Applying Computational Fluid Dynamics Simulations for Aerodynamic Studies of Long-span Bridges

In the design of long-span bridges, estimating the wind effects is a critical requirement to evaluate and eliminate the risks of wind-induced problems on structural safety and serviceability. This thesis aims to validate the use of Computational Fluid Dynamics (CFD) simulations in bridge aerodynamic studies. A comprehensive review of the literature on the use of wind tunnel testing and CFD modelling for wind issues on long-span bridges was prepared. The lack of confidence of the bridge industry in applying CFD simulations within the design process due to insufficient validation and standardized guidance for use was noticed. Also, the need for a robust workflow for CFD modelling of wind effects for long-span bridge design and bridge operation was identified. Hence, a workflow using open-source software was proposed, which can be conveniently adopted to investigate bridge aerodynamics in bridge engineering practice. A general guidance was also formed where examples were provided to demonstrate specific features. A group of three-dimensional (3D) Reynolds-Averaged Navier-Stokes (RANS) simulations and Detached Eddy simulations (DESs) were performed on sectional bridge deck models at a scale of 1/50. These simulations replicated wind tunnel tests conducted during the design of the Rose Fitzgerald Kennedy bridge in Ireland. Sensitivity studies of the mesh, domain sizes, and turbulence models were conducted during the development of the CFD model, and showed that these factors have appreciable effects on the results of the CFD simulations. When comparing aerodynamic coefficients calculated by simulations and wind tunnel tests, a general agreement was achieved though some discrepancies were found. Similar levels of discrepancies were witnessed in the literature and was most likely due to the sensor interference during wind tunnel tests. In addition, the effects of secondary structures on the aerodynamic coefficients of bridge decks were investigated and shown to be significant. This emphasized the necessity to include these secondary structures when investigating bridge aerodynamics, while existing studies in this area often neglect them. A group of full-bridge simulations were performed at full scale, which was the first to include boundary conditions mapped from Weather Research and Forecasting (WRF) simulations, a high-precision terrain model, and a full-bridge geometry with all the secondary structures. The pseudo-transient time history of wind velocities determined by CFD simulations achieved an extraordinary agreement field measurement data from sensors at multiple locations on the bridge, which was also unprecedented in this area. Results of these full-bridge simulations are also used to initialise simulations at local regions of bridge deck segments. Validations of these local simulations were completed through comparisons of wind velocities predicted by local simulations, full-bridge simulations, and field measurement data. Aerodynamic forces are calculated from these local simulations and compared with results determined by wind tunnel coefficients with field measurement data, Eurocode equations with field measurement data, and Eurocode equations with estimated wind velocities. Aerodynamic forces predicted by CFD simulations were shown to align with the philosophy of sustainable design. Overall, this thesis provided unprecedented validations for the use of CFD simulations at multiple geometric scales. It demonstrated the great performance of CFD simulations in investigating wind effects under realistic wind conditions, which are often challenging for wind tunnel tests to incorporate. The outcome of this thesis is being presented to the NSAI National Committee on revision of the National Annex of the Eurocode on Wind (EN 1991-1-4). But most importantly, great confidence can be drawn from the outcome of this thesis, in using CFD simulations as a robust approach to estimate wind effects on long-span bridges.