PSI - Issue 28

M.Z. Sadeghi et al. / Procedia Structural Integrity 28 (2020) 1590–1600 M.Z. Sadeghi et al./ Structural Integrity Procedia 00 (2020) 000–000

1597

The recorded longitudinal strains by the strain gauges for the two representative specimens show interesting results (Fig. 8). Firstly, for any of the specimens the strain on the front and back side shows a big difference. Apart from the geometric tolerances (in particular strain misalignment which will be later discussed), such a difference shows the unsymmetrical deformation of the joint (and consequently the damage initiation and propagation in the boded area) in tensile test. This can be attributed to different factors such as existing extra bending/torsion moments during the test (Solana et al., 2007) or void on the bonded area. In such circumstances, the strain variation on the both sides of the joint on the bonded area cannot be identical. On one side of the bonded area, the crack starts to initiate and with the propagation of the crack, the strain increases. However, the initiation of the crack on the other side for the reasons discussed above starts with delay and hence the reason the recorded longitudinal strain shows lower values with respect to other adherent. The variation of the longitudinal strain versus normalized force is shown for the experimental and the developed FE models. Comparing to the strain-displacement curve shown in Fig. 8, strain variation versus normalized force provides more accurate comparison between the FE and experimental results. As a whole, Fig. 9 provides similar trend with Fig. 8. To consider the effect of geometric tolerances as well as temperature change and strain gauge misalignment, especial approach was taken based on design of experiment (DOE) to combine the interaction effect of different parameters on the recorded measured strains.

Fig. 9. Variation of experimental longitudinal strain versus normalaized force compared with FE.

The motivation is to investigate the influence of geometric tolerances, strain gauge misalignment and temperature is on strain-based structural damage indicator (SDI) of the joint. In fact, by considering the possible joint tolerances existing in the assembly process as well as other influential factors, one can define upper and lower band for strain measurement on a location which previously showed high potential for damage detection (the following relation): ��� � �� � �� � in which t, O, T are the adhesive layer thickness and overlap length tolerances and temperature variation. Apart from geometric tolerances, the strain gauge misalignment from the so-called ZSP measured by FE model is a possible scenario to be happened while mounting the strain gauge on the adherents. In practice, the misalignment of strain gauge might happen in two different stages. The first stage is when the location is marked on the adherent and second is while mounting the strain gauge on the marked location. For either of these circumstances, the misalignment of 0.25 mm is likely possible. Precise scrutiny of the adhesive thickness over the bonded area showed the tolerance of ± 0.006 mm (forming an angle from one side with the adhesive layer thickness of 0.694 on one end (close to the clamped side)