GNGTS 2019 - Atti del 38° Convegno Nazionale

418 GNGTS 2019 S essione 2.2 CALIBRATION OF THE MACROSEISMIC VULNERABILITY MODEL AND DERIVATION OF FRAGILITY CURVES FOR URM BUILDINGS S. Lagomarsino, D. Ottonelli, S. Cattari Dept. of Civil, Chemical and Environmental Engineering, University of Genoa, Genova, Italy In the framework of seismic risk analyses at large scale, among the available methods for the vulnerability assessment the empirical and expert elicitation based ones still represent one of most widely used options. In fact, despite some drawbacks, they benefit of a direct correlation to the actual seismic behaviour of buildings and they are easy to handle also on huge stocks of buildings. The ReLUIS Research Unit of the University of Genoa has derived fragility curves for the masonry residential buildings in Italy from the macroseismic vulnerability model (originally proposed by Lagomarsino and Giovinazzi 2006 and further developed within this research). The method may be classified as heuristic, in the sense that: a) it is based on the expert judgment that is implicit in the European Macroseismic Scale (EMS 98), with some assumptions on the fuzzy definition of the binomial damage distribution; b) it is calibrated on the actual damage observed in Italy, available in Da.D.O.. This approach guarantees a fairly well fitting with observed damage but, at the same time, ensures physically correct results for both low and high values of the seismic intensity (for which observed data are incomplete or lacking), and a coherent distribution between the different damage levels. The IRMAplatform (Italian RiskMaps), developed by Eucentre, is based on the vulnerability classes (from A to D) and the five damage levels defined by the EMS 98 (D k , from D1 to D5). Fragility curves are associated to each damage level for each vulnerability class. Therefore, the vulnerability class is representative of a seismic behaviour (represented by the fragility curves), notwithstanding the building typology; indeed, buildings of different types may behave similarly (belong to the same vulnerability class) while buildings of the same type may behave differently. The information on the Italian building stock is taken from ISTAT 2001, in terms of material (masonry and r.c.), age and number of storeys. For masonry buildings, subtypes are defined in terms of age and height, associating to each one the percentage of buildings in the different EMS 98 vulnerability classes. The original macroseismic vulnerability model, described in details in Bernardini et al. (2010), represents the seismic behaviour of each vulnerability class by an analytical correlation between the macroseismic intensity ( I) and the mean damage level (m D ) in terms of two parameters: the vulnerability index ( V ), Tab. 1, and the ductility index ( Q ) (the latter equal to 2.3, when directly derived from EMS 98). The mean damage level is obtained from the damage levels observed in buildings of a specific class, hit by a given macroseismic intensity. The calibration of the model by the observed damage data in Da.D.O. has highlighted that Q is well correlated with V , thus providing a model defined by only one parameter: Table 1 - Vulnerability index for the EMS98 classes. Vulnerability Class A B C D V 0.99 0.80 0.61 0.42 The derivation of fragility curves, to be implemented in IRMA, requires: a) the identification of the mean damage level m Dk , which a probability of 50% is associated to the attainment of damage level k (assuming a binomial distribution: μ Dk = 0.9 k – 0.2); b) a correlation law between I and PGA , directly obtained from the INGV shake map of L’Aquila earthquake (2009):