PSI - Issue 33

R.F.P. Resende et al. / Procedia Structural Integrity 33 (2021) 126–137 Resende et al. / Structural Integrity Procedia 00 (2019) 000 – 000

135

10

The complete set of numerical/meshless P -  curves is presented in Fig. 9 for both von Mises (a) and EDP (b) yield criteria. Adhesive yielding is clearly denoted in all P -  curves, disregarding L O and the considered yield criteria. However, the transmitted loads are slightly higher for the EDP data compared to von Mises, due to accounting for the hydrostatic component of the stress state (thus giving lower stresses for the same applied load (Dean et al. 2004)), and the consideration of different tensile and compressive strengths; hence, the adhesive layer requires a higher load to reach failure. Due to the moderate ductility of this adhesive, a strain-based failure criterion (the maximum shear strain criterion). Actuall y, due to the adhesives’ behavior, this type of criteria usually enables a good representation of the experimental behavior (Lee and Lee 1992, Chai 1993). This statement is confirmed by the improved approximation of the numerical approximations to the experimental data, as further described in this work, taking place due to the adhesive’s ability to withstand higher deformations before the failure criterion is met.

16

16

12

12

8

8

P [N]

P [N]

4

4

0

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0

0.1

0.2

0.3

0.4

0.5

0.6

 [mm]

 [mm] 12.5 25 37.5 50

a)

b)

12.5

25

37.5

50

Fig. 9. P -  curves as a function of L O considering the vM (a) and EDP criteria (b).

16

12

8

P m [kN]

4

0

0

12.5

25

37.5

50

L O [mm]

Experimental

vM EDP

Fig. 10. Experimental P m and vM and EDP prediction vs. L O plot.

Fig. 10 presents the experimental/numerical P m data comparison, considering both yield criteria. The smallest deviations were found for L O =25 mm (1.4% for the vM criterion and 2.0% for the EDP criterion). On the other hand, the differences tended to increase by offsetting from this intermediate L O . Actually, a gradual error rise was found by increasing L O up to 50 mm (error of 20.0% for the vM criterion and 17.7% for the EDP criterion), but the predictions were worst for L O =12.5 mm (16.1% for the vM criterion and 41.2% for the EDP criterion). Comparing the vM and EDP criteria generally shows a relative difference to the experiments up to approximately 2%, except for L O =12.5 mm, in which case the difference ascended to 25.2%. Thus, generally both yield criteria present similar P m predictions and work reasonably well within the limitation of continuum mechanics-based failure criteria such as the one used in this work. Former works showed that the predictions tend to worsen as the adhesive’s ductility increases (Dean et al. 2004, Sánchez-Arce et al. 2021). Equally, the P m predictions offset from the real behavior for high L O , although in this work very small L O also led to discrepancies. Between yield criteria, the vM criterion is