Boundary condition focused finite element model updating for bridges

This paper pays specific attention to measuring and identifying the behaviour of the bridge bearings of a short span highway bridge, as well as the static and dynamic data commonly used for model updating. This is important because while it is widely accepted that correct simulation of boundary conditions in a Finite Element (FE) model is crucial to the accuracy of the model, few researchers have actually attempted to measure bearing movement as part of their model updating strategy. To demonstrate the approach and the benefits of tracking bearing movement two separate updated FE models of the bridge were developed; (i) was updated using dynamic performance information, and (ii) was updated using response to quasi-static loading. The inclusion of bearing behaviour data proved to be very important, as in (i) it was found that during ambient vibration testing with low level dynamic response to light traffic, the friction on the bridge bearings was such that they were effectively behaving as ‘pinned-pinned’, as opposed to ‘pinned-roller’ as indicated by the bridge drawings. Using this observation it was possible to get the updated model (i) to match very closely with the experimentally measured frequencies and mode shapes. Without this information, conventional model updating optimisation would likely have driven the system parameters (e.g. Young’s modulus, deck mass) to unrealistic values in order to get the FE predictions to match the experimentally observed frequencies. For the static model (ii) it was again observed that friction on the bearing was playing a significant role in the behaviour of the bridge and this was exploited to develop a simple but effective updated FE model that accurately predicted the bridge response during two separate static load tests. No single FE model could represent the bridge for both types of loading but in both cases the bearing performance data were critical in getting the relevant model to match the experimentally observed values.