A Full-Scale Study on Traffic Induced Vibrations of a Suspension Bridge

Abstract

This thesis focuses on the traffic-induced vibrations of a suspension bridge, and the modal properties embedded in the traffic-induced response data. Traffic has been observed at Lysefjorden bridge for five different days with low wind speeds. Observations were synchronized with the acceleration and wind data continuously acquired on the bridge. The purpose of the work has been to study the impact load response from heavy vehicles and estimate the modal damping ratios of the bridge. In an experiment performed in cooperation with truck driver Anette Ravndal, a 50 tonne truck has crossed the bridge with different velocities from different directions. Findings from this experiment suggest that impact load response is present, but with different magnitude and cause for different vibration modes. For the first modes, response appears to be relatively similar for both exiting and entering the bridge at high velocities, while for higher modes with, frequencies over 1Hz, the impact effect for the vehicle entering the bridge at high velocity appears to be more significant. Eight modal damping ratios for the bridge have been estimated using viscous damping assumption. This is done by isolating the free decay of the bridge after heavy vehicles have excited and exited the bridge. The results have relatively large variations, but are reasonable compared to other methods used for estimating modal damping ratio for Lysefjorden bridge as well as similar structures in existing literature. These large variations can be explained by variations in wind and temperature, but are also likely due to inaccuracies in filtering of the data. A spectral analysis of the acceleration time series has been performed using Frequency Domain Decomposition. The findings agreewith results fromprevious studies which showthat response from vehicles consists of a combination of both low and higher frequencymodes. Also, all identified frequencies agree with those found previously both using analytical methods and the fullscale response measurements with very little deviation.