Bayesian Networks for the derivation of probabilistic functionality loss curves for bridge systems
Abstract
In the context of infrastructure risk assessment, the application of fragility curves to elements such as bridges should mostly serve the purpose of quantifying the performance losses at the system level (e.g. disruption of traffic, additional travel times), since these losses usually outweigh the direct costs associated with the physical damage of infrastructure. To this end, a methodology is proposed for the derivation of probabilistic functionality curves for bridge systems: these curves directly provide the probability of exceedance of various loss metrics given the level of seismic intensity. The main steps of the proposed approach are the following:
- Identification of the failure modes for the various components of the bridge system (e.g. piers, bearings, deck, abutments, etc.).
- Derivation of specific component fragility curves for each component damage state.
- Estimation of the functionality losses that are associated with each component failure mode, through an expert-based survey.
- Construction of a Bayesian Network that describes the failure of the system, from the seismic intensity to the component damage states and the subsequent functionality losses.
- Use of the Bayesian Network to generate the joint probability of occurrence of various levels of functionality losses given the seismic intensity.
This approach is then applied to a generic multi-span simply-supported reinforced concrete bridge, for which component fragility curves are analytically derived through non-linear time-history analyses. The considered loss metrics are the repair duration, the proportion of closed lanes and the speed limit reduction, so that these parameters can be directly fed into traffic modelling tools for the computation of induced delays and the optimization of restoration strategies.
Origin | Files produced by the author(s) |
---|
Loading...