A class-oriented large scale comparison with postearthquake damage for Abruzzi Region

Abstract

The need to quickly and accurately estimate the number of involved people, dead and the injured, missing people and homeless represent a topic issue in the emergency management and disaster planning. Obviously, such an estimation is closely related to the evaluation of building damage scenario for the area of interest. To this aim, it may be used different methodological approaches for the evaluation of seismic vulnerability at large-scale, consisting of empirical or mechanical methodologies, more or less simplified and that require a different level of detail in input parameters. In present work, a simplified mechanical method - PushOver on Shear Type models (POST) - for seismic vulnerability assessment of infilled RC building is briefly recalled [1,2,3]. The methodology allows the definition of building structural characteristic through a simulated design procedure in compliance with design code prescriptions, professional practice and seismic classification of the area of interest at the time of construction. The evaluation of non-linear static response is performed through a simplified model, considering the contribution of both RC columns and infill panels to lateral resistance, based on Shear Type assumption. Seismic capacity assessment is made in the SPO2IDA framework [4] for different performance levels, defined according to the damage classification proposed by [5]. Finally, the use of a Monte Carlo simulation approach allows the derivation of fragility curves for different damage states (DSs). Therefore, predicted damage scenario is derived from POST methodology for a database consisting of 7597Moment Resisting Frame (MRF) residential RC buildings located in the Abruzzi region, which after the 2009 catastrophic earthquake have been charged to post-earthquake usability assessment procedure [6]. Due to the use of spatially extended and massive amount of data, the reference unit is the class of building. The choice of key parameters for the identification of classes is carried out evaluating their impact on the observed damage collected through post-earthquake survey. Then, within each class, the remaining parameters are assumed as random variables chosen in appropriate distributions evaluated from data collected during the post-earthquake survey. Correlation values among the variable of each class are evaluated. Predicted damage scenario is then compared with damage scenario based on post-earthquake data.