PSI - Issue 47

J.B.S. Nóbrega et al. / Procedia Structural Integrity 47 (2023) 408–416 Nóbrega et al./ Structural Integrity Procedia 00 (2023) 000–000

411

4

other joint geometries, was reported by (Faneco et al., 2017). After the curing time, the excess of adhesive on each specimen was removed by mechanical means. Prior to testing, each specimen was measured, and the dimensions recorded; then, the flexible adherend was peeled around 30 mm by hand, which allow the placement of the specimen in the testing jig. The final dimension of the bond line was also measured and recorded. The four specimens were tested using a Shimadzu AG-1 Universal Testing Machine (UTM) with a 10 kN load cell. The tests were performed using a testing jig compliant with the ASTM D3167, as depicted in Figure 1. The testing speed employed was 125 mm/min (Pereira et al., 2022). All the specimens were tested until the two adherends were separated.

Figure 1. Experimental setup using the ASTM D3167 testing jig.

2.3. Numerical modelling The geometries of the testing jig (ASTM-D3167) and the adhesive joint were modelled within the ABAQUS® environment (Abaqus 6.17, Dassault Systèmes, R.I. USA). Looking at the testing jig (Figure 1), it can be observed that it rotates as the test progresses; therefore, numerical geometry must consider this rotation. Therefore, the centre of rotation and the relative position of the adhesive joint with respect to it was considered. The model started with the fulcrum point (centre of rotation) of the testing jig at the origin and then the trajectories of the rollers’ centres, depicted in dashed lines in Figure 2. Then, the centreline line of the flexible adherend must be aligned with the fulcrum, so after considering the thickness of this substrate and making it tangent to the left side roller, it was possible to determine the position of the two rollers. In addition, the left side roller supports both adherends and the adhesive layer while the right roller only supports the rigid adherend. Therefore, construction lines representing these distances were employed. As a result, the construction lines were used to draw the joint’s geometry following the dimensions stated in the standard (ASTM D3167), which were also employed for the specimens tested experimentally (Section 2.2). The final numerical geometry is represented in Figure 2 together with the testing jig for comparison. Taking advantage of the joint configuration, it was decided to use a two-dimensional model, allowing to reduce the computational cost. The use of two-dimensional models to analyse bonded joints has proven to be adequate (Adams and Peppiatt, 1973, Adams and Peppiatt, 1974). Furthermore, in this work, the model was considered as a plane strain case. The adhesive joint was modelled as a deformable body. The adherends were considered made of aluminium and were modelled as an elastic-plastic material (Table 1). The adhesive layer was modelled as a