PSI - Issue 33

R.V.F. Faria et al. / Procedia Structural Integrity 33 (2021) 673–684 Faria et al. / Structural Integrity Procedia 00 (2019) 000 – 000

679

7

Table 2. Parameters required for the XFEM models.

Parameter E [GPa] G [GPa]

Araldite ® 2015

Steel 204.3 78.6 347.5 36.4 347.5 200.0 36.4 100.0 110.0 230.0

1.85 0.70

max [MPa]

21.63 4.77 21.63 17.90 4.77 43.90 0.43 4.70

 0

max [%] 0 [MPa] 0 [MPa]

 0

t n t s

0 [%] 0 [%]

 n  s

G IC [N/mm] G IIC [N/mm]

4. Results 4.1. Failure modes

Experimentally, the specimens were visually checked to define the failure modes. In the purely welded joints, failure naturally took place by weld- nugget detachment, induced after adherends’ plastic deformation. On the other hand, in the bonded joints, damage initiated by cracking of the adhesive layer at x/L =0 (see Fig. 1), but much before P m was achieved. Subsequently, the crack continued to grow longitudinally in the specimens until reaching P m . Thus, fully cohesive failures in the adhesive layer were produced, being noticeable a layer of adhesive covering both separated adherends. In the hybrid joints, failure was essentially equal, although P m correspo nded to adherends’ separation by the weld-nugget. In these joints, crack initiated in the adhesive, which stopped when it reached the weld-nugget region. Following, P m was achieved the adherends separated in the nugget region. After P m was reached, the joint still supported low loads until failing, due to failure of the adhesive up to x / L =1 (see Fig. 1). Numerically, the XFEM technique was able to match the experimental observations for all joints, although the full response accuracy (including P m prediction) varied, depending on the chosen criteria (initiation and growth). Fig. 4 shows a software output variable (STATUSXFEM) for the bonded joint configuration, including two test instants. For clarity, a 10× factor was applied to the displacements. The STATUSXFEM variable is a discrete value giving the percentile damage amount in a given finite element, i.e., in the damage laws connecting the real and phantom nodes the element. STATUSXFEM=0 indicates absence of damage, 0

Fig. 4. STATUSXFEM plot at the P m instant (a) and for δ =0.3 mm (b) for the bonded joint.

4.2. Elastic peel stresses σ y stress distributions are evaluated in this section at the adhesive’s mid -thickness and at the lateral symmetry plane, in the course of the initial part of loading, more specifically in the 1 st increment of the analysis step, and for the same  for all joint types. The analysis showed that the other stress components in the adhesive are negligible, including τ xy , due to the pure peel load applied. Although minor variations on σ y stresses exist along t A and joint width, in this work stresses were taken at the middle of the adhesive, as regularly used in research works, due to