PSI - Issue 84

Angela Diana et al. / Procedia Structural Integrity 84 (2026) 623–629

626

supported

30 30 30 30 30

4.14 4.10 3.86 3.98 4.17 4.00 3.96 4.23

14.28 13.99 13.48 14.08 - 13.15 13.45 13.67

31.72 31.66 31.65

M Simply

3

7

13.03

supported

3. Methodology This study analyses data acquired through continuous monitoring systems installed on bridges, with the aim of interpreting the main features of the dynamic and structural behaviour. A central element of the procedure is Operational Modal Analysis, a technique that allows the modal parameters of the structure to be estimated using the dynamic responses recorded under operating conditions. In particular, the raw signals acquired by the accelerometers distributed along the deck were firstly subjected to a filtering process, necessary to isolate the frequency components of interest and reduce the noise associated with the measurements. Subsequently, the Cov-SSI algorithm (Tomassini et al., 2025) was applied to the filtered signals to estimate the natural frequencies, damping factors, and vibration modes of the analysed spans. Ultimately, first and the second bending natural frequencies were extracted and compared with the regression formulations reported in a previous study of the authors (). 3.1. Sensors Each infrastructure examined is equipped with a continuous monitoring system based on a set of MEMS single axis accelerometers, which are designed to operate throughout the entire service life of structure. The device integrates, within the same physical housing, both the signal conditioning module and MEMS accelerometric transducer. The measurements are directly provided in digital format via an EtherCat bus cable. From a software point of view, the system uses an integrated platform for acquisition management, responsible for automatic hardware recognition, channel time synchronisation and signal conversion into engineering units. In this context, the software is not limited to signal acquisition alone but constitutes the first level of the data management chain, providing consistent, traceable measurements that can be used directly in subsequent processing and analysis stages. Two main sensors configurations are adopted. The first configuration (Figure 1a) consists of six accelerometers installed along the deck edges, with three sensors on each side. The second configuration (Figure 1b), includes spans instrumented with seven or eight accelerometers, comprising six lateral sensors and one or two central sensors aligned along the deck axis. The local reference system of the sensors is orientated according to the global reference system of the structure. The complete setup also includes triaxial sensors at the piers and temperature sensors which are not considered in this study. 3.1. Acquisition With reference to the analysed dataset, the data are acquired using sampling frequencies of 100 Hz or 200 Hz, a range that encompasses the typical frequency content of the investigated bridges. Subsequently, a filtering process is carried out, including removal the constant component (Detrend) and reduction of the Nyquist frequency using Butterworth filter and downsampling to obtain the range that contains the typical modes of bridge decks (0-25 Hz) (Pierdicca & Mattiauda, 2021). The filtered signals are then associated to their position on the bridge, grouped in sets defined by structural function (pier/span/entire bridge) and go through an identification process described in the following subsection.