PSI - Issue 72

P.D.A. da Silva et al. / Procedia Structural Integrity 72 (2025) 52–60

56

Mesh construction was based on 0.2 mm quadrilateral elements, which matches t a . Moreover, since the combination of material, boundary conditions, and joint geometry was chosen to provoke adhesive failure, a higher mesh refinement was applied to the adhesive layer, as shown in Fig. 4, for the joint with L O =20 mm.

Fig. 4. Mesh refinement on the adhesive layer.

3. Experimental results 3.1. CZM validation

Experimental results and numerical reference values obtained by Valente et al. (2019) are taken as reference. For the adhesives AV138, DP8005, and XNR6852 E-2, a comparison between experimental and reference P m with the CZM numerical P m of this work is presented in Fig. 5.

40

AV138

DP8005

XNR6852 E-2

30

26.92

28.39

24.00

14.43 19.93

19.90

20

P m [kN]

11.75 16.79

13.61

10

0

Experimental

Numerical reference

Current numerical

Fig. 5. Comparison between experimental and numerical P m values.

For the AV138 joints, the observed discrepancy was acceptable and the numerical P m was only 6% lower than the numerical reference, with a 16% difference compared to experimental data. The DP8005 produced satisfactory results, with a P - δ curve closely replicating the reference data, though some irregularities were noted. For P m , the numerical result was nearly identical to the numerical reference, differing by only 1%. However, the experimental P m was about 15% lower than the numerical values. This disparity may stem from experimental limitations in replicating the numerical conditions. Using the adhesive XNR6852 E-2, P m from the numerical model differed by 5% from the numerical reference, but it was 16% higher than the experimental reference, as shown. In view of these results, the CZM procedure is considered validated for design purposes, as it always returns an acceptable representation of the expected behavior. 3.2.1. Overlap length The base L O is 10 mm, to be studied and compared with L O =20 mm and 40 mm, for the three adhesives applied on this study.  y and  xy stresses are graphically plotted as a function of the normalized overlap ( x / L O ) only for the tubular joints bonded with the adhesive AV138 (Fig. 6), which has a representative behaviour of the tested adhesives.  y and  xy stresses were divided by  avg , which is the average  xy in the adhesive layer. As shown in Fig. 6 3.2. Parametric analysis