PSI - Issue 51

P.A.R. Ferreira et al. / Procedia Structural Integrity 51 (2023) 115–121 P.A.R. Ferreira et al. / Structural Integrity Procedia 00 (2022) 000–000

119

5

which was not affected. Increasing t P to 2.4 mm had a smaller effect, being below ±2% in most cases. P max was reduced by 2.27% on the L O =80 mm, being the maximum reduction observed with this R and t P combination. Similarly, increasing t P to 3.6 mm caused a small increase in P max , 2.62% for L O =70 mm was the maximum observed. Regarding the joints with R =3000 mm, the joint with t P =1.2 mm and L O =20 mm was the strongest, 6.52% than the one with R =1000 mm, the other differences remained within ±3%. For the t P =2.4 mm joint cases, the effect was similar to the previous R , but in this case, P m was reduced in all cases, L O =80 mm was the weakest with a loss of 3.1%. Increasing t P to 3.6 mm, the L O =10 mm was the most benefited by 3.87%. The differences between the R =2000 mm and R =3000 mm were also small, most of them between ±2.5%.

Fig. 3. Percentual differences in joint strength ( P max ) due to the curvature radius (R) for all the tested overlap lengths ( L O ). The first row considers R=1000 mm as a reference, the second compares both larger radii together.

From the above, the effect of R on P max was small (below 5% in most cases) regardless of the combinations of t P and L O . However, this effect can be attributed to the model dimensions, i.e., limiting L T to 200 mm (Fig. 1) together with the large radii evaluated resulting in almost equal arc lengths for the three R values. This also led to similar imposed forces on all the models. Therefore, increasing R had no significant effect on P max for this configuration of adhesive joints. The failure of the cohesive layers (adhesive or interlaminar) was evaluated through the damage variable (SDEG of ABAQUS®). For the tested joint configurations, regardless of the value of R , all the joints with t P =1.2 mm presented cohesive failure while the remaining had an interlaminar failure of the substrates; therefore, in this regard, the thinner joints behave in a more controlled way. Regarding the stress distributions, a stress concentration was observed in the vicinity of the knee (Fig. 4). Although the peak magnitudes seem similar regardless L O , the increase in the latter causes a slight reduction in the peak stress magnitude (~8%), as also observed on flat SLJ (Nunes et al., 2016). In addition, the presence of stress peaks in the vicinity of the knee is common in JLJ (Taib et al., 2006). Furthermore, the load is transferred more uniformly along the bond line in those joints with shorter L O , as shown in Fig. 4.