A vibration-based approach for health monitoring of tie-rods under uncertain environmental conditions

This work proposes a vibration-based damage detection approach for axially loaded beam-type structures. The potential of the adopted approach is shown with an application to structural health monitoring of tie-rods: these elements, widely used in civil structures to balance horizontal forces in arches, are normally subject to a significant change of the axial load under environmental and operating conditions. The related changes in the modal parameters are usually greater than those caused by damage, at least at an early stage, making it hard to separate the different effects. In this work, the confounding effect of environmental factors is filtered out by considering more than one vibration mode at a time, thus framing damage detection as a multivariate outlier detection problem. Two damage indexes are presented, based on modal parameters (eigenfrequencies and mode shapes) and a multivariate metric (the Mahalanobis squared distance). The reasons behind the potential of the proposed framework are shown on simulated data first. Then, this strategy is validated on data coming from experimental tests, where two nominally identical tie-rods have been monitored for several months under the effects of realistic environmental and operational conditions. Both indexes proved to be weakly influenced by environmental variations, mainly related to temperature, and thus suitable for automatic damage detection.