PSI - Issue 44

L. Navas-Sánchez et al. / Procedia Structural Integrity 44 (2023) 418–425 L. Navas-Sánchez et al. / Structural Integrity Procedia 00 (2022) 000 – 000

420

3

can vary significantly if compared to the case of the bare structure (Crisafulli et al. 2000). For that reason, the infills were included to realize an enhanced numerical model, as described below. 2.3. Numerical model The structural model of the building (Fig.1) was recreated using the Software SAP2000. Columns and beams were modeled as 1D RC elements with the dimensions detailed in the layouts provided by the architects who conducted the reparation after the earthquake. Floors were modeled as rigid diaphragms composed by 2D finite elements. The compressive strength of the concrete of this building was considered equal to 17.5 MPa, as obtained by testing cylindrical specimens. The connections between the structural frames and the foundations (rigid footings) and basement walls were modeled as fixed restraints. As the model was built so as to reproduce the linear behavior, plasticity of structural elements was not introduced. Additionally, the slab of the stairs and the RC walls constituting the elevator structure were modeled.

a

b

c

Fig. 1. (a) 3D model of the case study: a RC residential building of Lorca (diagonal struts in red; stairs and wall of the elevator in blue). (b) First mode of vibration. (c) Location of the Tromino Device during the Ambient Vibration Tests carried out (Number of Floor in Roman Numbers). Moreover, in order to capture the global behaviour of the structure in the linear range, the infill panels, representing the NSE, were modelled by means of two equivalent diagonal struts . Previous research has shown that equivalent diagonal single-strut models provide reliable approximations of the global response of the structure, such as the natural frequencies of a building or the definition of the mode shapes (Asteris et al. 2013 and Bovo et al. 2020). Only the infills that directly affect the global behaviour were modelled, i.e., those located within a RC frame or connected to the main structure. Conversely, the contribution of interior or exterior infills which were not directly connected to any frame element (beams or columns) were neglected. Nevertheless, the masses of all the infill masonry walls were included in the model. The diagonal struts were modelled only with axial stiffness and with the following characteristics: an elastic modulus E equal to that of the masonry itself, assumed as 2995 MPa (Bovo et al. 2020), panel thickness t equal to the actual thickness of the infill divided by two as there are two struts, and an equivalent strut width w . Then, the cross sectional area of the equivalent strut is wt . Several studies to evaluate the strut width w can be found, as reported in Bovo et al. (2020). In this case, the initial tentative model considered an optimal width equal to 0.24d , according to Bovo et al. (2020), where d is the diagonal length of the equivalent axial strut. The exterior and interior infills were differentiated by their dissimilar thickness, 18 cm and 7cm, respectively. Next, a model updating process was performed, that is, the parameters of the finite element model were adjusted in order to obtain the predictions of the model - in terms of eigenvalues - in agreement with the measurements obtained by modal testing. Specifically, parametric analyses on the value of the infills w/d ratio were performed in order to reach the natural period obtained in the experimental campaign. The frequency of the numerical model was equal to that of the experimental analyses for a w/d ratio of 0.30 . With this value, the main period of the analytical model was