by W. Thomas, D. Nicholas, D. Staines, P J Tubby and M F Gittos
Published on the Internet June 17, 2004
Presented at the IIW Meeting on FSW in Nagoya 2004, Nagoya University, Japan, 9th July 2004
Friction stir welding (FSW) has gained increasing interest since its invention at TWI some 13 years ago. The continuing development of the process is described. The design of the tool is the key to the successful application of the FSW process and its ability to accommodate a range of aluminium alloys over a wider range of thickness. A key factor in tool technology is the ratio between the volume swept by the probe and the volume of the probe itself.
This paper will focus on fatigue performance of lap welds made with rotary Skew-stir TM and reversal Skew-stir TM techniques.
Friction Stir welding (FSW) was invented and patented  by TWI in 1991. Early in the development it was realised that the flow path of material around the probe was important in achieving sound welds with good mechanical properties and acceptable welding speeds. A key factor in achieving sufficient flow around the probe is having a greater volume of the probe swept during rotation (the dynamic volume) compared to the volume of the probe itself (the static volume). This ratio is affected by the probe geometry and the motion of the tool.
The lap-welded joint or combined lap/butt welded joint is often unavoidable in real components. The scope for the application of lap joints in aluminium alloys is potentially large. 
The level of confidence, however, for the manufacture of lap joints made by the FSW process does not match that for butt joints. The detailed morphology of the notch at the edge of the stirred zone either side of the weld is critical to joint mechanical performance. [3,4,5 & 6]
Thinning of the upper sheet, inadequate oxide disruption and remnant oxide layers, particularly on the retreating side of the weld, can seriously affect the static and dynamic properties of the lap weld. Similarly, adverse re-orientation of the joint interface (hooking) on the advancing side of the weld is cause for concern. Work at TWI has focussed on techniques that help to improve FSW technology such as Skew-stir TM
, Re-stir TM
, Com-stir TM
and Dual-rotation. [7,8 & 9]
Fatigue tests have been used to evaluate lap weld integrity for conventional Rotary FSW, Rotary Skew-stir TM
and Reversal Skew-stir TM [10 & 11]
and a number of fatigue studies have also been carried out on FSW butt welds which have helped to provide confidence in the technology. [12 & 13]
Skew-stir TM welds, made in 6mm thick 5083-H111 aluminium alloy, were fatigue tested using the following procedure:
Parallel-sided 50mm wide fatigue test specimens were cut from lap-welded panels (100mm away from the start and stop regions of 500mm long test specimens) in both configurations shown in Figure 1.
Fig.1. Lap weld joint configuration
a) Advancing side near the top sheet edge (ANE)
b) Retreating side near the top sheet edge (RNE)
In order to provide a comparative reference for the fatigue performance of the Skew-stir TM lap welds, an 'artificial lap' of similar geometry was made from parent material, see Figure 2. These 'artificial lap' test pieces were made from solid material. Wire Electro Discharge Machining cuts into the parent material were made to simulate, as near as possible, the notch geometry and effective width of a lap weld.
Fig.2. Artificial lap made from 5083-H111 aluminium alloy parent material
The nominal stress range for the tests was calculated using the applied load and the cross-sectional area of one of the laps, i.e. the width of the specimen multiplied by the thickness of one sheet. Prior to testing, the edges of each specimen were dressed round and smooth with a nominal radius of 1mm, in an attempt to avoid failure from the specimen edges. All the tests were performed at a stress ratio of R=0.5 to simulate the presence of high tensile residual stresses in a FSW component. The fatigue tests were performed in a 100kN Amsler Vibrophore at a test frequency of approximately 100Hz. Packing pieces were used in the machine grips to accommodate the lap offset and minimise secondary bending. The specimens were either tested to failure or 10 million cycles. The results have been analysed according to the following welding parameter variables; tool type, probe length, rotation speed and lap weld configuration.
The fatigue test results are shown as S-N curves, i.e. applied stress range versus number of cycles to failure in Figure 3-8 &12. Also shown are the results from the artificial lap weld as a reference. In addition, a conventional cylindrical threaded 'pin' type probe was used as a basis for further comparison with the Skew-stir techniques.
Conventional rotary motion FSW
Although FSW consistently gives high quality welds, proper process control of a number of parameters is needed to achieve this. An essential factor in ensuring weld quality is the use of a suitable tool.
In preliminary lap welding trials, a conventional cylindrical threaded 'pin' type probe gave a good as-welded appearance, however, 'S' bend testing showed the welds to be weak, due to excessive thinning of the top sheet and corresponding thickening of the bottom sheet. [14,15] These welds all failed on the retreating side (ANE joint configuration). Although the tool employed gave satisfactory welds when butt-welding, its use for lap-welding was inappropriate. Fatigue tests, therefore, were only carried out with the best joint configuration (RNE) lap welds.
The effect of cylindrical threaded pin type probes
The effect of pin type probes for comparable welding parameters is shown in Figure 3. The details of the probe length, rotational speed and joint configuration are given in the figure. The results show that the pin type probe achieved low fatigue performance, when compared to the artificial lap of similar material and joint geometry.
Fig.3. Fatigue results of welds carried out with a cylindrical threaded pin type probe
Rotary motion skew-stir TM
The Skew-stir TM technique achieves greater dynamic volume than the static volume of the probe by a skew motion. The resultant ratio between the dynamic and static volumes of the probe can also be further increased by any ovality, flats, or re-entrant features incorporated within the probe itself such as an A-Skew TM probe.
The potential advantages of the Skew-stir TM technique include:
- An improvement in process tolerance
- A wider weld zone (particularly advantageous for Lap 'T' and spot welds and when material processing by FSW)
- An improvement in weld quality (because of the increased forging action at the root of the weld).
Effect of rotation speed
The effect of rotation speed for comparable tools and welding parameters is shown in Figure 4 & 5. The details of the tool, probe length and welding parameters are given in the title of the S-N curve. The results in Figure 4 show that there is little effect of rotation speed on the fatigue performance of the 'ANE' joint configuration lap welds (see Figure 1b) made using these welding parameters. However, Figure 5 shows that with the 'RNE' joint configuration lap welds ( Figure 1a), increased rotation speed results in increases in fatigue strength for the welding parameters investigated. The weld surface appearance, however, is adversely affected when made at the higher rotational speed of 745 rev/min. (A 'Dual rotation' FSW technique for providing optimum probe rotational speed, while effectively reducing the shoulder rotation speed, is currently under development by Watts et al. 
Fig.4. Fatigue results of welds, carried out at different rotational speeds, at a travel speed of 4mm/sec, using the 'ANE' configuration lap joint ( Figure 1b)
Fig.5. Fatigue results of welds, carried out at different rotational speeds, at a travel speed of 4mm/sec, using the 'RNE' configuration lap joint ( Figure 1a)
Effect of lap configuration
The effect of lap configuration is shown in Figure 6 & 7, the details of the tool and welding parameters are given in the title of the S-N curve. The data in Figure 6 shows that 'RNE' joint configuration lap welds ( Figure 1a) were slightly better than 'ANE' configuration lap welds ( Figure 1b). However, Figure 7 shows that with a longer probe length and/or slower travel speed 'ANE' type joint lap configuration had better fatigue performance than the 'RNE' joint lap configuration.
Fig.6. Fatigue results of welds carried out with different lap configuration with a 7mm long probe
Fig.7. Fatigue results of welds carried out with different lap configurations with a 8.25mm long probe
Effect of probe length
The effect of probe length for comparable tools and welding parameters is shown in Figure 8, the details of the tool and welding parameters are given in the title of the S-N curve. It can be seen that the 9mm long probe achieved better fatigue performance than the 7mm long probe.
Fig.8. Fatigue results of welds carried out with long and short probes
Locus of Failure
Since the effective weld width of the Skew-stir TM welds, made using an A-Skew TM probe for this investigation, is nominally 180% of the sheet thickness, failure of the lap welds was unlikely to occur in shear across the weld region. Therefore, tensile fatigue failures for simple lap welds will almost certainly initiate from the notch at either side of the weld, where the secondary bending stress, resulting from the lap offset, is tensile, see Figure 9a&b. Widening the weld will not, in itself, increase fatigue strength, as the notch morphology is of fundamental importance.
Previously lap welds made with narrow, tapered probes failed through the weld region itself in shear or by sheet-thinning or hooking. Making welds of sufficient width means that, providing a solution to the sheet-thinning and hooking can be achieved, lap welds will not fail in shear across the weld region.
Fig.9. Regions of tensile stress for simple lap welds
a) Advancing side near the top sheet edge (ANE)
b) Retreating side near the top sheet edge (RNE)
The photomacrographs of typical fatigue-tested Skew-stir TM welds, made using an A-skew TM probe, confirm that fatigue failure occurs, essentially, in the sheet material and not through the weld region, see Figure 10. Furthermore, Figure 10 shows a 'RNE' joint configuration lap weld with failure occuring from regions stressed by both tensile and bending loads in the top and bottom sheets. These Skew-stir TM lap welds were markedly different from lap welds made with a conventional pin type probe.  Instead of a significant upturn in the faying surfaces towards the top surface of the weld, on the retreating side of the weld a slight upturn then a downturn towards the weld root is shown, see Figure 10b. A downturn is shown on the advancing side of Figure 10c&d but often an upturn and then downturn can also be present.  The notch at the edge of the weld was a much smaller feature than when a conventional pin type probe was used and the path deviated much less from the original sheet interface. [14&15] Figure 10 also shows that the use of 8.25mm long probes provided good penetration of the lapped sheet.
Fig.10. Lap weld (RNE configuration) made using Skew-stir TM with an A-skew TM probe (8.25mm in length) in 6mm thick 5083-H111 aluminium alloy at a welding speed of 3mm/sec (180mm/min)
b) Detail of fracture, bottom sheet retreating side
c) Detail of fracture top sheet advancing side
d) Detail of the form of the notch at the edge of the weld - advancing side
Fit-up and adequate tool penetration
Previous studies by Christner et al., have highlighted the importance of proper fit-up, sheet thickness tolerance, bottom sheet deformation and adequate depth of penetration of the probe into the lapped sheet to ensure good quality welds are made  . Based on improved fatigue results provided by comparatively longer probes, as indicated on the S-N curves (see Figure 8), a shorter A-skew TM probe (6.5mm long) was made to establish the effect of short probes. Figure 11 shows the lack of penetration and the inadequate stirring effect that a short probe has on the interfacial surface oxide of a lapped sheet. A short probe is unable to penetrate the bottom sheet owing to the deformation/collapse of the aluminium itself. The original interfacial surface oxide layers which, in 5083-H111 material, are known to be particularly tenacious, essentially remained intact when a short probe was used.
Fig.11. Macrosection showing deformation of the lapped sheet with substantially little penetration of the interfacial surface oxides made with a short A-Skew TM probe
Rotary reversal motion skew-stir TM
The effect of reversal motion on Skew-stir TM
Rotary reversal (Re-stir TM ) motion is a friction stir welding technique in which reversal is imposed after one or more revolutions and has been described in the literature.  The results in Figure 12 (details of the welding parameters and probe length are given in the title of the figure) show that very good fatigue performance has been achieved compared with the previous results shown in Figures 3-8. Moreover, the results show that Skew-stir TM , in combination with reversal motion, achieves fatigue properties in the lap-welds that are close for those achieved for the artificial lap.
Fig.12. Fatigue results of welds carried out with reversal motion Skew-stir TM
Fatigue fracture appearance
The fracture surfaces of fatigue samples from lap welds made with rotary friction stir welding with a conventional cylindrical pin type probe, rotary Skew-stir TM with a A-skew TM probe and reversal Skew-stir TM with the same A-skew TM probe, show significant differences, see Figure 13.
Fig.13. From left to right - fatigue test fracture face of welds - conventional pin type probe; Skew-stir/A-skew TM probe and the Skew-stir/A-skew TM probe with reversal motion.
Figure 13a, shows the original oxide layer on the lower side of the fracture face. This oxide layer corresponds with the hook region on the advancing side. Figure 13b shows a clean fracture face, across the full sheet thickness, free from any oxide layers. Figure 13c shows a stepped fracture face, across the full sheet thickness. It is believed that the steps on the fracture face probably correspond to cyclic variations in the detail of the geometry at the weld edge of the joint interface resulting from the cyclic reversal of the tool rotation directions. The stepped fracture path on figure 13c, together with the fatigue data, suggest that resistance to fatigue propagation in lap welds made with Skew-stir TM /reversal motion was improved. In addition it may be expected that welds made using the Skew-stir TM /reversal motion will have similar resistance to fatigue on both sides of the weld due to the symmetrical nature of the process. This is in contrast to conventional rotary FSW techniques, which usually show markedly different performance on the retreating and advancing sides. It should be noted that process parameters and probe length for reversal Skew-stirTM welding have yet to be optimised and much more development work needs to be carried out before this technique can be used commercially.
This paper describes the influence of rotation speed; lap configuration; probe length and tool motion on fatigue performance in Friction stir welded lap joints. Comparative results for the fatigue performance of 6mm thick 5083-H111 aluminium alloy lap welds were generated welds made using by testing an 'artificial lap' of similar geometry and dimensions machined from parent material.
Rotary motion Skew-stir TM lap welds and reversal motion Skew-stir TM lap welds gave an improvement in fatigue performance, and a reduction in upper sheet thinning compared to the cylindrical threaded pin type probe and over current known published practice. Moreover, it is expected that welding techniques that produce a symmetrical welding operation and cyclic patterned weld region, such as Re-stir TM , are likely to further improve the morphology of the notches either side of a lap weld and improve the fatigue performance of both butt and lap welds.
The importance of adequate probe length to ensure proper penetration and oxide fragmentation of the lapped sheet has been established.
This initial study has highlighted some of the important aspects of the FSW process that influence fatigue performance. Further trials are planned to investigate alternative tool types and motions and to quantify their influence on fatigue properties.
The authors wish to thank C S Wiesner, I M Norris, S Lockyer, P J Oakley, M J Dore, A Duncan, S B Jones, E R Watts, P D Evans, M V Dobinson and D Saul.
- Thomas W M, Nicholas E D, Needham J C, Murch M G, Temple-Smith P and Dawes C J (TWI): Improvements relating to friction welding. European Patent Specification EP 0 615 480 B1.
- Brooker M J, van Deudekom A J M, Kallee S W and Sketchley P D: 'Applying Friction Stir Welding to the Ariane 5 Main Motor Thrust Frame'. Fokker Paper, European Conference on Spacecraft Structures, Materials and Mechanical Testing, 29 November - 1 December 2000, Estec, Noordwijk, The Netherlands.
- Cederqvist L and Reynolds A P: 'Properties of Friction Stir Welding Aluminium Lap Joints'. TWI 2 nd International Symposium on Friction Stir Welding, 26-28 June 2000, Gothenburg, Sweden.
- Cederqvist L and Reynolds A.P: ' Factors affecting the properties of friction stir welded aluminum lap joints'. Welding Research Supplement, December 2001, pp.281-287.
- Matsumoto K and Sasabe S: 'Lap Joints of Aluminium Alloys by Friction Stir Welding'. TWI 3 rd International Symposium on Friction Stir Welding, Kobe, Japan, 27-28 September 2001.
- Christner B, 'Developments and testing of friction stir welding (FSW) as a joining method for primary aircraft structure', ICAS congress 2002.
- Thomas W M, Norris I M, Smith I J, & Staines D G, 'Reversal stir welding - Re-stir TM - Feasibility Study', 4 th International Symposium of Friction Stir welding, Park City, Utah, USA, 14-16 May 2003.
- Thomas W M, Staines D G, Johnson K I, & Evans P, 'Com-stir TM - Compound motion for Friction Stir Welding and Machining', published on the Internet 6 March 2003.
- Watts E R, Staines D G, Thomas W M, and Nicholas E D, 'Dual-rotation Stir welding - Preliminary trials', published on the Internet April 8 th 2004, TWI Ltd Granta Park, Great Abington, Cambridge, CB1 6AL, United Kingdom.
- Thomas W M, Braithwaite A B M and R John, 'Skew-Stir TM technology'. TWI 3 rd International Symposium on Friction Stir Welding, Kobe, Japan, 27-28 September 2001.
- Cantin G M D, David S A, Thomas W M, Lara-Curzio E, and Babu S S, 'Friction Skew-stir welding of lap joints in 5083-O Aluminium Part 1: Microstructural Characterisation', Science Technology Welding Joining, journal, to be published.
- Dalle Donne C, Raimbeaux G, Biallas G, Allehaux D, Palm F and Ghidini T, 'Fatigue properties of Friction Stir welded aluminium butt joints', ICAF 2003-Fatigue of Aeronautical Structures as an Engineering Challenge. The 22 nd Symposium, International Committee on Aeronautical Fatigue, Switzerland, May 2003.
- Vaidya W V, Angamuthu K, Ventzke V, Kocak M, Palm F, 'Fatigue and fracture behaviour of Friction Stir welded aluminium alloy AA6056-T6', ICAF 3003-Fatigue of Aeronautical Structures as an Engineering Challenge. The 22 nd Symposium, International Committee on Aeronautical Fatigue, Switzerland, May 2003.
- Thomas W M, Johnson K I, & Wiesner C S. 'Friction stir Welding - Recent develpments in Tool and Process Technologies, Advanced Engineering Materials 2003, 5 No. 7 C35.
- Dolby R E, Johnson K I, and Thomas W M, 'The Joining of Aluminium Extrusions', Aluminium Two Thousand, 5 th World Congress, Rome, 2003.