ABSTRACT
For a proper design of a structure in a specific site, it would be necessary to have two statistically significant sets of accelerograms, corresponding to historical earthquakes with a short and a long return period respectively (the first one to check the serviceability limit state of the structure, the second one to grant it against collapse). For research purpose, artificially generated accelerograms are often used, because they can correspond to a given elastic response spectrum and it is thus easy to obtain a large set of homogeneous accelerograms. However, the uncertainty on the reliability of artificial-quakes energy contents led us to prefer historical events. We therefore examined the Enea-Enel data base which collects more than one thousand of recordings of the earthquakes occurred in Italy during the last twenty years. Most of them were discarded, because of their weakness (PGA< 0.05 g). The other ones were divided into sets, basing on site soil, magnitude, PGA value, shape of response spectrum, energy spectrum. The number of recordings with high PGA was not sufficient to create a statistically significant set of strong earthquakes. For this reason we finally selected only one set of 30 accelerograms (tab. 1) which presented an average value of PGA equal to 0.16 g. The elastic response spectra of the selected accelerograms were statistically analysed, evaluating at each period T the mean (plotted in the average response spectrum, fig. 1), the standard deviation, the coefficient of variation and the frequency distribution. The shape of the mean spectrum is very similar to that of the elastic spectrum proposed by ECS for soil site A; the coefficient of variation often attains values greater than 0.5, showing a large dispersion of samples, in spite of our efforts for homogeneity; the frequency distribution proves to be approximately lognormal. Basing on the above set, we defined the set of strong earthquakes, scaling the actually recorded ground accelerations by a factor 2.2 so as to get a mean PGA equal to 0.35 g, and the set of weak earthquakes, reducing the strong ones by a factor 4, i.e. multiplying the actual values by 0.55. Some problems may arise in consequence of this procedure, because the energy contents of actual weak and strong earthquakes are usually not proportional to their PGA.
