Paul A Colegrove and Philip L Threadgill
Published on the Internet 13 February 2003
TWI and the University of Cambridge are conducting a project on the modelling of Friction Stir Welding (FSW). Modelling is considered important because it will increase understanding of the process, enabling more effective tools tobe designed.
The modelling work has used the commercial Computational Fluid Dynamics (CFD) package FLUENT to model both 2 and 3 dimensional flow. The model is unique in that it allows the material to slip as well as stick to the tool surface. This technique has enabled the comparison of tool profiles, which was not possible with the conventional 'stick' model. The implementation of this technique is described in Colegrove and Shercliff. 
Using this technique it was shown that the weld traversing force was minimised through the use of a convex featured tool or the 'Trivex TM' profile shown in Fig.1. The traversing force was least when the tool area to swept area ratio was between 70-80%. Two tools were produced based on this concept and are shown in Fig.2. The first ( Fig.2a) used no threads while the second ( Fig.2b) was threaded along the length. It was thought that the threads on the second tool would aid oxide disruption leading to a stronger joint. These two tools were then compared against one version  of the MX-Triflute TM family of tools  that used the same pin and shoulder dimensions.
Butt welds were made using 6.35mm 7075-T7351 aluminium alloy on the ESAB SuperStir TM machine. This machine permitted the measurement of traversing and down forces, which were reduced by 18-25% (see Fig.3) and 12% respectively with the new tools. The tensile strength of the welds made with the new tool matched those made with the MX-Triflute TM.
Furthermore, a thin copper strip was placed in the joint line prior to welding. The final position after welding was determined by X-ray and indicated the degree to which the oxide and at the interface was disrupted. The results in Fig.4 show that all three tools mixed the material well near the base, however near the top the copper flowed around the tool with very little disturbance. This is because the shoulder, which does not disrupt the flow, has a much greater influence near the top surface.
This work has demonstrated the potential benefits of the Trivex TM tool over one version of the MX-Triflute TM tool in lowering both the traversing and down forces. In particular, the Trivex TM tool without threads shows remarkable promise, because it is inherently much easier to manufacture than either the MX-Triflute TM or MX-Trivex TM tools and the weld strength is comparable. Avoiding the use of threads is also beneficial for fatigue resistance of the tool, as the threads act as crack initiation sites.
Further details of this work will be presented at the upcoming Friction Stir Welding Symposium in Salt Lake City in May. Future investigations of the Trivex TM tool concept will focus on thick section butt welds, lap welds and welds in other materials e.g. steel.
The authors wish to express their thanks to Adelaide University, Post-Graduate Training Partnership (PTP), and the Cambridge Commonwealth Trust for their financial support in undertaking this work. The advice given by Chris Dawesand Hugh Shercliff has also been greatly appreciated.
- Colegrove P A, Shercliff H R: 'Friction Stir Welding and Processing II', (ed. Edited by K.V. Jata, M. Mahoney and R. Mishra), 2003, TMS.
- Dawes C J: 'Friction Stir Welding of Transport Structures - Phase II - Final Report', Report No: 12030/13/01, TWI Ltd, Cambridge, UK, 2001.
- Thomas W M and Gittos M F: 'Development of friction stir tools for the welding of thick (25mm) aluminium alloys', TWI Core Research Report No: 692/1999, Cambridge, UK, 1999.