Perpustakaan Digital ITB

Aeroelastic phenomena should be considered during the design phase of long span bridge. One of the aeroelastic problems is flutter, the dynamic instability that may cause structural failure at a wind speed called the flutter speed. The prediction of flutter speed needs a thorough modeling of bridge stiffness, inertias, and especially its unsteady aerodynamic forces. The potential flow theory is not applicable to calculate unsteady aerodynamics of oscillating bridges due to their non-streamlined and complex geometry, and the non avoidable flow separation. A semi empirical model proposed by Scanlan is used. In this model, relation between unsteady aerodynamic forces and motion of the bridge is modelled using parameters called flutter derivatives. The flutter derivatives have to be identified from free vibration response of an elastic bridge at several wind-speeds. This thesis presents wind tunnel tests and flutter derivatives identification of aeroelastic bridge models. Three bridge sections were designed, built, and tested in a wind-tunnel. Data obtained from the tests were displacement of the bridge model at some wind speeds which was mixed with noises. A zero-phase band pass filter (BPF) which was designed in MATLAB successfully removed the noises from measured data. Then an identification method called Modified Ibrahim Time Domain method was employed to identify the modal parameters of the system at each wind speed. Simulations of the bridge responses by using the identified modal parameters could represent the measured data excactly. The flutter derivatives of each bridge section at a wind speed were then calculated from the difference of effective stiffness and damping matices to those at zero wind speed. The identified flutter derivatives were verified by using theoretical flutter derivatives of a flat plate calculated based on Theodorsen model. The identified flutter derivatives coefficients H1*, H3*, A1*, and A3* have similar trend to those of flat plates, while H4* and A4* had different trend than those of flat plates. For trapezium section, flutter derivatives coefficient H2* have also different trend than those of flat plates. The identified flutter derivatives of diamond section were also compared to results of other researcher, which had good agreements with those of flat plates due to the closeness of its slenderness ratio to that of flat plate. The comparison showed that the identified flutter derivatives have similar trends with those other researcher except for H4* and A4*. It could be concluded that the measurement procedure used in this research was able to produce flutter derivatives which are in good agreements with those produced by other researcher.