Journal of Structural and Construction Engineering (Jan 2021)
Investigation of the effects of value and modeling approach of natural damping on nonlinear responses of low rise moment-resisting steel frames
Abstract
Nonlinear time history analysis is recognized as the most appropriate tool for assessing the real behaviour of structures during seismic excitations. Validity rate of obtained results from this analysis significantly depends to the modeling, details and defined parameters. One of the main aspects of dynamic modeling is the consideration of natural viscous damping as well as value of damping ratio (ξ), which during last years evaluating the effects of these factors on nonlinear responses of moment-resisting steel frames has been limited, and Rayleigh damping model with ξ of 5% was used by researchers. Therefore, in this study attempt has been made to examine the impacts of value and modeling approach of natural damping on nonlinear responses of moment-resisting steel frames. For this purpose, a full scale 4 story steel structure which tested on shake table in 2007 has been considered as reference and modeled by OpenSees software. Natural damping has been defined with three methods; Rayleigh damping, mass-proportional damping and stiffness-proportional damping, for five different values of damping ratio (ξ=0.01, 0.02, 0.03, 0.04, 0.05). After conducting the nonlinear time history analyses, difference of obtained structural responses compared to the experimental responses has been investigated, and then errors of them have been extracted. Results indicate that reduce of damping ratio leads to the notable decrease of responses, specially, for story shear and overturning moment. The use of 5% damping ratio for nonlinear dynamic analysis of low rise moment-resisting frames is not appropriate and in most of cases leads to the underestimate and unreal results. Besides, it should be noted that the mass-proportional and Rayleigh damping models have higher accuracy in comparison with stiffness-proportional damping model, and these models show lower error.
Keywords
