Vortex-Induced Vibrations of Bluff Decks on Cable-Stayed Bridges Using Computational Fluid Dynamics

Abstract

For long and flexible bridges, dynamic wind loading is an important aspect of the design which, if not adequately considered, can impact not just the comfort of the bridge users but also their safety. The standards dealing with adverse wind effects include an amount of uncertainty which is mitigated by prescribing wind tunnel tests for the cases outside the standards' scope. In this research, a number of areas where the design standards need further development were identified and the work towards their improvement was undertaken. While wind tunnel tests are an essential part of the majority of long span bridge designs, they can adversely impact both the project timeline and expenditure. By gaining prior knowledge of what to expect from these tests, engineers can reduce the time spent on their execution. Computational fluid dynamics has been used in bridge engineering research in the past decade, but in practice, it is not routinely used in design. Furthermore, most of the published work in this field dealt with streamlined bridge deck sections, while bluff, i.e., aerodynamically more complex sections, have received little attention. The research work described here was complemented by the results from the wind tunnel testing campaigns conducted as part of the design of three cable-stayed bridges. The first part of this research explores the level of accuracy and practical applicability of computational models built in commercial CFD software when applied in the bridge design industry for bluff bridge decks. A selection of physical and geometric modelling choices was evaluated against the data available from the wind tunnel tests. The CFD models developed were of such size that a desktop workstation can provide sufficient computational power while delivering results in an acceptable time. It was found that 2D RANS models can be successfully used to deliver the static force coefficients and the vortex induced vibration response. Extending the models to 3D did not provide improved results, despite the much greater computational effort. However, it was found that having a simplified geometry which includes a representation of the deck barrier is essential, although this can be greatly simplified as long as the gross blockage is comparable to the target geometry. The flow structure interaction model which allowed the bridge deck to deflect dynamically in response to the aerodynamic loading was able to predict the VIV lock in range, but the oscillation amplitude was overestimated by approximately 40%. When a parametric study of structural damping is of interest, a prescribed motion model in combination with an energy balance approach can provide results in a computationally lighter manner. The second part of this work dealt with the effects of the overhang length on the bridge deck VIV response. The overall goal is to instigate further development of the bridge standards dealing with aerodynamics to include the overhang as a parameter for evaluation of the bridge's susceptibility to VIV. For this purpose, two sets of generic decks were developed, each comprising five geometries with different lengths of the overhang. Deck shapes with and without the barriers were considered. In the case of the bridge deck without the barrier, it was found that with increasing overhang length the maximum displacement amplitude drastically decreases. For two geometries with the largest overhang, an additional LCO region was observed which was attributed to the motion induced vortices. It was shown that the system can enter this LCO only with an initial kick in the displacement amplitude. When it comes to the width of the LCO region, it experienced a decrease with the increasing overhang length. A somewhat different behaviour was observed for the geometries with the barriers. Initially, the displacement amplitude increased but with further increase of the overhang length, the displacement amplitude reduced however it remained above the initial one. The width of the LCO region experienced similar behaviour with the increasing overhang length as in the case of the deck without the barrier. It was noted that for the case with the barrier, the second LCO region was not clearly visible although some similar characteristics were observed. It was shown that CFD can be used as a preliminary design tool to provide information valuable at the onset of the wind tunnel testing. Furthermore, the research showed that the design standards might be conservative when estimating the VIV response of the bridge decks with relatively long overhangs.