A collapse capacity prediction model based on ground motion duration

Abstract

A prediction model for the seismic collapse capacity, which explicitly considers the influence of strong motion duration, in addition to key structural properties, is presented in this study. The model is developed based on incremental dynamic analyses of single-degree-of-freedom systems, employing a set of 67 earthquake records. The selected ground motions are matched by scaling to a target Eurocode 8 response spectrum, to avoid spectral shape bias in the results, and have varying 5–75% significant duration from about 5 s up to nearly 70 s. The structural models exhibit vibration periods in the range of 0.2 s to 3.0 s, varying negative slope in their post-capping branch from 0.02 to 0.30, and ductility between 1.0 and 6.0. Each oscillator is assigned bilinear and pinching hysteresis. Correlation coefficients are computed between the collapse capacity and the structural properties examined, as well as three ground motion parameters, which include the duration, Arias Intensity, and Husid plot slope. The period, post-capping slope, and ductility are all shown to be strongly correlated to collapse resistance while, among the ground motion variables, the duration exhibits the highest correlation. Based on iteratively reweighted least squares, predictive models are fitted to the collapse capacity data, providing estimates for the median collapse capacity and the variance. The predicted collapse capacity associated with the maximum duration considered is found to be up to 67% lower than that corresponding to the minimum duration. The rate of collapse capacity reduction with duration is shown to depend on the period, P-Δ level, hysteresis type, and the duration value itself. The proposed model is compared to those from previous studies and is shown to provide a simple and computationally efficient method to obtain collapse capacity estimates for specific target levels of strong motion duration.