PSI - Issue 78

Marco Martino Rosso et al. / Procedia Structural Integrity 78 (2026) 301–308

307

fundamental frequency, for instance, that considering White Noise 2 and 3 is now about 2.4 Hz, about 31% higher than the Building 1 fundamental frequency of 1.83 Hz. Even in this second case, the progressive damage accumulation has been evidenced by the reduction of the natural frequencies among the various White Noise scenarios, with a total decrease of about 30% for first two founded fundamental frequencies of 2.37 Hz and 7.62 from White Noise 2 to White Noise 7, and even with a decrease of 36% for the natural frequency at 11.81 Hz. Even in this case, the damping ratios are in line with similar RC frame infilled structures, except for the third mode at White Noise 5, which delivered an uncommonly high value of 6.6% of the damping ratio. The coherence among modal results for this second building appeared to be noisier and more difficult to retrieve, especially for White Noise 2 and 3 where no reliable and fully stable poles alignments appeared except for the Table 4 reported values. Furthermore, despite the higher relative reduction in the natural frequency values in the retrofitted structure 2 against the un-retrofitted building 1, the final natural frequencies of White Noise 7 on building 2 are associated to the values equivalent to building 1 on White Noise 5, i.e., following an Irpinia earthquake scaled at 75% only. This demonstrated the effectiveness of the retrofitting system. Indeed, despite experiencing an Irpinia earthquake scaled at 125%, building 2 is still presenting a higher level of residual stiffness, whilst building 1 accumulated such damage to further reduce its first three natural frequencies by about 8% reduction on average. Experimental modal results in terms of natural frequencies are quite robust in general, and despite some aleatory factors and higher uncertainties typical of MEMS-based technologies, it was possible to provide a good agreement with the FEM reference solution and the previous studies. Damping ratios and mode shapes are, however, more sensitive to noise in data and therefore it is not always possible to deliver good estimates for them with traditional OMA approaches. Indeed, in this specific case, the effects of interpolation may cause error propagation that prevented providing a clear interpretation of obtained mode shapes. Furthermore, it is noteworthy that the sensor setup is affected by an oversimplification, since the last floor is not monitored by two tri-axial accelerometers but one of the sensor is in reality placed on a column. Thus, the oversimplification which can also affect the mode shape correct resolution may be related to this oversimplification of translating the resulting modal components and interpreting them as the last floor components. These latter factors therefore may contribute altogether, and they may be the main responsible for poorly identified modal parameter estimates, especially in the mode shape correct reconstruction.

Table 4. OMA experimental results for Building 2. White Noise 2 White Noise 3

White Noise 4 [Hz] [%] [Hz] [%] [Hz] [%] 2.37 2.41 2.37 1.36 2.19 2.21 - - - - 4.33 5.47 7.62 2.38 7.63 1.66 6.94 3.25 11.81 5.51 - - 10.53 3.36 [Hz] [%] [Hz] [%] [Hz] [%] 1.98 1.83 1.74 2.61 1.63 2.44 3.80 4.41 3.43 5.03 3.15 5.00 6.49 6.56 5.85 5.37 5.38 3.65 9.34 4.26 8.17 4.53 7.52 3.93 White Noise 5 White Noise 6 White Noise 7

4. Conclusions This study examines two laboratory-scale reinforced concrete (RC) buildings, one retrofitted and one not, excited by a sequence of different Irpinia waveform scaled time series and random white noise inputs on a shake table.