Improving the Procedure of Probabilistic Load Testing Design of Typical Bridges Based on Structural Response Similarities
This paper concerns load testing of typical bridge structures performed prior to operation. In-situ tests of a twospan post-tensioned bridge loaded with three vehicles of 38-ton mass each formed the input of this study. On the basis of the results of these measurements an advanced FEM model of the structure was developed for which the sensitivity analysis was performed for chosen uncertainty sources. Three uncorrelated random variables representing material uncertainties, imperfections of
... erfections of positioning and total mass of loading vehicles were indicated. Afterwards, two alternative FE models were created based on a fully parametrised geometry of the bridge, differing by a chosen global parameter – the skew angle of the structure. All three solid models were subjected to probabilistic analyses with the use of second-order Response Surface Method in order to define the features of structural response of the models. It was observed that both the ranges of expected deflections and their corresponding mean values decreased with an increase of the skewness of the bridge models. Meanwhile, the coefficient of variation and relative difference between the mean value and boundary quantiles of the ranges remain insensitive to the changes in the skew angle. Owing to this, a procedure was formulated to simplify the process of load testing design of typical bridges differing by a chosen global parameter. The procedure allows - if certain conditions are fulfilled - to perform probabilistic calculations only once and use the indicated probabilistic parameters in the design of other bridges for which calculations can be performed deterministically.