Impact of Seismic Vulnerability on Bridge Management Systems

Motivated by the potential vulnerability of their road infrastructure, many national authorities and local Departments of Transportation are incorporating seismic risk assessment in their management systems. This Dissertation aims to develop methods and tools for seismic risk analysis that can be used in a Bridge Management System (BMS); helping bridge owners to assess the costs of repair, retrofit and replacement of the bridges under their responsibility. More specifically, these tools are designed to offer estimates of: (1) the seismic risk to single components of bridges and their expected performance after an earthquake. (2) the impact a priori (i.e. before an earthquake) of a given earthquake on the operation of a road network, in terms of connectivity between different locations. (3) the damage a posteriori (i.e. after an earthquake) to road network operation, based on prior knowledge of network vulnerability and on the observed damage to a small number of single bridges. The effectiveness of these methods is tested and validated in a specific case study, the bridge stock of the Autonomous Province of Trento (APT) in Italy. To address the first point, I will first introduce the fragility curve method for risk assessment of individual bridges. The Hazus model is chosen as the most appropriate and is applied to the bridges of the APT stock. Once the fragility curves for all the bridges have been generated, a risk analysis is performed for three earthquake scenarios (with return periods of 72, 475 and 2475 years) and four condition states (operational, damage, life safety and collapse limit state). Next, I will extend the results of the component level analysis to the network level: the APT road network is modeled in the form of a graph and the problem of connectivity between two locations is analyzed. A shortest path algorithm is introduced and implemented to identify the best path between any two given places. Correlation in capacity and demand among bridges is not considered at this stage. After reiterating the fundamentals of probability theory, the theory of Bayesian Networks is introduced. The Bayesian Network approach is used to incorporate mutual correlation in capacity and demand, in risk assessment of a bridge stock. The concept is first formulated and illustrated on a simple case (the ‘twin bridge problem’), then extended to the general case of a full stock. I will show how the same framework can be used in post-earthquake assessment problems, where the evidence of the state of one or more bridges affects the prediction of the performance of another bridge. The outcomes and the limits of this work are discussed at the end of the Thesis.