PSI - Issue 68

Ivan Senegaglia et al. / Procedia Structural Integrity 68 (2025) 610–618 Ivan Senegaglia at al. / Structural Integrity Procedia 00 (2025) 000–000

616

7

increase the homogenized stiffness, causing an increase in the reaction forces with imposed displacements. AH XYZ model should describe better the interface regions, but their behavior in the experimental acquisition is masked by the non-conformal contact, resulting in a poor correlation. Otherwise, the AH Y model can describe the plasticity dominated part of the curve and the elastic cycles. In a specimen characterized by plasticity dominance, the overall compliance is significantly influenced by the softer layers of cells, similarly to a stacked spring system with varying stiffness. Specifically, the central cells underwent more pronounced plastic softening, thereby impacting the deformation of the entire specimen.

Figure 6: Experimental results versus the different FE models used. The experimental curve was divided in two parts, the first one is dominated by the contact between the interface and the specimen (as shown in the macroscopic deformation analysis). The second one is dominated by plastic deformation. The two FE models identify the region where the experimental data were expected to be due to the opposite transversal deformative behavior. The summary of the elastic cycling analysis is shown in Figure 7, where a representative stress-strain cycle was reported per each cycling loading carried out after reaching the phase-one load steps reported in Table 3. Despite significant plastic deformation leading to large permanent strain (up to 10%), the specimen's E* shows low variability, ranging from 15.08 to 16.38 GPa. Notably, only the first cycle exhibits substantial scatter, attributed to contact effects at lower loads. The ability to maintain a stable elastic response, even with major shape changes, can be explained by comparing two FE simulations that model the extremes of the specimen's mechanical behavior. In more detail, the AH Y model shows a discrepancy in the last cycle, underestimating the E* observed in the experiment. Even worse correlation was retrieved by the AH XYZ model, where the magnitude was overestimated and an increase in the E* can be observed in the higher-strain cycles. By combining the effects of the over-constrained and compliant models, the curve could potentially be represented, though this must be further investigated.