A method for improving the dynamic response of full bridge reduced-scale models in aeroelastic wind tunnel tests by using optimization algorithms

The quality of aeroelastic tests of long-span bridges is highly influenced by the ability of the reduced-scale models to reproduce the dynamic response of the real bridge. Hence, the design of the models is one important component in the process, as differences between the dynamic properties of the model and the prototype typically arise due to inherent limitations when constructing the reduced-scale model. This paper reviews the design process of full bridge reduced-scale models and explores sources of inaccuracies. A method based on optimization algorithms is proposed to reduce these inaccuracies in the model dynamic responses. This method is applied to a modified version of a real project consisting of a cable-stayed bridge with two diamond-shaped towers, single-box deck, 176 stays, and four piers. The optimization of the dynamic properties of the full bridge reduced-scale model is conducted in two phases: the first one optimizes the standalone tower, and the second one the complete bridge model. The results show the capabilities of the method, which nearly eliminates the inaccuracies in the natural frequencies of the model. The effects on other dynamic properties, such as variations in the generalized masses and in the modal shapes, are studied and commented upon.