Wind tunnel testing of bridge decks - Vindtunneltesting av brutverrsnitt

Abstract

First different designs of bridges were investigated to lay out an overview of what bridge designs there were on the table for production and testing. Different bridge cross sections seemed to be favourable for different types of bridges and span lengths. Using the CADing software Fusion 360 and a router a section model of a twin deck bridge girder was built, in a 1:50 scale, based on a suggestion from Multiconsult for the Sulafjorden bridge. The section model was built by carving pieces out from a material called Divinycell and reinforcing the section with an aluminium pipe for strength and stiffness. The model was foiled and held together by end plates using 3D-printed parts for attachment. After assembly the model was explored with regard to eigenfrequencies and results were compared to analytical approximations. The eigenfrequencies of the model proved to be a bit low as it was subject to resonance because of vortex shedding for wind speeds between 0 and 12 m/s. The cross section was tested for static coefficients and aerodynamic derivatives, the precision of the testing and theory was explored through comparing measured forces and theoretical approximated forces.\\

The gathered data showed that the cross section was resistant to galloping and torsional instability while being vulnerable to static divergence and flutter. It was shown the assumption of the self excited forces being linearly dependent on velocity and displacement is not valid, but yield acceptable results in the vertical and angular directions. The ADs extracted from harmonic motion histories was used to estimate self exited forces from random motion histories with frequency content from 0 to 3.5 Hz. The results showed good correlation between the measured and estimated SE forces in the vertical and angular directions, where the $R^2$ values were in range 85-95\%. There was almost no correlation in the horizontal direction. The findings imply that the linearity of the SE forces and the superposition principle can be used in the vertical and angular direction, but not in the horizontal direction. There was observed some Reynolds dependency and static coefficients decreased as wind speed increased. Methods of reducing vortex shedding vibration were explored and the countermeasure of two spoilers in this case proved to be highly effective while a spoiler and a TMD also alleviated some vibrations.