PSI - Issue 68

Ivan Senegaglia et al. / Procedia Structural Integrity 68 (2025) 610–618

615

6

Ivan Senegaglia at al. / Structural Integrity Procedia 00 (2025) 000–000

The thickness measurements of the TPMS lattice were conducted following the mechanical testing, enabling us to measure and replicate the actual dimensions of the final specimen in the numerical models of the RVE. Stereo image contouring was utilized to assess the effective relative density ρ r of the manufactured specimen. The cut sections, converted as binary images, were postprocessed in Python 3.12 environment to extract the vertical thicknesses, which are the least affected by the plastic deformation process. Hence, the medians of thickness distributions were calculated, and employed to determine the final ρ r , estimated at 0.38 (up from 0.25). This increase in lattice density is related to the cell’s walls’ average thickness of approximately 0.2 mm (from 0.34 mm to 0.56 mm) which is reasonable with the AM parameters used in the manufacturing process. The results are displayed in Figure 5f, where the thickness distributions show a higher density near the undeformed average value of 0.75 mm. The secondary density peak, particularly evident in b) plot, is related to the number of high connectivity regions in the slice, which value is between 1.5-1.7mm. The virtual slice thickness distribution has the secondary peak over 2 mm, in disagreement with the previous analysis. The different secondary peak is a consequence of the plastic deformation in the lattice walls intersection, which compression flattened all acute angles, according to the comparison between the deformation in the lower Y nodes versus the center Y ones. Finally, the corrected FE models were computed with the evaluated ρ r .

Figure 5: a), b), c) d); Different binary images of the contour of the slices; e) Reconstructed virtual slice with the slicing plane f) Violin plot of the distribution of vertical thickness in the referenced slices. Solid red circles show some examples of high connectivity regions, while dotted red circles highlight some of the lower connectivity regions. 3.2. Macroscopic tensile curve comparisons As shown in Figure 6, both FE models were unsuccessful in following the first part of the experimental curve. The interaction between the irregular interface surface of the lattice and the testing machine affects the whole specimen mechanical behavior, particularly in the first part. This phenomenon became increasingly less prevalent as plastic deformation in the specimen’s central part intensified, and the AH Y model accurately reproduces the specimen response in this last deformation regime. In more detail, a surrogate model based on finely detailed geometry can reproduce in detail the local features of a lattice cell, which are strongly correlated to the occurrence of plastic deformations. The differences between the two FE models are noticeable: the imposed constraints in the lateral faces