Seismic fragilities of curved concrete bridges via bayesian parameter estimation method

Abstract

In this paper, a Bayesian parameter estimation method is applied for a regional seismic risk assessment of curved concrete bridges. For this purpose, a class of three-frame concrete box-girder bridges is chosen as a case-study bridge. Numerical bridge models accounting for geometric and material uncertainties are simulated in dynamic analyses to construct multi-parameter demand models of bridge components, including various uncertainty parameters and an intensity measure. The demand models are established introducing a Bayesian parameter estimation method. Logistic regression is used to develop parameterized fragility curves comparing demands and capacities. The developed fragilities can be used to generate their bridge-specific fragilities using a specific value of uncertainty parameters. Additionally, bridge-class fragility curves conditioned only on the intensity measure can be developed by integrating the parameterized fragility functions over the domain of the uncertainty parameters to assess the vulnerability of this bridge class. The bridge-class fragility model significantly reduces the model error in comparison to the traditional fragility model.