Improved Bridge Evaluation

Licentiate thesis, 2009

Accurate structural models are required for optimal management of existing bridges and economical new ones. However, unavoidable modelling uncertainties and difficulty in modelling the non-linear response of concrete bridges can lead to structural models which do not fulfil this requirement. To estimate uncertain modelling parameters and obtain a more accurate finite element (FE) model, it is possible to improve an initial finite element model, using on-site measurements through FE model updating. A methodology for FE model updating of bridges is proposed. It starts by removing lower bound assumptions from the initial model, as these are commonly used for the design of bridges but are not appropriate for FE model updating. Furthermore, manual model refinements, a parameter study, updating by non-linear optimization, evaluation of model parameter changes, and preparation of the model for further analysis are included. The application of this methodology to the new Svinesund Bridge led to a significantly more accurate FE model. It was also possible to verify the designers’ assumptions of critical details of the bridge. For more frequent practical application of non-linear FE analysis, reliable, easy to verify, and computationally fast and stable analytical methods are needed. Hence, an overview of ideas, approaches and methods is given, which can be used to simplify general non-linear FE analysis. This includes simplifying element formulations which allow the use of material models of reduced dimensions, sectional force dependent stiffnesses, and sectional force-generalised strain relations to include the non-linear response on the element level, rather than on the material level, as well as constitutive modelling in the nodal force-nodal deformation space. Furthermore, ways to analyse numerous load cases in combination with non-linear FE analysis are discussed.