INTRODUCING A NEW SCALING METHOD FOR NEAR-FAULT GROUND MOTIONS BASED ON THE ROOT-MEAN-SQUARE OF SPECTRAL RESPONSES

Assessment of the seismic performance of a structure often requires conducting nonlinear dynamic analyses under a set of ground motion records scaled to a specific level of intensity. Recent researches have demonstrated that pseudo spectral acceleration at the first mode period of vibration, Sa(T1),which is commonly used as the seismic intensity scaling index, may introduce a large scatter in the estimated seismic demands under near-fault pulselike ground motions. Considering the need to provide more accurate estimation of seismic demands by using a smaller number of records, development of an improved scaling method that can reduce the variability in seismic demands becomes inevitable. In this paper, an improved intensity measure is developed based on the Root-Mean-Square (RMS) value of spectral responses, which is calculated over an appropriate period range. For this purpose, Incremental Dynamic Analyses (IDAs) are carried out for five generic frames of short to relatively long periods under 40 pulse-like earthquake records rotated to the faultnormal direction. Statistical study of the IDAs results is performed to determine the type of the response spectrum (i.e. pseudo acceleration, velocity or displacement) and the optimal period range for calculating the RMS value as the improved scaling method. Statistical evaluation reveals that the RMS of pseudo spectral accelerations,(Sa)rms, provides much superior results than RMS of spectral displacements, (Sd)rms, and RMS of spectral velocities, (Sv)rms. It is concluded that the optimal period range varies with the fundamental period of a structure, and that (Sa)rms, if calculated over the optimal period range, can predict the seismic demands with the overall dispersions that are generally reduced by the relative amount of 14 percent with respect to those of Sa(T1). The newly proposed parameter can also reduce the dispersion in predicted collapse capacities of the frames by the relative amount of 24 percent compared to Sa(T1) on average. A. Yahyaabadi, M. Tehranizadeh