Detection, localisation and quantification of stiffness loss in a bridge using indirect drive-by measurements

Drive-by health monitoring uses the measurements gathered on a vehicle while traversing a bridge to assess its condition. To date, drive-by monitoring has been mostly proposed as a pre-screening tool to detect damage under favourable conditions of low vehicle speeds and low road roughness. A major shortcoming is the potential degradation of the road profile, which is often indistinguishable from bridge damage. In this paper, the influence of vehicle dynamics, speed and road roughness is removed by applying cross-entropy optimisation to the response of individual crossings of the vehicle at different speeds. The proposed model-based algorithm locates and quantifies damage while remaining unaffected by changes in the road profile. In addition to providing the distribution of bending stiffness in the underlying bridge, the algorithm isolates the bridge deflections accurately making the reconstruction of the actual road profile on the bridge possible. The latter can be a useful feature for ensuring traffic safety as well as preventing a major dynamic amplification of the bridge response. The algorithm is successfully tested for a quarter-car driving on a 15 m simply supported beam bridge model at highway speeds of 30 m/s over a class 'B' road with a roughness coefficient of 64 x 10-6 m3/cycle.