A Bayesian approach to construction of probabilistic seismic demandmodels for steel moment-resisting frames

Abstract In this article, construction of the most precise possible probabilistic model for the estimation of seismic demand in steel moment-resisting frames, with different heights, using the Bayesian Statistic, is attempted. In this study, after proving that the best form for defining the model is a linear combination of the logarithm of Intensity Measure (IM) parameter, the best IM was selected based on the practicality, efficiency and sufficiency of models. Some particular results show that the precision of models with one single IM depends on the number of stories in the structure, so no model including the current one, i.e. the one with spectral acceleration in the first period as IM can be introduced as the best model of all frames with whatever height. Since making use of a combined IM including two spectral values as a solution is not practical due to the extreme difficulty of producing a multiple parameter seismic hazard curve, in this study it was shown that the best possible method for construction of the model is the use of two single IMs, i.e. the best model to cover all structural heights is the one with a linear combination of the logarithm of spectral acceleration at the first and second periods.