Safety assessment of short span high-speed railway bridges using a probabilistic approach

In this paper a probabilistic methodology for the safety assessment of short span railway bridges for high-speed traffic is presented. The proposed methodology combines simulation techniques with the extreme value theory in order to minimize the required computational time and guarantee accurate results. Stochastic simulation is employed as it enables reflecting the real variability of the parameters that characterize the dynamic response of the train-bridge system. The variability of the bridge, the track and the train was accounted for, as well as the existence of track irregularities. As a case study the safety of a short span filler beam railway bridge crossed by a TGV double train is analysed. The safety of the train-bridge system is assessed for the two most critical aspects of the dynamic response: the deck accelerations and the wheel-rail contact forces. Two stopping criteria for the simulation analysis were defined in order to direct the computational effort to the critical train speeds. For cases where the probability of failure is one order of magnitude apart from the safety threshold proved the efficiency of these criteria, as ruling them out as critical required a reduced number of simulations. Moreover, it was concluded that the track instability due to excessive deck accelerations is the governing aspect of the dynamic response and the 3/4 span the critical section, limiting the train speed over the bridge to 280 km/h. In terms of number of simulations it was observed that less than 25,000 simulations were needed to accurately estimate the probability of failure for the critical train speed for both safety criteria. Since the probability of failure threshold was set at 10-4 used, this is a good indicator of the efficiency of the proposed methodology.