Influence of code-based record selection methods on the seismic assessment of existing steel buildings

Abstract

The recent concerns regarding the seismic safety of the existing building stock have placed the review of current seismic assessment procedures on the top of the agenda. Alongside with the development of more advanced commercial software tools and computational capacities, nonlinear dynamic analysis is becoming, more and more, a common and preferable procedure in the seismic assessment of existing buildings. Besides the complexity associated with the formulation of the mathematical model, major issues arise related with the definition of the seismic action, which can lead to different levels of uncertainty in terms of local and global building response. Aiming to address this issue, a comparative study of different code-based record selection methods proposed by EC8-3, ASC41-06 and NZSEE is presented herein. The various methods are employed in the seismic assessment of four steel buildings, designed according to different criteria, and the obtained results are compared and discussed. Special attention is devoted to the impact of the number of real ground motion records selected (three and seven) in the response of the buildings and the inclusion of additional selection criteria based on the control of the spectral mismatch of each individual record with respect to the reference response spectrum adopted in the seismic assessment. The results obtained indicate that the Eurocode selection procedure slightly underestimates both the local and global response of the buildings in comparison with the American and New Zealand procedures, which conversely, lead to similar results. It is also concluded that record sets incorporating the additional selection criteria significantly reduce the level of uncertainty in the response.