Seismic Risk Assessment of Bridges in Quebec using Fragility Curves

Fragility curves are used to evaluate the seismic vulnerability of the multi span bridges in the Province of Quebec. Fragility curves are a probabilistic tool to evaluate the vulnerability of a structure. They express the probability of a bridge reaching a certain damage state for a given seismic event. Due to their probabilistic aspect, fragility curves enable to account for uncertainties in some properties of the bridges or in the seismic excitation. The seismic historical activity in the Province of Quebec demonstrates the need to consider seismic effects in the evaluation and retrofit of existing bridges. Earthquake damages in recent decades have revealed that the seismic vulnerability of a transportation system is mostly due to bridges. The bridge network in Quebec dates back more than 30 years. At the time it was designed and built, the technology and knowledge in this domain were far from their current state.;The seismic risk assessment method using fragility curves is first used to evaluate the seismic vulnerability of a specific highway bridge. The bridge studied is the Chemin des dalles bridge over Highway 55 located in Trois-Rivières in Quebec. The data from in-situ dynamic tests are used to calibrate a three dimensional nonlinear finite element model in OpenSees. A series of 180 synthetic ground motions time histories (GMTH) compatible with eastern Canada are used to capture the uncertainties related to the hazard. Time-histories analysis are performed with these GMTH and the results are used in a log-space linear regression to define the bridge components probabilistic seismic demand model (PSDM). The bridge components monitored, e.g. abutments, bearings and columns are chosen based on the types of bridge failures observed in past earthquakes. The observational data in the literature is also used to defined the limit states for abutments and bearings and for columns in addition to damage mechanics analysis. Finally, a set of fragility curves for the as-built bridge is developed combining the demand and capacity models using a Monte Carlo method.;Even, though the evaluation of a specific bridge is a valuable source of information it is limited to the singularity of the structure. To enable the assessment of the entire road network in Quebec, not one but all bridge shall be considered. Bridges in Quebec have been built for 300 years. Using the Transports Quebec bridges database, a total of 2672 multi-span bridges are identified. The multi-span bridge network population is presented and an average bridge is defined. Then, deterministic analysis with ground motion time histories are performed in this bridge illustrating the seismic hazard variability. Using the same average bridge and the GMTH that have the most impact in this bridge three types of bridge bents are tested to verify the significance of the bridge bent type. Two types of abutments are also evaluated, the seat type with shallow foundation and the other with piles.. Two bent foundation types: shallow foundation with footings and deep foundation with piles are also evaluated. Following the classification of soils in the CAN/CSA-S6-06 and relating with the classification found in the NBCC 2010 four types of soil are used to evaluate their influence in the behavior of the average bridge.;The bridge network population description continues with their classification and their significant parameters for a seismic analysis are determined. To classify the bridges into bridge classes or portfolios the multi span bridges are separated according to the type of bridge system and material. Moreover, other parameters related to geometry, material and other variations are evaluated to better describe each bridge class. An analysis of variance, ANOVA, is performed through a factorial design to enable the determination of the significant parameters that define the uncertainties for each typical bridge portfolio. A three dimensional nonlinear finite element numerical model is developed for each class and these models are submitted to a series of events. The responses of some bridge components are analyzed and a linear regression is performed to develop probabilistic seismic demand models (PSDM). The ensemble of these PSDM is compared to predefined limit states that define de probabilistic seismic capacity models (PSCM) to develop the bridge system fragility curves.;Keywords: Fragility Curves; Seismic Vulnerability; Highway Bridges; Risk Assessment.