SAFETY ASSESSMENT OF A SMALL SPAN HIGH-SPEED RAILWAY BRIDGE USING AN EFFICIENT PROBABILISTIC METHODOLOGY

The behaviour of small span railway bridges is known to be particularly difficult to predict due to the complexity of the coupled train-track-bridge system, as well as for being particularly sensitive to resonant phenomena. The objective of this paper is to present an efficient methodology to evaluate the safety of small span bridges in high-speed railway lines, bearing in mind the real variability of the parameters that influence the dynamic response of the train-track-bridge coupled system. This requires the development of adequate numerical models for the bridge, the track and the train subsystems, as well as the definition of the distributions and variability of all the variables related to the structure, the train, the track and also the wheel-rail contact. Track irregularities are also accounted for in the dynamic analysis. Canelas railway bridge, located in the North of Portugal, was selected as case study. The bridge has six simply supported spans of 12 m each, leading to a total length of 72 m. The deck is a composite structure with two half concrete slab decks with nine embedded rolled steel profiles HEB 500, each supporting one rail track. In order to assess the safety of the bridge two simulation methods were used: the Monte Carlo method and the Latin Hypercube method. Furthermore, both simulation methods are combined with two different approaches to enhance efficiency. One based on the extreme value theory that uses the Generalized Pareto Distribution to model the tail of the distribution. The other uses an approximation procedure based on the evaluation of the failure probabilities at moderate levels to estimate the target probability of failure by extrapolation. The track stability safety due to the deck vibrations level is used as the safety criterion to validate the proposed methodology. The results are extremely promising and indicate the feasibility of this methodology due to the very reasonable computational costs that are required.