PSI - Issue 44

Marialuigia Sangirardi et al. / Procedia Structural Integrity 44 (2023) 1602–1607 M. Sangirardi et al./ Structural Integrity Procedia 00 (2022) 000 – 000

1603

2

1. Introduction The safe management of the building stock requires effective structural health monitoring tools. The aging of existing structures, the increasing load demand, the occurrence of damage due to material deterioration and extreme events, such as earthquakes, and the need of complying with rising safety requirements make monitoring essential for periodic condition assessment and maintenance ranking. Amongst the various available methodologies, the analysis of vibration test data has been extensively used in the last decades, to derive indication on the modification of global structural behaviour. The reduction of the fundamental frequency, which is associated with damage development, is routinely calculated from acceleration (or velocity) time histories recorded in some points of the structure by means of accelerometers (or velocimeters). Advanced processing techniques allow identifying damage from modal shapes (Lofrano et al. 2020a) even with a reduced number of sensors, thanks to sensor location optimization (Lofrano et al. 2020a). These approaches are still the preferred option in most cases. Nonetheless, dynamic monitoring based on devices physically fixed to the structure have some inherent drawbacks, related to the cost for purchase, installation, and maintenance, to the need of cables for electricity supply and data transmission (there are wireless sensors, of course, as part of an even more expensive equipment), and to possible damage. As a result, data are often recorded at a limited number of locations, which, moreover, are to be selected a priori . In order to overcome these limitations, contactless methods based on the analysis of digital videos have been recently proposed (Wadhwa et al. 2013; Feng and Feng, 2018; Bhowmick et al. 2020; Silva et al. 2022). They have proven reliable for measuring displacement time histories and, from them, providing modal parameters, and are cheaper and faster than standard techniques. Structural health monitoring applications of computer-vision-based techniques under environmental excitations (such as wind or traffic), which generally entail small oscillations, often include motion magnification algorithms, representing a crucial pre-processing step when dealing with small amplitude vibrations, which amplify displacements within a frequency range to enhance accuracy (Yang et al. 2017; Fioriti et al. 2018; Fioriti et al. 2019, Civera et al., 2020). This paper describes the application of a computer-video-based methodology for the structural health monitoring of a reinforced concrete elevated water tank situated in Rome, Italy. The method makes use of principal component analysis to calculate the natural frequency of the structure and has already been validated through laboratory tests on small scale specimens (Sangirardi et al. 2021) and shake table tests on real scale masonry walls (Sangirardi et al. 2022). The frequencies of the elevated tank were estimated from digital videos recorded with commercial equipment under environmental noise and compared to those provided by an accelerometer placed on the tank, to discuss advantages and limitations of the proposed dynamic identification approach and identify future research challenges. 2. Case Study The structure analysed in this paper is an elevated water tank comprising four reinforced concrete columns, having 2.20 m  0.65 m cross section, with a brickwork outer layer, whose thickness was detected as 5 cm through georadar inspections, leading to an overall cross section of 2.30 m  0.75 m (Figure 1a). The columns are 16.30 m high and, on top, support a cylindrical tank with 8.30 m diameter and 9.60 m height. The estimated thickness of the side wall of the tank is 0.30 m and that of the base slab is 0.40 m (original design documents and field tests are unavailable). At the height of the tank base slab, the cross section of the columns reduces to 1.30 m  0.65 m, which then keeps constant up to the top of the structure, at an overall height of 27.90 m, where they support a circular roof provided with cross braces (Fig. 1b). A central column houses the piping system (Fig. 1c). The structure has a reinforced concrete foundation slab, partially emerging from the ground. A preliminary estimate of the natural frequency of the structure was performed with a finite element model, which provided a first fundamental frequency of between 4 Hz and 5 Hz, associated with a torsional mode, consistently with the seismic behaviour of elevated water tanks reported in the literature (Murty and Jain 1996; Dutta et al. 2001a;b; Soroushnia et al. 2011; Hirde et al. 2011), and a second frequency between 5 Hz and 6 Hz, associated with two translational modes, one for each of the two principal directions in the horizontal plane, parallel to the sides of the cross sections of the columns. named as x and y . Note that information on the mechanical behaviour of the materials were unavailable, so plausible values were assumed, also considering the possible variability within reasonable ranges. The details of these analyses are omitted here for the sake of brevity.