VIV mechanisms of a non-streamlined bridge deck equipped with traffic barriers

Vortex-induced vibration (VIV) has been addressed in the literature mostly for quasistreamlined and shallow π-deck sections, typical of long-span bridges, since the latter are particularly prone to wind-induced oscillations. In contrast, although full-scale observations demonstrate that even steel-box girder bridges, usually characterized by a shorter span length if compared to suspension and cable-stayed bridges, can experience a violent VIV response, systematic studies for these bluffer cross-section geometries are less frequent. In addition, the aerodynamic optimization of nonstructural additions (barriers, screens, fairings) is rarely carried out for this bridge typology. Therefore, a wind tunnel investigation is conducted on a non-streamlined box-girder sectional model (inspired by the Volgograd Bridge, Russia) equipped with two typologies of traffic barriers giving rise to a large ratio of barrier height to deck width. A realistic range of angles of attack (from −3° to 3°) are considered, and static forces, aeroelastic vibrations and wake velocity fluctuations are measured. A large and even unexpected variability in the vibration amplitude and lock-in curve pattern is found, emphasizing the possible existence of competing excitation mechanisms. Indeed, low-porosity barriers can alter the characteristics of vortex shedding, in particular creating a cavity on the upper side of the deck, which is known to foster the impinging-shear-layer instability, as in H-shaped sections. This vortex-shedding mechanism may co-exist with Kármán-vortex shedding and may be responsible for significant anticipation of the VIV onset compared to the predictions based on the Strouhal number measured during static tests. The intensity of a secondary excitation mechanism and its interaction with the dominant mechanism strongly depend on the angle of attack and is largely responsible for profound changes in the VIV bridge response, both in terms of qualitative pattern and peak amplitude. In some cases, the tracks of these competing vortex-shedding mechanisms are even clearly visible in the VIV response curves of the tested bridge model. Finally, the wind tunnel results are also reconsidered based on the quasi-steady theory, highlighting some, even qualitative, discrepancies.