Post-impact fatigue behavior

Figure 8.10-a shows the S-N curves of post-impacted friction riveted joints. The impact energies 5 J and 20 J were selected once they induced BVID and VID respectively. Post-impact fatigue testing defines the sensitivity of the structure to impact damage growth [247]. With increasing impact energy a significant decrease in fatigue resistance was observed in the friction riveted joints.

This was to be expected, because the quasi-static strength of 20 J impacted joints was lower, as discussed in Section 8.3. Joints impacted with 20 J underwent higher peeling stresses, leading to the premature adhesive failure of the squeezed material, which no longer contributed to joint strength.

Moreover, the slope of the S-N curve for the lower impact energy was steeper than for the higher impact energy, indicating that fatigue performance was more dependent on the stress amplitude. As reported in the literature [220], during fatigue testing peel stresses are generated at the edges of single lap joints. Therefore, it is believed that the 5 J impacted joints withstood higher peel forces, owing to the contribution of adhesion forces between the composite parts established by the undamaged squeezed material. Moreover, the stiffness degradation rate over the fatigue life indicates the initiation and propagation of different types of damage, and would help understanding the post-impact fatigue behavior of the joints [71]. For this reason, the loss of joint stiffness was monitored for 5 J and 20 J impacted friction riveted joints compared with undamaged specimens, as shown in Figure 8.10-b. The joints were loaded with 66 % of their respective ULSF, because at such a level the undamaged friction riveted joints had already withstood the 10⁵ cycles that is used for certification purposes in the aircraft industry [225].

Figure 8.10 a) S-N curves and b) stiffness degradation of friction riveted joints impacted with 5 J and 20 J.

The stiffness degradation was evaluated for specimens loaded with 66 % of their ULSF, to values as follows:

ULSF0J (6.6 ± 0.4) kN, ULSF5J (6.0 ± 0.3) kN, ULSF20J (4.8 ± 0.3) kN.

As expected, the damage introduced in the composite by impact testing decreased the joint strength and impaired its fatigue resistance at low fatigue cycles. Moreover, the typical four-stage degradation curves of 0 J (undamaged) friction riveted joints became gradually increasing curves, almost a plateau with the impacted joints, leading to an unremarkable stiffness degradation towards final failure.

The joint stiffness degraded faster with 20 J impact energy in comparison to 5 J, which can be explained by the type of damage, its extent, and the effect of peel stresses induced by the impact testing. As discussed in the Section 8.2, the 20 J impacted joints presented shear-driven delamination, which extended to the metal-composite interface, along with failure of the squeezed material driven by peel stresses. According to Shahkhosravi et al. [250], delamination is a damage that propagates unstably and fast under fatigue testing. Therefore, it is believed that with low cycle fatigue the metal-composite interface failed entirely, compromising load transfer between the materials and hence inhibiting additional mechanisms that dissipate energy. The multiple forms of impact damage observed throughout the composite thickness in Figure 8.5 for 20 J may also work as stress concentration sites that trigger faster damage propagation during fatigue testing. On the other hand, for 5 J impacted joints in which no evidence of delamination and peeling defects were observed, it is expected that the squeezed material failed at low cycles, driven by shear stresses, followed by evolution of the matrix and fiber cracking into shear-driven delamination.

In addition, at high fatigue cycles, where a plateau-like curve was established for both impact energy scenarios, unvarying stiffness degradation might indicate the added contribution of the rivet to withstand fatigue cycles. As the squeezed material and composite were progressively and prematurely degraded at low fatigue cycles, the undamaged rivet would partially arrest the propagation of cracks in the composite and metal-composite interface. Therefore, one can assume

that the rivet plays a more important role in dissipating energy during cyclic testing in damaged joints compared to undamaged joints, where the triggering of other failure mechanisms in the composite may occur.

By inspection of the fracture analysis shown in Figure 8.11, it can be withdrawn that the final failure in BVID joints under cyclic loading was not driven by impact damage, despite it causing a reduction in their fatigue life (see Figure 8.10). As highlighted by an arrow in Figure 8.11-a, the composite underwent high plastic deformation, which enlarged the edges of the hole and assisted rivet removal from the composite part. This failure behavior is typical for undamaged friction riveted joints under quasi-static and cyclic loading, as described in Section 7.2.2. Additionally, hackles in the squeezed material confirms the plastic deformation of the squeezed material triggered by shear during fatigue testing (Figure 8.11-e) [234]. For the VID joints a more catastrophic failure of the joint was evidenced by the low plastic deformation bored by the edges of the composite hole (Figure 8.11-b) and intense cracking in the hole’s surface (Figure 8.11-d), mainly around the rivet tip, where the stress concentrations are believed to be higher under shear loading [251]. Figure 8.11-f highlights the tearing of elongated fibrils in the squeezed material, confirming the influence of impact-induced peel stresses, which led to premature failure of the squeezed material prior to cyclic loading.

Figure 8.11 Typical cross-section of failed friction riveted joint, impacted with a) 5 J, and b) 20 J; overview of composite hole from c) 5 J, and d) 20 J impacted joints; microstructure of the squeezed material from e) 5 J, and f) 20 J impacted joints, highlighting shear-induced hackles and tearing of fibrils, respectively.

The impacted specimens that survived one million cycles were further tested under quasi-static loading to assess their residual strength. According to the recommendations proposed for the damage tolerance analysis of composite aircraft structures, by EASA [69], it is necessary to ensure that such a structure is not exposed for an excessive period, when it has a residual strength that is less than its design limit for loads. Figure 8.12 illustrates the comparison between initial quasi-static and residual strength of the impacted joints after one million cycles. For unimpacted joints, lower and higher impact energies their residual strength was lower compared to their initial strength by 9 %, 2 %, and 4 % respectively. Although the fracture analysis showed a possible extension of the impact damage upon the fatigue cycles, their narrow variations in strength and standard deviation suggest that the effects of cyclic loading on the initiation of additional damage and its propagation to the BVID and VID were not critical to the mechanical integrity of the joint.

Figure 8.12 Residual strength compared with the joint strength after one million fatigue cycles of undamaged friction riveted joints and joints impacted with 5 J and 20 J.