Structural Seismic Demand Analysis: Consideration of "Collapse"

Abstract

It is well known that a structure generally responds well beyond its elastic capacity at high seismic loading; nonlinear time-history analysis is therefore the most accurate procedure for the seismic demand calculations. Such analyses are being used systematically, for example, in the current SAC steel momentresisting frame project. There is apparently no methodology currently available to characterize the seismicdemand calculations when we observe that in some cases structures, especially tall structures, “collapse” because of strength degradation and/or significant effects. In this paper we will propose a novel way to calculate the seismic demand by considering a three-parameter probability distribution model. The parameters of the distribution at any ground-motion intensity level can be calculated easily from the response results when the ground-motion records are scaled to that intensity level. More generally, the parameters can be estimated economically from (1) a conventional-regression analysis of the deformations/drifts of a structure versus the intensity of ground motion conditioned on no collapse and (2) a binary-regression analysis of observations of “collapse” or “no collapse” versus intensity level. We illustrate the seismic-demand calculations through an example problem of a 20-story steel structure that collapses in a subset of the nonlinear time-history analyses under high-intensity ground motions. These regression results can be used with the conventional seismic hazard analysis to calculate the annual response-exceedance probabilities.