GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 2.2 329 PHYSICS-BASED GROUND SHAKING SCENARIOS FOR EMPIRICAL FRAGILITY STUDIES: THE CASE-STUDY OF THE 2009 L’AQUILA EARTHQUAKE C. Smerzini 1 , A. Rosti 2 , R. Paolucci 1 , A. Penna 2 , M. Rota 3 1 Department of Civil and Environmental Engineering, Politecnico di Milano, Milano, Italy 2 Department of Civil Engineering and Architecture, University of Pavia, Pavia, Italy 3 EUCENTRE Foundation, Pavia, Italy Introduction. With the advancement of high-performance computing resources, three- dimensional physics-based numerical simulations (PBS) have started to play a role in providing realistic, region- and site-specific predictions of seismic shaking and they can be employed, in addition to or in place of empirical approaches for groundmotion prediction. In the recent past, the issues related to seismological (i.e. in terms of ground motion waveforms and intensity measures) and engineering validation (i.e. in terms of engineering demand parameters) of PBS results have already found considerable attention (e.g., Taborda and Bielak 2013; Lee et al., 2020; Paolucci et al. , 2021; Petrone et al. , 2021). However, in view of the seismic risk applications at urban scale (see e.g., Smerzini and Pitilakis 2018; Stupazzini et al. , 2021; Riaño et al., 2021), it is also of paramount importance to verify whether a PBS scenario is suitable to characterize the distribution of ground motion intensity for the calibration of empirical fragility curves. As a matter of fact, in standard approaches, the ground shaking scenario is given either by ShakeMaps (Wald et al., 2021; Michelini et al., 2020), in case a sufficient number of records is available, or by empirical Ground Motion Models (GMM), as it is often the case for historical earthquakes with no instrumental records. However, in all such cases, there is a large level of uncertainty when a ground motion level is associated to the specific site where an earthquake effect is observed. Besides, the estimated ground motion is typically available only through its peak values, with no information on other parameters related to the time history itself, such as duration and frequency content. Instead, the PBS ground motion scenario, upon validation, may provide a complete picture of the variability of the ground motion waveforms, supporting the derivation of empirical fragility curves with a wider set of intensity measures (IM). The main aim of this work is to explore and validate the use of ground shaking scenarios generated by means of 3D PBS for empirical fragility studies. To this end, the case study of the Mw6.2 April 6, 2009 L’Aquila earthquake is considered because of the availability, on one side, of a comprehensive database of post-earthquake damage surveys (Dolce et al., 2019; Rosti et al., 2021a; 2021b), and, on the other, of a validated numerical model for PBS (Evangelista et al., 2017). Case study and methodology. In this work, the case-study of the 2009 L’Aquila earthquake is considered (see Fig. 1). The 3D spectral element model of the L’Aquila earthquake derives from a previous study (Evangelista et al., 2017), which was focused on the calibration and the validation of the numerical model by the SPEED code (Mazzieri et al., 2013) on the available recordings (see top right panel in Fig. 1). To overcome the frequency limit of the numerical model, the ANN2BB technique proposed by Paolucci et al. (2018) and further improved in Paolucci et al. (2021), is used to enrich the PBS signals at high frequency. In order to derive empirical fragility curves, the up-to-date version of the L’Aquila damage database (Rosti et al., 2018; 2020), now available in the Observed Damage Database - DaDo (Dolce et al., 2019), is considered. For the objective of this study, which is to compare the effect of different approaches for the characterization of ground shaking in the empirical fragility analysis, we considered a subset of the damage dataset corresponding to the detailed study area depicted in Fig. 1. The considered dataset, hereafter referred to as detailed dataset, counts 7987 residential buildings, 4564 (57%) of which refer to masonry, whereas 3423 (43%)