Identification of efficient groups of design earthquake inputs

In recent years, Performance Based Seismic Design (PBSD) has played a central role for research works in the field of seismic engineering. The core idea of the PBSD is the probabilistic assessment of the structural performances due to seismic action [Bertero and Bertero 2002, Moehle and Deierlein 2004, Zhang et al. 2004]. Within this framework a central role is played by the so called “Probabilistic Seismic Demand Hazard Analysis (PSDHA) [Zhang et al. 2004]. As pointed out in recent scientific contributions [Giovenale et al. 2004] the main shortcoming which prevent the effective application of such a approach currently lies in the large dispersion (i.e. large variability) of the structural response (as evaluated through properly chosen Damage Measure DM or Estimated Damage Parameters EDP) obtained from non-linear dynamic analysis which uses as input selected earthquake acceleration time histories. A number, of scientific work have recently focused upon the identification of the “optimal” procedure for the identification of representative (i.e. characterized by a given probability of occurrence) groups of earthquake inputs which lead to a reduced dispersion of the structural response (“efficient” groups of earthquake). This paper, referring to a general procedure for the identification of representative groups of earthquake inputs proposed by the authors: (a) compares the results which can be obtained following different rules for the application of the methodology proposed, (b) identifies a practical proposal for the effective application of the methodology based upon the use of a vector-valued IM composed of Peak Ground Acceleration (PGA) and Peak Ground Velocity (PGV), (c) proposes the use of “representative” EPI groups characterised by a (1) a vectorial IM composed of PGA and PGV and (2) a specific relative ratio of near-field records to far-field records.