No abstract

The state-of-the-practice in seismic network reliability assessment of highway bridges often ignores bridge failure correlations imposed by factors such as the network topology, construction methods, and present-day condition of bridges, amongst others. Additionally, aging bridge seismic fragilities are typically determined using historical estimates of deterioration parameters. This research presents a methodology to estimate bridge fragilities using spatially interpolated and updated deterioration parameters from limited instrumented bridges in the network, while incorporating the impacts of overlooked correlation factors in bridge fragility estimates. Simulated samples of correlated bridge failures are used in an enhanced Monte Carlo method to assess bridge network reliability, and the impact of different correlation structures on the network reliability is discussed. The presented methodology aims to provide more realistic estimates of seismic reliability of aging transportation networks and potentially helps network stakeholders to more accurately identify critical bridges for maintenance and retrofit prioritization.

The Bridge Reliability in Networks (BRAN) methodology introduced in the companion paper is applied to evaluate the reliability of part of the highway bridge network in South Carolina, USA, under a selected seismic scenario. The case study demonstrates Bayesian updating of deterioration parameters across bridges after spatial interpolation of data acquired from limited instrumented bridges. The updated deterioration parameters inform aging bridge seismic fragility curves through multidimensional integration of parameterized fragility models, which are utilized to derive bridge failure probabilities. The paper establishes the correlation structure among bridge failures from three information sources to generate realizations of bridge failures for network level reliability assessment by Monte Carlo analysis.Positive correlations improve the reliability of the case study network, also predicted from the network topology. The benefits of the BRAN methodology are highlighted in its applicability to large networks while addressing some of the existing gaps in bridge network reliability studies.

No abstract

No abstract