Joint design optimization

The improvement of joint strength was further assessed by the joint design, considering the rivet parameters of tightening torque (TT), outer diameter of external washer (Dw) and the geometry of the friction riveted joint, including edge distance (e) and specimen width (W). The optimized set of joining parameters (RS: 15000 rpm, FFI: 5 kN, FFII: 10 kN, DF: 7.5 mm, CP: 0.2 MPa) was used to join the single lap specimens, which were then tested under tension. To attach a washer and nut, threads were cut along the free length of the rivet. It is worth mentioning that all the joint designs tested in this section failed by the bearing of the composite and rivet pull-through, as is reported by Borba et al. [115] characteristic for direct friction riveted joints of CF-PEEK and Ti6Al4V.

Therefore, the aircraft requirement of a failure mode that is non-catastrophic [225] was fulfilled for all the joint geometries. The damage evolution prior to this failure mode will be described in detail in Section 7.1.2.

6.6.1 Effect of washer size

Figure 6.21-a shows the influence of washer size on the quasi-static mechanical behavior of friction riveted joints, with other variables fixed (tightening torque: 1 N·m; W/D ratio: 7; e/D ratio:

3.5). There was an increase of approximately 33 % in ULSF from untightened joints to joints assembled with a washer of any size. However, no significant change to ULSF was observed by

increasing the outer diameter of the washer. As reported in the literature [220], the washer distributes the pre-load of tightening up the joint, decreasing the stress concentration at the fastener surroundings and therefore increasing the composite’s bearing capability. Moreover, the washer promotes lateral constraints to the area underneath it and may affect the secondary bending of the joints [33], as shown in Figure 6.21-b. Although no improvement in ULSF was observed, larger washers seem to affect the eccentricity of friction riveted joints, leading to higher out-of-plane displacement. Additionally, larger washers may decrease the pressure applied by the pre-tightening torque, which could impair joint stiffness and consequently also contribute to higher out-of-plane displacement. For these reasons, a washer outer diameter of 2D (10 mm) was selected as optimum, which also led to a weight saving among the geometries evaluated in this work.

Figure 6.21 a) Ultimate lap shear force (ULSF) as a function of outer diameter of the washer (Dw); b) out-of-plane displacement of joints tightened using different sizes of washer, measured by digital image correlation

(DIC).

6.6.2 Effect of tightening torque

Figure 6.22 shows the effect of tightening torque on friction riveted joint strength for both untightened and tightened joints with three levels of torque, with other variables fixed (W/D ratio: 7;

e/D ratio: 3.5). As expected, by increasing the tightening torque, an improvement of approximately 30 % of joint strength was observed up to 1 N·m. Similar to bolted joints [33,189,220], changes in tightening torque promotes friction between the washer and the composite surface, which leads to an additional mechanism of load transfer during quasi-static and cyclic mechanical testing. For the untightened joints, the load is transferred mostly through the fastener in contact with the composite, while for tightened joints the load is partially transferred by friction [220]. Therefore, higher

tightening torque leads to higher friction force, which releases the stress concentration near the edge of the hole, which in turn produces stronger joints. However, after 1 N·m, a 20 % decrease of friction riveted joint strength was observed, which is explained by premature failure of the metal-composite interface at the rivet tip. Owing to excessive torsion transmitted to the rivet anchoring over the joint tightening, the adhesion between the materials was compromised, as shown in Figure 6.22-c, leading to premature debonding of the rivet from the composite hole. Therefore, a tightening torque of 1.5 N·m was considered the break loose torque, which should be avoided. For further analysis, 1 N·m was used as the optimized tightening torque for CF-PEEK/ Ti6Al4V friction riveted joints produced with the joining parameters established in this work.

Figure 6.22 a) Ultimate lap shear force (ULSF) as a function of tightening torque; X-ray µ-computed tomographs of joints tightened with b) 1 N·m, and c) 1.5 N·m, showing the damage of the composite-metal

interface when joints are tightened with a torque higher than the break loose torque.

6.6.3 Effect of joint width and edge distance

Altmeyer [116] assessed the influence of edge distance (e/D) and specimen width (W/D) on the strength and failure behavior of CF-PEEK/Ti gr.3 friction riveted joints. The author reported that joints with e = 3D and W > 4D presented higher strength and failed by progressive plastic deformation of the edges of the composite hole. Similar findings were reported by Cooper and Turvey [189] for pultruded glass fiber reinforced polyester joined with M10 steel bolts. Following these previous investigations, three levels of e and W were evaluated for CF-PEEK/Ti6Al4V friction riveted joints, including recommended levels of e = 3D and W > 4D. The other variables were kept constant (TT:

1 N·m; Dw/D: 2). Figure 6.23-a shows the average ULSF for all the joint designs. The ULSF varied between (5.3 ± 0.3) kN and (6.2 ± 0.2) kN, with no indication of dependence between joint geometry

and strength. To statistically analyze the significance of these parameters on ULSF, ANOVA was performed, and the main effect plot is shown in Figure 6.23-b. Although the increase of e and W showed a small effect on the joint strength, the flat slope of the tendency lines indicates a weak influence of the parameters on the response, and therefore they are statistically insignificant.

Additionally, in contrary to the literature [116,189], CF-PEEK/Ti6Al4V friction riveted joints underwent only bearing of the composite without any change on the failure behavior. It is reasonable to assume that the W/D and e/D investigated in this work were larger enough to compensate the compressive stress concentration at the edges of the composite hole, imposed by the rivet. Thus, catastrophic tensile failure as shear-out and cleavage, commonly observed for composite bolted joints [33,189], were avoided.

Figure 6.23 a) Effect of edge distance (e/D) and joint width (W/D) on the ultimate lap shear force (ULSF) of friction riveted joints; b) main effect plot of e/D and W/D on the ULSF, calculated from ANOVA.

As the levels of e and W investigated in this work did not influence the joint strength, and owing to better clamping of larger specimens in the clamping system (Figure 5.5-a), friction riveted joints were produced according to the recommendations of joint geometry in ASTM D5961 [176]

(see Section 5.2.1).