Fragility-based seismic decision making for highway overpass bridges

Transportation network seismic loss modeling is a multifaceted procedure for calculating monetary losses due to transportation link damage, specifically bridge damage, in an earthquake. This dissertation addresses probabilistic techniques for engineering decisions regarding earthquake damage to highway overpass bridges. For each individual bridge, the calculation of losses occurred on two levels. Direct costs included damage to bridge components and their return to functionality. Indirect costs included interruption to the flow of goods, services, and people.; Costs were assessed by defining loss fragilities, the probability of exceeding decision thresholds, given the earthquake intensity. Fragilities were defined using the Pacific Earthquake Engineering Research Center's performance-based earthquake engineering framework. Decision variables were related to earthquake intensity through a series of disaggregated models (demand, damage, and loss). Intermediate quantities were the damage measure and engineering demand parameter. The proposed probabilistic loss-modeling procedure for a typical highway overpass bridge class in California is intended to improve current seismic loss estimation tools.; The demand model was formulated using finite element analysis. The optimum choice amongst conventional intensity measures was the first-mode spectral displacement. The optimum intensity measure was the average of spectral- or bandpass filtered quantities over a natural-period centered band. Intensity measures were coupled with local, intermediate, and global engineering demand parameters to formulate probabilistic demand models.; Two damage models were formulated. Component damage models utilized experimental data to predict response levels at which observable damage states were reached. System damage models utilized finite element reliability analysis to predict the loss of lateral and vertical load-carrying capacity. The best of several improved methods (Method D) for computing system damage incorporated both maximum and residual displacements into the formulation.; Similarly, two loss models were formulated. Component damage states were described in terms of repair costs of returning bridges to full functionality. System load-loss states were described in terms of bridge traffic capacity and collapse-prevention. System loss fragilities were enhanced using the same improved methods developed for damage models. Interim models were combined using numerical and graphical methods. The proposed design procedure incorporates uncertainties at each level of demand, damage, and decision modeling.