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

133

8

at the overlap edges. However, at the middle of the adhesive layer, these stresses significantly decrease. This is due to the shear-lag phenomenon caused by the finite longitudinal stiffness of the adherends and the gradual loading within the adherends at the overlap. For the shorter L O , 12.5 and 25 mm,  xy stresses have a steady evolution along L O . However, for L O equal to 37.5 and 50 mm, these stresses show a small disruption at the middle of the adhesive layer, thus indicating that the layer has not yielded along its entire length. Using the EDP yield criteria, it was found that this region extends further along L O , leading to the conclusion that the adhesive layer did not yield over a larger value of L O . Fig. 6 reports normalized peel stresses (  xy /  avg ) at P m , equally averaging over  avg , for the two equated yield criteria. For both criteria,  y stresses at the middle region of the adhesive are either null or compressive (near the edges), while high magnitude peaks are formed at the overlap edges. This is due to the rotation of the joint, induced by the applied load eccentricity, which depends on the adhesive stiffness, causing separation at the overlap edges and compression at the interior of the overlap.  y stresses, in general, cause a decrease in the adhesive joint strength, eventually leading to a premature failure (Campilho et al. 2007). By comparing the obtained data with different yield criteria,  y stresses obtained by the EDP present higher peak values. In both criteria, it should be noted that, as L O increases,  y stress values at the middle of the adhesive layer tend to be zero. Comparing the peel stresses results with the effective plastic strain values, it is concluded that the peel stresses tend to be null when the effective plastic strain values are null.

5

5

4

4

3

3

2

2

σ y / τ avg

σ y / τ avg

1

1

0

0

-1

-1

0

0.2

0.4

0.6

0.8

1

0

0.2

0.4

0.6

0.8

1

Normalised L O (mm/mm) 12.5 25.0 37.5 50.0

Normalised L O (mm/mm) 12.5 25.0 37.5 50.0

a)

b)

Fig. 6. Peel stresses in the adhesive layer at P m using the von Mises (a) and the Drucker-Prager yield criteria (b).

Through the analysis of the obtained results for both  xy and  y stresses, it is possible to conclude that the stresses obtained by the EDP yield criteria are lower than those obtained by the vM yield criteria, thus suggesting a higher joint capacity before reaching failure. This phenomenon occurs due to the introduction of the hydrostatic stress component in the EDP yield criteria, a component to which adhesives are subjected at the time of yield, and which is not taken into account in the vM yield criterion. Comparing the stresses obtained according to the two criteria, maximum  xy and  y stresses obtained by the EDP criterion are lower. This effect leads to the conclusion that the joint has a greater capacity to support load before reaching failure. As a result, it can be affirmed that this criterion represents better the behavior of the adhesive along its L O . 4.2. Strength prediction After performing the experimental tests, all failures were found to be cohesive in the adhesive layer, including a visible layer of adhesive in both bonding surfaces of the adherends, without any evidence of adherend plasticization. This failure mode confirms that the correct procedures were undertaken during the joints’ fabrication. Fig. 7 shows the evolution of experimental P m with L O for the SLJ with the adhesive Araldite ® 2015, together with the global yielding (GY) analytical prediction, due to its moderate ductility. A significant P m improvement was found in the experimental P m values with this adhesive, which is a typical behavior on account of its ability to sustain plastic strains, and to support loads after the adhesive layer’s edges attain the limiting stresses. However, since the ductility