Automated Operational Modal Analysis of a steel truss railway bridge employing free decay response

The efficiency and resilience of transportation networks depend significantly on the integrity of bridges, which are increasingly threatened by ageing, traffic, and extreme climate events. Traditional visual inspections have notable limitations, necessitating the adoption of more objective methods like Structural Health Monitoring (SHM). This study explores the application of Operational Modal Analysis (OMA) to estimate the modal parameters of railway bridges, specifically using the Covariance-based Stochastic Subspace Identification (SSI-COV) algorithm. The case study involves a steel Warren truss bridge monitored over 20 months. The research demonstrates that SSI-COV, typically requiring stationary random input, can effectively utilise the bridge’s free decay responses following train passages. This approach strongly improves signal-to-noise ratio, which is vice-versa critical for railway bridges ambient vibrations due to the very low input energy, enabling precise modal parameter estimation with shorter time windows and lower-performance sensors. Results were validated against the Peak-Picking (PP) and the Enhanced Frequency Domain Decomposition (EFDD) methods, with SSI-COV identifying three additional natural frequencies and exhibiting lower dispersion in frequency estimates throughout the monitored period. Statistical analysis further indicated that using multiple free decays enhances the accuracy and reduces variability for challenging modes, while dominant modes are reliably estimated with minimal decay data. These findings endorse the combination of SSI-COV and free decays as a robust tool for detailed and long-term bridge monitoring, offering a valuable and potentially low-cost alternative to ambient vibration-based OMA techniques. • Train-induced free decays for operational modal analysis. • Comparison between SSI-COV and Peak-Picking for modal identification. • Demonstration that SSI-COV identifies more natural frequencies and achieves lower frequency dispersion than Peak-Picking. • Finds improved accuracy with more free decay samples. • Long-Term Monitoring Feasibility: Validates SSI-COV for reliable modal parameter estimation over a 20-month period. • Effiicient Data Use: Achieves stable frequency estimates even with limited decay data, optimizing data processing needs.