PSI - Issue 53

Costanzo Bellini et al. / Procedia Structural Integrity 53 (2024) 227–235 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

230

4

Fig. 2. Lattice cored specimens for three-point bending tests.

The construction of a virtual geometrical model was the initial step in the production of the specimens, after that all of the geometry parameters had been established. The Materialise Magics software, which allowed for creating a lattice structure in a particular volume, was used for this purpose. Therefore, all that was required were the parameters of the parallelepiped that represented the core, the cell type, and the strut diameter. After the lattice had been created, the skins were generated on two other computer-aided design (CAD) files before being integrated into the lattice model. For this step, a 0.2 mm core penetration into the skin was considered to ensure a good skin-core interface. The 3D printer slicing software was used to export the finished CAD model after it was complete. The ARCAM A2X EBM system and the ARCAM Build Processing slicing program were used for this project. The machine was then set up for the manufacturing run, with the process parameters calibrated and the powder hoppers loaded. The manufacturing chamber air was pumped out, the electron beam was calibrated, and the chamber was preheated to 700 °C. The specimens were constructed in accordance with the conventional procedure of a powder bed additive manufacturing process as soon as the required temperature was met. After the procedure was complete, the chamber was cooled to room temperature and the samples were removed from the powder block and cleaned using sandblasting equipment and a pressurized air chamber. An ultrasonic bath was then employed for a deeper powder removing procedure. The manufactured specimens were subjected to the test in accordance with ASTM C393, which is the industry standard for assessing sandwich construction flexural behaviour. Actually, a sandwich structure can be deemed similar to the flat specimens taken into consideration in this work. Although 3D printing is frequently used for complex shape parts, for this mechanical behavior investigation, a flat shape was used. The test design was a standard three point bending flexure, where the specimen was supported by two points and loaded in the middle by a loading nose. The span length of 200 mm and the loading speed of 5 mm/min were both selected. Every sample was loaded until its failure. As concerns the numerical model, the entire FE model was composed of beam and shell elements to improve computational efficiency. The first one was used to discretize each of the cell trusses, while the second one was utilized to simulate the skins. Additionally, a particular interface method completely eliminated the requirement for adding multi-point constraints between the shell and beam elements in order to maintain structural continuity. Therefore, regardless of the topology of the structure to be studied, the skins and the lattice core were joined by interface nodes sharing. As visible in Fig. 3, only one-quarter of the specimen was simulated due to the use of the geometry and load symmetries; as a result, two planes of symmetry were employed to replicate the three-point bending test. These planes were parallel to the lateral faces of the specimens and perpendicular to its primary plane, which is the plane parallel to the skins. All nodal displacements orthogonal to the symmetry planes were prohibited in order to replicate the symmetry requirement, but in-plane displacements were permitted. Moreover, since the considered elements possessed 6 degrees of freedom at the nodes (three displacements and three rotations), it was also necessary to block out-of-plane rotations. All of these factors improved the results correctness, reduced any convergence issues, and, on the other side, decreased the computing expense. By preventing the displacement of the nodes located at 100 mm from the surface, that represented the centre of the specimen, in the Z direction, the support was reproduced in the numerical model. Instead, the loading was replicated by applying the displacement of the nodes that were positioned at the point where the YZ plane and the top surface converged. Fig. 4 shows the boundary conditions mentioned. The titanium alloy was numerically modelled as a multi-