COMPUTATIONAL FLUID DYNAMICS SIMULATION OF RAIN-WIND-INDUCED VIBRATION OF STAY CABLES

Due to the large amplitude, frequent occurrence and severe consequences, the rain-wind induced vibration (RWIV) of stay cables in cable-stayed bridges has been investigated extensively by researchers around the world. However, the underlying excitation mechanism is still unclear. Recently, computational fluid dynamics (CFD) simulation has been widely applied in the research on structure vibrations due to dynamic wind loads, with great potential as an alternative tool for wind tunnel testing. It has also been adopted to investigate the RWIV, but only a few studies have been conducted so far. Furthermore, most of the CFD simulations reported in literatures are two-dimensional, and a circular cylinder has been widely adopted to represent the stay cables. Thus, only the wind flow normal to the stay cable axis has been taken into account. Such simplifications may be inappropriate, since some key factors are completely ignored, such as the wind flow parallel to the stay cable axis, the spatial orientation of stay cable, etc. Therefore, the CFD method has also been chosen in this dissertation, but with a new modeling approach-using a two-dimensional skewed elliptical cylinder to represent an inclined stay cable. Extensive simulations have been conducted for this case to investigate the role of the upper rivulet in the RWIV, by examining the related aerodynamic forces acting on the cable models. Two key parameters- airflow speed and the rivulet location have been focused on. The simulation results show that the upper rivulet may play an important role in the RWIV of stay cable. When the rivulet is located within a certain critical range, the aerodynamic force coefficients change significantly not only in amplitude but also in frequency. In particular, a kind of “beat” phenomenon has been extensively observed on the time histories of the aerodynamic force coefficients, especially on the drag coefficients. Some extremely low frequency components, which are in a magnitude of only one-half or even one-third of the corresponding classical Kármán vortex shedding frequency, have been found to be present extensively with remarkable spikes in the PSD plots, even act as the dominant frequency. Moreover, they are quite close to the natural frequencies of the porotype of stay cables in lower order modes, and thus may account for the RWIV. In addition, the elliptical cylinder without rivulet was simulated too and similar low-frequency components have been identified too, which may explain the “dry-cable vibration” observed on site. It has also been found that the Strouhal number drops to a certain degree for the rivulet within this range, and more importantly it can drop abruptly to a significantly low level for the rivulet at some critical positions within the range of 60°≤ θ ≤ 75° with wind speed U = 8.0~16.0 m/s, which is consistent with the experimental results reported by other researchers. The airflow around the elliptical cylinder were also reviewed too. The instantaneous vorticity contours demonstrate that the vortex pattern in the downstream has been significantly affected by the upper rivulet not only in strength but also in period. Therefore, the RWIV and the “dry-cable vibration” may be explained as a type of vortex-induced vibration dominated by low frequency due to either the spatial orientation of stay cable or the upper rivulet at some critical locations, or both. In addition to the new modelling approach, the regular modeling approach frequently adopted by other researchers, i.e., using a two-dimensional circular cylinder to represent the stay cable, has also been used to simulate based on the same cable model and rivulet for contrast. Comparisons have been made between the simulation results based on the two approaches, as well as with the experimental results. The simulation results based on both two approaches agree with the experimental results in certain aspects, such as the variation pattern of the aerodynamic coefficients versus the rivulet location. However, there are large discrepancies in amplitude between the simulation results and the experimental results, and it may be attributed to the three-dimensional spatial effect between the cable model and incoming wind, which is not included in the current simulations. Although low frequency components have also been found in the simulation results based on the circular cylinder, they are not as extensive as in the results based on the elliptical cylinder. Moreover, they are much larger and therefore may not be able to explain the RWIV of stay cables or the “dry cable vibration”