Process optimisation and mechanical properties of friction stir lap welds of 7075-T6 stringers on 2024-T3 skin
L. Dubourga, A. Meratib, M. Jahazic
aInstitut Maupertuit, Assembly department, Campus Ker Lann, Bruz, France
bStructure and Materials Performance Lab, Institute for Aerospace Research, National Research Council, Canada
cAerospace Manufacturing Technology Centre, Institute for Aerospace Research, National Research Council, Canada
This paper focuses on the results of process optimisation and mechanical tests that were used to ascertain the feasibility of using friction stir welding (FSW) to join stringers to skin. The effects of process parameters on weld quality of 1.5-mm 7075-T6 stringers lap-joined on 2.3-mm 2024-T3 skins were investigated.
Advancing and retreating side locations on the joint configuration were alternated to determine optimal design arrangements. The effects of travel and rotation speeds on weld quality and defect generation were also investigated. Weld quality was assessed by optical microscopy and bending tests. It was found that: (i) the increase of the welding speed or the decrease of the rotational speed resulted in a reduction of the hooking size and top plate thinning but did not eliminated them, (ii) double pass welds by overlapping the advancing sides improved significantly the weld quality by overriding the hooking defect, and (iii) change of the rotational direction for a counter clockwise with a left-threaded probe eliminated the top sheet thinning defect. Subsequently, FSW lap joints were produced using optimum conditions and underwent extensive mechanical testing program. Several assembly configurations including discontinuous and continuous welds as well as single and double pass welds were produced.
The results obtained for cyclic fatigue performance of FSW panels are compared with riveted lap joints of identical geometry. S–N curves, bending behaviour, failure locations and defect characterisation are also discussed. It was found that: (i) the tensile strength of FSW joints approached that of the base material but with a significant reduction in the fatigue life, (ii) the probe plunge and removal locations served as the key crack nucleation sites in specimens with discontinuous welds, and (iii) double pass welds with overlapping advancing sides showed outstanding fatigue life and very good tensile properties. The present work provided some valuable insight into both the fabrication and application of FSW on stringer/skin lap joints.
Key-words: friction stir welding, welding process, lap weld, tensile test, fatigue resistance
1 – Introduction
Aircraft structural elements are typically fabricated from high strength 2xxx and 7xxx series aluminium alloys. These alloys are often considered very difficult to weld using conventional welding techniques [1,2]. This difficulty is one of the factors that have made riveting the most popular joining technique for aircraft structural elements. FSW technology is seen by designers as an alternative to riveting that has proven to be effective in welded joints of 2024-T3 [3,4], 7075-T6 [3,5] and joints between 2024-T3 and 7075-T6 alloys [1,6]. Aircraft fuselage panels are generally composed of 7xxx stringers and frames riveted on a 2xxx skin. For manufacturing of aircraft panels [7,8], spaceflight vehicle components [9] and aerospace beam assemblies [10], FSW could be the technology of choice as it can lead to lighter and more aerodynamic structures at lower manufacturing costs. However, skin to stringer joints pose a significant problem for FSW because of the defects associated with manufacturing FSW lap welds, such as, top plate thinning, voids, hooking and kissing bonds [11,12]. Thus, reliable design of FSWed joints requires a quantification of the influence of process parameters on weld defect generation and their impact on mechanical properties of stringer/skin joints. The influences of rotational speed [6], tool size [7] and welding speed [15] on weld quality, joint size, microstructure, ultimate tensile strength (UTS) and fatigue resistance have been reported in the literature.
Extended studies have been performed on the impact of tool shape, such as a double-shouldered tool [13], a skew shape [14–16] or 3-flat machined pin [17]. Moreover, some authors have been studied the influence of welding sequence on lap weld quality, such as a reversal stir motion [18], or using a partially [19] or totally double pass sequence welding sequence [6,20–25]. However, among these only few are related to the impact of process parameters on lap weld properties.
In the first part of the present manuscript, the results on the influence of process parameters and weld configurations on weld quality of 1.5-mm 7075-T6 stringers lap-joined on 2.3-mm 2024- T3 skins will be reported. The location of the advancing and retreating side relative to the joint configuration was alternated to determine optimal design arrangement. A satisfactory set of processing parameters to obtain sound lap joints was developed and validated through optical microscopy and fractography. In the second part of the investigation, FSW lap joints produced using the determined optimum conditions underwent an extensive mechanical testing. The results were compared with riveted lap joint of identical geometry mainly in terms of cyclic strength performance.
2 – Experimental set up
FSW trials were performed on a MTS I-STIR machine. Position control welding mode was used during the process optimisation (forge depth of 0.1 mm), while force control mode was used to produce samples for mechanical characterisation (forge force of 19 kN). The force control mode resulted in more repeatable weld quality than position control welding mode. Weldments were produced with FSW tools made of H13 tool steel. The tool was composed of a cylindrical left hand threaded probe of 6.3 mm diameter and 2 mm pitch and a scroll shoulder of 19 mm diameter. 450 50 mm2 7075-T6 stringers were joined to 450 190 mm2 2024-T3 skin sheets of 2.3 mm thickness in lap configuration (Fig. 1a). For process optimisation, the influences of welding speed (50–1000 mm min1), rotational speed (500–2000 rpm) and FSW tool rotation direction on defect generation and mechanical properties were investigated. The feed ratio expressed in mm per rotation is also introduced as a critical process parameter that allows predicting the quality of the joint. The magnitude of the feed ration parameter corresponds to the ratio of the welding speed to the rotational speed. For mechanical characterisation, the following welding parameters were selected: welding speed of 700 mm min1, tool rotational speed of 700 rpm and counter clockwise tool direction. Using the above parameters, the following four weld configurations (Fig. 1b) were manufactured (total of 20 panels of 450 mm wide):
(i) Single pass continuous (SPC). Only one weld pass was carried out over the entire length of the panel.
(ii) Double pass continuous (DPC). The advancing side of the first weld was overlapped by the second pass using a transversal shift of 5 mm (rastering process). Consequently, the retreating sides of the two welds were placed outside the joint. Here, the double weld was performed along the entire length of the panel.
(iii) Double pass discontinuous (DPD). Four 100 mm long welds were made along the entire length of the panel (Fig. 1a). At the end of each 100 mm portion, the FSW tool was extracted from the coupon. This welding sequence allowed studying the influence of tool entry and exit on mechanical properties.
(iv) Riveting. Four rivets, 19 mm far from each other, were place along the 95-mm wide coupons.
Figure 1: (a) 1.5 mm 7075-T6 stringers lap-joined to 2.3 mm 2024-T3 skins, (b) the four different weld configurations : single pass continuous (SPC), double pass continuous (DPC), double pass discontinuous (DPD) and conventional riveted joints.
For microstructure observation, the cross section of the welds was polished to a mirror like finish (diamond paste with a grain size of 0.5 lm) and etched with Keller’s reagent. All hardness measurements were made using the Vickers hardness scale over a length of 20 mm with a sampling step of 1 mm and a load of 50 g applied during 15 s. For bending and tensile characterisation, 12.5-mm wide coupons perpendicular to the welding direction were machined out from the welded panel. Bending tests were manually conducted by bending the stringer as indicated in Fig. 2. This was done solely to determine the fracture location and no attempt was made to measure the force. For each welding condition, five sub size tensile coupons were prepared according to the ASTM E8 M-01 standard. Tensile loads were applied on the 2024-T3 skin as shown in Fig. 2. Cross sections of 12.5-mm width and 50-mm gauge length were used for all-welded samples. The tensile specimens were ground and polished to help the identification of the failure location. For fatigue analysis, 95-mm wide coupons perpendicular to the welding direction were machined out from the welded panel. Assessment of the fatigue behaviour of lap welds was made by applying statistical analysis to produce fatigue
strength data as represented in the S–N diagrams. The fatigue performance was evaluated using an MTS hydraulic testing machine.
Fatigue tests were conducted in load control, using a sinu-soidal waveform at 4 Hz and R = 0.1. Fatigue stress levels of 69, 83, 103 and 138 MPa were applied on the 2024-T3 skin as shown in Fig. 2. Maximum stress and the stress ratio and fatigue test frequency were 0.1 and 5 Hz, respectively. The failed coupons were examined using optical and scanning electron microscopy (SEM) to determine the fatigue crack behaviour and the influence of the weld microstructure on the nucleation sites. Micrographs of the weld regions were taken from each coupon prior to and after failure to determine which features weakened the coupons.
Figure 2 – Diagram of loads applied during mechanical testing
3 – Results and discussion
3.1 FSW process optimisation
The microstructure of the lap welds consisted of a nugget, a thermomechanically affected zone (TMAZ) and a heat affected zone (HAZ). The nugget region for both materials was composed of equiaxed grains that were approximately 2–7 lm in diameter. The transition region between the TMAZ and the nugget was much smaller on the advancing side than on the retreating side. The difficulties associated with stirring of the interface between the two materials were clearly visible on the root of the retreating side (RS). As shown in Fig. 3, three main defects could be observed in FSW lap joints of 7075-T6/2024-T3 sheets: kissing-bond, hooking and top sheet thinning. These defects have been already reported in the literature [11,12]. Kissing bond occurs when the interface between the two joining metals is insufficiently heated and stirred, which results in a remnant oxide layer. At the interface, there could be little or no bond, raising the local stress in the weld during mechanical solicitation. In the case of lap welding, the oxide disruption at the sheet interface is more difficult than for butt weldingdue to the orientation of the joint interface with respect to the FSW tool. Indeed, in the case of butt weld, the sheet interface is parallel to the tool probe, increasing the plastic strain applied to the oxide layer. By contrast, in the case of lap weld, the sheet interface is radial to the tool probe, reducing the plastic strain applied on the oxide layer thereby reducing the chances to fragment it during the FSW process. As shown in Fig. 3, this phenomenon resulted in a remnant oxide layer located at the RS. Hooking defect is formed when the interface between the two joining sheets is stirred up into the top sheet, which effectively reduces the cross sectional area of the top sheet. The hooking defect occurs in the TMAZ of the AS. Moreover, the form and the vertical direction of the notch can reduce the mechanical properties by acting as crack initiation site during transversal loading [12,26]. Finally, top sheet thinning is observed on both sides. As the hooking defect, this thinning is due to the pulling up of the interface into the top sheet thereby reducing its cross section [6,13].
Accéder à l’article scientifique complet
Vous venez de lire un extrait de ce document scientifique sur le soudage par friction-malaxage. Pour accéder à l’intégralité du document au format PDF, veuillez remplir le formulaire ci-dessous.
Section snippets
3.2 – Mechanical characterisation
4 – Conclusions
This research provided some valuable insight into both the fabrication and the application of FSW on stringer/skin lap joints. Important observations obtained from the research include: (i) the increase of the travel speed or the decrease of the rotational speed caused a reduction of the hooking size. However, this defect may not be fully eliminated by varying the travel and rotational speeds. (ii) Double pass welds by overlapping the advancing sides improved significantly the weld quality by overriding the hooking defect. (iii) Change of the rotational direction from clockwise to counter clockwise inhibited the top plate thinning defect. This has been confirmed by the lack of failure during bending test. (iv) FSW parameters leading to highest weld quality were selected to perform the mechanical characterisation: travel speed of 700 mm min-1, rotational speed of 700 rpm, counter clockwise tool direction. (v) Hardness distribution across nugget depends on the location of measurement and varies between hardness level of AA7075 and AA2024 base materials. (vi) The tensile strength of FSW joints approached that of the base material. The scatter in UTS was much lower for double pass welds than for single pass welds. (vii) Double pass continuous welds with overlapping advancing sides showed higher fatigue life than riveted structures and single pass welds. (viii) The FSW probe entry and exit holes served as the key crack nucleation sites in specimens with discontinuous welds. However, plugging the exit holes had a negligible effect on the fatigue life. This resulted in lower fatigue life of the discontinuous welds than continuous ones. (ix) Bending, tensile and fatigue tests showed the detrimental effects of the hook defect. This defect resulted in failures during bending test, lower UTS and lifetime higher scatter in UTS during tensile test.
Acknowledgments
The authors thank M. Guérin for welding tests, A. Gagnon and M. Gallant for metallography and mechanical testing. This work is part of an extensive multi-pronged study, 46NM-7RFSW, which includes structural design, process optimisation, mechanical properties, NDE and robotic processing. This presentation only reports part of the results of process optimisation and mechanical properties. The authors appreciate the financial support from “New Initiative Funding” of Institude for Aerospace Research, National Research Council Canada.
