According to the results of conventional wind tunnel tests on rectangular cross sections with side ratios of B/D = 2–8 (B: along-wind length (m), D: cross-wind length (m)), motion-induced vortex excitation was confirmed. The generation of motion-induced vortex excitation is considered to be caused by the unification of separated vortices from the leading edge and secondary vortices at the trailing edge [1]. Spring-supported test for B/D = 1.18 was conducted in a closed circuit wind tunnel (cross section: 1.8 m high×0.9 m wide) at Kyushu Institute of Technology. Vibrations were confirmed in the neighborhoods of reduced wind speeds Vr = V/fD = 2 and Vr = 8 (V: wind speed (m/s), f: natural frequency (Hz)). Because the reduced wind speed in motion-induced vortex excitation is calculated as Vr = 1.67×B/D = 1.67×1.18 = 2.0 [1], vibrations around Vr = 2 were considered to be motion-induced vortex excitation. According to the smoke flow visualization result for B/D = 1.18 which was carried out by the authors, no secondary vortices at the trailing edge were formed, although separated vortices from the leading edge were formed at the time of oscillation at the onset wind speed of motion-induced vortex excitation, where aerodynamic vibrations considered to be motion-induced vortex excitation were confirmed. It was suggested that motion-induced vortex excitation might possibly occur in the range of low wind speeds, even in the case of side ratios where secondary vortices at trailing edge were not confirmed.
In this study, smoke flow visualizations were performed for ratios of B/D = 0.5–2.0 in order to find out the relation between side ratios of rectangular cross sections and secondary vortices at trailing edge in motion-induced vortex excitation. The smoke flow visualizations around the model during oscillating condition were conducted in a small-sized wind tunnel at Kyushu Institute of Technology. Experimental Reynolds number was Re = VD/v = 1.6×103. For the forced-oscillating amplitude η, the non-dimensional double amplitudes were set as 2η/D = 0.02–0.15. Spring-supported tests were also carried out in order to obtain the response characteristics of the models.