Multi-mode flutter and buffeting analysis of the Akashi-Kaikyo bridge

Abstract

Flutter and buffeting are both issues of utmost concern in the wind-resistant design of long-span bridges. These aeroelastic phenomena are usually investigated by wind tunnel testing and/or an analytical approach. Since multi-mode behaviors have been reported with increased center-span length of bridges, considering the coupling effect among modes is necessary; and therefore, wind-tunnel testing with an aeroelastic full-model or a multi-mode analysis method is desirable. The Akashi-Kaikyo bridge, which is the world's longest suspension bridge with a center span of 1990 m, was opened in April 1998. Since the bridge is very susceptible statically as well as aerodynamically to wind due to its flexibility, wind-resistant design of the bridge was a crucial issue in the design process. For example, the bridge deflects laterally by about 30 m accompanied by large deck rotation of about −3° at the design wind speed. This requires true three-dimensional consideration in evaluating the bridge stability under wind. Wind tunnel testing was performed in Japan using a 1/100-scaled aeroelastic full-model of the bridge. In this paper, the flutter and buffeting responses of the Akashi-Kaikyo bridge are analyzed. The analytical method used here was a multi-mode analysis in the frequency domain in which mode responses obtained by a spectral analysis were summed including consideration of full aeroelastic and aerodynamic coupling among modes. The analytical results were compared with the wind tunnel test data. Comparison showed that there was good agreement between the multi-mode flutter analysis and the measurement, and that there was indeed a significant coupling among modes. Multi-mode buffeting analysis showed excellent agreement between the analysis and measurements in vertical and torsional response. However, the analysis over estimated lateral response. Significant coupling among modes was also observed in the buffeting analysis, and the multi-mode analysis predicted the measurements better than an equivalent single-mode analysis method.