Experimental investigation into amplitude-dependent modal properties of an eleven-span motorway bridge

This paper examines experimentally the effect of forcing and response amplitude on the variability of modal parameters of a bridge. An eleven-span motorway prestressed concrete off-ramp bridge was subjected to multiple dynamic tests with varying excitation levels by using eccentric mass shakers exerting forces in the vertical and lateral direction. The frequency sweeping technique with small increment steps in the vicinity of resonant frequencies was employed to construct frequency response functions at different shaking levels from which the natural frequencies, damping ratios and mode shapes were identified for several vertical, mixed vertical-torsional and lateral modes. Softening dynamic forcedisplacement relationships were observed for all the modes, and the natural frequencies showed clear and consistent decreasing trends with increasing response amplitude. Modal damping ratios initially increased with increasing response amplitude, but later, for the modes where experimental data were available, stabilised at elevated levels. A finite element (FE) model of the bridge was also created and the experimental modal properties compared to the numerical ones. A good agreement was generally noticed for the lower modes but the higher modes had more error. The FE model was used to assess the likely levels of structural damage that would have a similar effect on the natural frequencies as the amplitude dependence. One numerical damage scenario indicated that a reduction of 20% of stiffness in the middle of the main span would cause larger frequency shifts of some modes but amplitude dependent effects will dominate in other modes. Another numerical damage scenario was a reduction by 50% of stiffness at the bottom of the highest pier, and it was shown this type of damage would result in only one third of the frequency drop caused by the amplitude effects in a single, most affected mode.