Typological fragility curves for RC buildings: influence of damage index and building sample selection

• Empirical fragility curves for RC frame typological classes were calibrated for Abruzzo and Emilia Romagna regions. • A locality completeness ratio was introduced to quantify the amount of surveyed buildings over the total number of buildings in each locality. • The post-seismic damage database has been incremented by adding data related to unsurveyed buildings estimated from Census data. • The role of selected building damage indices on the fragility curves shapes was shown. The recent earthquakes occurred in the last years in Italy registered a great number of casualties and large economic losses. In this perspective, the effective seismic risk assessment has become crucial for the Government mitigation policies that have to allocate the limited economic sources for the risk reduction. In this paper, data collected after the 2012 Emilia and the 2009 L’Aquila, Italy, earthquakes were used to derive novel fragility curves for RC frame buildings classes. Compared to previous studies, different aspects affecting the fragility curves parameters were investigated. First of all, data are completed accounting for the presence of undamaged buildings that are generally uninspected after the seismic events, considering alternative criteria to complete the database. Then, the influence of the building damage index definition was considered by selecting four damage indices available in the literature that allow converting the damage data collected with the post-earthquake form into one of the five damage states, according to the European macroseismic scale. Separate sets of fragility curves, as a function of peak ground acceleration, were developed for the building stock of the two regions. The results reveal that the database correction with the uninspected buildings is relevant to avoid the overestimation of the exceedance probability, especially at lower peak ground acceleration values, resulting in less steep curves. Then, the building damage indices accounting for the damage of both structural and non-structural elements led to a higher exceedance probability with respect to damage indices based on damage of only structural elements, especially for the lower damage states.