PSI - Issue 78

Antonio Sánchez López-Cuervo et al. / Procedia Structural Integrity 78 (2026) 1791–1798

1793

(b)

(c)

(a)

500

250

280

(d)

Accelerometer Strain Gauge

Fig. 1: (a) Frame scheme (dimensions in mm); (b) strain gauge; (c) accelerometer; (d) DC motor.

The study analyses the modal properties identified through OMA using di ff erent types of sensors. Specifically, the frame was instrumented with 17 SGs and 10 accelerometers (Fig. 1(b) and Fig. 1(c)), with data acquisition performed using two Arduino MEGA boards. The sensors and controller locations are shown in Fig. 1(a). Additionally, due to the low ambient vibration levels recorded in the laboratory, a 12V DC motor with an eccentric mass was installed at the top level of the frame to provide artificial excitation (Fig. 1(d)). The motor was controlled by an Arduino Nano microcontroller, which activated it at random time intervals with randomly varying rotational speeds.

2.2. Damage scenarios

The first part of this study analyses the e ff ect of damage on the modal properties of the structure as identified by both monitoring systems. To this end, the process began with an undamaged condition (referred to as D.S.0), followed by progressive damage (without restoring previous conditions) at various column-to-slab joints, resulting in a total of 11 damage scenarios (D.S.1 through D.S.11). For each one, an Ambient Vibration Test (AVT) was conducted. The damage introduced at each stage consisted of the removal of two bolts from a joint. Fig. 2(a) shows an undam aged joint, while Fig. 2(b) illustrates the damaged condition. In total, 12 columns were a ff ected, corresponding to 11 damage scenarios. The location of the damaged joints is depicted in Fig. 2. Subfigure (c) illustrates the progression of damage from D.S.1 to D.S.4; specifically, the frame configuration shown with four damaged columns corresponds to D.S.4. Simi larly, subfigure (d) displays D.S.5 to D.S.7, and subfigure (e) shows D.S.8 to D.S.11. For each vibration record, the modal properties were identified by OMA. Prior to system identification, basic signal preprocessing was applied, including detrending, removal of electrical spikes, and a Butterworth band-pass filter. The modal identification was performed using the COV-SSI algorithm implemented in the MOVA software (Garc´ıa–Mac´ıas and Ubertini, 2020). The results are presented in Section 3.1.

2.3. Multi-objective optimization

The second part of the study involved the updating of a FEM in SAP2000 based on the modal data obtained from the undamaged condition (D.S.0). Methodologically, a multi-objective optimization problem was formulated,