One of the key objectives of performance-based earthquake engineering is the avoidance of collapse. Hence, accurate estimation of the seismic collapse potential is a crucial task in the design and assessment of new or existing structural systems. As a result, this has been the subject of various previous investigations, based on single-degree-of-freedom (SDOF) or multi-degree-of-freedom (MDOF) systems. Due to the large computational cost involved, research efforts have also aimed to develop simple, yet reliable, prediction models for the collapse capacity. However, the majority of these studies have employed short-duration records, thus completely disregarding the influence of strong motion duration. In addition, the strength and stiffness degradation of MDOF systems has not been properly modelled in most cases. Therefore, the aim of this thesis is to further investigate the role of duration in the seismic response, with particular emphasis on the state of collapse, and incorporate its effects in predictive relationships for the collapse capacity.
Among the available strong motion duration metrics, the 5-75% significant duration is used in this research, based on recommendations from previous studies. Its effect on the seismic collapse is firstly investigated on SDOF systems, through incremental dynamic analyses (IDAs) with 101 spectrally equivalent pairs of long and short records. For each duration type, predictive models for the 50th, 16th, and 84th collapse capacity are developed. The importance of pulses in quantifying the duration effects is also examined by means of a modification procedure that generates spectrally and pulse equivalent records. In order to directly incorporate the duration in collapse predictive models, IDAs are performed on SDOFs, using 67 spectrally matched ground motions with varying duration. Regression analyses are then carried out to derive relationships for the mean and variance of the collapse capacity. The research is extended to steel moment-resisting frames (MRFs), designed according to the current and upcoming Eurocode 8. Accordingly, an MDOF-based model for the collapse capacity is developed, incorporating the duration, in addition to the fundamental period, the P−Δ level, the plasticity resistance ratio, and the first-mode participation factor. Finally, the models proposed in this thesis and other similar predictive relationships from the literature are compared through an illustrative case study of a typical 6-storey MRF. The treatment of duration in current codes is also discussed, and recommendations are made to incorporate its effects in collapse assessment procedures.