MS17.4 - Structural Health Monitoring

Railway bridge KW51 in Leuven, Belgium, has been monitored since October 2018 with the aim of validating various structural health monitoring techniques. The displacement and strain measurements on the structure show a non-linear behaviour, which is attributed to friction in the pot bearings. Therefore, displacement sensors have been installed at the bearings, measuring the longitudinal bearing deformation, i.e. the relative displacement between concrete abutment and main bridge girders. At low excitation amplitudes, the bearings designed as roller supports act as fixed bearings while at higher amplitudes bearing deformation does occur. This contribution describes and validates a methodology that allows incorporating the non-linear frictional behaviour of the bearings in a finite element model of the bridge. This is important for understanding and reproducing the bridge behaviour under combined train and thermal loading as in e.g. virtual sensing applications. The response of the bridge subjected to a train passage is simulated and the resulting computed bearing displacements are compared to the measurements. It is investigated to what extent the resulting model enables describing the response of the railway bridge to train and thermal loading. A comparison between the measured and the predicted bearing displacements under train loading shows a significant improvement of the response prediction in comparison with the case where the bearings are modelled as roller supports. In addition, it is also shown that the model enables a qualitative description of the thermal bridge response.

The number of applications of structural health monitoring techniques to buildings is steadily increasing, fostered by the diffusion of low cost accelerometric sensors based on MEMS. In many cases these sensors may represent an alternative to more expensive transducers, such as piezoelectric accelerometers, even if they typically have lower signal to noise ratios. Their higher noise levels may limit the capacity of identifying modal parameters of structures when small amplitude vibrations are recorded. The present paper presents the results of an experimental campaign in which two SHM systems were installed on a masonry building built at the CIRI Buildings and construction Laboratory of the University of Bologna, recording accelerations for more than six months. One system was based on uniaxial piezoelectric accelerometers and the other on triaxial MEMS. The data recorded by the two systems were used as input for output-only structural identification algorithms. Using the results obtained from the piezoelectric accelerometers as the reference, the capacity of identifying the modal features of the building from the MEMS data is analysed, with focus also on the long-term stability of the estimates of the modal parameters and the effect of temperature variations.