Abstract Summary
The wind induced vibration of flexible structures is an important topic in the field of wind engineering in terms of human comfort and safety, and of global instability of cable structures. In particular, cable structures used for roofs, for example, cable net and membrane tensile structures, are very sensitive to wind induced vibrations [1] because the axial force in a cable depends on its geometry. When the cable changes its geometry during the vibrations, it may either lose its tension or the cable tension may exceed the cable’s material strength. The wind structure interaction on flexible roofs is most frequently investigated through numerical analyses using forces time histories calculated by aerodynamic tests on rigid models. However, this approach fails to predict the effective dynamic deformation of the roof. This paper discusses results obtained from aeroelastic tests in a wind tunnel on flexible roofs. Accelerometric signals were acquired using very small and lightweight accelerometers (lighter than 1 g). The tested geometry was a saddle roof that recreates a cable net of a membrane hyperbolic paraboloid roof. The plan shape of the roof was square. The sum of the upward cables’ sag and downward cables’ sag was 1/10 of the roof span. The test model was constructed using a rigid steel structure for the supporting walls and very tiny steel ropes to simulate the cable net. In total, 39 × 39 ropes were used in the model. The membrane was recreated with silk fabric and linked to the net nodes. The wind induced accelerations were measured on 39 points located on the roof cable net nodes. Each cable was prestressed using a mechanical approach based on stayed bolts. Three different wind angles of attack were investigated and accelerations were measured under seven different inflow velocities. A significant flutter of the roof was observed under the highest velocity tested in the wind tunnel, even if collapses of either cables or the membrane were not observed. Signals were acquired with a sampling frequency of 1600 Hz for 1 minute, which corresponds to 3840 s in a real life scale (i.e. time scale it was equal to 0.0156). Signals were divided into 6 slots of 10 minutes (at the prototype scale) and the statistics of peaks were estimated and discussed through the cumulative distribution function plot. Quantiles of 79%, 93% and 95% of the wind induced acceleration were compared with the values provided in literature, codes and standards. The comparison showed that the values of accelerations provided in the literature are underestimated. In particular, electric and thermal installations suspended to the roof should be carefully designed taking into account the roof vibration. Codes and standards tend to neglect this aspect, even if, for sports arenas or meeting rooms, installations are large enough to pose a considerable risk in the event of a fall.
