2007-01-1704 Fatigue behaviour of friction stir joined aluminium alloy NG5754 and AA6111 sheets L. Han*, N. Blundell Warwick Manufacturing Group, University of Warwick, Coventry, CV4 7AL, UK Z. Lu, M. Shergold Jaguar & Land Rover, Gaydon Engineering centre, Lighthorne, CV35 0RR, UK A. Chrysanthou School of Aerospace, Automotive and Design Engineering, University of Hertfordshire, Herts, AL10 9AB, UK ABSTRACT A study examining the fatigue behaviour of spot friction stir joined (SFJ) aluminium NG5754 and AA6111 sheets was carried out. Different friction stir holding times were applied to the NG5754 alloy, whilst paint bake cycles were applied to the AA6111 joints in order to examine the effects of these variables on the fatigue behaviour of the joints. Static and fatigue tests were performed on all samples and the results were compared. It was demonstrated that the material flow characteristics were affected by the holding time leading to different joint attributes, which in turn contributed to different fatigue behaviour. Paint bake cycle affected the fatigue life, particularly at low applied loads. INTRODUCTION The increase in use of light weight materials, in particular aluminium alloys, for vehicle body applications has impelled the development of new joining techniques. Traditional joining methods such as spot-welding and arc-welding are being challenged. Self-piercing riveting, as one of the alternatives to spot-welding, has been introduced and used in vehicle body assembly [1]. Its superior advantages such as, no pre-drilled hole, capable of joining similar or dissimilar materials and combinations have attracted considerable interest from the automotive industry. However, the use of rivets adds weight and cost to the process. Manufacturers are therefore looking for other alternatives. Spot Friction Stir Joining (SFJ) as another alternative to spot-welding has thus emerged. SFJ is derived from the established technology of linear Friction Stir Welding (FSW), which was invented at The Welding Institute (TWI) in 1991. It is a solid phase process and particularly suitable for joining lightweight sheet materials, such as aluminium, copper, lead and plastics to produce straight-line welds. The principle of the SFJ process is similar to FSW and has all the advantages of the process. A backing anvil is used to replace the backing plate for FSW. The pin is rotated and then plunged into the sheet material to be joined. Instead of moving in a straight-line making a butt joint, the pin is kept in the same place as it is continuously rotating. At the same time a relatively high pressure is applied between the pin and anvil to hold the work-pieces together. Frictional heat is generated and the materials are therefore heated and softened. The high pressure brings the softened material together as the material flows. The pin is released and a permanent point joint is formed. There are four major stages that can be simply described as Rotate, Plunge, Hold and Release, involved in the process, as shown in Fig. 1. The process is attracting interest from the automotive industry due to a number of advantages including low cost, low power demand, no emission of fumes and radiation [2, 3, 4]. In addition, as a solid-phase joining method, many of the problems associated with liquidphase welding of aluminium alloys are avoided in SFJ. The surface oxide layer can be effectively broken and dispersed throughout the weld due to the weld action. SFJ has been reported to join thin sheet material for automotive closures [5]. It is also reported by Lin et al that a fatigue crack growth model based on the Paris law was created and adopted to predict the fatigue lives of spot friction stir joined AA61111-t4 sheets for automotive application [6, 7]. However, as an emerging technology, the mechanical behaviour of SFJ joints has not been well understood, in particular the effect of the process variables on the fatigue behaviour of such joints. Freeney et al [8] reported the effect of the rotation rate and plunge depth on the static behaviour of spot friction stir joined 5052 aluminium alloy. The present authors reported the effect of paint bake cycle on static behaviour of such joints [9]. The aim of the study that is reported here was therefore to examine the effect of holding time and paint bake cycles on the fatigue behaviour of SFJ joints. Rotate Plunge Hold was employed as the applied load. The stress ratio was R=0.1 and a frequency of 5~15Hz were used. Three load levels that had maximum axial loads ranging from 50-80% of the static ultimate shear load were used. In total 12 samples were tested for each group. The failure criterion was that the specimen reached 1.0mm extension or to 2,000,000 cycles, which was defined as run out. Release Figure 1: Schematic Spot friction stirring process EXPERIMENTAL PROCEDURES The materials used for this study were NG5754 and AA6111 aluminium alloy sheets in 2.0mm thickness. The NG5754 alloy sheets were previously pre-treated with a thin chromate-free film and a wax-based solid lubricant. The AA6111 alloy sheets were in the T4 condition and had a mill finish surface only. The nominal compositions and mechanical properties of the NG5754 and AA6111 aluminium alloy sheet materials are listed in Table 1 and 2 respectively. All the samples were made using a robot mounted SFJ machine built by Kawasaki Heavy Industries and supplied by Kawasaki Robot UK, featuring 2.2kW motors for both the rotation and plunge of the weld tool. The welding is controlled via a force control system, where a programmed force is applied for a pre-determined amount of holding time. The weld tool used was a Kawasaki standard SFJ tool designated ‘KN07’. All samples were manufactured using two coupons with dimensions of 40mm x 100mm. The specimen dimensions for both lap shear and fatigue tests are the same and are shown in figure 2. For the NG5754 alloy sheets, three groups of samples were created and designated as 4s, 5s and 6s, that simply indicated the holding time of 4, 5 and 6 seconds respectively. The holding pressure and rotation speed for all three groups of samples remained the same. Thus the results obtained from the three groups of samples would only reflect the effect of holding time on the fatigue behaviour of the joints. For the AA6111 alloy sheets, one group of samples was manufactured under an identical setting condition and then it was divided into two groups. One group of samples was subjected to a paint bake cycle at 180°C for 30min, whilst the other group was not. At least 5 samples were tested in the shear condition from each group in order to obtain the ultimate shear load under static conditions. The strength values presented in the paper are average values of the 5 samples. Fatigue tests were carried out on an Instron servo-hydraulic test machine with a load capacity of 10kN at room temperature. A cyclic tension–tension load with a sinusoidal waveform and constant load amplitude Figure 2: Geometry and dimensions of shear and fatigue samples Table 1: Compositions and mechanical properties of NG5754 alloy Young’s Modulus (GPa) 70 Si 0-0.40 MECHANICAL PROPERTIES Tensile Elongation Hardness strength (HV) (MPa) 240 22% 63.5 nominal composition(Balance Al) Fe Cu Mn Mg 0-0.40 0-0.10 0-0.50 2.60-3.60 Table 2: Composition and Mechanical Properties of AA6111 Young’s Modulus (GPa) 70 Si 0.201.70 MECHANICAL PROPERTIES Tensile Elongation Hardness strength (HV) (MPa) 308 26% 93 Nominal compositions Fe Cu Mn Mg Al 0.70 0.90 0.80 0.101.40 Balance RESULTS AND DISCUSSION Figure 3 shows the shear test results for the NG5754 and AA6111 samples. For the NG5754 samples, the 5s group had the highest shear strength, whilst the 4s group exhibited the lowest strength of all three groups. In addition, the 5s group had the lowest strength variation and the 6s group had the greatest strength variation of the three groups of samples. These results indicated the effect of holding time on shear behaviour and the longer holding time does not necessarily provide the joints with higher shear strength. For the AA6111 group, the baked samples had higher shear strength than the unbaked samples indicating that the paint baking cycle led to an increase in the shear strength of the SFJ joints. Shear Ultimate Load (N) 4500 4s 5s 6s Baked Unbaked 4000 3500 3000 2500 2000 calculated at any load amplitude. At a load amplitude Fa of 0.5kN, the baked samples could sustain 1,166,090 cycles before failure, whilst the unbaked samples could reach 1,622,631 cycles until failure. The paint bake cycle led to a 28% reduction in the fatigue life of the samples at this load level. However, at 1.0kN, following paint baking, the fatigue life was increased by 10% from 50,269 to 55,179 cycles. It has to be noted here that, statistically, the experimental data did not differ significantly. Although based on the power law, the fatigue life for the two groups can be calculated at any load amplitude. Due to the feature of scattered data for fatigue test and the use of progression method for the power law, the calculation can only indicate a trend. 1500 1000 500 0 NG5754 samples AA6111 samples Figure 3: Shear test results for NG5754 and AA6111 samples Failure examination revealed that following fatigue testing only one failure mode was observed for the AA6111 samples. As shown in Figure 5, nugget failure with an eyebrow crack formed at the top sheet was observed for both the baked and unbaked AA6111 samples. Fretting indicated by black fretting debris was also observed to occur for all the samples. FATIGUE OF THE AA6111 SAMPLES Figure 4 shows fatigue test results for the AA6111 samples. Although the fatigue data are scattered indicating that process optimization is necessary, the trend is that the unbaked samples had longer fatigue life than the baked samples. The lower shear strength for the unbaked samples is not accompanied with lower fatigue strength. This suggested that for a SFJ joint, high shear strength is not necessarily accompanied with high fatigue strength and the paint bake cycles had a minor effect on the fatigue strength. 1 Unbaked -0.2272 Load amplitude [KN] y = 11.949x -0.1995 y = 8.6678x 6111(2.0)/6111(2.0) Un-baked 6111(2.0)/6111(2.0) Paint-baked 0.1 10000 100000 1000000 10000000 N [cycles] Figure 4: Fatigue results for the AA6111 samples Baked As shown in Fig. 4, within the life interval of 10,000 and 2,000,000 cycles the fatigue life for both baked and unbaked samples can be expressed by a power law: Fa = CN m where Fa is the load amplitude, N is fatigue life in cycles, C and m are constants as shown in Fig. 4. Based on the power law, the fatigue life for the two groups can be Figure 5: Fatigue failure mode that occurred for the AA6111 baked and unbaked samples FATIGUE OF THE NG5754 SAMPLES Based on the shear test results, a range of maximum applied loads from 1.0kN to 2.7kN were adopted in fatigue tests for the NG5754 samples. Figure 6 shows the fatigue test results for the three groups of samples. Similar to the AA6111 group, the data are scattered and statistically the differences between samples with different holding time are not significant. However, the 6s group of samples still exhibited the longest fatigue endurance, whilst the 4s group had the shortest fatigue life of all three groups. The increase in the holding time led to an increase in the fatigue life, although the effect is not significant. and some of the 5s group of samples, the main failure mode was fracture of the sheet material, as shown in Fig. 7b. Fretting was also observed at the interface between the two joined sheets for all samples following the fatigue test. 1 Load amplitude [KN] y = 3.9523x -0.146 y = 4.24x -0.1568 -0.1591 y = 4.2046x Bottom sheet Botto Top sheet (a) 3 12 5754(2.0)/5754(2.0) 4s 5754(2.0)/5754(2.0) 5s 9 5754(2.0)/5754(2.0) 6s 0.1 10000 6 100000 1000000 10000000 N [cycles] Figure 6: Fatigue test results for the three groups of samples Figure 6 also shows evidence that within the life interval of 10,000 and 2,000,000 cycles, the fatigue life of all three groups of samples can also be expressed by a power law relationship: Fa = CN m Based on the power law, the fatigue life for the three groups can be calculated. At 1.0kN load amplitude, the 4s, 5s and 6s groups had a fatigue life of 8324 cycles, 10024 cycles and 12247 cycles respectively. The increase by 1s holding time from 4s to 5s led to a 20% increase in fatigue life, whilst from 5s to 6s increase in the holding time resulted in 22% increase in fatigue life. The increase in the holding time led to a corresponding increase in the fatigue life. From 4s – 6s, the longer the holding time, the longer the fatigue life. This is different from what was observed during shear test, in which the 5s group had the highest shear strength, rather than 6s group. It has to be emphasised again, similar to the AA6111 group, due to great scatter of the data and the use of progression method for the power law, any increase in the fatigue life as the holding time increased only indicated a trend. Examination of the failed samples reveals that two failure modes were involved following the fatigue test. For the 4s group and most of the 5s group of samples interface failure or nugget failure accompanied with eyebrow crack on the bottom sheet of the joint was the main failure mode, as shown in Fig. 7a. For the 6s group Bottom sheet Top sheet (b) Figure 7: Failure modes that occurred in fatigue tests Figure 8 shows the surface of a fractured nugget with the top and bottom sheets for a sample from the 4s group. The nugget fractured along the interface between the two joined sheets leaving a fractured nugget in both the joined sheets. Fretting debris appeared around the 12 o’clock position and part of the fractured nugget around the 12 o’clock position was also covered by fretting debris. The eyebrow crack was observed to initiate at the 12 o’clock position and propagated outwards. It seemed that for the nugget failure, the crack initiated at the 12 o’clock position of the nugget and propagated outwards along the keyhole circumference and into the shank of the nugget. This led to fracture of the nugget and eventually separation of the two jointed sheets. For sheet material failure, the top sheet fractured along the 9 and 3 o’clock positions. It seemed that the crack initiated at the 9 and 3 o’clock positions and propagated outwards leading to fracture of the top sheet, whilst the other half of the nugget and part of the top sheet remained intact, as shown in Fig. 7(b). the baked AA6111 samples, than for the NG5754 curves. Consequently, as the applied load decreased, the difference in the fatigue life between the AA6111 and NG5754 samples was reduced. 12 9 3 6 Top sheet 12 9 3 6 Bottom sheet Figure 8: Fractured surface of a 4s group of samples COMPARISON OF FATIGUE PERFORMANCE FOR THE NG5754 AND AA6111 SAMPLES 1 -0.2272 y = 11.949x -0.1995 Load amplitude [KN] y = 8.6678x y = 4.24x -0.1568 -0.1591 y = 4.2046x y = 3.9523x -0.146 5754(2.0)/5754(2.0) 4s 5754(2.0)/5754(2.0) 5s 5754(2.0)/5754(2.0) 6s 6111(2.0)/6111(2.0) Un-baked 6111(2.0)/6111(2.0) Paint-baked 0.1 10000 100000 1000000 10000000 N [cycles] Figure 9: Comparison fatigue results for the NG5754 and AA6111 groups Fig. 9 shows the comparison of fatigue strength for the NG5754 and AA6111 samples. Although the data are scattered as mentioned earlier, at high applied load, the AA6111 group exhibited longer fatigue life than the NG5754 samples. However, the S-N curves for the AA6111 group had more negative slopes, in particular Fatigue fracture usually occurs on a cross-sectional discontinuity due to high stress concentration. For a spot friction stirred joint, the keyhole creates a cross-sectional discontinuity leading to high stress concentration around the keyhole circumference. Fatigue cracks therefore are likely to initiate in this region. Under the fatigue loading, the top sheet was subjected to the highest tensile stress at the 3 and 9 o’clock positions. This may lead to crack initiation and eventually failure of the top sheet at the 3 and 9 o’clock positions. In addition, secondary bending, which is an inherent feature of single lap joints like those that were used throughout this project, also played an important role in the fatigue failure mechanism. Secondary bending introduces a bending stress, which can locally exceed the axial stress by several times at cross-sectional discontinuity to the joined sheet material, whilst a tensile force is also introduced to the nugget. This resulting bending stress at the cross-sectional discontinuity combined with normal tensile stress led to crack initiation and propagation at the cross-sectional discontinuity, which were the 6 and 12 o’clock positions. Additionally, the resulting tensile stress for the nugget was the highest at the 12 o’clock position for the top sheet and the 6 o’clock positions for the bottom sheet. This led to crack initiation at the 12 o’clock position for the nugget. Following the crack initiation, a new interface at the nugget was generated. Fretting fatigue, as a kind of wear and corrosion phenomenon, occurs when two contacting components are subjected to an oscillatory load. The appearance of fretting debris on part of surface of the fractured nugget indicated the occurrence of fretting. Under continuous fatigue loading, the crack propagated along the circumference of the nugget and penetrated into the shank of the nugget leading to fretting damage and fracture of the nugget. The AA6111 alloy has higher strength than the NG5754 alloy. If the sheet material fracture led to failure of the joints then the AA6111 group would have higher fatigue strength than the NG5754 group. However, from the experimental results, most of the samples failed by nugget fracture suggesting that the sheet material was stronger than the nugget itself. The fact that the AA6111 group exhibited longer fatigue life than the NG5754 group at high applied load and failed by nugget fracture only indicated that the nugget for the AA6111 group was stronger than the NG5754 group. This was proved true by the shear test results in which the AA6111 group had higher shear strength than the NG5754 group. The high strength of the AA6111 alloy may have contributed to this result. However, at low applied load, the fatigue strength of the AA6111 and NG5754 groups tended to be similar. This may be attributed to the heat-treatable feature, which allows participant hardening to occur during the joining process leading to even more brittle of the nugget. For the NG5754 group of samples, longer holding time produced not only deeper welds and thinner upper sheet, but also changed the position of triple point, where the two sheets meet with the stir zone boundary. These contributed to the different fatigue life and failure mode. As shown in Fig. 10, for the 4s group the triple point, as circled, was almost at the same level of the interface between the two joined sheets. A very sharp interface was formed leading to sharp notch effect. The cross-section of the nugget at the interface between the two joined sheets sustained most of the loading. Therefore fracture of the nugget occurred leaving broken nugget in both the joined sheets for the 4s group of samples. In contrast, for the 6s group, this triple point was much higher than the interface between the two joined sheets and smoother than that for the 4s group. The sharp notch effect was therefore minimized and a weak point of the top sheet at the triple point was created due to thinning of the top sheet during nugget formation. As a result, the top sheet material fracture occurred along the weak points leading to failure of the joints for the 6s group of samples. The geometry for the 5s group was similar to the 4s group leading to nugget failure. 4s 6s Figure 10: Micrograph of etched sections of the 4s and 6s group of samples CONCLUSION The results reported here are from initial primary trials. The scattered data suggested that further process optimization is necessary in order to improve process sustainability. Based on this experimental result, the following trends may be predicted: 1. The paint bake cycle – of 180°C for 30min - had a minor effect on the fatigue performance of the spot 2. 3. 4. friction joined AA6111 sheets, in particular at low applied loads. Nugget failure was the only failure mode that occurred for the AA6111 joints tested during this investigation. The holding times of 4-6s affected the fatigue life of the SFJ joined NG5754 sheets, but the effect was not significant during this examination. Both nugget failure and sheet material fracture were observed to occur during fatigue testing for the NG5754 samples due to the effect of holding time on material flow and geometric formation of the nugget. ACKNOWLEDGMENTS The authors wish to thank Advanced West Midlands for funding the project. REFERENCES 1. L. Han, Y. K. Chen, A. Chrysanthou and J. M. O’Sullivan, “Self-pierce riveting – a new way for joining structures”, ASME, 2002; PVP-Vol. 446(2): 123-127. 2. Pan, T., Joaquin, A., Wilkosz, D. E., Reatherford, L.and Nicholson, J. M., 2004, Spot Friction Welding for Sheet Aluminum Joining. Proceedings of the 5th International Symposium of Friction Stir Welding, Metz, France, Sep. 1416, 2004. 3. Kallee. S. W and Nicholas. E. D, “Application of friction stir welding to lightweight vehicles”, SEA Technical Paper, 982362, 1998. 4. Waldron. D. J, Roberts. R. W, Dawes. C. J, Tubby. P. J, “Friction stir welding-A revolutionary new joining method”, SAE Technical Paper, 982149, 1998. 5. Sakano R, Murakami K, Yamashita K, Hoye T, Fujimoto M, Inuzuka M, Nagao Y, Kashiki H, “Development of Spot FSW Robot System for Automobile Body Members”, Proceedings of the 3rd International Symposium of Friction Stir Welding. Kobe, 2001. 6. Lin, P.-C., Pan, J. and Pan, T., “Fracture and Fatigue Mechanisms of Spot Friction Welds in Lap-Shear Specimens of Aluminum Alloy 6111T4 Sheets”, SAE Technical Paper No. 2005-011247. 7. Lin, P.-C., Pan, J. and Pan, T., “Fatigue Failures of Spot Friction Welds in Aluminum 6111-T4 Sheets Under Cyclic Loading Conditions”, SAE Technical Paper, 2006-01-1207. 8. T. A. Freeney, S. R. Sharma and R. S. Mishra, “Effect of welding parameters on properties of 5052 aluminum friction stir spot welds”, SAE Technical Paper, 2006-01-0969. 9. N. Blundell, L. Han, K. W. Young and R. Hewitt, “The influence of Paint Bake Cycles on the Mechanical Properties of Spot Friction Joined Aluminium Alloys”, SAE Technical paper, 2006 – 01 – 0967 CONTACT Corresponding author: Dr. L. Han Warwick Manufacturing Group, University of Warwick, Coventry, CV4 7AL, UK Tel: +44(0)2476575385, Fax: +44(0)2476575366, E-mail: li.han@warwick.ac.uk