GNGTS 2021 - Atti del 39° Convegno Nazionale

331 GNGTS 2021 S essione 2.2 SITE-DEPENDENT FRAGILITY CURVES FOR CORRECT SCENARIO AND RISK ASSESSMENT R. Rahmat Rabi, G. Monti Department of Structural Engineering and Geotechnics, Sapienza Univ. of Rome, Italy Fragility curves (FCs) of a certain construction embed essential information about the seismic hazard function at the site, the local stratigraphy, and the local topography. For this reason, two twin constructions at different sites – or even at the same site but on different soils – have different FCs. This is true even if their dynamic response is governed by their fundamental mode. Therefore, when developing FCs for a given construction, the reference site, the stratigraphic, and the topographic conditions should always be specified. All FCs are site-dependent and this has no trivial consequences. For example, certain standard constructions – e.g., a bridge with standard piers and deck spans, or a standard industrial building – cannot be assigned a single FC, rather, different site-dependent FCs should be developed whenever the construction is placed at a different site (Fig. 1). The same considerations apply to observational FCs. These are usually obtained by elaborating data collected from earthquake-damaged constructions. A common, yet usually overlooked, shortcoming of this approach is that FCs are sistematically constructed from available evidence, regardless of the site features the constructions are built on. This is certainly one of the factors explaining their usually large variability. However, it should be recognized that disaggregating available data by different site conditions would significantly reduce their statistical significance, especially for the less common soil types, where the lack of available data is endemic. The final goal of this study is therefore the following: how tomodify a FC pertaining to a certain site, with given hazard, stratigraphy and topography, to adapt it to a different site, with different hazard, stratigraphy and topography. In short, how can FCs be made spectrum consistent. The analytical derivations account in a comprehensive manner of all the modifications induced by both the hazard function, the soil stratigraphy and the topography. Figure 2 shows a sample comparison between a spectrum-consistent FC and a FC obtained through a simplistic approach that amounts to shifting the median based on the stratigraphic amplification coefficient at the site. The comparison shows an unacceptable underestimation error that may significantly alter the outcomes of risk and scenario studies. Fig. 1 - Fragility curves including soil type variation at: a) light damage; b) severe damage; c) near collapse.