GNGTS 2019 - Atti del 38° Convegno Nazionale

448 GNGTS 2019 S essione 2.2 EVALUATION OF THE SEISMIC PERFORMANCE OF SLOPES BASED ON A PROBABILISTIC APPROACH F. Rollo, S. Rampello Dipartimento di Ingegneria Strutturale e Geotecnica, University of Rome La Sapienza, Italy Prediction of sliding displacements of slopes induced by earthquakes represents a crucial parameter to evaluate the seismic performance of a slope. In this work a probabilistic approach to evaluate the earthquake-induced slope displacements is presented and, stemming from the Italian strong motion database updated by Gaudio et al . (2019), some results obtained for different sites in the national territory are illustrated. The database includes 954 records of 208 seismic events recorded by 297 stations located throughout the national territory referred to a temporal period ranging from 14/06/1972 to 24/04/2017. Therefore, the updated version of the Italian seismic database includes the seismic records of the recent destructive earthquakes occurred in L’Aquila (2009), Emilia (2012) and the 2016 Central Italy seismic sequence. Using the updated set of acceleration time histories, Gaudio et al . (2019) performed a series of parametric analyses for different values of the yield seismic coefficient k y to evaluate the permanent displacements. To this purpose, they followed the approach proposed by Rampello and Callisto (2008) and Rampello et al . (2010), performing Newmark-type calculations for the simple scheme of rigid block sliding on a horizontal plane (Newmark, 1965). Therefore, for three different subsoil groups and four acceleration levels, updated upper-bound semi-empirical relationships have been obtained, in which the seismic-induced permanent displacement is expressed as a function of the slope resistance through the yield seismic coefficient. The method provides semi-empirical relationships linking the permanent displacements to the yield coefficient and some selected ground motion parameters. The displacement models predict the natural log of permanent horizontal displacement d as a function of the natural log of one or more ground motion parameters ( GM ) or earthquake magnitude. For the single ground motion parameter model it is: (1) where a 0 and a 1 are regression coefficients and d is expressed cm. The efficiency of the semi- empirical relationships is described by the standard deviation of the natural log of displacement ( σ ln ) (Cornell and Luco, 2001). According to Saygili and Rathje (2008), the smallest standard deviation is obtained using the Arias intensity I a as it encompasses the intensity, the frequency content and the duration of the ground motion. Nevertheless, in this study the peak ground acceleration PGA is considered as single motion parameter due to the limited availability of I a ground motion models. The coefficients in Eq. (1), equal to 3.04 and 1.64 for a 0 and a 1 , respectively, are evaluated for k y = 0.12 and only taking into account displacements greater than 1 cm, as lower displacements are not relevant for an engineering perspective. Eq. (1), in combination with the standard deviation σ ln = 0.806, represent the input parameters to evaluate the performance of a slope within a probabilistic framework. The results of a probabilistic analysis are synthesised in terms of seismic displacement hazard curves, providing the mean annual rate of exceedance λ d for different levels of displacements given a yield seismic coefficient and a specific site. According to Rathje and Cho (2019) and Rathje et al . (2014), λ d for the single ground motion PGA displacement model can be computed as: (2) where P [ d > x | PGA i ] is the probability of the displacement exceeding a specific value x given the peak ground acceleration PGA i and P [ PGA i ] is the annual probability of occurrence of the ground motion level PGA i . The first term is evaluated assuming a cumulative lognormal