PSI - Issue 72

H.G.E. da Silva et al. / Procedia Structural Integrity 72 (2025) 26–33

32

a)

b)

Fig. 7. 3PB tests results for Configuration 2: a) Tsai-Wu criterion results extracted in the skin and b)  xy stress retrieved on the sandwich core.

3.4. Strength prediction and analysis P -  curves of 3PB tests were numerically obtained for each configuration. The following figures compare the numerical and experimental curves to assess the predictive capability of the numerical models. The point at which the sandwich fails was chosen as the point at which the core shear stress equals the average value of the core compression stress, obtained experimentally. Fig. 8 (a) compares two experimental curves and the respective numerical curve for configuration 1. The point of the numerical curve with a displacement of 7.3 mm, corresponding to a load of 1.04 kN, was considered as the failure point. Thus, it can be considered that the numerical models replicate the experiments with accuracy (deviation to the average experimental P m of 16.5%). Stiffness evolution also shows consistency, although a slightly stiffer numerical P - d curve than the experiments. Fig. 8 (b) compares two experimental and the respective numerical curve for configuration 2. The failure point is attained at a displacement of 6.9 mm, which corresponds to a force applied to the loading punch of 1.05 kN. Consequently, it can be considered that the numerical models replicate the experiments with accuracy (deviation to the average experimental P m of 11.7%).

0 0.2 0.4 0.6 0.8 1 1.2 1.4

0 0.2 0.4 0.6 0.8 1 1.2 1.4

1.03938

1.04653

P [kN]

P [kN]

0

2

4

6

8

10

0

2

4

6

8

10

 [mm]

 [mm]

P1

P2

Numerical

P1

P2

Numerical

a)

b)

Fig. 8. Experimental and numerical P-  curves during 3PB tests for configurations 1 (a) and 2 (b).

Table 6 presents the numerical and experimental values of  and  f , and respective relative deviations for each configuration during 3PB tests. Analyzing the obtained results, one can notice that, while experimentally configuration 1 promotes higher  f stresses when compared to configuration 2 (18.4%), numerically the opposite is verified (although with a difference of 0.1%). The numerical results show that configurations 1 and 2 achieved equal values for τ . Experimentally, τ are 8.1% higher for configuration 2. Thus, it can be concluded that, if the layer at 0˚ in contact with the punch does not fail, the difference between the orientation of two intermediate layers of the laminate does not have a significant contribution to the overall laminate behavior. When comparing the numerical and experimental values, configuration 1 presented a smaller relative difference between the experimental and numerical  and  . Configuration 1 lead to a deviation between the analytical and numerical values of τ and σ of approximately 16.2 and 18.3%, respectively. In configuration 2, the relative deviations of τ and σ were 7.5 and 40.2%, respectively. When comparing the values of the Tsai-Wu criterion between configurations, (Fig. 6 (a) and Fig. 7 (a)) it is possible to observe that, in configuration 2, the laminate is closer to failure than in configuration 1. The stresses in the core are similar between the configurations (Fig. 6 (b) and Fig. 7 (b)).