PSI - Issue 84

Fulvio Busatta et al. / Procedia Structural Integrity 84 (2026) 797–804

799

evolution of the modal estimates is expected to be stationary until the structural condition does not change, whereas non-stationarities should happen if any structural anomaly occurs. However, in real-world applications, variations of modal estimates might occur not only because of structural changes, but also due to the environmental and operational variability (EOV). Hence, the removal of the EOV effects should be performed before conducting any novelty analysis. The paper presents and discusses the dynamic monitoring of the Torrente Reggia PSC bridge. About 1-year monitoring was carried out by applying the automated modal identification tools available in the Python-based ANAS platform, with eleven vibration modes being identified and tracked. Subsequently, the modal estimates were exported to MATLAB to apply the following procedures: (a) the control charts based on the Hotelling T 2 statistics (Hotelling, 1947) and Mahalanobis distance (Mahalanobis, 1936), with frequency residuals being evaluated through PCA regression (Sharma, 1995), and (b) the Cointegration technique (Cross et al., 2011), not requiring any removal of EOVs and based on a linear combination of non-stationary monitored features. The performance of those procedures was compared by using a 9-month long training period. The paper outline is as follows. Section 2 describes the PSC bridge and the monitoring system installed in the infrastructure. Section 3 summarizes the modal estimates identified through OMA that were used as a reference for the continuous dynamic monitoring. Section 4 shows some selected outcomes from 1-year continuous SHM of the bridge deck. Section 5 focuses on novelty detection. Section 6 concludes the paper. 2. Description of the bridge and monitoring system The Torrente Reggia bridge is located in the Umbria Region, middle Italy, on the State Road named SS3bis Tiberina. The bridge deck is a 5-span continuous PSC box girder with span sequence of 45.0 m + 90.0 m + 3×75.0 m and overall length of 360 m. The bridge deck, where the carriageways are located, is approximately 15 m wide. Hence, the span length-to-deck width ratio falls into the range 3-6. Since this ratio is significantly larger than 1, the continuous box girder is expected to exhibit classical beam-like bending and/or torsion vibration modes rather than plate-like modes. Consequently, a standard wired monitoring system was deployed on the bridge deck with 6 uni-axial (vertical) MEMS accelerometers per span located along the girder edges at mid-span, as well as at one-quarter and three-quarter span cross-sections (Fig. 2). Additionally, 4 bi-axial MEMS accelerometers were placed at the top of piers, and 1 tri axial MEMS accelerometer was installed at the abutment for seismic monitoring purposes. Moreover, 3 thermocouples were installed on the girder to support the EOV effect estimation. Overall, the dynamic monitoring system consists of 35 accelerometers, with 41 channels continuously collecting data signals.

Fig. 2. Longitudinal and plan views of the monitoring system installed on the Torrente Reggia bridge (dimensions in m).