PSI - Issue 78

D. Scocciolini et al. / Procedia Structural Integrity 78 (2026) 769–776

774

the three monitoring systems are discussed here. Table 2 provides a comparison of the natural frequencies obtained fromdi ff erent monitoring systems, based on both the EFDD and SSI-CoV / IHCA methods. The modal frequencies of the identified modes are in the range 0-30 Hz, with four modes identified based on the accelerations acquired by the piezoelectric sensors and only three modes identified for MEMS and FBG sensors. Consistency among the identified modal frequencies for di ff erent sensor types and for di ff erent identification techniques is observed from Table 2. How ever, the second bending mode is not identified using MEMS and FBG sensors, probably due to the limited number of sensors installed and the higher noise level compared to that of the piezoelectric accelerometers. Considering the piezoelectric system as the reference, Table 3 highlights the relative di ff erence among the the identified modal fre quencies. The largest di ff erence is obtained for the third mode (local bending) using the EFDD method with MEMS sensors, but it is fairly limited, namely lower than 3%. Finally, Figure 5 illustrates the mode shapes for the piezoelectric and FBG sensors, identified for both sensors by means of the EFDD method. Modes identified by means of the SSI-CoV / ICHA method are not represented for the high similarity with those of Figure 5. With reference to piezoelectric sensors, the four identified modes are represented in Figures 5a-5d. The first mode at 4.37 Hz corresponds to the first global bending mode of the structure (Figure 5a); the second mode at 9.27 Hz is the first torsional mode (Figure 5b); the third one in Figure 5c involves local bending of the central span, while the fourth mode is a global second bending mode (Figure 5d). With regard to FBG sensors, some of the modes extracted from the combined datasets of SetUp01 and SetUp02 are presented. If a comparison with the previously illustrated modes is performed, there is consistency in terms of frequencies and mode shapes for the first (Figure 5e), the second (Figure 5f) and the third (not represented in the figure) identified modes. The correlation between the modes extracted using piezoelectric sensors and the modes extracted with MEMS and FBG sensors is quantified in Table 4. The largest MAC value is obtained for the first bending mode, as expected. Slightly lower MAC values are found for the other two modes, which are still greater than 0.80, confirming the consistency between the modes identified with di ff erent sensors. Table 2: Comparative analysis of the identified modal frequencies obtained through di ff erent monitoring systems and dynamic identification meth ods.

Mode

Piezoelectric sensors

MEMS

FBG sensors

EFDD( Hz )

SSI-CoV / IHCA( Hz )

EFDD( Hz )

SSI-CoV / IHCA( Hz )

EFDD( Hz )

SSI-CoV / IHCA( Hz )

Bending (1st)

4.37 9.27

4.38 9.29

4.37 9.35

4.38 9.33

4.38 9.31

4.38 9.32

Torsional

Local bending Bending (2nd)

15.70 30.27

15.79 30.27

16.15

15.99

16.06

15.83

–

–

–

–

Table 3: Relative di ff erence of identified modal frequencies for MEMS and FBG sensors compared to piezoelectric accelerometers.

Mode

MEMS

FBG sensors

EFDD(%)

SSI-CoV / IHCA(%)

EFDD(%)

SSI-CoV / IHCA(%)

Bending (1st)

0.00 0.86 2.87

0.00 0.43 1.26

0.23 0.43 2.29

0.00 0.32 0.25

Torsional

Local bending

5. Conclusions

The paper presents the results of the dynamic identification of the ”Ponte delle Grazie”, a reinforced concrete Gerber-type bridge located in Northern Italy. The results refer to preliminary tests carried out with the aim of char-