R Johnson and P L Threadgill
Paper presented at RINA Conference: Advanced Marine Materials: Technology and Applications. October 2003
Friction stir welding of aluminium was first demonstrated in 1991, and most of the early industrial applications were in the marine industry, joining 6xxx extrusions for use in decks, bulkheads, on a variety of vessels. Since then, significant progress has been made in improving productivity, for example early welding speeds were about 1m/min or less in 6mm 6082-T6, and speeds of 6m/min are now possible.
There has been growing interest in the friction stir welding of steel, especially for naval applications. Although the process has been successfully demonstrated, further development is needed to improve tool material technology and process control, and to obtain more data on the performance of welds. Process economics for welding steel have also not been fully established, and the robustness of the process for shipyard applications needs further consideration.
The paper reviews current progress and potential of friction stir welding of both aluminium and steel for marine applications, and provides a comment on the state of the art, and what may be achieved in the near future.
Since its invention at TWI in 1991, friction stir welding has made an increasing impact in the welding of aluminium in many industry sectors, notable shipbuilding, aerospace, and railways, and is growing rapidly in other industries, for example automotive. The reasons for the rapid uptake of the technology lie in the basic simplicity and flexibility of the process, and its ability to weld just about any aluminium alloy, including those not considered weldable by more established fusion processes.
This paper reviews some of the major features of the process, and its application and further potential for the marine industries.
2. The friction stir process
The principle of the process is shown in Figure 1. FSW can be thought of as a process of constrained extrusion. Friction between the rotating tool and the workpieces generates heat, and the high normal pressure under the tool causes a plasticised zone of material to form around the probe. As the tool is traversed along the joint line, the workpiece material is heated, plasticised, and extruded around the tool probe. The extruded material forms a solid-phase joint behind the tool as it passes.
The advantages of the process can be summarised as follows:
- Solid phase process
- Single pass process (<1mm to >50mm per pass in Al alloys)
- Mechanised process
- No special pre-weld edge profiling or cleaning
- No shielding gas or filler wire required for most materials (gas shield usually required for steels)
- Low distortion and shrinkage
- 1-D, 2-D or 3-D, any position
- Excellent mechanical properties
- No welding fume or spatter hazards
- No UV or electromagnetic radiation hazards
- Very low energy consumption
Fig. 1. Principal of friction stir welding
Of these advantages, low distortion, zero consumable requirement, full automation and very low defect incidence have probably been the most attractive to marine fabricators. Studies of mechanical properties have shown that tensile strengths are generally above those obtainable by fusion processes, and welding speeds are generally comparable or better in the thickness range of interest. There are numerous publications which show fatigue properties to exceed those of MIG welding, although as yet this is not reflected in design codes.  Several Classification Societies have approved the process, generally on a case by case basis, but this has opened the way for the process to make significant inroads into the shipbuilding industry.
2.2 Application to aluminium
The process is clearly at its greatest advantage when welding long straight welds, as this allows maximum exploitation of the fully mechanised nature of the process. For this reason, the overwhelming majority of marine applications are for flat structures, such as decks, bulkheads, floors, superstructures etc. which do not contain complex curves. Hull profiles are not known to be welded commercially in three-dimensional profiles, although such efforts have been demonstrated in laboratory trials. One exception to this was an interesting experiment in Australia in the mid 90's, where flat friction stir welded plates were explosively formed into hull profiles for a high speed tourist vessel designed for reef viewing. Although the process was a technical success, funding for the whole vessel development was cancelled, and this activity has not been picked up elsewhere. In many ways this is a pity, as the project required the design and fabrication of a mobile friction stir welding rig, which gave very encouraging results.
Most of the structures welded by this process for marine applications are AA5xxx plate and AA6xxx extrusions, although the process is also used with success to weld plate to extrusions. Friction stir welding can successfully weld any of the normal range of aluminium alloys, whether these be plate, extruded sections, castings or forgings, and the process has no difficulty in welding the so called 'unweldable' high strength alloys form the AA2xxx, 7xxx and 8xxxseries. However, most aluminium alloys used in marine structures are weldable by a variety of processes, and so friction stir welding must compete on technical and economic terms, and show clear advantages over competing processes.
Fig. 2. High speed aluminium ferry containing numerous friction stir welded components.(Hydro Aluminium).
Despite the attributes of friction stir welding, it will not displace more established processes from all applications. One advantage of the process, which is the lack of requirement for a filler, is also in some ways a disadvantage, as it means that the process cannot easily be used to make fillet welds, or other geometries where extra material must be added to the weld. Thus, marine structures containing friction stir welds to join long extrusionsfor decks and bulkheads typically also contain MIG welded transverse stiffeners, which require fillet welds.
Friction stir welding is seen as one of a family of welding processes, ideally suited for some joint types, and generally impractical for others. Figure 3. 'Seven Seas Navigator', a cruise ship with extensive friction stir welding in the superstructure.
Shipyards, and suppliers to shipyards are always seeking methods to improve productivity, and this is manifested by an obsession with welding speed. When friction stir welding was first introduced commercially, extrusions such asAA6082-T6 were welded at about 0.7 to 0.8m/min with a joint line thickness of 5 to 6mm. 6xxx alloys are generally the easiest to friction stir weld, and constant improvements in tool design and process control have permitted the welding speed to be drastically improved. Laboratory trials now claim welding speeds of around 10m/min, and increase of an order of magnitude. It will be some time before such speeds become a commercial reality, but it does demonstrate the enormous scope for continuous improvement which exists in this still very young process. In AA6xxx alloys, it is possible to compete with high speed processes on more or less even terms in the laboratory, and still maintain a high quality of weld. However, the alternative high speed processes such as laser welding are also undergoing constant improvement in productivity and quality, and the results of this have to be seen as a benefit to industry, even it industry is not yet able to exploit these developments.
Figures 2 and 3 show typical vessels which have made extensive use of friction stir welding.
Fig. 3. 'Seven Seas Navigator', a cruise ship with extensive friction stir welding in the superstructure
2.3 Properties of aluminium welds
There is a wealth of data in the literature on mechanical properties of friction stir welded materials, although this has not been comprehensively reviewed for some time. Data quoted in this paper are typical of that in many other publications.
A section through a typical friction stir weld in an aluminium alloy is shown in Figure 4. In aluminium alloys, these welds are characterised by a distinct nugget at the weld centre, and in this area the material has been severely hot worked, resulting in a very fine, equiaxed recrystallised structure. The grain size is typically 5-10µ. Just outside this area, it can be seen that the material has suffered severe deformation, with the banding in the rolled material often being turned through almost 90°. This area is defined as the thermo-mechanically affected zone (TMAZ). Outside this area, a conventional HAZ exists, where there is no visible plastic deformation.
Fig. 4. Macrosection through a friction stir welded 6mm thick aluminium alloy
In alloys such as AA6082-T6, the thermo-mechanical cycle experienced by the material during welding results in a wide range of microstructures, from solution treated to overaged. Hardness traverses show a drop in hardness in the nugget, and more severely in the HAZ. This loss of strength can be significantly recovered by an additional T6 heat treatment, although in practice this is seldom applied. When compared to MIG welds, friction stir welds typically give a narrower overall weld width, and a lower reduction in hardness.
In alloys which are hardened by mechanical work (e.g.5xxx alloys), the strength of the weld nugget is usually greater than the base metal in lightly hardened conditions. However, in the hardest grades, the heat from the process will temper back the hardness to something approaching the O condition.
Examples of hardness data for 5083-H111 and 6082-T6 are shown in Figure 5(a) and (b).
Cross weld tensiles in friction stir welded AA6082-T6 normally fail in the HAZ, in the area of minimum hardness. Elongations can appear low, but this is due to the strain at fracture being concentrated in a very small area, giving a small elongation. Examination of the fracture face will almost always show a significant reduction in area, confirming that the ductility in the failed region is high. Typical data is shown in Figure 6.
Fig. 5. Hardness traverses for friction stir welded AA6082-T6  and AA5083-H111 alloys 
Fig. 6. Static strength of AA6082-T6 for friction stir and MIG welds 
Many studies have shown very encouraging fatigue results. Typically, fatigue data on butt welds is slightly lower than the parent material values, but higher than data obtained from fusion welds. Such observations have been reported by many authors, and typical data from Kumagai  is shown in Figure 7. Although data in other joint configurations is sparse, the results are nevertheless encouraging.
Fig. 7. Fatigue data for FSW and MIG welds on 6082-T6 
2.4 Recent developments
Although shipbuilding is a major user of the process in Europe, other sectors are also forcing the pace of development of the process. In particular, aerospace applications in Europe and America, and railcar building in Japan, and to a lesser extent in Europe. Welding machines are often built for specific applications, and recently very sophisticated equipment has been manufactured for the aerospace industry. One machine is a full 6-axis machine for welding complex three dimensional structures for the Eclipse executive aircraft which is made in the USA  . This has demonstrated the possibility of designing machines for more complex joint geometries in other industries. The Eclipse aircraft has replaced many rivets with friction stir welds, resulting in a substantial saving in weight, manufacturing time and cost. Manufacturers of larger commercial aircraft are also investing heavily in process development and application, and the usage if the process is certain to increase significantly.
In Japan, railcar manufacturers have built multi-head machines for welding extrusions for railcars, in lengths up to 25m. Here the low defect rate, and low distortion are generally claimed as the main reasons for the investment. Some welds made in Japan are of such quality that the root of the weld is exposed, unpainted, and is virtually impossible to see.
Current research work is investigating new tool designs, process control systems and manufacturing methods which will further improve the quality of welds, reduce their cost, and broaden the range of applications. Recent reports from ESAB indicate that very high welding speeds can now be obtained in the laboratory (up to 10m/min in 5mm 6xxxx alloys). Such speeds can compete with any process, but the ability of such high speed procedures to withstand industrial implementation remains to be demonstrated. However, friction stir welding is here to stay, and it is expected that it will continue to make an impact in the building of aluminium vessels, in particular structures which require long straight welds.
3. Friction stir welding of steel
3.1 Current developments
Although the friction stir welding process has been developed initially for joining aluminium, it was inevitable that interest would emerge in applying it to other materials. An increasing volume of work is appearing on friction stir welding of materials such as magnesium, copper, titanium and steels. It has been known for some time that many steels can be welded by the process, including C-Mn steels used in shipbuilding, but the process is not yet ready for shipyard use. The main reason for this lies in the tools used to make the weld. In aluminium, the process seldom achieves temperatures above 500°C, and there are therefore a number of materials which can be used which will operate well at these temperatures. When welding C-Mn steels, much higher temperatures of between 1000°C and 1200°C are achieved, and the options for tool materials are greatly reduced, especially when the high forces present are considered. It is essential that the tool does not degrade by wear, deformation, microstructural instability, reaction with the workpiece or fracture.
So far, two tool material types have given promising results, although neither meets all of the criteria for an industrial tool material. These materials are refractory alloys based on the tungsten - rhenium system, and a ceramic solution based on polycrystalline cubic boron nitride (PCBN).  Tungsten - rhenium alloys exhibit excellent high temperature strength and toughness, but are prone to wear and deformation, in particular when higher strength steels are used. However, they are much cheaper than PCBN, and have an advantage in that they can be machined relatively easily. PCBN is a very hard material (boron nitride is thought to be the second hardest material known), and therefore wear rates are very low. However, the fracture toughness of the material is, like almost all ceramics, very low, and the tools are prone to cracking and sudden failure unless handled with great care. The extreme hardness of the materials means that they are very difficult to process to more complex shapes. The tools undergo a very complex manufacturing process, and are therefore rather expensive.
3.2 Properties of steel welds
Rapid progress is being made to improve the characteristics and performance of both tool materials, and indeed further work is underway to identify alternative tool materials.
Fig. 8. Sections through two pass and single pass friction stir welds in a 12mm structural steel 
Examples of single and two pass welds in a ferritic structural steel (EN10025:1993 S355) are given in Figure 8.  Unlike aluminium welds, friction stir welds in C-Mn steels do not show a distinct nugget region, and microstructural regions are more difficult to distinguish, as the extent of plastic deformation is generally difficult todetermine. The microstructure in the weld centre is generally similar to the HAZ of a conventional fusion weld, although the maximum austenite grain size is generally less. As distance from the original bond line increases, themicrostructure exhibits the normal features of fine grained supercritical HAZ, intercritical HAZ and subcritical HAZ. Hardness values can vary, but high hardness levels are unusual. Charpy data are rather variable, and as expectedlower than parent values in the weld region. Data from a C-Mn steel (EN10025:1993 S355) are shown in Figure 9.  No direct comparative data with more established welding processes are known. Direct comparison may be difficult, as friction stir welds are normally made in one or two passes when welding material of 10-12mm thick, whereascompeting processes may have several passes, allowing tempering of earlier passes by later passes. Additional data has been generated on shipbuilding steels such as DH36 and HSLA-65 in the USA, [7,8,9] but the data obtained generally reflect that illustrated above.
Fig. 9. Charpy data for friction stir welded C-Mn structural steel 
No data on fracture toughness as measured by more sophisticated techniques such as CTOD are available, but it is clear that the generation of a considerable volume of such data will be needed to give industry the confidence to use the process. A similar lack of data exists for fatigue in friction stir welded steel, and the same argument regarding the need for such data is valid. At present, most research efforts are directed towards solving the main challenges, which are the development of improved tool technology and possible improved process technology.
This paper has summarised the potential for friction stir welding of aluminium and steel for marine industries. The principal conclusions are as follows:
- Friction stir welding of aluminium, in particular AA5xxxx and AA6xxxx grades is well established, and it is expected that use of the process will continue to grow.
- Friction stir welding of C-Mn steels has been demonstrated, but further work is required to improve the tool characteristics, and also to demonstrate the mechnaical properties of the welds.
Some of the work described here was funded as part of TWI's Core Research Programme.
- Threadgill P L: 'Friction stir welding - the state of the art'. TWI Members report 678/1999.
- Ericsson M, Sandström R and Hagström J: 'Fatigue of friction stir welded AlMgSi alloy 6082'. Proc 2 nd International Friction Stir Welding Symposium, Gothenburg, Sweden, June 2000.
- Kumagai M and Tanaka S: 'Properties of aluminium wide panels by friction stir welding'. Proc 1 st International Friction Stir Welding Symposium, Thousand Oaks CA, USA, June 1999.
- Christner B, Hansen M, Skinner M and Sylva G: 'Friction stir welding system development for thin-gauge aerospace structures'. 4 th International Friction Stir Welding Symposium, Park City UT, USA, May 2003.
- Sorensen C D and Nelson T W: 'Progress in Polycrystalline Cubic Boron Nitride Friction Stir Welding Process'. 4 th International Friction Stir Welding Symposium, Park City UT, USA, May 2003.
- Johnson R and dos Santos J F: 'Mechanical Properties of Friction Stir Welded S355 C-Mn Steel Plates'. 4 th International Friction Stir Welding Symposium, Park City UT, USA, May 2003.
- Lienert T J, Tang W, Hogeboom J A and Kvidahl L G: 'Friction Stir Welding of DH36 Steel'. 4 th International Friction Stir Welding Symposium, Park City UT, USA, May 2003.
- Konkol P J, Mathers J A, Johnson R and Pickens J R: 'Friction Stir Welding of HSLA-65 for Shipbuilding'. 3 rd International Friction Stir Welding Symposium, Kobe, Japan, September, 2001.
- Posada M, DeLoach J J, Reynolds A P, Fonda R W and Halpin J P: 'Evaluation of Friction Stir Welded HSLA-65'. 4 th International Friction Stir Welding Symposium, Park City UT, USA, May 2003.
7. Authors biographies
Richard Johnson is a Principal Research Engineer in the Friction and Forge Processes section at TWI, and Philip Threadgill is currently R&Dmp;D Manager for the same section. Both authors have worked extensively on friction stir welding of various materials, in particular aluminium alloys and various grades of steel, and on the development of improved tool materials technology.