PSI - Issue 66

Domentico Ammendolea et al. / Procedia Structural Integrity 66 (2024) 350–361 Author name / Structural Integrity Procedia 00 (2025) 000–000

356

7

The process begins by introducing input data, including the geometry and material properties of the masonry being studied, as well as parameters related to reinforced concrete (RC), such as the homogenized constitutive tensor C and the first failure surface of the RC masonry. Next, the model carries out two sets of discretizations for the computational domain. The first set involves discretizing the domain with a finite number of macro-elements, having the same shape and size as the RC and arranged in a rectangular array. This represents the masonry in a coarse-scale manner, similar to a macro-modeling approach. On the other hand, the second set involves a fine-scale representation of the domain, achieved through the discretization of many finite elements, in line with a pure micromechanical modeling approach. After both discretizations are performed, the code superimposes coarse- and fine-scale meshes and activates all the macro-elements. It then imposes the boundary conditions and initiates the analysis. To conduct the failure analysis of a masonry structure, the proposed method utilizes a displacement-controlled numerical analysis, in which the external loads are adjusted based on the displacements of a monitored control point that consistently increases during the numerical simulation.

Fig. 2. A schematic of the relevant steps involved in the proposed adaptive concurrent multiscale model: (a) initial step; (b) intermediate step – refinement of the computational domain; (c) collapse.

In the initial stages of the analysis, the masonry behaves in a linear elastic manner. The macroscopic modeling approach, using macro elements, effectively replicates the mechanical behavior of the masonry (Fig. 2-a). The code monitors each macro-element at each step to assess the response within the elastic domain. When a macro-element reaches the elastic limits, the code deactivates it and activates the surrounding micromechanical region (Fig. 2-b). Once a macro-element is deactivated, the computational domain consists of both macro- and micro-scale discretization. The code imposes constraint conditions on the hanging node along the micro-to-macro interfaces to ensure continuity in the displacement field. The analysis then continues, and activation conditions are assessed for each step. If the conditions are satisfied, macroelements are progressively substituted by micromechanical regions. Therefore, the computational model tends toward the micromechanical model as the analysis progresses, up to the whole collapse of the structure under investigation (Fig. 2-c). 4 Results This section reports numerical results devoted to assessing the accuracy and computational efficiency of the proposed adaptive concurrent multiscale model for the failure analysis of masonry structures. Particularly, the proposed strategy is applied to the failure analysis of a masonry beam under a three-point bending test (Fig. 3-a). The beam consists of brick units of length L b = 230 mm, height H b = 76 mm, and thickness t b = 110 mm bonded with vertical and horizontal mortar joints of t m = 10 mm.