incomplete with reference to undamaged buildings, their number was estimated considering also census data. A total of about 60000 masonry buildings were analyzed. In order to define fragility models, damage data must be combined with the ground-shaking intensity at the location of each building. To this aim, various ground-motion intensity measures among those proposed in the literature (e.g. PGA, spectral accelerations, Housner spectrum intensity), were considered. Specific ground-motion attenuation relationships and spatial correlation models were calibrated for all the intensity measures taken into account, using data from the strongest earthquakes of the 2012 Emilia seismic sequence. Shake maps were then computed using a multivariate log-normal approximation for the ground-motion intensity spatial distribution. Finally, a probit generalized linear model was employed in order to fit parametric fragility models. The paper discusses the fragility models obtained and analyzes the correlation among different ground-motion intensity measures ad damage