Bridge Condition Monitoring Using Frequency Domain Decomposition Method

Modal identification techniques play a pivotal role in assessing the dynamic characteristics of critical structures like bridges, offering essential insights for structural health monitoring. These methodologies have evolved over the years, keeping pace with advances in measurement and computational efficiency. Traditionally, input-output modal identification methods have been used for modal analysis. However, their dependence on external excitation presents challenges, particularly for large and complex structures. In response to these limitations, outputonly modal identification techniques have emerged as an alternative. The output-only techniques utilize the structure's dynamic responses to operational loads, eliminating the need for external excitation. This study delves into the effectiveness of output-only techniques in monitoring the condition of a railway steel truss bridge, employing the frequency domain decomposition method to analyze dynamic properties. Specifically, the study focuses on the predominant mode shape vector related to the predominant frequency and evaluates its temporal variation across each sensor position over an observation period. To carry out this investigation, the Pamban bridge was equipped with 20 bi-axial accelerometers to record acceleration responses at bottom nodes on the bridge. Analysis of temporal variation reveals that the contribution of the predominant mode shape vector across different sensor positions remains unaffected mainly by gradual, consistent changes in the truss bridge's cross-sectional area, which may occur over time due to ongoing corrosion. However, slight deviations become apparent when retrofitting introduces non-uniform changes in the cross-section area. Thus, using this approach to identify potential issues in steel bridges needs caution.