Developing a multi-variable vulnerability function for a class of multi-span continuous concrete box-girder highway bridges with emphasis on near-field earthquakes

Abstract

The most common tools in seismic performance assessment of highway bridges are single-variable fragility curves. However, fragility curves are not confidently reliable due to utilizing only a single intensity measure (IM). Alternatively, single- and multi-variable vulnerability functions, of which the latter is scarce and rather complex, have been proposed for that purpose, yet without considering the effects of the strong vertical component of near-field earthquakes. This study aimed to develop a multi-variable vulnerability function for multi-span continuous concrete box-girder bridges, accentuating near-field earthquakes. Ten bridge models corresponding to the 1971-1990 design era were considered and analyzed, subjected to three-component records of 164 earthquakes using OpenSees framework. First, 28 IMs related to horizontal and vertical ground motions, and 12 engineering demand parameters were utilized, and a set of potentially optimal IMs were determined through optimality investigation. Then, to ensure that identified IMs are authentically optimal candidates, Lasso regression was employed, introducing 12 IMs as optimal. Finally, Step-wise regressions were conducted to reduce the complexity of the proposed equation. Results revealed peak ground acceleration, peak ground velocity, and velocity spectrum intensity of the horizontal component of ground motion, and acceleration spectrum intensities of both horizontal and vertical components as optimal ones.