Spatio-temporal distribution of aerodynamic forces and their associated vortex drift patterns around a closed-box girder during torsional vortex-induced vibration

Torsional vortex -induced vibrations (VIVs) pose potential risks to prototype bridges with increasing spans and lower damping levels. However, the underlying mechanism remains unclear. Wind tunnel tests revealed that torsional VIVs occurred at an initial attack angle of + 3 degrees. To unravel the VIV mechanism throughout the lock -in range, encompassing the ascent stage, amplitude extremum, and descent stage, the spatio-temporal distribution of aerodynamic forces around the bridge girder was analyzed. The results demonstrated the presence of traveling pressure waves with approximately 2.5 wavelengths and three uniformly distributed separated vortices over the upper side throughout the lock -in range. The primary source of these travelling pressure waves stemmed from the vortex drift patterns, with the aerodynamic wave travelling velocity equating the vortex drift velocity. The formation of separated vortices, known as the "double -vortex model", occurred at the windward barrier on the upper side and windward maintenance rail on the lower side. These vortices directly imposed travelling pressure waves on the girder, thereby inducing torsional VIVs. Among them, the former vortices contributed more to the VIVs than the latter. Therefore, the formation of separated vortices and their related aerodynamic waves in the windward edge with the 2nd SVM/AWM modes predominantly governed the occurrence of torsional VIVs. Furthermore, both the ratio of the wave travelling time to the vibration period and the mean value of the wave travelling velocity ratio exhibited an inverse relationship with the VIV responses. However, the actions exerted by the vortex drift patterns were proportional to the VIV responses on both the upper and lower sides. Therefore, there existed evolutionary characteristics of aerodynamic waves and their related vortex drift patterns throughout the lock -in range, which was responsible for the "lock -in" phenomena.