Search

Virtual sensing enables a model-based extrapolation of the response measured in a limited number of locations (accelerations, strains, etc.) to other points of interest on the structure. One possible application of virtual sensing consists of estimating the stresses in critical but unmeasured details for assessing fatigue accumulation of structures exposed to cyclic or dynamic loading. Since virtual sensing involves the combined use of response data and a model of the structure, the outcome is inherently prone to modeling errors. This paper investigates the use of virtual sensing for stress/strain monitoring on steel railway bridges. The analysis is based on strain measurements obtained from an in situ monitoring campaign on railway bridge KW51 in Leuven, Belgium. It is found that the quasi-static response of the structure is largely affected by the behavior of the pot bearings, which cannot be easily modeled. A methodology is proposed to avoid the incorporation of the bearings in the model adopted in the virtual sensing, where the strain data are extended by longitudinal displacement measurements at the bearings. It is shown that this methodology enables accurate strain prediction, where in this case similar results are obtained for a model based on the blueprints only and a model calibrated based on experimental modal data. • Virtual sensing enables monitoring fatigue in unmeasured details. • The results, however, are prone to modeling errors. • A validation is performed using data obtained from strain monitoring on railway bridge KW51. • Accurate strain estimation requires explicitly accounting for unknown bearing deformation. [ABSTRACT FROM AUTHOR]

In this paper, the feasibility of structural health monitoring based on natural frequencies is investigated for a steel bowstring railway bridge in Leuven, Belgium. The data used in the study are obtained from an ongoing long-term monitoring campaign on the railway bridge and include acceleration measurements on the bridge deck and the arches. During the monitoring period, the railway bridge has been retrofitted, resulting in data for two distinct states of the structure. Particular attention is paid to removing the effects of environmental conditions, such as temperature, which affect the modal characteristics of the structure and therefore may lead to false-positive or false-negative damage detection. A comparison is made between standard linear regression and robust principal component analysis (PCA), two black-box modeling techniques which are adopted to remove natural frequency variations resulting from changes in the environmental conditions. In order to assess the success rate of these techniques, a receiver operating characteristic (ROC) curve analysis is performed, considering the actual retrofit as well as a number of more subtle structural changes, which are modeled using a detailed finite element model of the structure. The state transition can be observed for the actual retrofit as well as for smaller structural modifications that result in relatively small natural frequency shifts. • Railway bridge KW51 is monitored before, during, and after retrofitting. • Natural frequency data are used to validate vibration-based SHM. • The results obtained from linear regression and robust PCA are compared. • Finite element simulations are adopted to introduce more subtle structural changes. [ABSTRACT FROM AUTHOR]