On the Application of Laser Shock Peening as a Manufacturing and Repair Process to Improve the Fatigue Performance of Refill Friction Stir Spot-Welded AA2024-T3 Joints

The material used for all tests was AA2024 in the T3 heat treatment condition in the form of rolled sheets of 2 mm thickness. A specimen geometry for overlap joints was selected based on the specifications of DIN EN ISO 14324. This standard provides procedures for the fatigue testing of spot welds. Within the tolerance specified in this standard, two sheets 2 mm thick, 190 mm long and 60 mm wide were joined together via refill FSSW. Fig. 1a shows the geometry of the welded specimens with the overlap area of the sheets 45 mm and the spot weld located in the center of the overlap area. In addition, 5 mm diameter holes were drilled at the ends of the fatigue specimens for clamping in the mechanical testing machine.

Figure 1.

a) Geometry of the welded overlap joint and b) schematic representation of the refill FSSW joint: BM – base material, SZ – stir zone, HAZ – heat-affected zone, TMAZ – thermo-mechanically-affected zone.

An RPS 200 welding machine from Harms & Wende GmbH was used for the refill FSSW process. The tool consisted of a clamping ring with an outer diameter of 17 mm, a shoulder with an outer diameter of 9 mm and a probe with a diameter of 6 mm. The welding parameters (welding time of 5.8 s, penetration depth of 2.35 mm and rotation speed of 1800 rpm) were determined in this study based on preliminary tests on the same material and the same specimen geometry (Brzostec et al., 2018). A sketch of the refill FSSW joint with the identification of the typical joint characteristics is shown in Fig. 2b.

Figure 2.

a) Schematic illustration of the application of the LSP treatment to the base material specimens and to the welded specimen with refill FSSW joint in the case of b) LSP treatment on one side and c) LSP treatment on both sides. The starting point of each LSP sequence was in the lower right-hand corner. The direction of the LSP sequence is parallel to the y-direction. The step-position of each LSP sequence is in the negative direction of the x-axis. All dimensions are in mm.

A 5 J Q-switched Nd:YAG laser with a wavelength of 1064 nm, a pulse duration of 20 ns (full width at half maximum) and a frequency of 10 Hz was used for the LSP experiments. A diffractive optical element was used to focus a square laser spot (1 mm × 1 mm) onto the specimen surface. During the LSP treatment, a laminar layer of water with a thickness between 1 mm and 2 mm flowed over the specimen surface. The LSP treatment was carried out without a protective film on the surface of the aluminum sheets and welds. For residual stress analysis, aluminum sheets of base material as well as overlap welded joints were LSP-treated. The dimensions of the peening area placed in the center of the sheets were 18 mm × 18 mm (Figure 2a). The direction of the peening sequences was orthogonal to the rolling direction of the material for both the base material specimens (Figure 2a) and the welded overlap joints (Figure 2b-c). The sequence steps were in the negative direction with respect to the x-axis. The LSP treatment of the fatigue test specimens was carried out as shown in Figures 2b-c, with the center of the LSP-treated area located in the center of the weld. Two types of LSP-treated specimens were produced for the fatigue tests, with the LSP treatment applied either on one side only (welding side, Fig. 2b) or on both sides (Fig. 2c). Each LSP shot was positioned with no overlap between the neighboring shots. In all cases, 2 layers of LSP treatment were applied and the energy of the laser beam was set at 3 J.

The residual stress analysis was carried out using the PRISM measurement system from Stresstech GmbH. PRISM uses electronic speckle pattern interferometry to record the deformation of the specimen surface caused by drilling. The residual stresses are then calculated using the integral method. This hole drilling measurement technique has been described in detail by Steinzig and Ponslet (2003). Holes of 2 mm diameter and 1 mm depth were incrementally drilled for all measurements performed. Each hole was drilled in 15 increments. The increments near the surface were smaller than those below, according to the expected residual stress gradient.

The uniaxial fatigue tests were performed at room temperature on a 10-kN servohydraulic testing machine. In all tests the force was applied in a sinusoidal shape and the stress ratio R was set to 0.1. The test frequency for all tests was approx. 20 Hz. Failure of the specimens with overlap joints was defined as complete separation of the sheets. The specimens were clamped in the testing machine with counter-support plates to prevent additional bending loading during the fatigue tests. The fatigue test results as a function of the maximum cyclic load F_max on the number of cycles to failure were analyzed using the Basquin equation (Schijve, 2001) to calculate the 10%, 50% and 90% survival probabilities and the Basquin fatigue strength at 2 × 10⁶ cycles.

The fatigue tests on pre-damaged specimens were intended to demonstrate whether the LSP treatment could extend their fatigue life. To this end, the specimens were first subjected to a fatigue test up to a predefined number of cycles. This number was based on the results of fatigue testing of as-welded specimens. This enabled the service life of an as-welded specimen to be estimated for a given load level. Different pre-damage levels were then selected for the two load levels based on this. These levels were set to between 50% and 85% of the average service life of tests on as-welded specimens. After pre-damaging the specimens, they were LSP-treated on both sides and subjected to fatigue testing. The load levels corresponded exactly to those used in the pre-damage tests. The pre-damaged and LSP-treated specimens were subjected to cyclic loading until failure, and the number of cycles to failure was calculated from 0.

The microstructure of the welded specimens was analyzed by optical microscopy. (Leica DMI5000 M) The overlapped weld was ground in several steps (SiC 500-SiC 2500) and polished (3μm and OPS) and etched with Keller's reagent. The surfaces of the fractured specimens were examined using a stereomicroscope (Leica M80) and a scanning electron microscope (SEM) (JEOL JSM-6490LV).

Preliminary studies have shown that LSP treatment at 5 J, 20 ns pulse duration and using a 1 mm × 1 mm spot size with 2 layers does not result in a significant increase in fatigue strength compared to LSP treatment at 3 J with the same parameters (Kashaev et al., 2023). This can be attributed to the fact that the dielectric breakdown was already achieved at a laser energy level of 5 J, which in turn limits the possible peak pressure (Fairand et al., 1974, Berthe et al., 1999). The peculiarity of dielectric breakdown is that it leads to the generation of plasma in the air outside the substrate surface, which absorbs the incoming laser pulse and limits the energy to generate a shock wave (Montross et al., 2002). It was therefore decided to set the laser energy for the experiments in this study at 3 J for the 1 mm × 1 mm spot size used. The focus was to investigate whether a significant increase in fatigue strength could be achieved by applying LSP on both sides of the welded specimens.

To illustrate the effect of the applied LSP parameters on the residual stress distribution, base material specimens of AA2024-T3 were peened in an area of 18 mm × 18 mm (Fig. 3a). The results of residual stress analysis using the speckle interferometry (hole drilling technique) are shown in Fig. 3f. Up to a depth of approx. 0.4 mm the residual stress profiles show a noticeable difference between the residual stresses obtained in the longitudinal (σ_xx) and the transverse (σ_yy) directions. The non-equibiaxiality in residual stress profiles could be attributed to the LSP treatment sequence and/or material texture, as typically observed for the LSP treatment of Alalloys (Toparli & Fitzpatrick, 2019, Kallien et al., 2019). The absolute value of maximum compressive residual stresses in the longitudinal direction is approximately 320 MPa, which refers to approximately 84% of the yield strength of the base material (approximately 380 MPa). According to Chupakhin et al. (2016), the absolute values of the residual stresses determined are overestimated by more than 10% once they exceed 80% of the yield strength of the material. Nevertheless, the hole-drilling technique provides useful qualitative results. Significant compressive residual stresses can be generated up to a depth of 1 mm.

Figure 3.

a) Sketch of specimen with positions for three drilled holes used for the residual stress analysis, b) weld side of specimen in as-welded condition with position of drilled hole, c) back side of specimen in as-welded condition with position for drilled hole, d) weld side of specimen welded and LSP-treated with position for drilled hole, e) back side of specimen welded and LSP-treated with position for drilled hole. Depth-resolved residual stress profiles obtained for f) BM and LSP-treated AA2024 sheet and g) AA2024 lap-joint in as-welded as well as welded and LSP-treated condition.

In the case of AA2024 lap joints, residual stresses were determined at the center of the weld for both as-welded and welded and peened specimens. The holes were placed in the center of the welds on the upper and lower parts (Fig. 3b-e) and the results are presented as the average of three measurements from three specimens for each depth value (Fig. 3g). It should be noted that in the case of the specimens investigated, the hole-drilling technique provides only a qualitative analysis of the residual stresses, as a significant gradient of residual stresses is expected not only in the depth direction, but also in longitudinal (x-direction) and transverse (y-direction) directions. This could be a reason for the high scatter of the values obtained. High gradients of residual stresses in all directions are typical for welded joints. The results for the as-welded specimens show significant residual tensile stresses below the critical value of 80% of the yield strength of the material. As expected, higher residual tensile stresses were found on the upper part of the welded specimen (weld side). For the specimens after LSP treatment, deep compressive residual stresses are preferentially present on both sides of the specimen deep below the surface, with their maximum at a depth of about 0.2 mm, with absolute values of about 350 MPa and about 300 MPa for the upper part (weld side) and lower part (back side) measurements, respectively. This qualitative result shows that unfavorable tensile residual stresses in the weld region, which are critical for fatigue, can be converted into more favorable compressive residual stresses via the application of LSP.

The fatigue test results for specimens with refill FSSW joints in the as-welded condition and after LSP treatment are shown in Fig. 4. The fatigue strength of the lap joint was significantly increased compared to the as-welded condition, both for the finite life region and for the fatigue limit, due to the compressive residual stresses generated by the LSP treatment in the weld area. In terms of Basquin fatigue strength at 2 × 10⁶ cycles, the LSP treatment resulted in an improvement by a factor of 1.51 and 2.82 for the one- and two-sided LSP-treated specimens, respectively (Fig. 4). Therefore, the application of LSP treatment to refill FSSW joints as an additional post fabrication step can significantly increase the fatigue strength from 15% of the ultimate lap shear strength (Brzostek et al., 2018) to levels of 23% and 42% for the one- and two-sided LSP-treated specimens, respectively. The application of the LSP treatment on one side already results in a comparable fatigue behavior to the classical riveted joint with a fatigue strength of 22% of the ultimate lap shear strength (Kashaev et al., 2015), and the application of the LSP treatment on both sides even results in a significantly improved performance, making the combination of refill FSSW with LSP an attractive technique to replace classical riveting in the aircraft industry.

Figure 4.

Fatigue test results for as-welded and welded and LSP-treated.

Two exemplary load levels of 1.5 kN and 3.5 kN were selected to investigate the use of LSP as a potential repair method and the results are presented in Fig. 5. At 1.5 kN, two specimens were tested which had been previously pre-damaged to 75% and 80% of their original life. Despite the pre-damage, the specimens were able to complete 10 million cycles after the application of LSP on both sides, thus reaching the infinite life criterion.

Figure 5.

Fatigue test results for pre-damaged specimens.

Three levels of pre-damage (51%, 75% and 83% of the average fatigue life of the as-welded specimen) were tested at a load level of 3.5 kN (see Fig. 5). For the lowest level of pre-damage (51%), there was a significant difference in the lifespans of the two specimens tested. Therefore, an additional specimen was tested at this level, bringing the total number of specimens tested to three. While two specimens achieved between 8.4 and 9.1 million cycles to failure, one specimen failed after approximately 8.56 × 10⁵ cycles. With a pre-damage level of 75% of the average life of as-welded specimens, the results of the two tests were approximately 2.0 and 3.2 million cycles, respectively. Compared to the lower pre-damage level of 51%, there was a significant reduction in the number of cycles performed after repair at the 75% pre-damage level. Final tests at the load level of 3.5 kN were carried out with 83% pre-damage. After LSP application, the specimens failed at approximately 2.0 and 2.5 million cycles, respectively. There was also a slight reduction in the number of cycles at this level compared to the lower pre-damage level of 75%.

The results described above are in good agreement with the study of the application of LSP for the retardation of surface fatigue cracks developed in a fastener hole of a riveted joint (Sikhamov et al., 2020). For a fastener hole with a fatigue crack, LSP significantly extends the fatigue life of specimens with an initial fatigue crack, but the effect of LSP depends on the crack length. The larger the crack length, the weaker the effect of the subsequent LSP treatment. Considering that the size of the fatigue crack developed in the weld is smaller in the case of lower pre-damage level for refill FSSW joints compared to the joints with higher pre-damage level, the positive effect of LSP on fatigue life repair is higher at lower pre-damage level. However, even at higher predamage levels of 75% and 80%, a fatigue life restoring effect of LSP was demonstrated in relation to the as-welded undamaged specimens, assuming that any internal fatigue crack that may have developed was completely placed within the compressive residual stress field generated by the LSP treatment.

To study the damage evolution during the fatigue tests, exemplary cross-sections (Fig. 6a) were taken from as-welded specimens tested at 3.5 kN, achieving the number of cycles of 6.5 × 10³ and 1.4 × 10⁴, see Figures 6b-c and 6d-f respectively. A crosssection of the welded and LSP-treated specimen on both sides that was tested at 3.5 kN was also taken for comparison (Fig. 6g-h). The LSP-treated specimen showed no visible surface damage after 1.0 × 10⁷ cycles. Figure 6b shows an overview of the entire crosssection without etching of the as-welded specimen subjected to 6.5 × 10³ cycles. On the surface of the upper sheet, notches can be seen at the outer ends of the weld. It can also be seen that two cracks have developed within the sheet. These are located at the outer edges of the weld (positions 1 and 2). In the specimen that has completed 1.4 × 10⁴ cycles, these cracks are clearly advanced (Fig. 6d). On the left side of both as-welded specimens, there is a transcrystalline crack propagation originating from the unbonded sheets outside the weld. The etched microsection in Figure 6c shows that the crack propagates between TMAZ and SZ. After 6.5 × 10³ cycles, it has not yet reached the surface of the upper sheet (Fig. 6b), although this crack is already visible on the upper surface after 6.5 × 10³ cycles (Fig. 6d).

Figure 6.

a) Schematic representation of the cross section of the overlap joint with the weld and positions 1 and 2 indicating the two crack origins in the cross section (marked with the red box). b) Cross-section of the as-welded specimen tested at 3.5 kN and subjected to 6.5 × 10³ cycles with cracks starting at positions 1 and 2, and c) the magnification of the crack at position 1. d) Cross-section of the as-welded specimen tested at 3.5 kN and subjected to 1.4 × 10⁴ cycles with cracks starting at positions 1 and 2, and e) the magnification of the crack at position 2 located near the interface between the upper and lower parts and f) the magnification of the position 2 near the outer surface of the lower part (crack tip). g) Cross section of the welded and LSP-treated specimen on both sides tested at 3.5 kN and subjected to 1.0 × 10⁷ cycles with cracks starting at positions 1 and 2, and h) the magnification of the crack at position 2.

The second crack was developed on the right side of the weld (position 2 in Figs. 6b and d). This is a crack that originates outside the weld at the fine grain/coarse grain interface (Fig. 6e). It propagates in an arc around the fine grain structure. This structure is formed because the sleeve has penetrated to this depth. The partial bonding zone is located at the boundary between the SZ and TMAZ. It can be identified by its oxide deposits. The structure in the partial bonding zone is somewhat coarser grained in comparison to the base material. There is also a crack path that extends towards the lower sheet. This is a transcrystalline crack. The orientation of the crack is about 45° to the normal force applied. Figure 6f shows the tip of this crack grown after 1.4 × 10⁴ cycles.

These two fatigue cracks are also present in the cross-section of the LSP-treated specimen treated on both sides (Figures 6g-h), which was also tested at 3.5 kN and showed no visible damage on surface after 10 million cycles (Figure 6g). The morphological characteristics of the crack initiation and propagation are comparable to the as-welded specimens, as can be seen from the example of the crack on the right side of the specimen in position 2 (compare Figs. 6d and 6g). The significantly smaller distance between the crack flanks around the weld can also be seen in the LSP-treated condition (Fig. 6h) compared to the specimen in the as-welded condition (Fig. 6e). Therefore, these results confirm the positive effect of compressive residual stresses induced by LSP on the significant retardation of fatigue crack propagation, resulting in a significantly higher number of cycles to failure. Since in this example there are signs of fatigue in the cross-section of the specimen after 10 million cycles, the definition of “run out” in the case of this weld is only to be understood as a test criterion that does not mean the absence of developed damage in the weld. Thus, the successfully demonstrated repair method using LSP in Section 3.3 will play an important role when it is not possible to guarantee the absence of developed damage in the welds during operation, and yet safe operation without failure must be ensured.

During the fatigue tests, various fracture patterns were identified that led to the failure of specimens. These were observed in both as-welded and LSP-treated specimens. Basically, the fatigue cracks leading to failure can be divided into two types. On the one hand, there are specimens in which a crack through the sheet metal led to failure (Fig. 7a). The location of the crack also allows for two subtypes. The crack can run through the upper or lower sheet. On the other hand, there are specimens that fail due to a crack in the welding spot, which leads to the sheets shearing off (Fig. 7b).

Figure 7.

Categorization of the mechanisms responsible for the failure of specimens in the fatigue test. a) Fatigue fracture failure “crack through the sheet” and b) fatigue fracture failure “shearing the weld”.

The results showed that the “crack through the sheet” failure mechanism can occur at lower maximum levels of cyclic loading (Fig. 7a). For as-welded specimens, this failure mechanism occurs at maximum loads of up to 2 kN. This “crack through the sheet” failure mechanism is associated with the highest load cycles to failure.

At higher maximum loads (more than 2 kN for the as-welded condition) with a lower number of load cycles, the “shearing the weld” fracture pattern occurs more frequently (Fig. 7b). Treatment with LSP shifts the boundary between the fracture forms, so the fatigue fracture “crack through the sheet” exists for load levels of less or equal than 3.5 kN and 5 kN for the specimens welded and LSP-treated on one side and both sides, respectively (Fig. 7a-b). At this point it must be pointed out that mixed forms also exist. For example, some specimens with the “shearing the weld” form also exhibit cracks in the sheet, but these were not the cause of the eventual failure. It was also observed, particularly in the case of LSP-treated specimens, that the “through the sheet” fracture pattern can also result in the weld spot being torn out.

An as-welded specimen tested at 1.5 kN was selected for the fracture surface analysis. This is a specimen in which the failure was caused by a fatigue crack propagated in the upper sheet. Fig. 8a-d shows the images of the corresponding SEM examination. Fig. 8a presents an overview of the fracture surface. The fatigue fracture took place between the TMAZ and SZ. Outside the welding spot, the fracture surface was created through the base material. The structure at the welding spot can be roughly divided into two areas. The first area marked with 1 is smooth, indicating little bonding capacity. The second area, which lies deeper in the sheet, has a much rougher appearance. This indicates a higher bonding strength. This area is shown enlarged in Fig. 8b). At the lower edge of the upper plate there are places where fatigue cracks were originated (marked with white arrows in Fig. 8b). Area C, see Fig. 8c, is in the center of the sheet and shows fatigue lines that indicate a propagating fatigue crack. There is also a secondary crack, which is marked. Fig. 8d shows the transition area between the bond-rich and bond-poor structure. A honeycomb-like surface can be seen directly below the low-bond structure. This indicates a transcrystalline-ductile residual fracture surface. It can be assumed that the crack propagates from the lower edge of the upper plate towards the surface of the upper plate. The crack then propagates into the base material outside the welding spot.

Figure 8.

Fracture surfaces of a)-d) as-welded specimen tested at 1.5 kN and subjected to 1.41 × 10⁶ loading cycles and e)-h) welded and LSP-treated specimen on both sides tested at 4.0 kN and subjected to 1.60 × 10⁶ loading cycles.

Figs. 8e-h show SEM images of a welded specimen treated on both sides with LSP and tested at 4.0 kN. A fatigue crack propagating through the top sheet caused the specimen to fail. Figure 8e shows an overview of the fracture surface. Based on the location of the fracture surface, it can be assumed that crack propagation occurred between the TMAZ and the SZ. Several crack initiation sites can be seen in the overlapping area of the upper and lower sheets, marked with white arrows. The structure is fine in the center of the fracture surface and becomes coarser towards the outer sides. A section of this fine structure is shown in Figure 8f. Fatigue fracture paths and partially broken fatigue lines, which can be considered as secondary cracks, can be seen. Figures 8g-h show the fracture surface on the respective outsides of the weld, which surfaces have been treated with LSP. The structure is partially deformed in these areas. Partial deformation can be caused by frictional contact between the crack flanks. The typical characteristics of a fatigue crack are not visible. However, the outer faces of the sheet outside the LSP treated area show signs of fatigue fracture.

The contact between the fracture surfaces and the resulting crack closure is the main retardation mechanism for fatigue crack growth due to the presence of compressive residual stresses introduced by the LSP treatment (Kashaev et al., 2017). Therefore, the crack closure effect due to the presence of high compressive residual stresses could be also the main mechanism for the significant retardation of fatigue crack growth in the case of the welded and LSP-treated specimens, resulting in a comparable fatigue life even at the higher load level of 4.0 kN (LSP-treated specimen on both sides) compared to the as-welded specimen tested at 1.5 kN.