PSI - Issue 44

Davide Arezzo et al. / Procedia Structural Integrity 44 (2023) 2098–2105 D. Arezzo at al./ Structural Integrity Procedia 00 (2022) 000 – 000

2102

5

M1 masonry consists of two unconnected outer layers of regular-textured bricks containing a rubble infill. This type of masonry is used to build the perimeter walls of the façade. M2 masonry is a multi-layered stone masonry with irregular courses, and is used to build the inner walls of the façade and few walls of the body of the church, especially those with a high thickness. M3 masonry is a double-layer masonry with rubble filling and, as M1, the connection between the different layers is almost absent. This typology is adopted for most of the walls and, consequently, presents various thicknesses. Finally, M4 masonry is composed of two layers: the inner layer consists of brickwork, while the outer layer consists of rough stone blocks with irregular courses. This type is used for the walls of the tiburium. Figure 5a shows volumes and location of each masonry typology within the body of the church. The collapsed parts of the church (rear part of the tiburium and lateral sides of the façade) were filled with steel lattice systems, indicated as the Braced-up regions in Figure 5a, which are considered as additional material to those previously described. The structure was modelled with solid elements (SOLID186 and 187) by means of the ANSYS software. Five different regions of homogeneous and isotropic material relating to each masonry type described in the previous section were modelled. Furthermore, the external steel system was included in the model because it can contribute to the whole structural dynamic response, being in contact with the church façade; beam elements are used to model both the steel trusses and the steel cables surrounding the entire church. Figure 5b shows the geometric model and the FEM. At this point, by defining an error function that takes into account the distance between the real dynamic behaviour and that of the numerical model (see Eq. 1), the model parameters were automatically calibrated using the PSO algorithm. Figure 6a shows the modal parameters of the calibrated model and Figure 6b the MAC matrix between the numerical and experimental modal shapes. The last step is the definition of the optimal position of the sensors of the monitoring system, i.e. the selection of the most informative degrees of freedom to monitor. ( ) = ⁡(1 + | ( )− | + (1 − , . ( )) where: = [ 1 ⁡ 2 ⁡ 3 ⁡ 4 ⁡ 5 ] (1)

Fig. 4. AVTs results: (a) identified modal parameters; (b) AutoMAC matrix.