Effects of temperature, analysis and modelling uncertainties on the reliability of base-isolated bridges in Eastern Canada

The seismic response of base-isolated bridges depends strongly on the seismic isolation system (SIS) properties, which typically vary under the effect of multiple conditions, such as temperature. The North American Bridge Design Codes and the EN-Eurocode 8 recommend a bounding analysis method in order to consider the effect of SIS properties variation with different conditions. However, this method is not founded on a rigorous probabilistic approach, and thus does not impact the reliability of base-isolated bridges. This paper presents an estimate of the seismic reliability and its variation range for a typical two-span concrete base-isolated bridge located in Montreal, a typical site for Eastern Canada. Temperature, bidirectional horizontal seismic hazard and the main parameters affecting the resistance of bridge components, specifically, dimensions, material properties of the bridge pier and SIS properties are all modelled as Random Variables (RVs). Bridge failure associated with two limit states is examined: (1) damage at the pier base by lack of bending moment capacity and (2) SIS failure due to a lack of displacement capacity. The Monte-Carlo method is used to evaluate each limit state's reliability and probability of failure. In earlier studies, epistemic uncertainties (EUs) associated with the modelling and seismic analysis method were either ignored or not thoroughly considered. A main innovation of the present study is to include EUs associated with the seismic response analysis method and the lead core heating of the lead rubber bearing SIS are modelled as RVs. Their margins of variation are estimated, and their probabilistic distribution parameters are identified. The effects of these EUs on the reliability of the bridge are investigated. The effect of the natural rubber material type on the reliability of the bridge is also studied. Preliminary results show that, due to any of these uncertainties, the global reliability index is within a range of [3.36-3.62], whereas the reliability index when these EUs are not included is 3.49. When including the epistemic uncertainty associated with the response analysis method, the reliability index of the pier base flexural behaviour limit state varies between 3.52 and 3.76(as compared to a value of 3.56 obtained without including the above EUs). Similarly, when including the epistemic uncertainty associated with the heating effect, the reliability index of the SIS displacement limit state varies between 3.31, for an extreme effect, and 3.49 for a more representative effect (as compared to a value of 3.51 obtained without including the above EUs). Earlier studies considering fewer variability sources showed that the global seismic reliability varies between 3.47 and 3.52 for the same typical two-span base-isolated bridge in Montreal. In these studies, the SIS material and geometrical properties were considered deterministic. Other sites in Quebec have shown comparable reliabilities to those at the studied site, with a maximum variation of 3% between sites.

Automated summary

This paper investigates the seismic reliability and its variation range of a typical base-isolated bridge in Montreal. Random variables such as temperature, seismic hazard, and parameters affecting bridge resistance are modeled to evaluate the reliability and failure probability. Epistemic uncertainties associated with seismic response analysis method and heating effect are included, and their effects on the bridge reliability are investigated. The results show that these uncertainties have an impact on the overall reliability of the bridge.