Evaluation of damping modification factors for floor response spectra via machine learning model

Abstract

In the last few decades, the seismic performance of Non-structural components (NSCs) has been a subject of extensive research. The dynamic behavior of NSCs, can cause damage to buildings, resulting in severe financial losses, injuries, and death. The necessity of assessing the seismic demand on NSCs in seismic design cannot be overstated. Current building regulations include overly simple relationships for defining design inertia forces on non-structural acceleration-sensitive components. Building code formulations are frequently unable to forecast realistic accelerations of NSCs. Floor Response Spectra (FRS) approach determines the seismic demands on acceleration-sensitive NSCs better. In this study, elastic FRS at a floor level of a linear single-degree-of-freedom system is investigated under near-field earthquakes. Damping modification factors (DMFs) for the elastic FRS are calculated for the various damping ratios of the NSCs ranging from 0.1% to 30%, the tuning ratio (i.e., the ratio between the NSC vibration period, Ts, to the building structural period, Tp) ranging from 0.1 to 3, and the building structural period ranging from 0.5 s to 4 s with 0.5 s interval. Also, artificial neural networks (ANNs) were employed to generate an empirical expression to determine the DMFs for elastic FRS. The DMFs were observed to be highly dependent on the building structural period, damping ratio and the tuning ratio of the NSC. The proposed expression is then compared with the existing relations in the literature. The expression is further validated by comparing the predicted DMFs to those obtained from time-history analysis with different ground motion records used to build the model.