Item

Abstract

The relation between the response of an elastic single degree of freedom oscillator subject to seismic shaking and the seismological characteristics of the input waveforms represented in terms of the Fourier amplitude spectrum (FAS) has become a topic of growing interest in the engineering seismological community in recent years. As simulations of ground motions assume a greater prominence in engineering design and databases of both strong and weak motion records expand exponentially, the need to reconcile the influence of the controlling seismological properties of the motions with their potential impacts on structural response is becoming ever more important. Prediction of ground motion in the Fourier amplitude domain has several key advantages when compared to that of the response spectrum, namely a closer relationship to the physical seismological properties of the source, path and site that can be inferred from more abundant small magnitude and weak motion data, as well as maintained linearity of site response at all frequencies. Recognising this, new empirical ground motion models have been developed in terms of FAS [1, 2], in addition to an inter-frequency correlation model that can facilitate the definition of conditional spectrum compatible empirical and simulated ground motion records for design [3]. In spite of these advantages, the usage of FAS in seismic hazard and risk analysis remains limited to the scaling of simulated ground motions in the development of median ground motion models using random vibration theory (RVT). One major factor behind this is that translation from FAS to response spectra via RVT requires joint characterization not only of ground motions across a range of frequencies but also of strong motion duration. However, this limitation could potentially be overcome if fragility functions were to be derived directly in terms of FAS. It is for this purpose that a comparison is made in this paper between the efficiency of intensity measures based on the FAS and those based on conventional response spectra for a set of simple building fragility models of the type commonly used in seismic risk analysis. The feasibility of achieving end-to-end loss estimation exclusively in the Fourier amplitude domain is subsequently explored. While its full range of benefits and limitations will require further study, the potential for embedding seismological theory and data more deeply into engineering applications is appealing for the future practice of seismic design and risk analysis.