GNGTS 2018 - 37° Convegno Nazionale

400 GNGTS 2018 S essione 2.2 from each set are compared. Jayaram and Abrahamson (2012) and Bijelic et al. (2014) assumed that the difference between the sample mean of simulated and recorded accelerograms follows a normal distribution and that this difference is significant if it falls outside the range between the 2.5 th and 97.5 th percentile of this distribution. Here, a more stringent percentile range is used, namely from 16 th to 84 th percentile. In Fig. 2a to Fig. 2d comparisons between two sets of natural records (NAT1 and NAT2) and the five sets of synthetic accelerograms are shown adopting Inter-Storey Drift (SDR) and Peak StoreyAcceleration (PSA) as EDPs. All tests give a positive result, confirming that there is no systematic difference between the two means. Hence, if used to look for maximum values of instantaneous parameters (as suggested by code-based procedures), NDSHA synthetic accelerograms provide reliable results. A second test consists in the definition of the Site-Specific Maximum Credible Seismic Input (MCSI SS ) at the site of Norcia (NRC) and its comparison with results obtained with the records of the event of the 30 th of October 2016 (Mw=6.5). Fig. 3a shows the calculated MCSI “cloud” (percentile from 5 th to 95 th ) as reported in Fasan et al. (2016), along with a comparison with recorded spectral accelerations of the event. As it can be seen, MCSI spectral accelerations are really close to the recorded ones and site-dependent amplifications are well taken into account, meaning that NDSHA-MCSI can be a valuable tool to predict real accelerations of rare, yet possible, strong events. Similar results are obtained using the records of the events of 24 of August and 26 of October 2016 and can be found in Fasan (2017). Synthetic accelerograms used to define MCSI are also used to perform NLTHA on the frame model introduced above. From all the accelerograms contributing to the definition of MCSI, 31 are selected to match the 95 th MCSI percentile at the first vibrational period of the frame (T 1 =1.5 s). The average acceleration response spectrum of these accelerograms is called Conditional-MCSI (Fasan, 2017) and is shown in Fig. 3b. This percentile was chosen because it is the value suggested to be used as a hazard target when using MCSI (see Fasan et al. , 2015). Indeed, the aim is to verify if even in the non-linear field the expected demands using MCSI are consistent with those due to the real record. Fig. 3c and 3d shows the comparison between the structural demands (Inter Storey Drift and Peak Storey Acceleration) due to the records of the 30 th of October 2016 and due to the synthetic records selected on the basis of the NDSHA-MCSI hazard assessment. The mean demands of the simulations are close to those from the real records, confirming that synthetic accelerograms used to define MCSI can also be used to run time history analysis of structures, allowing to capture site specific ground motion characteristics. The same tests reported in this section were also carried out on a 2-storey frame in order to investigate the possible influence of the structural periods, confirming what is reported here. Further tests are to be performed to check for cumulative parameters such as cumulative ductility, hysteretic energy dissipation or the equivalent number of yield cycles. Conclusions. In thiswork, natural records are used to performnon-linear time history analysis on a selected steel Moment Resisting Frame. Results obtained with natural accelerograms are then compared with those obtained with different sets of NDSHA synthetic records selected in order to be consistent with the characteristics of the natural accelerograms following code-based instructions. These comparisons show that synthetic accelerograms selected to match any target response spectrum (both deterministic and probabilistic) provide structural demands that are equivalent to those obtained with natural records selected to match the same spectrum, at least when looking for maximum values of instantaneous parameters. A second test consists in using synthetic accelerograms to evaluate the Maximum Credible Seismic Input at the site of Norcia in the framework of the Neo Deterministic method. A comparison between MCSI acceleration response spectrum and the recorded spectral acceleration during the event of the 30 th of October 2016 shows that the method well predicted the accelerations. Finally, a comparison between structural demands due to synthetic records selected following the Conditional Maximum Credible Seismic Input as proposed by Fasan (2017), and the demands due to the record of the 30 th of October shows that this method also effectively predicted the structural demand. Hence,