PSI - Issue 78

Gennaro Magliulo et al. / Procedia Structural Integrity 78 (2026) 1847–1854

1849

(such as technical rooms or radiology departments). Beyond the potential for direct damage, their failure could cause dangerous obstructions to escape routes or damage other attached elements (such as cables, pipes, or shelving).

(a)

(b)

(c)

Fig. 1. Anomalies and vulnerabilities of non-structural elements observed during field inspections.

Following this selection, two main categories of elements underwent dynamic on-site testing: piping systems and partition walls. Tests were conducted on representative samples for each category: for piping systems, a fire suppression system (Fig. 2a) and a hot water distribution network (Fig. 2b) were tested; for partition walls, two masonry configurations were examined, with thicknesses of 15 cm and 37 cm respectively. The dynamic identification tests provided data useful for defining the dynamic properties of these components, forming the basis for subsequent modeling and analysis phases.

(a) (b) Fig. 2. (a) Fire-fighting system and (b) heat water piping system tested.

2.2. In-situ dynamic identification tests To acquire information on the dynamic properties of the selected non-structural elements, an experimental campaign was conducted based on modal identification tests using ambient vibrations (AVT – Ambient Vibration Testing). This non-invasive technique allowed the dynamic response of the elements to be analyzed directly under their operating conditions, without interrupting the functioning of the systems. The experimental procedure involved three main phases: recording acceleration signals using single-axis accelerometers installed at key points of the analyzed components during their normal operation; processing the recorded signals through frequency analysis (FFT) to obtain power spectral density (PSD) spectra; applying Operational Modal Analysis (OMA) methods, including Frequency Domain Decomposition (FDD) for frequency domain analysis and stochastic subspace techniques for time-domain analysis, aimed at identifying the natural