Tuning of Finite Element Models of Multi-girder Composite Structures

Abstract

Dynamic load testing is an important part of the acceptance process for new bridges in Italy. This paper is an overview of a part of a field-testing program carried out to investigate the dynamic properties of the five main new viaducts along the Brescia-Milano highway (BreBeMi) before their operation. Among them, the focus of the paper is on the Muzza Bridge and the VX1 Bridge: they are examples of continuous multi-girder composite structures. VX01 Bridge has a total length of 112 m with three continuous spans while Muzza spans approximately 80 m with a significant skew angle. Structural analysis was performed with the commercial FE software named Midas Gen. Modal parameters were obtained from experimental testing and were then employed in the calibration of the numerical models. The experimental evaluation of the performances of bridges proves very advantageous since it provides a benchmark for the validation of the numerical simulations, which often exhibit an inherent uncertainty. The presence of simplifications and assumptions in the numerical analysis may lead to results that don’t accurately predict the service life conditions of the bridge. In this case study, a comparative discussion of experimental results and numerical predictions is carried out with reference to the two different, seismic isolated, highway bridges mentioned above, both of which were tested using both environmental excitation and forcing: a large set of data was thus collected and an extensive model tuning activity could be carried out, allowing a thorough sensitivity analysis of a number of modelling parameters. The effects of different assumptions used when modelling some peculiar features of composite bridges, such as diaphragms, stiffeners, skew angle, expansion joints, rubber bearings etc., on the prediction of the dynamic properties of the composite viaducts, were investigated. At the same time, a comparison between the experimental results provided by ambient and forced vibration test results was carried out, based on their effectiveness in providing a reliable and useful benchmark for model tuning. Some conclusive suggestions based on the case study are finally addressed to structural engineers needing to set up an efficient procedure to perform similar tests and computer analyses.