PSI - Issue 78

Carlo Rainieri et al. / Procedia Structural Integrity 78 (2026) 426–432

429

managed by the S2-DDA commercial software (S2X 2021), while the automatic estimation and monitoring of the modal parameters of the monitored structure is carried out by the S2-SHM commercial software (S2X 2023). The 1800 sec long acceleration records of the ambient vibration response of the building are continuously acquired at a sampling frequency of 100 Hz, and they are stored in distinct files. Each new file is automatically loaded and processed to extract the fundamental modal properties of the building. Thus, every thirty minutes new estimates of the fundamental modal properties of the structure are automatically extracted from the measured acceleration time series by the S2-SHM software, which also shows the time series of the identified modal properties and can be used to compensate the environmental influence and for anomaly detection. In addition to continuous data acquisition, a threshold-based data acquisition has been set to store data related to possible seismic events in separate files. A manual OMA has been carried out right after the installation of the SHM system in order to check its functionality and to get reference estimates of the fundamental modal parameters to validate the automatic outcomes of the SHM system. The preliminary modal parameter identification has been carried out by analyzing an 1800 s long record of the ambient vibration response of the structure. The acceleration time histories have been filtered and decimated by a factor of 10. Modal parameter estimates have been obtained by applying the Covariance driven Stochastic Subspace Identification (Cov-SSI) method (Peeters and De Roeck 2001), as implemented in the commercial software S2-OMA (S2X 2022). The identified fundamental modes are listed in Table 1. The obtained results are also consistent with those obtained from the previous monitoring system installed on the structure about two decades ago (Rainieri et al. 2011). Table 1. OMA results and comparisons with the reference values (denoted by the subscript ref ) reported in Rainieri et al. (2011). Mode f (Hz) x (%) f ref (Hz) x ref (%) I (bending, long side) 0.93 1.2 0.92 1.1 II (bending, short side) 0.98 1.0 0.98 1.1 III (torsion) 1.30 0.8 1.30 0.8 IV (bending, long side) 2.09 2.9 N.A. N.A. V (bending, short side) 2.24 2.7 N.A. N.A. VI (torsion) 2.60 2.8 N.A. N.A. 2.2. Continuous monitoring results The vibration-based SHM system of the School of Engineering Main Building started operating on July 12 th , 2024. Here the results in the period between November 4 th , 2024 and March 5 th , 2025 are considered, characterized only by a single major interruption in the period between December 17 th and December 18 th , 2024, due to the need of performing some planned maintenance operations. Results of continuous monitoring in terms of natural frequency and modal damping ratio time series for the three fundamental modes are shown in Fig. 2. The consistency with the results of the preliminary manual OMA can be appreciated, demonstrating the reliability of the SHM system in tracking the fundamental modal properties of the monitored structure without any human intervention. The plots in Fig. 2 also allow to recognize a significant environmental and operational influence on the natural frequencies and, to some extent, on the associated modal damping ratios. In addition, the occurrence of seismic events in the area, whose effects on the structural response were also recorded by the SHM system, is remarked by the red dashed lines in the same figure. The monitored structure, indeed, is located in the area of Campi Flegrei, which has been hit by several small or moderate earthquakes over the past months as a result of bradyseism. Thus, particular attention has been devoted in this study to the analysis of the patterns of natural frequencies to discriminate between environmental and operational effects, on one hand, and the possible effects of seismic input, on the other hand. Localized drops in the natural frequency time series corresponding at the occurrence of an earthquake are the first indication of a possible structural damage (Rainieri et al. 2018, Cieri et al. 2024). However, looking at the natural frequency time series in Fig. 2 across the main seismic events (M D >3.0) occurred in the area of Campi Flegrei during the monitoring period, it is possible to recognize that there were no clear drops caused by the earthquake loadings. However, environmental and operational variables (EOVs) can hide minor damage effects. Thus, a more accurate anomaly detection has been carried out by applying the Principal Component Analysis (PCA) (Yan et al. 2005) to