PSI - Issue 44

Ylenia Di Lallo et al. / Procedia Structural Integrity 44 (2023) 488–495 Y. Di Lallo et al. / Structural Integrity Procedia 00 (2022) 000–000

492

5

set-up and loading protocol, was numerically reproduced through the MUDis approach, thus extending for the first time the proposed procedure to a full-scale masonry wall with openings. 3.3. Modelling strategy and software implementation The numerical model was implemented in the non-linear software MIDAS-FEA (Midas FEA User Manual 2016), which allows considering the “combined cracking-shearing-crushing” model for the interface elements. As for the elastic units, six-node triangular shell elements with three Gaussian integration points were employed; concerning the interfaces, two-node/zero-thickness elements with general formulation were adopted.

(a) (b) Fig. 2. Geometry of the considered benchmark case study: (a) plan configuration (b) front view of wall D. Adapted from Magenes et al. (1995) To replicate the behaviour of the analysed wall as best as possible, some modelling assumptions were necessary. For instance, to model the interaction between adjacent modules, during the assembly process the interface elements were placed not only among the four units composing the modules, but also along the outer boundaries of each elastic module. In addition, to avoid vertical sliding between modules, a rigid constraint was imposed in the vertical direction along these boundaries. Subsequently, the kinematic constraints were defined: the structure was fixed to the ground through a rigid beam placed at the base in order to allow the correct deformation of the structure under lateral loads. A rigid beam element was also inserted at the top of the structure to ensure a correct distribution of the vertical loads, thus avoiding the generation of stress concentrations that would lead to numerical problems. When analysing a structure with openings, attention must be paid to the adequate modelling of the opening geometry to catch their possible influence on the overall response. To this end, lintels were introduced in the model to replicate the details of the real structure. The modelling of these elements was implicitly included by assigning a suitable value of the elastic modulus, without any interface surfaces, thus simulating a continuous and sufficiently clamped element. The choice concerning the size of the modules was also crucial. For the present case study, a set of modules with a side length of 250 mm - a size consistent with the dimensions of the structure - was used. Yet, the Authors do not exclude the possibility of carrying out future sensitivity studies with modules of different sizes to prove whether or not the adoption of larger modules might lead to more efficient models without compromising the accuracy of the results. The numeric idealization of the façade with the MUDis strategy is shown in Fig. 3. Particularly, Fig. 3(a) illustrates the wall subdivision into modules of 250 mm edge as well as the position of the elastic regions featuring different Young’s moduli (upper and lower beams and lintels). Fig. 3(b) depicts the model implemented and discretized into the software.