Computation of aerodynamic derivatives by various CFD techniques

Aerodynamic derivatives are a very useful tool when evaluating the aeroelastic stability of bridges by means of computational fluid dynamics methods. Rather than modelling the complete elastic bridge structure in a CFD code, only a segment of the bridge girder is simulated and is forced to move as a rigid body in vertical bending and torsion modes. Aerodynamic derivatives thus obtained can then be combined with the structural inertia and modal frequencies of the bridge in a conventional flutter routine to determine the critical wind speed of the bridge. This method has been used for about 8 years by the present authors to evaluate the aeroelastic consequences of changing deck layouts during early design. To this end discrete vortex simulations have been quite successful and produced flutter wind speeds in very good agreement with experimental flutter speeds obtained from section model and full bridge model tests. Discrete vortex simulations are to the authors' knowledge only practical in two dimensions which, for most practical bridge decks, are quite adequate for flutter predictions. However, in recent years, twin deck structures connected by lateral cross beams and furnished with complex parapets and wind screens are emerging, with Stonecutters Bridge presently under construction in Hong Kong and the Messina Bridge to be constructed in Italy being prime examples. Such bridge deck structures are inherently three-dimensional, and proper CFD simulations must thus involve three-dimensional computations in order to capture the impact of the structural three-dimensionality on flutter stability.