PSI - Issue 79

Mansi Gupta et al. / Procedia Structural Integrity 79 (2026) 259–265

262

Then, production of aggregate starts from the largest size and placed in the domain. A checked is performed to ensure that the aggregate is placed within the domain, the placed aggregate does not overlap with existing aggregates, and the specific minimum boundary conditions are achieved. If any one of these checks fails, the aggregate is discarded and a new aggregate is placed. This loop repeats until the aggregates of same dimensions are placed within the domain boundary. Afterwards, the process is repeated for next smaller aggregate size. Further details for the take and place method are provided in (Kargari et al. (2023)). For the present study, an in-house python scripting has been done for the aggregate generation. Three different aggregate shapes namely: circular, elliptical and irregular polyhedrons as shown in Fig. 1 are used for generating aggregates, with volume fraction as 0.75 as specified in Zhu et al. (2024). For the volume fraction study, the irregular polyhedrons are used with percent packing of 50%, 605 and 75%, as represented in Fig. 2. Similarly, for the size effect study, irregular aggregates with 75% packing fraction are used for mesoscale modelling.

Fig. 1. Mesostructure geometry for (a) circular, (b) elliptical and (c) irregular polyhedron aggregates

Fig. 2. Polyhedron aggregates with volumetric content (a) 50%, (b) 60%, (c) 75% 2.4. Finite element modelling of mesostructure

The generated mesoscale geometry is imported in Abaqus for the subsequent FE modelling. The mesoscale material properties for bulk aggregate and mortar geometries are provided as per Table 2 and 3-noded triangular plane stress (CPS3) elements with mesh size of 1 mm are used (Zhu et al. (2024), Trawi ń ski et al. (2016)). Next, the input file is written as exported to Python environment for inserting zero-thickness cohesive elements. The insertion of zero-thickness cohesive elements in mesostructure is done in two stages. First, the meshed data from Abaqus input file is parsed to identify the common nodes in mortar and aggregate sets. The shared boundaries between mortar-mortar elements and aggregate-mortar elements are then identified and all the shared nodes are