Summary
Bridge performance under earthquake loading can be significantly influenced by the interaction between the structure and the supporting soil. Even though the frequency dependence of the interaction mentioned in this study has long been documented, the simplifying assumption that the dynamic stiffness is dominated by the mean or predominant excitation frequency is still commonly made, primarily as a result of the associated numerical difficulties when the analysis has to be performed in the time domain. This study makes use of the advanced lumped parameter models recently developed in order to quantify the impact of the assumption on the predicted fragility of bridges mentioned in this study. This is achieved by comparing the predicted vulnerability for the case of a reference, well studied, actual bridge using both conventional, frequency‐independent, Kelvin–Voigt models and the aforementioned lumped parameter formulation. Analysis results demonstrate that the more refined consideration of frequency dependence of soil–structure interaction at the piers and the abutments of a bridge not only leads to different probabilities of failure for given intensity measures but also leads to different hierarchy and distribution of damage within the structure for the same set of earthquake ground motions even if the overall probability of exceeding a given damage state is the same. The paper concludes with the comparative assessment of the effect for different soil conditions, foundation configurations, and ground motion characteristics mentioned in this study along with the relevant analysis and design recommendations. Copyright © 2016 John Wiley & Sons, Ltd.